Download textbook Ai Ia 2016 Advances In Artificial Intelligence Xvth International Conference Of The Italian Association For Artificial Intelligence Genova Italy November 29 December 1 2016 Proceedings 1St Edition Gio ebook all chapter pdf

AI IA 2016 Advances in Artificial
Intelligence XVth International

Conference of the Italian Association
for Artificial Intelligence Genova Italy
November 29 December 1 2016
Proceedings 1st Edition Giovanni
Adorni
Visit to download the full and correct content document:
https://textbookfull.com/product/ai-ia-2016-advances-in-artificial-intelligence-xvth-inter
national-conference-of-the-italian-association-for-artificial-intelligence-genova-italy-no
vember-29-december-1-2016-proceedings-1st-edition-gio/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
AI IA 2019 Advances in Artificial Intelligence XVIIIth

International Conference of the Italian Association for
Artificial Intelligence Rende Italy November 19 22 2019
Proceedings Mario Alviano
https://textbookfull.com/product/ai-ia-2019-advances-in-
artificial-intelligence-xviiith-international-conference-of-the-
italian-association-for-artificial-intelligence-rende-italy-
november-19-22-2019-proceedings-mario-alviano/
AI IA 2013 Advances in Artificial Intelligence XIIIth

Artificial Intelligence Turin Italy December 4 6 2013
Proceedings 1st Edition Mario Alviano
artificial-intelligence-xiiith-international-conference-of-the-
italian-association-for-artificial-intelligence-turin-italy-
december-4-6-2013-proceedings-1st-edition-mario-alvian/
AI IA 2017 Advances in Artificial Intelligence XVIth

Artificial Intelligence Bari Italy November 14 17 2017
Proceedings 1st Edition Floriana Esposito
artificial-intelligence-xvith-international-conference-of-the-
italian-association-for-artificial-intelligence-bari-italy-
november-14-17-2017-proceedings-1st-edition-floriana-esp/
Artificial Intelligence XXXVI 39th SGAI International

Conference on Artificial Intelligence AI 2019 Cambridge
UK December 17 19 2019 Proceedings Max Bramer
https://textbookfull.com/product/artificial-intelligence-
xxxvi-39th-sgai-international-conference-on-artificial-
intelligence-ai-2019-cambridge-uk-
Artificial Intelligence XXXVII 40th SGAI International
Conference on Artificial Intelligence AI 2020 Cambridge
UK December 15 17 2020 Proceedings 1st Edition Max
Bramer
https://textbookfull.com/product/artificial-intelligence-
xxxvii-40th-sgai-international-conference-on-artificial-
intelligence-ai-2020-cambridge-uk-
december-15-17-2020-proceedings-1st-edition-max-bramer/
Advances in Artificial Intelligence 18th Conference of

the Spanish Association for Artificial Intelligence
CAEPIA 2018 Granada Spain October 23 26 2018
Proceedings Francisco Herrera
https://textbookfull.com/product/advances-in-artificial-
intelligence-18th-conference-of-the-spanish-association-for-
artificial-intelligence-caepia-2018-granada-spain-
october-23-26-2018-proceedings-francisco-herrera/
Multi disciplinary Trends in Artificial Intelligence

10th International Workshop MIWAI 2016 Chiang Mai
Thailand December 7 9 2016 Proceedings 1st Edition
Chattrakul Sombattheera
https://textbookfull.com/product/multi-disciplinary-trends-in-
artificial-intelligence-10th-international-workshop-
miwai-2016-chiang-mai-thailand-december-7-9-2016-proceedings-1st-
edition-chattrakul-sombattheera/
AI 2019 Advances in Artificial Intelligence 32nd

Australasian Joint Conference Adelaide SA Australia
December 2 5 2019 Proceedings Jixue Liu
https://textbookfull.com/product/ai-2019-advances-in-artificial-
intelligence-32nd-australasian-joint-conference-adelaide-sa-
australia-december-2-5-2019-proceedings-jixue-liu/
Agents and Artificial Intelligence 8th International

Conference ICAART 2016 Rome Italy February 24 26 2016
Revised Selected Papers 1st Edition Jaap Van Den Herik
https://textbookfull.com/product/agents-and-artificial-
intelligence-8th-international-conference-icaart-2016-rome-italy-
february-24-26-2016-revised-selected-papers-1st-edition-jaap-van-
Giovanni Adorni · Stefano Cagnoni
Marco Gori · Marco Maratea (Eds.)
AI*IA 2016:
LNAI 10037
Advances in
Artificial Intelligence
XVth International Conference
of the Italian Association for Artificial Intelligence
Genova, Italy, November 29 – December 1, 2016, Proceedings
123
Lecture Notes in Artificial Intelligence 10037
Subseries of Lecture Notes in Computer Science
LNAI Series Editors

Randy Goebel
University of Alberta, Edmonton, Canada
Yuzuru Tanaka
Hokkaido University, Sapporo, Japan
Wolfgang Wahlster
DFKI and Saarland University, Saarbrücken, Germany
LNAI Founding Series Editor

Joerg Siekmann
DFKI and Saarland University, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/1244
Giovanni Adorni Stefano Cagnoni
•
Marco Gori Marco Maratea (Eds.)

•
AI*IA 2016:
Advances in
Artificial Intelligence
XVth International Conference
of the Italian Association for Artificial Intelligence
Genova, Italy, November 29 – December 1, 2016
Proceedings
123
Editors
Giovanni Adorni Marco Gori
University of Genoa University of Siena
Genova Siena
Italy Italy
Stefano Cagnoni Marco Maratea
University of Parma University of Genoa
Parma Genova
Italy Italy
ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Artificial Intelligence
ISBN 978-3-319-49129-5 ISBN 978-3-319-49130-1 (eBook)
DOI 10.1007/978-3-319-49130-1
Library of Congress Control Number: 2016956538
LNCS Sublibrary: SL7 – Artificial Intelligence
© Springer International Publishing AG 2016

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissions that may have been made.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
This volume collects the contributions presented at the XV Conference of the Italian
Association for Artificial Intelligence (AI*IA 2016). The conference was held in
Genova, Italy, from November 28 to December 1, 2016. The conference is organized
by AI*IA (the Italian Association for Artificial Intelligence), and it is held annually.
The conference received 53 submissions. Each paper was carefully reviewed by at
least three members of the Program Committee, and finally 39 papers were accepted for
publication in these proceedings.
Following the 2013 and 2015 editions of the conference, we adopted a “social”
format for the presentations: The papers were made available to conference participants
in advance. Each paper was shortly presented at the conference, then assigned a time
slot and a reserved table where the authors were available for discussing their work
with the interested audience. The aim of this format is to foster discussion and facilitate
idea exchange, community creation, and collaboration.
AI*IA 2016 featured exciting keynotes by Pietro Leo, Executive Architect, IBM
Italy, CTO for Big Data Analytics; Giorgio Metta, Vice Scientific Director of IIT; and
Dan Roth, University of Illinois at Urbana-Champaign.
The conference program also included seven workshops: the Second Italian
Workshop on Artificial Intelligence for Ambient Assisted Living (AAL 2016), the
Third Workshop on Artificial Intelligence and Robotics (AIRO 2016), the 10th Italian
Workshop on Artificial Intelligence for Cultural Heritage (AI*CH 2016), the 5th Italian
Workshop on Machine Learning and Data Mining (MLDM.it 2016), the 23rd RCRA
International Workshop on Experimental Evaluation of Algorithms for Solving Prob-
lems with Combinatorial Explosion (RCRA 2016), and the First Italian Workshop on
Deep Understanding and Reasoning: A Challenge for Next-Generation Intelligent
Agents (URANIA 2016), plus a Doctoral Consortium.
The chairs wish to thank the Program Committee members and the reviewers for
their careful work in selecting the best papers, the chairs of the workshops and of the
Doctoral Consortium for organizing the corresponding events, as well as Angelo
Ferrando, Frosina Koceva, and Laura Pandolfo for their help during the organization
of the conference.
September 2016 Giovanni Adorni

Stefano Cagnoni
Marco Gori
Marco Maratea
Organization
AI*IA 2016 was organized by AI*IA (in Italian, Associazione Italiana per l’Intelli-
genza Artificiale), in cooperation with the Department of Informatics, Bioengineering,
Robotics and Systems Engineering and the Polytechnic School of the University of
Genoa (Italy).
Executive Committee
General Chair
Giovanni Adorni Università di Genova, Italy
Program Chairs
Stefano Cagnoni Università di Parma, Italy
Marco Gori Università di Siena, Italy
Marco Maratea Università di Genova, Italy
Doctoral Consortium Chairs

Ilaria Torre Università di Genova, Italy
Viviana Mascardi Università di Genova, Italy
Local Chairs
Alessio Merlo Università di Genova, Italy
Simone Torsani Università di Genova, Italy
Program Committee
Matteo Baldoni Università di Torino, Italy
Stefania Bandini Università Milano-Bicocca, Italy
Roberto Basili Università di Roma Tor Vergata, Italy
Nicola Basilico Università di Milano, Italy
Federico Bergenti Università di Parma, Italy
Stefano Bistarelli Università di Perugia, Italy
Luciana Bordoni ENEA, Italy
Francesco Buccafurri Università Mediterranea di Reggio Calabria, Italy
Amedeo Cappelli CNR, Italy
Luigia Carlucci Aiello Sapienza Università di Roma, Italy
Amedeo Cesta CNR, Italy
Antonio Chella Università di Palermo, Italy
Carlo Combi Università di Verona, Italy
Gabriella Cortellessa CNR, Italy
Stefania Costantini Università dell’Aquila, Italy
VIII Organization
Giuseppe De Giacomo Sapienza Università di Roma, Italy

