bookBC PDF

Descărcați ca pdf sau txt
Descărcați ca pdf sau txt
Sunteți pe pagina 1din 10

Departamentul de Informatica

Universitatea din Craiova, Romania

Mihaela Colhon

Baze de cunostinte

Note de curs

2014
Cuprins

1 Aspecte de reprezentare a cunostintelor 11


1.1 Tipuri de cunostiite . . . . . . . . . . . . . . . . . . . . . . . . 11
1.1.1 Metode pentru reprezentarea cunostintelor . . . . . . . 13
1.1.2 Cunostinte relationale simple . . . . . . . . . . . . . . 15
1.1.3 Cunostinte care permit mostenirea . . . . . . . . . . . 15
1.1.4 Cunostinte inferentiale . . . . . . . . . . . . . . . . . . 16
1.1.5 Cunostinte procedurale . . . . . . . . . . . . . . . . . . 16
1.2 Sisteme de Reprezentare si Procesare . . . . . . . . . . . . . . 17
1.2.1 Sisteme bazate pe cunostinte . . . . . . . . . . . . . . . 18
1.2.2 Motor de inferenta . . . . . . . . . . . . . . . . . . . . 20
1.2.3 Reprezentarea si procesarea cunostintelor n Prolog . . 21
1.2.4 Sisteme Expert . . . . . . . . . . . . . . . . . . . . . . 24

2 Reprezentarea cunostintelor bazata pe cadre 29


2.1 Baze de cunostinte cu frameuri . . . . . . . . . . . . . . . . . 30
2.2 Inferenta n sistemele cu frameuri . . . . . . . . . . . . . . . . 34
2.2.1 Sisteme cu frameuri. O singura mostenire . . . . . . . 34
2.2.2 Sisteme cu frameuri. Mosteniri multiple . . . . . . . . 34
2.2.3 Functia de raspuns . . . . . . . . . . . . . . . . . . . . 35

3 Sisteme Lindenmayer 41
3.1 Sisteme de rescriere . . . . . . . . . . . . . . . . . . . . . . . . 41
3.1.1 Grafica Turtle . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.2 Rescrierea arcelor si rescrierea nodurilor . . . . . . . . 46
3.1.3 Sisteme Lindenmayer cu paranteze . . . . . . . . . . . 50
3.2 Trasatura de autosimilaritate. Fractali . . . . . . . . . . . . . 51
3.3 Curbe FASS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.4 Implementarea n Prolog a sistemelor
DOL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

3
4 CUPRINS

4 Reprezentarea si procesarea cunostintelor n sisteme de


productie 59
4.1 Reguli if-then . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.2 Motorul de inferenta . . . . . . . . . . . . . . . . . . . . . . . 60
4.2.1 Formalizarea rationamentului nainte . . . . . . . . . . 62
4.2.2 Formalizarea rationamentului napoi . . . . . . . . . . 64
4.2.3 Strategii de control . . . . . . . . . . . . . . . . . . . . 65
4.3 Arbori de decizie . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.3.1 Interpretarea fotografiilor spatiale . . . . . . . . . . . . 66
4.4 Modulul de generare a explicatiilor . . . . . . . . . . . . . . . 69
4.5 Incertitudine n sistemele de reguli . . . . . . . . . . . . . . . 69
4.5.1 Factori de certitudine. Intarirea ncrederii . . . . . . . 72
4.6 Sisteme expert bazate pe reguli . . . . . . . . . . . . . . . . . 72

5 Multimi Rough 75
5.1 Indiscernabilitate . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2 Apartenenta nuantata . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Aplicatii ale multimilor rough . . . . . . . . . . . . . . . . . . 79

6 Reprezentarea cunostintelor n structuri de tip graf 83


6.1 Grafuri conceptuale . . . . . . . . . . . . . . . . . . . . . . . . 83
6.2 Retelele semantice . . . . . . . . . . . . . . . . . . . . . . . . 85
6.2.1 WordNet . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.2.1.1 Relatii semantice n WordNet . . . . . . . . . 94
6.3 Grafuri stratificate etichetate . . . . . . . . . . . . . . . . . . 96
6.3.1 Grafuri startificate etichetate n sintaza de imagini . . 99
6.4 Scheme semantice . . . . . . . . . . . . . . . . . . . . . . . . . 101

7 Reprezentarea cunostintelor n Procesarea Limbajului Natu-


ral 105
7.1 Clase de gramatici si limbaje . . . . . . . . . . . . . . . . . . . 108
7.1.1 Reprezentarea gramaticilor n Prolog . . . . . . . . . . 110
7.2 Reprezentari bazate pe structuri de tip graf . . . . . . . . . . 114
7.2.1 Retele de tranzitie . . . . . . . . . . . . . . . . . . . . 115
7.2.2 Traducatoare finite . . . . . . . . . . . . . . . . . . . . 118
7.3 Analiza sintactica bazata pe constituenti . . . . . . . . . . . . 121
7.4 Analiza sintactica cu relatii de dependenta . . . . . . . . . . . 126
7.5 Dezambiguizarea automata a sensului cuvintelor . . . . . . . . 135
7.5.1 Dezambiguizarea bazata pe cunostinte . . . . . . . . . 138
Cuvant nainte