Nicola Di Mauro Università di Bari, Italy
Francesco Donini Università della Tuscia, Italy
Agostino Dovier Università di Udine, Italy
Floriana Esposito Università di Bari, Italy
Stefano Ferilli Università di Bari, Italy
Salvatore Gaglio Università di Palermo, Italy
Patrick Gallinari University of Paris 6, France
Marco Gavanelli Università di Ferrara, Italy
Georg Gottlob Oxford University, UK
Nicola Guarino CNR, Italy
Luca Iocchi Sapienza Università di Roma, Italy
Evelina Lamma Università di Ferrara, Italy
Nicola Leone Università della Calabria, Italy
Chendong Li Dell, USA
Francesca Alessandra Lisi Università di Bari, Italy
Sara Manzoni Università Milano-Bicocca, Italy
Paola Mello Università di Bologna, Italy
Alessio Micheli Università di Pisa, Italy
Alfredo Milani Università di Perugia, Italy
Michela Milano Università di Bologna, Italy
Stefania Montani Università del Piemonte Orientale, Italy
Alessandro Moschitti Università di Trento, Italy
Angelo Oddi CNR, Italy
Andrea Omicini Università di Bologna, Italy
Maria Teresa Pazienza Università di Roma Tor Vergata, Italy
Alberto Pettorossi Università di Roma Tor Vergata, Italy
Roberto Pirrone Università di Palermo, Italy
Piero Poccianti Consorzio Operativo Gruppo MPS, Italy
Gian Luca Pozzato Università di Torino, Italy
Luca Pulina Università di Sassari, Italy
Daniele P. Radicioni Università di Torino, Italy
Francesco Ricca Università della Calabria, Italy
Fabrizio Riguzzi Università di Ferrara, Italy
Andrea Roli Università di Bologna, Italy
Salvatore Ruggieri Università di Pisa, Italy
Fabio Sartori Università Milan-Bicocca, Italy
Ken Satoh National Institute of Informatics and Sokendai, Japan
Andrea Schaerf Università di Udine, Italy
Floriano Scioscia Politecnico di Bari, Italy
Giovanni Semeraro Università di Bari, Italy
Roberto Serra Università di Modena e Reggio Emilia, Italy
Giovanni Squillero Politecnico di Torino, Italy
Pietro Torasso Università di Torino, Italy
Leonardo Vanneschi Università Milan-Bicocca, Italy
Eloisa Vargiu Euracat, Spain
Organization IX
Marco Villani Università di Modena e Reggio Emilia, Italy

Giuseppe Vizzari Università Milano-Bicocca, Italy
Additional Reviewers
Giovanni Amendola Alessandra De Paola Marco Morana

Giuliano Armano Luca Di Gaspero Paolo Naggar
Marco Alberti Antonino Fiannaca Mirco Nanni
Eneko Agirre Claudio Gallicchio Laura Pandolfo
Cristina Baroglio Laura Giordano Giovanni Pani
Ludovico Boratto Michele Gorgoglione Fabio Patrizi
Annalina Caputo Hiroyuki Kido Andrea Pazienza
Federico Capuzzimati Yves Lesperance Zohreh Shams
Martine Ceberio Tingting Li Maria Simi
Giuseppe Cota Marco Lippi Francesco Santini
Rosario Culmone Giuseppe Loseto Pierfrancesco Veltri
Bernardo Cuteri Marco Manna Gennaro Vessio
Berardina Nadja Federico Meschini Riccardo Zese
De Carolis Marco Montali
Sponsoring Institutions
AI*IA 2016 was partially funded by the Artificial Intelligence journal, by the
Department of Informatics, Bioengineering, Robotics and Systems Engineering, and
the Polytechnic School, by the Istituto Nazionale di Alta Matematica “F. Severi” —
Gruppo Nazionale per il Calcolo Scientifico, by Camlin Italy, by the city of Genoa, and
by the Italian Association for Artificial Intelligence.
Contents
Optimization and Evolutionary Algorithms
Understanding Characteristics of Evolved Instances for State-of-the-Art

Inexact TSP Solvers with Maximum Performance Difference . . . . . . . . . . . . 3
Jakob Bossek and Heike Trautmann
On-line Dynamic Station Redeployments in Bike-Sharing Systems . . . . . . . . 13

Carlo Manna
Optimized Word-Size Time Series Representation Method Using a Genetic

Algorithm with a Flexible Encoding Scheme . . . . . . . . . . . . . . . . . . . . . . . 26
Muhammad Marwan Muhammad Fuad
Efficient Search of Relevant Structures in Complex Systems . . . . . . . . . . . . 35

Laura Sani, Michele Amoretti, Emilio Vicari, Monica Mordonini,
Riccardo Pecori, Andrea Roli, Marco Villani, Stefano Cagnoni,
and Roberto Serra
Classification, Pattern Recognition, and Computer Vision
Flat and Hierarchical Classifiers for Detecting Emotion in Tweets . . . . . . . . . 51

Giulio Angiani, Stefano Cagnoni, Natalia Chuzhikova, Paolo Fornacciari,
Monica Mordonini, and Michele Tomaiuolo
Spam Filtering Using Regularized Neural Networks with Rectified

Linear Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Aliaksandr Barushka and Petr Hájek
User Mood Tracking for Opinion Analysis on Twitter . . . . . . . . . . . . . . . . . 76

Giuseppe Castellucci, Danilo Croce, Diego De Cao, and Roberto Basili
Using Random Forests for the Estimation of Multiple Users’ Visual

Focus of Attention from Head Pose. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Silvia Rossi, Enrico Leone, and Mariacarla Staffa
Multi-agent Systems
An Analytic Study of Opinion Dynamics in Multi-agent Systems

with Additive Random Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Stefania Monica and Federico Bergenti
XII Contents
Combining Avoidance and Imitation to Improve Multi-agent

Pedestrian Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Luca Crociani, Giuseppe Vizzari, and Stefania Bandini
Knowledge Representation and Reasoning
Educational Concept Maps for Personalized Learning Path Generation. . . . . . 135

Giovanni Adorni and Frosina Koceva
Answer Set Enumeration via Assumption Literals . . . . . . . . . . . . . . . . . . . . 149

Mario Alviano and Carmine Dodaro
On the Application of Answer Set Programming to the Conference

Paper Assignment Problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Giovanni Amendola, Carmine Dodaro, Nicola Leone,
and Francesco Ricca
A Subdivision Approach to the Solution of Polynomial Constraints

over Finite Domains Using the Modified Bernstein Form . . . . . . . . . . . . . . . 179
Federico Bergenti, Stefania Monica, and Gianfranco Rossi
I -DLV: The New Intelligent Grounder of DLV . . . . . . . . . . . . . . . . . . . . . . . 192

Francesco Calimeri, Davide Fuscà, Simona Perri, and Jessica Zangari
Abducing Compliance of Incomplete Event Logs . . . . . . . . . . . . . . . . . . . . 208

Federico Chesani, Riccardo De Masellis, Chiara Di Francescomarino,
Chiara Ghidini, Paola Mello, Marco Montali, and Sergio Tessaris
Boosting the Development of ASP-Based Applications in Mobile

and General Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Francesco Calimeri, Davide Fuscà, Stefano Germano, Simona Perri,
and Jessica Zangari
Relationships and Events: Towards a General Theory of Reification

and Truthmaking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Nicola Guarino and Giancarlo Guizzardi
A Self-Adaptive Context-Aware Group Recommender System . . . . . . . . . . . 250

Reza Khoshkangini, Maria Silvia Pini, and Francesca Rossi
A Model+Solver Approach to Concept Learning . . . . . . . . . . . . . . . . . . . . . 266

Francesca Alessandra Lisi
Machine Learning
A Comparative Study of Inductive and Transductive Learning

with Feedforward Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Monica Bianchini, Anas Belahcen, and Franco Scarselli
Contents XIII
Structural Knowledge Extraction from Mobility Data. . . . . . . . . . . . . . . . . . 294

Pietro Cottone, Salvatore Gaglio, Giuseppe Lo Re, Marco Ortolani,
and Gabriele Pergola
Predicting Process Behavior in WoMan . . . . . . . . . . . . . . . . . . . . . . . . . . . 308

Stefano Ferilli, Floriana Esposito, Domenico Redavid,
and Sergio Angelastro
On-line Learning on Temporal Manifolds. . . . . . . . . . . . . . . . . . . . . . . . . . 321

Marco Maggini and Alessandro Rossi
Learning and Reasoning with Logic Tensor Networks . . . . . . . . . . . . . . . . . 334

Luciano Serafini and Artur S. d’Avila Garcez
Semantic Web and Description Logics
Probabilistic Logical Inference on the Web. . . . . . . . . . . . . . . . . . . . . . . . . 351

Marco Alberti, Giuseppe Cota, Fabrizio Riguzzi, and Riccardo Zese
Probabilistic Hybrid Knowledge Bases Under the Distribution Semantics . . . . 364

Marco Alberti, Evelina Lamma, Fabrizio Riguzzi, and Riccardo Zese
Context-Awareness for Multi-sensor Data Fusion in Smart Environments . . . . 377

Alessandra De Paola, Pierluca Ferraro, Salvatore Gaglio,
and Giuseppe Lo Re
Reasoning about Multiple Aspects in Rational Closure for DLs . . . . . . . . . . 392

Valentina Gliozzi
A Framework for Automatic Population of Ontology-Based

Digital Libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
Laura Pandolfo, Luca Pulina, and Giovanni Adorni
Reasoning About Surprising Scenarios in Description Logics of Typicality. . . 418

Gian Luca Pozzato
Natural Language Processing
A Resource-Driven Approach for Anchoring Linguistic Resources

to Conceptual Spaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
Antonio Lieto, Enrico Mensa, and Daniele P. Radicioni
Analysis of the Impact of Machine Translation Evaluation Metrics

for Semantic Textual Similarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
Simone Magnolini, Ngoc Phuoc An Vo, and Octavian Popescu
XIV Contents
QuASIt: A Cognitive Inspired Approach to Question Answering

for the Italian Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
Arianna Pipitone, Giuseppe Tirone, and Roberto Pirrone
Spoken Language Understanding for Service Robotics in Italian . . . . . . . . . . 477

Andrea Vanzo, Danilo Croce, Giuseppe Castellucci, Roberto Basili,
and Daniele Nardi
Planning and Scheduling
DARDIS: Distributed And Randomized DIspatching and Scheduling. . . . . . . 493

Thomas Bridi, Michele Lombardi, Andrea Bartolini, Luca Benini,
and Michela Milano
Steps in Assessing a Timeline-Based Planner . . . . . . . . . . . . . . . . . . . . . . . 508

Alessandro Umbrico, Amedeo Cesta, Marta Cialdea Mayer,
and Andrea Orlandini
Formal Verification
Learning for Verification in Embedded Systems: A Case Study . . . . . . . . . . 525

Ali Khalili, Massimo Narizzano, and Armando Tacchella
Learning in Physical Domains: Mating Safety Requirements

and Costly Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
Francesco Leofante and Armando Tacchella
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553

Optimization and Evolutionary
Algorithms
Understanding Characteristics of Evolved
Instances for State-of-the-Art Inexact TSP
Solvers with Maximum Performance Difference
Jakob Bossek(B) and Heike Trautmann
Information Systems and Statistics Group, University of Münster, Münster, Germany

{bossek,trautmann}@wi.uni-muenster.de
Abstract. State of the Art inexact solvers of the NP-hard Traveling

Salesperson Problem (TSP) are known to mostly yield high-quality solu-
tions in reasonable computation times. With the purpose of understand-
ing different levels of instance difficulties, instances for the current State
of the Art heuristic TSP solvers LKH+restart and EAX+restart are pre-
sented which are evolved using a sophisticated evolutionary algorithm.
More specifically, the performance differences of the respective solvers
are maximized resulting in instances which are easier to solve for one
solver and much more difficult for the other. Focusing on both optimiza-
tion directions, instance features are identified which characterize both
types of instances and increase the understanding of solver performance
differences.
Keywords: Transportation · Metaheuristics · Combinatorial optimiza-

tion · TSP · Instance hardness
1 Introduction
In the Traveling Salesperson Problem (TSP) we aim to find a minimal cost
roundtrip tour in an edge-weighted graph, which visits each node exactly once
and returns to the starting node. A plethora of algorithmic approaches for this
famous NP-hard combinatorial problem was developed in the past decades. Inex-
act solvers for the TSP are known to produce high-quality solutions in reasonable
time compared to exact solvers such as Concorde [1]. Recently, the EAX solver [2]
was shown to be competitive to the well-known State of the Art LKH algorithm
[3], more specifically respective restart variants LKH+restart and EAX+restart
even improve the original versions [4] on the Euclidean TSP. However, there is
no single inexact solver which operates best on all possible problem instances
regarding solution quality. In this work, we investigate performance differences
of the current State of the Art TSP solvers on specifically evolved instances.
Efficient algorithm selection approaches [5] in this field are conducted in
a feature- and instance-based fashion. TSP features, e.g. in [6–9]1 , are com-
puted on benchmark instances and related to algorithm performance allowing for
1
These feature sets are available in the R-package salesperson [10].

c Springer International Publishing AG 2016
G. Adorni et al. (Eds.): AI*IA 2016, LNAI 10037, pp. 3–12, 2016.
DOI: 10.1007/978-3-319-49130-1 1
4 J. Bossek and H. Trautmann
constructing algorithm selection models for unseen instances based on machine

learning techniques.
Understanding which instance characteristics pose a specific level of difficulty
onto high-performing TSP solvers is an active research field, see e.g. [4,11]. In
this paper, we specifically address LKH+restart compared to EAX+restart as
the two current State of the Art TSP solvers with potential for improving their
standalone application by means of a portfolio approach [4]. We are specifically
interested in instances on which both solvers exhibit maximum performance
difference, i.e., which are much harder to solve for one of the solvers, while we
focus both directions. Thus, the performance ratio is used as fitness function of
a sophisticated evolutionary algorithm for evolving instances which was already
used in a similar fashion for single solvers in [7,8,12,13]. Two variants of solver
performance are contrasted. The classical mean par10 score is supplemented by
focussing on the median solver runtime over a fixed number of runs diminishing
the influence of timeouts in individual runs. Moreover, the influence of rounding
instances to a grid structure is analysed systematically. Additionally, we contrast
characteristics of instances which are much harder or much easier for one solver
w.r.t. the other.
Section 2 details the evolutionary algorithm. Experimental results are then
presented in Sect. 3. Conclusions and an outlook on future research are given in
Sect. 4.
2 EA for Evolving Instances
Algorithm 1 reflects the process of the evolutionary algorithm in terms of

pseudocode. The core parameter of the EA is the kind of fitness function used.
As the EA aims at generating instances with maximum performance difference
of two solvers, we define the fitness function as the performance ratio P(A,B) (I)
for a pair of solvers A and B, i.e.
PA (I)
P(A,B) (I) =
PB (I)
on a specific instance I, where PA (I) and PB (I) are the solver performances
of solver A and B on instance I. Solver performance in our scenario is either
determined by the standard indicator penalized average runtime or by the penal-
ized median runtime. The former repeatedly measures the runtime of the solver
on an instance until the optimal tour (pre-computed by Concorde) has been
found and computes the arithmetic mean subsequently. In case the cutoff time
timeLimit is reached, ten times the cutoff time is used for further computations
as a penalty. However, inside the EA, the actual cutoff time is used ensuring
that the probability of removal of such a solution at later algorithm stages is
not unreasonably low. The evaluation at the final generation uses the classical
par10 score with the full penalty. The median score is much more insensitive to
outliers and maximum ratio in medians is much harder to obtain.
Understanding Characteristics of Evolved Instances 5
Algorithm 1. Evolving EA
1: function EA(fitnessFun, popSize, instSize, generations, timeLimit, cells,
rnd=true)
2: poolSize = popSize / 2
3: population = generateRandomInstances(popSize, instSize) in [0, 1]2
4: while stopping condition not met do
5: for i = 1 → popSize do
6: fitness[i] = fitnessFun(population[i])
7: end for
8: matingPool = createMatingPool
9: 2-tournament-selection
10: offspring[1] = getBestFrom
11: CurrentPopulation 1-elitism
12: for i = 2 → popSize do
13: Choose p1 and p2 randomly from the
14: mating pool
15: offspring[i] = applyVariations(p1, p2)
16: Rescale offsspring to [0, 1]2 by dimension
17: if rnd then
18: Round each point to nearest cell grid
19: end if
20: end for
21: population = offspring
22: end while
23: end function
The initial population of size popSize is randomly generated in [0, 1]2 for
instances of size instSize and the performance ratio is computed. Distances are
scaled by multiplying with a factor of 100 and afterwards rounded to the nearest
integer. This step is neccassary since EAX expects integer distances. The EA is
then run for a fixed number of generations and the evolutionary loop is executed
as follows: The mating pool is formed by 2-tournament selection supplemented by
the best solution of the current population (1-elitism). Two different mutation
operators are applied to each combination of randomly drawn instance pairs
of the mating pool. Uniform mutation replacing coordinates of selected nodes
with new randomly chosen coordinates is applied with a very low probability
possibly followed by gaussian mutation adding normal noise to the selected point
coordinates. Therefore, global as well as local changes can come into effect. In
the current version the EA does not use any recombination operator. A single
EA generation ends after rescaling the instance to completely cover [0, 1]2 and,
if rnd = true, rounding the points to the nearest cell grid. The latter relates to
important relevant structures in practice such as the structural design of circuit
boards.
3 Experiments
3.1 Experimental Setup
In total 200 TSP instances were evolved. For all four considered opti-
mization directions, i.e. P(LKH,EAX) , P(EAX,LKH) , P(LKH+restart,EAX+restart)
and P(EAX+restart,LKH+restart) , in each case 25 instances were generated
with activated and deactivated rounding respectively. Based on prelimi-
nary experiments and experimental results of [8,12] the EA parameters
were set as follows: timeLimit = 120, popSize = 30, generations =
5000, unif ormM utationRate = 0.05, normalM utationRate = 0.1,
normalM utationSD = 0.025 and cells = 100. We used the reference imple-
mentation LKH 2.0.7 based on the former implementation 1.3 [14], the original
EAX implementation as well as specific restart variants as described in [4]. The
solvers were repeatedly evaluated, three times inside the EA due to a limited
computational budget but ten times for final evaluations. As described in Sect. 2
either the par10 score or the median score was computed for the final instances.
For comparison and practical validation, performance ratios of the respec-
tive solvers on TSPLIB instances2 of comparable size, i.e. 200 ≤ instSize ≤ 400
were computed for both kinds of performance measures. Moreover, 100 random
instances in [0, 1]2 were generated while the same rounding strategy of the dis-
tance matrix was applied as used inside the EA for the evolved instances. All
experiments were run on the parallel linux computer cluster PALMA at Univer-
sity of Münster, consisting of 3528 processor cores in total. The utilized compute
nodes are 2,6 GHz machines with 2 hexacore Intel Westmere processors, totally
12 cores per node and 2 GB main memory per core.
Fig. 1. Average (left) and median (right) par10 scores (log-scale) of LKH+restart and
EAX+restart on evolved, random and TSPLIB instances. A specific symbol visualizes
whether instances were rounded to a grid structure (rnd) or not (nrnd).
2
TSPLIB-Instances: a280, gil262, kroA200, kroB200, lin318, pr226, pr264, pr299,
rd400, ts225, tsp225.
3.2 Results
Figures 1 and 2 give an overview about the performance scores of the considered
solver pairs, i.e., both for the original as well as the restart variants. Evolved
instances are visualized together with random and TSPLIB instances.
Fig. 2. Average (left) and median (right) par10 scores (log-scale) of LKH and EAX on
evolved, random and TSPLIB instances. A specific symbol visualizes whether instances
were rounded to a grid structure (rnd) or not (nrnd).
It becomes obvious that in both pairings the presented EA successfully gen-

erated instances with much higher performance differences of both solvers than
usually present in random as well as TSPLIB instances. Whether the instance
was rounded to a grid structure inside the EA does not have a structural influ-
ence on the relation of the performance scores. Moreover, we see that generating
easier instances for LKH+restart compared to the EAX+restart is a much harder
task than in case of considering the opposite direction. Specifically, EAX+restart
timeouts did not occur here. On the contrary, for variants without restart, this
effect cannot be observed. In addition, in some cases, the EA did not converge in
that instances are of similar difficulty for both solvers. This behaviour, however,
is due to a part of solver runs resulting in timeouts as reflected by the location
of the points in case the median scores are considered (Fig. 2, lower right part).
In general, evaluating with median scores diminishes the influence of timeouts.
Maximum median scores can only be obtained in case at least fifty percent of
solver runs on a specific instance result in a timeout. Therefore, there could be
potential of using this kind of performance measure inside the EA. Results are
presented further down.
The previous observations are reflected in Fig. 3 as well. Here, boxplots of
performance scores on the evolved instances for each solver depending on the
optimization direction as well as the rounding activations are given. Supplemen-
tary to Figs. 1 and 2 all solvers have been evaluated on the respective instances.
Not surprisingly, instances specifically generated for the restart variants do not
result in such extreme performance differences for the classical variants and the
other way round. However, the basic tendency can be observed here as well.
Fig. 3. Mean (left) and median par10 scores (right) of the four solvers depending on
rounding (rnd) and type of optimization (log-scale).
Evaluating with median scores shows that especially the pairing (LKH,EAX),
i.e. generating easier instances for LKH, does not show the desired performance
effects. Figure 4 explicitly provides boxplots of the performance ratios P(A,B) and
thus summarizes all effects previously listed, in particular the huge differences
in performance ratios compared to random and TSPLIB instances.
Understanding Characteristics of Extreme Instances. Next we try to under-

stand which structural TSP characteristics, termed TSP instance features, of
the evolved instances are suitable to distinguish between easy and hard to solve
instances respectively. Those features could be used in algorithm selection sce-
narios to select the best solver out of a portfolio of solvers. A classification app-
roach was used in order to separate the respective groups of the most extreme
instances by means of well-known established TSP features as introduced in
[6,8]. The R-package salesperson [10] was used for this purpose. However, we
did not consider expensive features such as local search, clustering distance based
features as well as branch and cut techniques in order to avoid much computa-
tional overhead, especially regarding the possible influence on future algorithm
selection models as e.g. in [4].
In case of the restart variants, the ten most extreme instances w.r.t. perfor-
mance ratios on mean par10 scores are selected, i.e. the ten best EA results of
both optimization directions for the solver pairings. A random forest was used
to distinguish between both instance sets combined with a feature selection app-
roach based on leave-one-out cross-validation and an outer resampling resulting
in a median misclassification error of zero and a respective mean misclassification
error of 0.3 regarding all folds. Thus, we are able to separate both instance sets in
a satisfying way including an indication which features are of crucial importance
here. Figure 5 shows the respective feature importance plot created by means of
the R-package flacco [15]. Red dots reflect features which are used in at least
70 % of the folds, orange labeled features at least in 30 % and black ones at
least once. In this regard the median distance of the minimum spanning tree is
Fig. 4. Performance ratios based on mean par10 scores of the four considered algo-
rithms on all considered instance sets.
identified as the crucial feature separating both instance classes. The results coni-
cide with the results of [8] where the mean distance of the minimum spanning
tree was identified as a separating feature between easy and hard instances for
TSP approximation algorithms. This result is promising with respect to future
work in the algorithm selection context: The computation of minimum spanning
tree characteristics is an computationally cheap task and we strive for cheap fea-
ture, since wasting a lot of runtime for the feature computation before actually
solving the TSP itself is senseless.
The same analysis was conducted for the original solver variants. However,
as evolved instances are much denser in the lower right and upper left corner in
Fig. 2 than in the restart case, we only selected the respective five most extreme
instances. In this case different features play a key role in explaining solver per-
formance differences including nearest-neighbor based features as visualized in
Fig. 5. Again, the median misclassification error vanishes while the mean mis-
classification error is 0.2, i.e. only two out of the ten instances are misclassified.
Median Scores as EA Internal Performance Measures. In order to investigate

possible potential of using the median par10 score inside the EA for performance
evaluation, we conducted a smaller experiment focussed on the restart variants
with inactive rounding as these solvers in our view are most interesting. This
lead to fifty evolved instances, i.e. 25 for each optimization direction.
Figure 6 gives an overview about the resulting median par10 scores on the
newly evolved instances together with the random instance and TSPLIB results.
We see that the EA is not successful in improving the performance ratio of both
Fig. 5. Variable Importance Plot of Random Forest distinguishing the (left): ten most
extreme instances w.r.t. performance ratio for the restart variants, (right): five most
extreme instances w.r.t. performance ratio for the original algorithm variants.
Fig. 6. Median par10 scores (log-scale) of LKH+restart and EAX+restart on evolved

(median score fitness), random and TSPLIB instances.
solvers even further compared to the resulting median evaluation on the instances
originally generated inside the EA using mean par10 scores. The same is true for
comparing the mean par10 scores on both scenarios (see Fig. 7). However, slight
improvements are visible in case easier instances for LKH+restart are evolved.
Most probably the median alone does not provide enough differentiation between
varying solver results over the repetitions.
However, in our view an adequate combination of mean and median scores
inside the EA fitness function is promising in order to get deeper insights into
solver variance on the considered instances. We will investigate this issue in
further studies together with increasing the number of solver repetitions along
the evolutionary loop.
Fig. 7. Comparison of mean and median par10 ratios (log-scale) of instance sets evolved
for LKH+restart and EAX+restart either by using the mean par10 or the median score
inside the EA as fitness function.
4 Conclusions
This work focusses on the two current State of the Art inexact TSP solvers LKH
and EAX together with their respective restart variants. In order to increase
understanding of performance differences of both solvers, a sophisticated evolu-
tionary algorithm was used to evolve instances which lead to maximum perfor-
mance difference of both solvers on the specific instances. Both directions are
analyzed, i.e. we generated instances which are easier for solver A but much
harder for solver B as well as the opposite case. In this regard we observed sub-
stantial differences in solver performance ratios compared to random or TSPLIB
instances on the evolved instances. By feature-based analysis of the most extreme
instances in terms of performance ratio crucial features are identified for both
solver pairings which are indicated to have an influence on solver-specific prob-
lem difficulty. Moreover, we contrasted the classical mean par10 score with a
respective median version to even increase the challenge of evolving instances
with high solver performance differences.
Future studies will focus on generalizing the results to higher instance sizes
and on designing a more sophisticated fitness function inside the EA to even
increase solver performance differences on the evolved instances.
Acknowledgments. The authors acknowledge support by the European Research

Center for Information Systems (ERCIS).
References
1. Applegate, D.L., Bixby, R.E., Chvatal, V., Cook, W.J.: The Traveling Salesman
Problem: A Computational Study. Princeton University Press, Princeton (2007)
2. Nagata, Y., Kobayashi, S.: A powerful genetic algorithm using edge assembly
crossover for the traveling salesman problem. INFORMS J. Comput. 25, 346–363
(2013)
3. Helsgaun, K.: General k-opt submoves for the Lin-Kernighan TSP heuristic. Math.
Program. Comput. 1, 119–163 (2009)
4. Kotthoff, L., Kerschke, P., Hoos, H., Trautmann, H.: Improving the state of the
art in inexact TSP solving using per-instance algorithm selection. In: Dhaenens,
C., Jourdan, L., Marmion, M.-E. (eds.) LION 2015. LNCS, vol. 8994, pp. 202–217.
Springer, Heidelberg (2015). doi:10.1007/978-3-319-19084-6 18
5. Kotthoff, L.: Algorithm selection for combinatorial search problems: a survey. AI
Mag. 35, 48–60 (2014)
6. Hutter, F., Xu, L., Hoos, H.H., Leyton-Brown, K.: Algorithm runtime prediction:
methods & evaluation. Artif. Intell. 206, 79–111 (2014)
7. Smith-Miles, K., van Hemert, J.: Discovering the suitability of optimisation algo-
rithms by learning from evolved instances. Ann. Math. Artif. Intell. 61, 87–104
(2011)
8. Mersmann, O., Bischl, B., Trautmann, H., Wagner, M., Bossek, J., Neumann,
F.: A novel feature-based approach to characterize algorithm performance for the
traveling salesperson problem. Ann. Math. Artif. Intell. 69, 1–32 (2013)
9. Pihera, J., Musliu, N.: Application of machine learning to algorithm selection for
TSP. In: Fogel, D., et al. (eds.) Proceedings of the IEEE 26th International Con-
ference on Tools with Artificial Intelligence (ICTAI). IEEE press (2014)
10. Bossek, J.: salesperson: Computation of Instance Feature Sets and R Interface
to the State-of-the-Art Solvers for the Traveling Salesperson Problem. R package
version 1.0 (2015)
11. Fischer, T., Stützle, T., Hoos, H.H., Merz, P.: An analysis of the hardness of
TSP instances for two high-performance algorithms. In: Proceedings of the 6th
Metaheuristics International Conference, Vienna, Austria, pp. 361–367 (2005)
12. Mersmann, O., Bischl, B., Bossek, J., Trautmann, H., Markus, W., Neumann, F.:
Local search and the traveling salesman problem: a feature-based characterization
of problem hardness. In: Hamadi, Y., Schoenauer, M. (eds.) LION 6. LNCS, vol.
7219, pp. 115–129. Springer, Heidelberg (2012)
13. Nallaperuma, S., Wagner, M., Neumann, F., Bischl, B., Mersmann, O., Trautmann,
H.: A feature-based comparison of local search and the christofides algorithm for
the travelling salesperson problem. In: Foundations of Genetic Algorithms (FOGA)
(2013) (accepted)
14. Lacoste, J.D., Hoos, H.H., Stützle, T.: On the empirical time complexity of state-
of-the-art inexact tsp solvers. Optimization Letters (to appear)
15. Kerschke, P., Dagefoerde, J.: flacco: Feature-Based Landscape Analysis of Contin-
uous and Constraint Optimization Problems. R package version 1.1 (2015)
On-line Dynamic Station Redeployments
in Bike-Sharing Systems
Carlo Manna(B)
Insight Research Centre for Data Analytics, University College Cork, Cork, Ireland
[email protected]
Abstract. Bike-sharing has seen great development during recent years,

both in Europe and globally. However, these systems are far from per-
fect. The uncertainty of the customer demand often leads to an unbal-
anced distribution of bicycles over the time and space (congestion and/or
starvation), resulting both in a loss of customers and a poor customer
experience. In order to improve those aspects, we propose a dynamic
bike-sharing system, which combines the standard fixed base stations
with movable stations (using trucks), which will able to be dynamically
re-allocated according to the upcoming forecasted customer demand dur-
ing the day in real-time. The purpose of this paper is to investigate
whether using moveable stations in designing the bike-sharing system
has a significant positive effect on the system performance. To that end,
we contribute an on-line stochastic optimization formulation to address
the redeployment of the moveable stations during the day, to better
match the upcoming customer demand. Finally, we demonstrate the util-
ity of our approach with numerical experiments using data provided by
bike-sharing companies.
Keywords: On-line combinatorial optimization · Uncertainty · Smart

cities
1 Introduction
Bike-sharing systems (BSS) are in place in several cities in the world, and are
an increasingly important support for multi-modal transport systems [1]. BSS
are widely adopted with 747 active systems, a fleet of over 772,000 bicycles
and 235 systems in planning or under construction [2]. A BSS typically has a
number of base stations scattered throughout a city. At the beginning of the
day, each station is stocked with a pre-determined number of bikes. Users with a
membership card can pick up and return bikes from any designated station, each
of which has a finite number of docks. At the end of the work day, trucks are used
to move bikes around so as to return to some pre-determined configuration at the
beginning of the day. Due to the individual movement of customers according to
their needs, there is often congestion (more than required) or starvation (fewer
than required) of bikes at certain base stations. According to CapitalBikeShare
Company [3], in a city like Washington, at a minimum, there are around 100

c Springer International Publishing AG 2016
G. Adorni et al. (Eds.): AI*IA 2016, LNAI 10037, pp. 13–25, 2016.
DOI: 10.1007/978-3-319-49130-1 2
14 C. Manna
cases of empty stations and 100 cases of full stations per day and at a maximum
there are about 750 cases of empty stations and 330 cases of full stations per day.
As demonstrated in [4] this can result in a significant loss of customer activity.
Such loss in demand can have two undesirable outcomes: (a) loss in revenue; (b)
increase in carbon emissions, as people resort to less environmentally-friendly
modes of transport. To address such a problem, the majority of the proposed
solutions aim to find more efficient methods for dynamic rebalancing of the
number of bicycles on the base stations taking into account the uncertainty of
the customer demand or predicting the customer demand at each station [5,6].
Basically, by using trucks a number of bicycles are transferred from one station
to another to accomplish the upcoming demand. This operation takes place ones
a day (or more in some specific situations).
The aim of this paper is to study a totally different approach, in which
the fixed base stations are augmented by a number of movable stations (using
trucks) with the dual purpose of both: (1) dynamically adding/re-allocating dock
stations in city areas to match customer demand in real-time and, (2) a dynamic
re-balancing of the number of bicycles in some particular fixed station where a
redeployment of a docking station is unnecessary.
Particularly, we consider a problem in which the day time is partitioned in
time intervals of equal length. We suppose that for each of those time periods,
the probability distributions of the travel demands between different locations
are known. At start of the day, we compute the best possible locations for the
dock stations in each time period, taking into account the stochastic nature of
the demand, with the aim to maximize the number of customers. This is an
on-line problem. That means, although the solution is a sequence of decisions
for each time period, only the immediate decision is actually taken (i.e. the
station allocations for the incoming time period). While this time period, a new
computation is performed, using more updated travel demand predictions, and a
new decision for the next period is taken from the re-computed decision sequence.
This carry on until the termination of the time horizon.
The main advantage of such a system is that the bike stations configuration
is not fixed, but it can change adaptively with the travel demand day by day
and, in each day, it can change during a number of time periods, in the respect of
specific time constraints (i.e. a new configuration must be computed in advance
with enough time to allow the repositioning of the stations). Finally, updating
the decisions each time periods, allow reducing the uncertainty in the predictions.
This because the majority of the travel demand prediction techniques are based
on auto-regressive models, which make use of the most recent known data to
predict future outcomes [7].
The key distinction from existing research on bike sharing is that we consider
the dynamic redeployment of bicycle stations (instead of just rebalancing the
number of bicycles in the existing stations). This approach from one side does
not exclude the possibility to rebalancing as in the existing research. However
this extends such a research to a novel approach to the BSS by which is possible
to dynamically change the configuration of the dock stations in real-time to
On-line Dynamic Station Redeployments in Bike-Sharing Systems 15
maximize the potential customer demand. In doing this, while numerous models
exist in literature, in this paper the potential customer demand is based on a
primary concept: the distance decay. That is “the attenuation of a pattern or a
process with distance” [8]. In other words, people are less willing to use a facility
allocated too far from them. This is a focal concept in a variety of modelling
contexts, such as transporation, migration and location theory [9].
Specifically, our key contributions are as follows:
– A mixed and linear programming (MILP) formulation to maximize the

expected demand assigned to the moveable bicycle stations while simulta-
neously address the distance decay;
– An on-line stochastic optimization formulation to address the dock stations
re-allocation problem during the day to accomplish the estimated customer
demand in real-time;
– A potentially novel approach to the design of a bike-sharing system, along
with numerical results for an initial investigation.
Extensive numerical simulations using datasets of two bike-sharing compa-

nies, namely Capital Bikeshare (Washington, DC) [10] and Hubway (Boston,
MA) [11] show that the proposed approach can improve the customer usage of
the bike-sharing system and that the computation time is reasonably fast to be
used in real-time.
2 Related Work
Although bike sharing systems are relatively new, they have been studied exten-
sively in the literature. For this reason, we only focus on threads of research that
are of relevance to this paper. However, on the best of our knowledge, there is
no any previous work for on-line stochastic redeployment of moveable stations
in bike-sharing.
The first thread of research focus on the bicycles rebalancing between the
stations. Particularly, [12–14] focus on the problem of finding routes at the end
of the day for a set of carriers to achieve the desired configuration of bikes
across the base stations. They have provided scalable exact and approximate
approaches to this problem by either abstracting base stations into mega sta-
tions or by employing insights from inventory management or by using variable
neighbourhood search based heuristics. Those works assume there is only one
fixed redeployment of bikes that happens at the end of the day. In contrast,
[15] predict the stochastic demand from user trip data of Singapore metro sys-
tem using poisson distribution and provide an optimization model that suggests
the best location of the stations and a dynamic bicycles redeployment for the
model to minimize the number of unsatisfied customers. However, they assume
that redeployment of bikes from one station to another is always possible with-
out considering the routing of carriers, which is a major cost driver for the
bike-sharing company. In [16] they overcome this problem, developing a mixed
16 C. Manna
integer programming formulation which includes the carrier routes into the opti-
mization model. Finally, other relevant works have been proposed in [17] to deal
with unmet demand in rush hours. They provide a myopic redeployment policy
by considering the current demand. They employed Dantzig-Wolfe and Benders
decomposition techniques to make the decision problem faster. [18] also provides
a myopic online decisions based on assessment of demand for the next 30 min.
The second thread of research is complementary to the work presented in
this paper is on demand prediction and analysis. [19] provides a service level
analysis of the BSS using a dualbounded joint-chance constraints where they
predict the near future demands for a short period of time. Finally, in [20], the
BSS is represented as a dual markovian waiting system to predict the actual
demand.
As we already highlighted, all the aforementioned works differ from the one
proposed in this paper as we consider a dynamic re-allocation of a certain number
of bicycle stations during the day. This lead to a formulation of the problem
which is different from all the previous provided in literature.
3 Problem Description
In this section we formally describe the bike-sharing system with dynamic rede-
ployment. It is compactly described using the following tuple: T, A, K, S, P, D,
X, X̂, δ, l, where A represents the set of areas for which the demand has to be
covered, K represents the set of possible locations for the dock stations with
K ⊆ A, T is the time horizon. S is a binary vector representing the totality
of decisions on the allocated stations, with Skt denotes the decision on whether
or not enabling a dock station in k ∈ K at time t ∈ T . Furthermore, P is a
binary vector representing the distribution of the service coverage between areas
t
and dock stations. In particular, Pa,k denotes whether or not the area a ∈ A is
served by the station in k ∈ K, at time t ∈ T . D is a vector of the distances
(or travel time) between all areas and station locations, with da,a (da,k ), the
distance/travel time between the area a and a (between the area a and the sta-
tion in k). X is a vector of the probability distribution of the potential customer
demand between different areas, with xta,a denoting the potential travel demand
from the area a to the area a (with (a, a ) ∈ A), at time t ∈ T . The potential
customer demand at time t denotes the maximum number of possible customers
to be served. This is expressed in form of probability distribution. Similarly, X̂
is a vector of the expected covered customer demand between different areas,
with x̂ta,a denoting the expected travel demand from the area a to the area a
(with (a, a ) ∈ A), at time t ∈ T . Finally δ ∈ [0, 1] is a distance decay parame-
ter [8], through which we take into account the aforementioned distance decay
concept. Finally l is the maximum number of moveable stations. The expected
covered customer demand denotes the number of customers which is expected
to use the bike-sharing system. Hence, given a potential demand xta,a , it holds
x̂ta,a xta,a for each a ∈ A and t ∈ T . X̂ depends on various factors, such as the
On-line Dynamic Station Redeployments in Bike-Sharing Systems 17
total distance/travel time between the customer location and the dock station,
and from the distance decay parameter δ.
Given the potential customer demand X instantiated from a known prob-
ability distribution, at each time step t ∈ T , the goal is to maximize a profit
function W (·) of the overall expected covered demand X̂ over the total time
horizon T, subject on some problem specific constraints C(·) (both are specified
in the following paragraph). This is achieved through finding the best possible
sequence of decisions S 1 , . . . , S T , concern the locations of the dock stations
over the entire time horizon T.
Finally, as reported in the relevant literature [21], in defining the solution
approach, we have made the following assumptions:
1. the expected covered demand decrease with the total distance travelled on
foot (i.e. the distance between the starting point and the pick up station,
and the distance between the return station and the arrival point), given a
predefined decay parameter δ.
2. each area can be served by only one station.
The assumption (1) states that the distance travelled on bicycle between the
stations does not have any negative impact on the demand. This because the
potential demand X is already estimated considering this factor. Conversely,
the distance travelled on foot does have a negative impact on the service usage.
The assumption (2) states that the totality of the demand in an area a ∈ A,
considering a either starting or arrival point, can be served by only one station
located at k ∈ K. This hypothesis is one of the most used in many facility
location models, which assumes that the customer always use the closest facility.
4 Solution Approach
We propose an on-line stochastic combinatorial optimization approach for the

system described in the last section. Consequently, we first outline the solution
approach for the deterministic case (in which we consider no uncertainty into
the customer demand) and then, we remove such hypothesis and extend this
formulation to the on-line stochastic case.
4.1 Deterministic Case
The deterministic model is based on a single scenario. That means considering

X as a single per-determined scenario values for the potential customer demand
between different areas. The solution is expressed as a sequence of decisions to
be taken at each time step. We do not need to recompute the solution at each
time step, since the demand remain the same in each period. Hence the problem
does not have any on-line feature.
To address the deterministic case we propose a MILP formulation with the
following decision variables:
Another random document with
no related content on Scribd:
in South Africa, and to produce what, thank God! he had failed
in producing—a racial war." Mr. Chamberlain retorted that Sir
William Harcourt's attitude was unpatriotic and injurious to
the cause of peace. He denied aggressiveness in the policy of
the government, asserting that the South African Republic had
been spending millions on armaments imported from abroad, in
view of which the strengthening of the British garrison at the
Cape by an additional regiment and three batteries was no
unreasonable measure. Mr. Balfour, also, begged the House and
the country to believe that the troops were sent only as a
measure of precaution, to maintain admitted rights.
{478}
SOUTH AFRICA: The Transvaal: A. D. 1897 (May-October).

The British assertion of suzerainty and declination of
proposal to arbitrate disagreements.
On the 7th of May, 1897, the Acting State Secretary of the

South African Republic addressed to the British Agent at
Pretoria a communication of great length, reviewing the
positions taken by Mr. Chamberlain in his several arraignments
of the government of the Republic for alleged violation of the
London Convention of 1884, and proposing an arbitration of the
questions involved. "The complaint," he wrote, "which Her
British Majesty's Government has advanced in an unmistakably
pronounced manner over an actual or possible breach of the
Convention has deeply grieved this Government, as it thinks
that it has fulfilled its obligations. It sees in the
fulfilment of the mutual obligations under the Convention one
of the best guarantees for the maintenance of a mutual good
understanding and for the promotion of reciprocal confidence.
To this good understanding and that confidence, however,
severe shocks have been given by events which cannot be
lightly forgotten. And if it were not that this Government
wishes to guard itself against adopting a recriminating tone,
it might put the question whether, for example, the incursion
of Dr. Jameson, whether considered as a breach of the
Convention or a grievance, is not of immeasurably greater
importance than the various matters adduced by Her British
Majesty's Government would be, even if the contention that
they constitute breaches of the Convention could be accepted.
There should, in the view of this Government, be a strong
mutual endeavour to restore the shocked confidence and to calm
the excited spirit which this Government with sincere regret
sees reigning throughout almost the whole of South Africa.
This Government is anxious to co-operate for this end, for the
desire of the Republic, with the maintenance of its
independence and rights, is for peace, and where for the
reasons given it has been unable to entertain the proposal of
Her British Majesty's Government in the matter of the Aliens
Law,—and it appears very difficult to arrive at a solution of
the question by means of correspondence,—it wishes to come to
a permanent good understanding along a peaceful course, not
only with respect to its undisturbed right to make an alien
law, but also with regard to all points touching the
Convention which are referred to in the two Despatches under
reply by Her British Majesty's Government. While it respects
the opinion of Her British Majesty's Government, it takes the
liberty, with full confidence in the correctness of its own
views, to propose to Her British Majesty's Government the
principle of arbitration with which the honourable the First
Volksraad agrees, in the hope that it will be taken in the
conciliatory spirit in which it is made. …
"Although this Government is firmly convinced that a just and

impartial decision might be obtained even better in South
Africa than anywhere else, it wishes, in view of the
conflicting elements, interests, and aspirations, which are
now apparent in South Africa, and in order to avoid even the
appearance that it would be able or desire to exercise
influence in order to obtain a decision favourable to it, to
propose that the President of the Swiss Bondstate, who may be
reckoned upon as standing altogether outside the question, and
to feel sympathy or antipathy neither for the one party nor
for the other, be requested to point out a competent jurist,
as has already oftener been done in respect of international
disputes. The Government would have no objection that the
arbitrator be subject to a limitation of time, and gives the
assurance now already that it will willingly subject itself to
any decision if such should, contrary to its expectation, be
given against it. The Government repeats the well-meant wish
that this proposal may find favour with Her British Majesty's
Government, and inasmuch as the allegations of breaches of the
Convention find entrance now even in South Africa, and bring
and keep the feelings more and more in a state of suspense,
this Government will be pleased if it can learn the decision
of Her Majesty's Government as soon as possible."
Mr. Chamberlain's reply to this proposal was not written until

the 16th of the following October, when he, in turn, reviewed,
point by point, the matters dealt with, in the despatch of Mr.
Van Boeschoten. With reference to the Jameson raid he said: "Her
Majesty's Government note with satisfaction that the
Government of the South African Republic see in the fulfilment
of the mutual obligations under the London Convention one of
the best guarantees for the maintenance of a mutual
understanding and for the promotion of reciprocal confidence.
Her Majesty's Government have uniformly fulfilled these
obligations on their part, and they must strongly protest
against what appears to be an implication in the Note under
consideration that the incursion of Dr. Jameson can be
considered as either a breach of the Convention by Her
Majesty's Government or a grievance against them. That
incursion was the act of private individuals unauthorised by
Her Majesty's Government, and was repudiated by them
immediately it became known. The immense importance to the
Government of the South African Republic of that repudiation,
and of the proclamation issued by the High Commissioner under
instructions from Her Majesty's Government, is recognised
throughout South Africa. Her Majesty's Government maintain
strongly that since the Convention of 1881 there has never
been any breach or even any allegation of a breach on their
part of that or the subsequent Convention, and, as the subject
has been raised by the implied accusation contained in the
Note under consideration, Her Majesty's Government feel
constrained to contrast their loyal action in the case of the
Jameson raid with the cases in which they have had cause to
complain that the Government of the South African Republic
failed to interfere with, if they did not countenance,
invasions of the adjacent territories by its burghers in
violation of the Convention, and they feel bound to remind the
Government of the Republic that in one of these cases Her
Majesty's Government were compelled to maintain their rights
by an armed expedition at the cost of about one million
sterling, for which no compensation has ever been received by
them."
Concerning the proposal of arbitration, the reply of the

British colonial secretary was as follows: "In making this
proposal the Government of the South African Republic appears
to have overlooked the distinction between the Conventions of
1881 and 1884 and an ordinary treaty between two independent
Powers, questions arising upon which may properly be the
subject of arbitration.
{479}
By the Pretoria Convention of 1881 Her Majesty, as Sovereign
of the Transvaal Territory, accorded to the inhabitants of
that territory complete self-government subject to the
suzerainty of Her Majesty, her heirs and successors, upon
certain terms and conditions and subject to certain
reservations and limitations set forth in 33 articles, and by
the London Convention of 1884 Her Majesty, while maintaining
the preamble of the earlier instrument, directed and declared
that certain other articles embodied therein should be
substituted for the articles embodied in the Convention of
1881. The articles of the Convention of 1881 were accepted by
the Volksraad of the Transvaal State, and those of the
Convention of 1884 by the Volksraad of the South African
Republic. Under these Conventions, therefore, Her Majesty
holds towards the South African Republic the relation of a
suzerain who has accorded to the people of that Republic
self-government upon certain conditions, and it would be
incompatible with that position to submit to arbitration the
construction of the conditions on which she accorded
self-government to the Republic. One of the main objects which
Her Majesty's Government had in view was the prevention of the
interference of any foreign Power between Her Majesty and the
South African Republic, a matter which they then held, and
which Her Majesty's present Government hold, to be essential
to British interests, and this object would be defeated by the
course now proposed. The clear intention of Her Majesty's
Government at the time of the London Convention, that
questions in relation to it should not be submitted to
arbitration, is shown by the fact that when the delegates of
the South African Republic, in the negotiations which preceded
that Convention, submitted to Her Majesty's Government in the
first instance (in a letter of the 26th of November, 1883,
which will be found on page 9 of the Parliamentary Paper C.
3947 of 1884) the draft of a treaty or convention containing
an arbitration clause, they were informed by the Earl of Derby
that it was neither in form nor in substance such as Her
Majesty's Government could adopt."
Great Britain, Papers by Command.

C.—8721, 1898.
SOUTH AFRICA: Natal: A. D. 1897 (December).

Annexation of Zululand.
See (in this volume)

AFRICA: A. D. 1897 (ZULULAND).
SOUTH AFRICA: Cape Colony: A. D. 1898.

The position of political parties.
The Progressives and the Afrikander Bund.
"The present position of parties at the Cape is as unfortunate

and as unwarranted as any that the severest critic of
Parliamentary institutions could have conjured up. … The Cape
has always had the curse of race prejudice to contend with.
Time might have done much to soften, if not to expunge it, if
home-made stupidities had not always been forthcoming to goad
to fresh rancour. The facts are too well known to need
repetition. It is true not only of the Transvaalers that 'the
trek has eaten into their souls,' and up to the time of
emancipation and since, every conceivable mistake has been
committed by those in authority. Thus, when the breach was, to
all appearances, partly healed, the fatal winter of 1895 put
back the hands of the clock to the old point of departure. As
Englishmen, our sympathies are naturally with the party that
is prevalently English, and against the party that is
prevalently Dutch; but to find a real line of political
difference between them other than national sentiment requires
fine drawing. … According to our lines of cleavage both
Bondsmen [Afrikander Bund] and Progressives are Conservatives
of a decided type. Practically they are agreed in advocating
protective duties on sea-borne trade, although in degree they
differ, for whilst the Bond would have imposts as they are,
the Progressives wish to reduce the duties on food stuffs to
meet the grievance of the urban constituencies, and might be
induced to accord preferential treatment to British goods. On
the native question neither party adopts what would in England
be considered an 'advanced' programme, for education is not
made a cardinal point, and they would equally like, if
possible, to extend the application of the Glen Grey Act,
which, by levying a tax on the young Kaffirs who have not a
labour certificate, forces them to do some service to the
community before exercising their right of 'putting the
spoon,' as the phrase is, 'into the family pot.' Neither party
wishes to interfere with the rights of property or the
absolute tenure of land under the Roman-Dutch law. A tax on
the output of diamonds at Kimberley has been advocated by some
members of the Bond as a financial expedient, but it is
understood to have been put forward rather as a threat against
Mr. Rhodes personally than as a measure of practical politics.
Questions of franchise are tacitly left as they are, for no
responsible politicians wish to go back upon the enactment
which restricted the Kaffir vote to safe and inconsiderable
limits. The redistribution of seats was the subject of a Bill
upon which the last House was dissolved, after the rebuff that
the Ministry received upon a crucial division, but it has been
dealt with rather for practical than theoretical reasons. Two
schemes of redistribution have been formulated, and each has
been proposed and opposed with arguments directed to show the
party advantage to be derived. For political reform, in the
abstract, with or without an extension of the suffrage, there
is no sort of enthusiasm in any quarter. Railway
administration furnishes, no doubt, an occasional battle-field
for the two sides of the House. Roughly, the Progressives
favour the northern extension, and are willing to make
concessions in rates and charges to help on the new trade with
Rhodesia; whilst the Bond declare themselves against special
treatment of the new interests, and would spend all the money
that could be devoted to railway construction in the farming
districts of the colony itself. Mr. Rhodes, however, has
warned the Cape that any hostile action will be counteracted
by a diversion of traffic to the East, and it is unlikely that
any line of policy will be pursued that is likely to injure
the carrying trade of the southern ports. Between the
followers of Mr. Rhodes and the followers of Mr. Hofmeyr there
is no wide divergence of principle on public affairs of the
near future, so far as they have been or are to be the subject
of legislation; where the difference comes in is in the
attitude they severally assume towards the two republics and
the territories of the north, but when talk has to yield to
action it is improbable that there will be much in their
disagreement."
N. L. W. Lawson,
Cape Politics and Colonial Policy
(Fortnightly Review, November 1898).
{480}
SOUTH AFRICA: The Transvaal: A. D. 1898 (January-February).
Re-election of President Kruger.

Renewed conflict of the Executive with the Judiciary.
Dismissal of Chief-Justice Kotze.
The Presidential election in the South African Republic was

held in January and February, the polls being open from the 3d
of the former month until the 4th of the latter. President
Kruger was re-elected for a fourth term of five years, by
nearly 13,000 votes against less than 6,000 divided between
Mr. Schalk Burger and General Joubert, who were opposing
candidates. Soon afterwards, the conflict of 1897 between the
Judiciary and the Executive (see above: A. D. 1897,
JANUARY-MARCH), was reopened by a communication in which
Chief-Justice Kotze, of the High Court, called the attention
of the President to the fact that nothing had been done in
fulfilment of the agreement that the independence of the Court
and the stability of the Grondwet should both be protected by
law against arbitrary interference, and giving notice that he
considered the compromise then arranged to be ended. Thereupon
(February 16) President Kruger removed the judge from his
office and placed the State Attorney in his seat. Justice
Kotze denied the legality of the removal, and adjourned his
court sine die. In a speech at Johannesburg, some weeks
afterwards, he denounced the action of President Kruger with
great severity, saying: "I charge the President, as head of
the State, with having violated both the constitution and the
ordinary laws of the land; with having interfered with the
independence of the High Court; and invaded and imperilled the
rights and liberties of everyone in the country. The
guarantees provided by the constitution for the protection of
real and personal rights have disappeared, and these are now
dependent on the 'arbitrium' of President Kruger."
SOUTH AFRICA: Rhodesia and the British South Africa Company:

A. D. 1898 (February).
Reorganization.
In February, the British government announced the adoption of

plans for a reorganization of the British South Africa Company
and of the administration of its territories. The Company,
already deprived of military powers, was to give up, in great
part, but not wholly, its political functions. It was still to
appoint an Administrator for Rhodesia south of the Zambesi,
and to name the majority of members in a council assisting
him, so long as it remained responsible for the expenses of
administration; but, by the side of the Administrator was to
be placed a Resident Commissioner, appointed by the Crown, and
over both was the authority of the High Commissioner for South
Africa, to whom the Resident Commissioner made reports. At
home the status of the Board of Directors was to be
considerably altered. The life directorships were to be
abolished, and the whole Board of Directors in future to be
elected by the shareholders,—any official or director removed
by the Secretary of State not being eligible without his
consent. The Board of Directors was to communicate all
minutes, etc., to the Secretary of State, and he to have the
power of veto or suspension. Finally, the Secretary of State
was to have full powers to inspect and examine all documents;
Colonial Office officials named by him were, in effect, to
exercise powers like those of the old Indian Board of Control.
SOUTH AFRICA: Cape Colony: A. D. 1898 (March-October).
Election in favor of the Afrikander Bund.

Change in the government.
Elections to the Upper House of the Cape Parliament, in March,
gave the party called the Progressives, headed by Mr. Rhodes, a
small majority over the Afrikander Bund—more commonly called
the Bond. The Parliament opened in May, and the Progressive
Ministry, under Sir Gordon Sprigg, was defeated in the Lower
House in the following month, on a bill to create new
electoral divisions. The Ministry dissolved Parliament and
appealed to the constituencies, with the result of a defeat on
that appeal. The Bond party won in the elections by a majority
of two, which barely enabled it to carry a resolution of want
of confidence in the government when Parliament was
reassembled, in October. The Ministry of Sir Gordon Sprigg
resigned, and a new one was formed with Mr. Schreiner at its
head.
SOUTH AFRICA: The Transvaal: A. D. 1898-1899.

Continued dispute with the British Government
concerning Suzerainty.
During 1898 and half of 1899, a new dispute, raised by Mr.

Chamberlain's emphatic assertion of the suzerainty of Great
Britain over the South African Republic, went on between the
British Colonial Office and the government at Pretoria.
Essentially, the question at issue seemed to lie between a
word and a fact and the difference between the disputants was
the difference between the meanings they had severally drawn
from the omission of the word "suzerainty" from the London
Convention of 1884. On one side could be quoted the report
which the Transvaal deputation to London, in 1884, had made to
their Volksraad, when they brought the treaty back, and
recommended that it be approved. The treaty, they reported,
"is entirely bilateral [meaning that there were two sides in
the making of it] whereby your representatives were not placed
in the humiliating position of merely having to accept from a
Suzerain Government a one-sided document as rule and
regulation, but whereby they were recognized as a free
contracting party. It makes, then, also an end of the British
suzerainty, and, with the official recognition of her name,
also restores her full self-government to the South African
Republic, excepting one single limitation regarding the
conclusion of treaties with foreign powers (Article 4). With
the suzerainty the various provisions and limitations of the
Pretoria Convention which Her Majesty's Government as suzerain
had retained have also, of course, lapsed."
On the other side, Mr. Chamberlain could quote with effect

from a speech which Lord Derby, then the British Colonial
Secretary, who negotiated the Convention of 1884 with the Boer
envoys, made on the 17th of March, that year, in the House of
Lords. As reported in Hansard, Lord Derby had then dealt with
the very question of suzerainty, as involved in the new
convention, and had set forth his own understanding of the
effect of the latter in the following words: "Then the noble
Earl (Earl Cadogan) said that the object of the Convention had
been to abolish the suzerainty of the British Crown. The word
'suzerainty' is a very vague word, and I do not think it is
capable of any precise legal definition.
{481}
Whatever we may understand by it, I think it is not very easy
to define. But I apprehend, whether you call it a
protectorate, or a suzerainty, or the recognition of England
as a paramount Power, the fact is that a certain controlling
power is retained when the State which exercises this
suzerainty has a right to veto any negotiations into which the
dependent State may enter with foreign Powers. Whatever
suzerainty meant in the Convention of Pretoria, the condition
of things which it implied still remains; although the word is
not actually employed, we have kept the substance. We have
abstained from using the word because it was not capable of
legal definition, and because it seemed to be a word which was
likely to lead to misconception and misunderstanding."
Great Britain,
Papers by Command: C. 9507, 1899, pages 24 and 34.
SOUTH AFRICA: The Transvaal: A. D. 1899 (March).
Petition of British subjects to the Queen.
A fresh excitement of discontent in the Rand, due especially

to the shooting of an Englishman by a Boer policeman, whom the
Boer authorities seemed disposed to punish lightly or not at
all, led to the preparation of a petition to the British
Queen, from her subjects in the South African Republic,
purporting to be signed in the first instance by 21,684, and
finally by 23,000. The genuineness of many of the signatures
was disputed by the Boers, but strenuously affirmed by those
who conducted the circulation of the petition. It set forth
the grievances of the memorialists at length, and prayed Her
Majesty to cause them to be investigated, and to direct her
representative in South Africa to take measures for securing
from the South African Republic a recognition of their rights.
The petition was forwarded to the Colonial Office on the 28th
of March.
Great Britain, Papers by Command: 1899, C. 9345.
SOUTH AFRICA: The Transvaal: A. D. 1899 (May-June).

The Bloemfontein Conference between President Kruger and
the British High Commissioner, Sir Alfred Milner.
There seems to be no mode in which the questions at issue

between the British and the Boers, and the attitude of the two
parties, respectively, in their contention with each other,
can be represented more accurately than by quoting essential
parts of the official report of a formal conference between
President Kruger and the British High Commissioner in South
Africa, Sir Alfred Milner, which was held at Bloemfontein, the
capital of the Orange Free State, during five days, May
31-June 5, 1899. The meeting was arranged by President Steyn,
of the Orange Free State, with a view to bringing about an
adjustment of differences by a free and full discussion of
them, face to face. In the official report of the
conversations that occurred, from which we shall quote, the
remarks of President Kruger are given as being made by the
"President," and those of the High Commissioner as by "His
Excellency." The latter, invited by the President to speak
first, said:
"There are a considerable number of open questions between Her

Majesty's Government and the Government of the South African
Republic on which there is at present no sign of agreement. On
the contrary, disagreements seem to increase as time goes on.
… In my personal opinion the cause of many of the points of
difference, and the most serious ones, arises out of the
policy pursued by the Government of the South African Republic
towards the Uitlander population of that Republic among whom
many thousands are British subjects. This policy, the bitter
feeling it engenders between the Government and a section of
Uitlanders, and the effect of the resulting tension in South
Africa, and the feeling of sympathy in Great Britain, and even
throughout the British Empire generally, with the Uitlander
population, creates an irritated state of public opinion on
both sides, which renders it much more difficult for the two
Governments to settle their differences amicably. It is my
strong conviction that if the Government of the South African
Republic could now, before things get worse, of its own motion
change its policy towards the Uitlanders, and take measures
calculated to content the reasonable people among them, who,
after all, are a great majority, such a course would not only
strengthen the independence of the Republic but it would make
such a better state of feeling all round that it would become
far easier to settle outstanding questions between the two
Governments. … The President, in coming here, has made a
reservation as to the independence of the Republic. I cannot
see that it is in any way impairing the independence of the
Republic for Her Majesty's Government to support the cause of
the Uitlanders as far as it is reasonable. A vast number of
them are British subjects. If we had an equal number of
British subjects and equally large interests in any part of
the world, even in a country which was not under any
conventional obligations to Her Majesty's Government we should
be bound to make representations to the Government in the
interests of Her Majesty's subjects, and to point out that the
intense discontent of those subjects stood in the way of the
cordial relations which we desire to exist between us. I know
that the citizens of the South African Republic are intensely
jealous of British interference in their internal affairs.
What I want to impress upon the President is that if the
Government of the South African Republic of its own accord,
from its own sense of policy and justice, would afford a more
liberal treatment to the Uitlander population, this would not
increase British interference, but enormously diminish it. If
the Uitlanders were in a position to help themselves they
would not always be appealing to us under the Convention. …
"President.—I shall be brief. I have come with my commission,

in the trust that Your Excellency is a man capable of
conviction, to go into all points of difference. … I should
like His Excellency to go point by point in this discussion,
so that we can discuss each point that he thinks requires
attention, not with a view to at once coming to a decision,
but to hear each side, and we can go back on any point if
necessary, and see if we can arrive at an understanding. I
would like to give concessions as far as is possible and
practicable, but I want to speak openly, so that His
Excellency may be able to understand. I should like to say
that the memorials placed before Her Majesty's Government came
from those who do not speak the truth. I mean to convey that
we do give concessions wherever we think it practicable to do
so, and after we have discussed it in a friendly way Your
Excellency will be able to judge whether I or the memorialists
are right. I have said that if there are any mistakes on our
side, we are willing to discuss them. Even in any matter
concerning internal affairs I would be willing to listen to
his advice if he said it could be removed in this way or that
way. But when I show him that by the point we may be
discussing our independence may be touched, I trust he will be
open to conviction on that subject. …
{482}
"His Excellency.—I think the point which it would be best to

take first, if the President agrees, … would be the Franchise.
… There are a number of questions more or less resting upon
that. … I should like to know a little more about the
President's views. I want to know more because if I were to
begin and say I want this, that, and the other, I know I
should be told this was dictation. I do not want to formulate
a scheme of my own, but I can, if necessary.
"President.—As long as I understand that it is meant in a

friendly manner, and you mean to give hints, I won't take it
that they are commands. It has already been arranged that you
give me friendly hints and advice, and I will not take it as
dictation, even though it should be on points on which I
should consider you have no right to interfere. … I would like
you to bear one point in view, namely, that all kinds of
nations and languages, of nearly all powers, have rushed in at
the point where the gold is to be found. In other countries …
there are millions of old burghers, and the few that come in
cannot out-vote the old burghers, but with us, those who
rushed in to the gold fields are in large numbers and of all
kinds, and the number of old burghers is still insignificant;
therefore we are compelled to make the franchise so that they
cannot all rush into it at once, and as soon as we can assure
ourselves by a gradual increase of our burghers that we can
safely do it, our plan was to reduce the time for anyone there
to take up the franchise, and that is also my plan. … As His
Excellency doubtless knows, I have proposed to the Volksraad
that the time should be reduced by five years, and gradually
as more trusted burghers join our numbers, we can, perhaps, go
further. There are a number who really do not want the
franchise, but they use it as pretext to egg on people with
Her Majesty. … You must remember, also, on this subject, that
the burghers in our Republic are our soldiers, who must
protect the land, and that we have told these men to come and
fight when we have had difficulties with the Kaffirs. They
wanted the vote, but they would not come and fight. Those who
were willing to help obtained the franchise, but it appears
that many do not want to have it.
"His Excellency.—They did not want to take the obligations

without the rights of citizenship, and in that I sympathize
with them. If they should obtain that right, then naturally
they would have to take those burdens upon them.
"President.—Those who want the franchise should bear the

burdens.
"His Excellency.—Yes. Immediately they get the franchise they

take upon themselves the obligations connected therewith."
[From this the talk wandered to the subject of commandeering,

until the High Commissioner brought it back to the franchise
question.]
"His Excellency.—If I made a proposal to admit strangers under

such conditions as to swamp the old burghers it would be
unreasonable. But the newcomers have, at present, no influence
on the legislation of the Republic, which makes an enormous
difference. They haven't got a single representative. The
First Volksraad consists of 28 members, and not one member
represents the feelings of the large Uitlander population.
"President.—Men from any country could after two years vote

for the Second Volksraad, and after two years more sit in the
Second Raad. There are Englishmen who have obtained the full
franchise in that way, and are eligible for the Volksraad. And
now I have proposed to shorten the last ten years of the
period required for the full franchise and make it five years.
"His Excellency.—There are a great many objections of the

gravest kind to the process by which men may now obtain
burgher rights. First of all, before he can begin the process
of gradually securing burgher rights—which will be completed
in 14 years at present, and in 9 years according to the
President—he has to forswear his own allegiance. Take the case
of a British subject, which interests me most. He takes the
oath, and ceases to be a British subject by the mere fact of
taking that oath; he loses all the rights of a British
subject, and he would still have to wait for 12 years, and
under the new plan 7 years, before he can become a full
citizen of the Republic. British subjects are discouraged by
such a law from attempting to get the franchise. Even if they
wanted to become citizens, they would not give up their
British citizenship on the chance of becoming in 12 years
citizens of the Republic.
"President.—The people are the cause of that themselves. In

1870 anyone being in the land for one year had the full
franchise.
"His Excellency.—That was very liberal.
"President.—In 1881, after the war of independence, some of

our officials and even members of our Raad then said that they
were still British subjects, although they had taken the oath
of allegiance, and I had to pay back, out of the £250,000,
what I had commandeered from them. That was the reason the
oath had to be altered. …
"His Excellency.—In 1882, after all this had happened, there

was a franchise law in the Transvaal, which demanded five
years' residence, but it did not require the oath that is now
taken. It required a simple declaration of allegiance to the
State, though all this that the President refers to happened
before. Why was not it necessary to introduce this alteration
then?
"President.—The people who, before the annexation, had taken

that oath, but had not forsworn their nationality, 1887, sent
a lying memorial, as they are sending lying memorials now, to
say that everybody was satisfied, as they now say that
everybody is dissatisfied.
"His Excellency.—I think I must just explain a little more

clearly my views on the point we are now discussing. … I think
it is unreasonable to ask a man to forswear one citizenship
unless in the very act of giving up one he gets another, and I
think it is also unnecessary to ask him to do more than take
an oath of fealty to the new State, of willingness to obey its
laws and to defend its independence, when it is known and
certain that the taking of that oath deprives him of his
existing citizenship. I think the oath should be a simple oath
of allegiance, and that it should not be required of a man
until the moment he can get full rights in a new State. Now
that was the position under the law of 1882, and all these
reasons which the President has been giving are based on what
happened before that.
{483}
Why were they not considered and acted upon when the law of
1882 was made? … As for the period required to qualify for the
full franchise, I do not see why the length of time should be
longer in the South African Republic than in any other South
African State. They are all new countries. In the new country
which is springing up in the north, and which is getting a new
Constitution this year, the period is one year. The people who
have conquered that country for the white race may find that
the newcomers are more numerous than they are. But I do not
expect that anything like that will be done in the South
African Republic; something far short of that would be
reasonable. What I do think and desire, and that is the object
of my suggestion, is this: that the numerous foreign population
engaged in commerce and industry—to which the country, after
all, owes its present great position in wealth and influence—
should have a real share in the government of the Republic,
not to over-rule the old burghers—not at all—but to share the
work of Government with them, to give them the benefit of
their knowledge and experience, which is in many cases greater
than that of the old burghers, so that through their gradual
co-operation a time may come when, instead of being divided
into two separate communities they will all be burghers of the
same State. It is not enough that a few people should be let
in. It is obvious, however, that you could not let in the
whole crowd, without character or anything—I do not ask
it—but you want such a substantial measure that in elections
of members of the Volksraad the desires of the new industrial
population should have reasonable consideration. They have not
got it now, and when the questions that interest them come
before the Volksraad it is too evident that they are discussed
from an outside point of view. The industrial population are
regarded as strangers. … I do not want to swamp the old
population, but it is perfectly possible to give the new
population an immediate voice in the legislation, and yet to
leave the old burghers in such a position that they cannot
possibly be swamped.
"President.—I hope you will be open to conviction on that

point. I would like to convince you on the subject, and to
show you that it would be virtually to give up the
independence of my burghers. In the Republic the majority of
the enfranchised burghers consider they are the masters. Our
enfranchised burghers are probably about 30,000, and the
newcomers may be from 60,000 to 70,000, and if we give them
the franchise to-morrow we may as well give up the Republic. I
hope you will clearly see that I shall not get it through with
my people. We can still consult about the form of oath, but we
cannot make the time too short, because we would never get it
through with the people—they have had bitter experience. I
hope His Excellency will think about what I have said, and
weigh it well.
"His Excellency.—I see your point, and want to meet it.
"President.—I will think over what has been said, and will try
and meet every difficulty.
At the opening of the Conference on the second day the

President spoke of reports of an increase of British forces in
South Africa, which the High Commissioner assured him were
untrue. The latter in turn referred to accounts that had
appeared of an extensive purchase of arms in the Transvaal;
and was assured by the President that the armament of the
burghers was only for their proper preparation to deal with
the surrounding natives. The President then produced a
memorial purporting to be signed by 21,000 Uitlanders,
contradictory of the representations contained in the memorial
sent to the Queen in March (see above). After discussion upon
this, the conversation returned to the question of the
franchise.
"His Excellency.—What makes this whole discussion so difficult

is the intense prejudice on the side of the present burghers,
and their intense suspicion of us. They think Her Majesty's
Government wants to get their country back in one way or
another. Her Majesty's Government does not; but what it does
desire is that it should have such a state of rest in the
country as will remove causes of friction and difficulty
between the Republic and Her Majesty's possessions in South
Africa, and the whole of the British Empire, and my
suggestions here are directed to that end. I do not want to
say it over and over again, I say it once for all. …
"President.—I should like to make a slight explanation to His

Excellency. His Excellency yesterday mentioned that in some
States those going in from outside speedily got burgher
rights, but he must not forget, as I said before, they are

Download textbook Ai Ia 2016 Advances In Artificial Intelligence Xvth International Conference Of The Italian Association For Artificial Intelligence Genova Italy November 29 December 1 2016 Proceedings 1St Edition Gio ebook all chapter pdf

Uploaded by

Copyright:

Available Formats

Download textbook Ai Ia 2016 Advances In Artificial Intelligence Xvth International Conference Of The Italian Association For Artificial Intelligence Genova Italy November 29 December 1 2016 Proceedings 1St Edition Gio ebook all chapter pdf

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Download textbook Ai Ia 2016 Advances In Artificial Intelligence Xvth International Conference Of The Italian Association For Artificial Intelligence Genova Italy November 29 December 1 2016 Proceedings 1St Edition Gio ebook all chapter pdf

Uploaded by

Copyright:

Available Formats

AI IA 2016 Advances in Artificial

Intelligence XVth International

AI IA 2019 Advances in Artificial Intelligence XVIIIth

AI IA 2013 Advances in Artificial Intelligence XIIIth

AI IA 2017 Advances in Artificial Intelligence XVIth

Artificial Intelligence XXXVI 39th SGAI International

Advances in Artificial Intelligence 18th Conference of

Multi disciplinary Trends in Artificial Intelligence

AI 2019 Advances in Artificial Intelligence 32nd

Agents and Artificial Intelligence 8th International

Subseries of Lecture Notes in Computer Science

LNAI Series Editors

LNAI Founding Series Editor

Marco Gori Marco Maratea (Eds.)

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Library of Congress Control Number: 2016956538

LNCS Sublibrary: SL7 – Artiﬁcial Intelligence

© Springer International Publishing AG 2016

Printed on acid-free paper

This Springer imprint is published by Springer Nature

September 2016 Giovanni Adorni

Doctoral Consortium Chairs

Giuseppe De Giacomo Sapienza Università di Roma, Italy

Marco Villani Università di Modena e Reggio Emilia, Italy

Giovanni Amendola Alessandra De Paola Marco Morana

Optimization and Evolutionary Algorithms

Understanding Characteristics of Evolved Instances for State-of-the-Art

On-line Dynamic Station Redeployments in Bike-Sharing Systems . . . . . . . . 13

Optimized Word-Size Time Series Representation Method Using a Genetic

Efficient Search of Relevant Structures in Complex Systems . . . . . . . . . . . . 35

Classification, Pattern Recognition, and Computer Vision

Flat and Hierarchical Classifiers for Detecting Emotion in Tweets . . . . . . . . . 51

Spam Filtering Using Regularized Neural Networks with Rectified

User Mood Tracking for Opinion Analysis on Twitter . . . . . . . . . . . . . . . . . 76

Using Random Forests for the Estimation of Multiple Users’ Visual

An Analytic Study of Opinion Dynamics in Multi-agent Systems

Combining Avoidance and Imitation to Improve Multi-agent

Knowledge Representation and Reasoning

Educational Concept Maps for Personalized Learning Path Generation. . . . . . 135

Answer Set Enumeration via Assumption Literals . . . . . . . . . . . . . . . . . . . . 149

On the Application of Answer Set Programming to the Conference

A Subdivision Approach to the Solution of Polynomial Constraints

I -DLV: The New Intelligent Grounder of DLV . . . . . . . . . . . . . . . . . . . . . . . 192

Abducing Compliance of Incomplete Event Logs . . . . . . . . . . . . . . . . . . . . 208

Boosting the Development of ASP-Based Applications in Mobile

Relationships and Events: Towards a General Theory of Reification

A Self-Adaptive Context-Aware Group Recommender System . . . . . . . . . . . 250

A Model+Solver Approach to Concept Learning . . . . . . . . . . . . . . . . . . . . . 266

A Comparative Study of Inductive and Transductive Learning

Structural Knowledge Extraction from Mobility Data. . . . . . . . . . . . . . . . . . 294

Predicting Process Behavior in WoMan . . . . . . . . . . . . . . . . . . . . . . . . . . . 308

On-line Learning on Temporal Manifolds. . . . . . . . . . . . . . . . . . . . . . . . . . 321

Learning and Reasoning with Logic Tensor Networks . . . . . . . . . . . . . . . . . 334

Semantic Web and Description Logics

Probabilistic Logical Inference on the Web. . . . . . . . . . . . . . . . . . . . . . . . . 351

Probabilistic Hybrid Knowledge Bases Under the Distribution Semantics . . . . 364

Context-Awareness for Multi-sensor Data Fusion in Smart Environments . . . . 377

Reasoning about Multiple Aspects in Rational Closure for DLs . . . . . . . . . . 392

A Framework for Automatic Population of Ontology-Based

Reasoning About Surprising Scenarios in Description Logics of Typicality. . . 418