Prezentarile din aceasta carte au la baza studiile pe care le-am realizat


n cei 15 ani care s-au scurs decand am devenit cadru didactic al Facultatii
de Matematica si Informatica (actualmente, Facultatea de Matematica si
Stiinte ale Naturii) si pana acum. Activitatea mea de cercetare s-a axat pe
dezvoltarea de noi mecansime de reprezentarea si procesare a cunostintelor.
De asemenea, domeniul n care mi-am sustinut teza de doctorat cat si tema de
cercetare abordata se ncadreaza n acest domeniul de cercetare, titlul tezei
mele de doctorat fiind Metode de tip graf pentru reprezentarea si procesarea
cunostintelor. Formalisme si reprezentari. In ultimii 5 ani accentul s-a pus
analiza de texte n limbaj natural.
Tot de 15 ani predau Baze de cunostinte stundetilor: pana n 2011 am
predat seminariile acestei discipline (titular de curs fiind Prof. dr. Nicolae
Tandareanu) iar din 2011 predau si cursul. Tematica actuala abordata la
curs include prezentarea detaliata a celor mai importante sisteme de repre-
zentare a cunostintelor: reprezentarea si procesarea cunostintelor n sisteme
cu reguli de productie, reprezentarea si procesarea cunostintelor structurate
(sisteme cu cadre, retele semantice), sisteme de rationament incert, repre-
zentarea cunostintelor n procesarea limbajului natural. In cadrul cursurilor
ce trateaza reprezentarea prin retele semantice, am inclus doua formalisme
derivate din aceste reprezentari: Grafuri Statificate Etichetate si Scheme
Semantice dar si prezentarea unui exemplu de retea semantica extrem de
folosita n aplicatiile inteligentei artificiale n general, cu precadere n sub-
domeniul procesarii limbajului natural. Este vorba de reteaua semantica
WordNet, datorita careia procesarea cunostintelor a dobandit noi dimensiuni.
Dincolo de aspectele teoretice ale conceptelor enumerate anterior, n timpul
seminariilor am urmarit cu precadere tratarea aplicatiilor practice, cautand
sa perfectionez abilitatile de programare ale studentilor pentru realizarea si
utilizarea eficienta a sistemelor de rationament, propunandu-le spre analiza
implementari variate n diverse limbaje de programare cum ar fi Prolog, Lisp
sau Java.
Aceasta carte sintetizeaza activitatea mea de cercetare n domeniul repre-
zentarii si procesarii cunostintelor precum si materialele didactice si aplicatiile

5
6 CUPRINS

dezbatute cu studentii n timpul orelor de curs si seminar/laborator.


Deoarece aceasta carte se adreseaza n primul rand studentilor interesati
de cursul de Baze de cunostinte, fundamentele teoretice si detaliile de forma-
lizare si implementare ale modelelor de reprezentare pe care le-am inclus n
carte si care sunt rodul activitatilor mele de cercetare sunt tratate pe scurt.
Pentru cititorul interesat listez mai jos lucrarile n cauza care mi-au servit la
redactarea acestei carti.

Reprezentari de cunostinte n limbaj natural


Badica, C, Colhon, M., Sendre, A. Sentiment Analysis Of Tourist Reviews:
Data Preparation And Preliminary Results In Mihaela Colhon, Adrian Iftene,
Verginica Barbu Mititelu, Dan Cristea, Dan Tufis (eds.) Proceedings of the
10th International Conference Linguistic Resources And Tools For Proces-
sing The Romanian Language, pp. 135-142 (2014)
Colhon, M., Sendre, A., Badica, C. Relating the Opinion Holder and the
Review Accuracy in Sentiment Analysis of Tourist Reviews, Proceedings of
The 7th International Conference on Knowledge Science, Engineering and
Management, KSEM 2014 (2014)
Bibiri, A., Colhon, M., Diac, P., Cristea, D. Statistics Over A Corpus Of
Semantic Links: QuoVadis. In Mihaela Colhon, Adrian Iftene, Verginica
Barbu Mititelu, Dan Cristea, Dan Tufis (eds.) Proceedings of the 10th In-
ternational Conference Linguistic Resources And Tools For Processing The
Romanian Language, pp. 33-44 (2014)
Cristea, D., Gfu, D., Colhon, M., Diac, P., Bibiri, A.-D., Maranduc, Scu-
telnicu, L.-A. Quo Vadis: A Corpus of Entities and Relations. In N.Gala,
R.Rapp and G.B.Enguix (eds.): Language Production, Cognition, and the
Lexicon, Springer International Publishing Switzerland (2014)
Colhon, M., Diac, P., Maranduc, Perez, A. QuoVadis Research Areas - Text
Analysis. In Mihaela Colhon, Adrian Iftene, Verginica Barbu Mititelu, Dan
Cristea, Dan Tufis (eds.) Proceedings of the 10th International Conference
Linguistic Resources And Tools For Processing The Romanian Language,
pp. 45-56 (2014)
Danciulescu, D., Colhon, M. Splitting the structured paths in stratified gra-
phs. Application in Natural Language Generation, An. St. Univ. Ovidius
Constanta, 22(2) pp. 59-69 (2014)
Colhon, M. Automatic Lexical Alignment Between Syntactically Weak Rela-
ted Languages. Application for English and Romanian, Proceedings of ICCCI
2013, vol. 8083, Lecture Notes in Computer Science series (2013)
Hristea, F., Colhon, M. Feeding Syntactic Versus Semantic Knowledge to
a Knowledge-lean Unsupervised Word Sense Disambiguation Algorithm with
an Underlying Naive Bayes Model, Fundamenta Informaticae Journal, ISSN:
CUPRINS 7

0169-2968 vol. 119(1), pp. 61-86, DOI 10.3233/FI-2012-727 (2012)


Colhon, M. Acquiring Syntactic Translation rules from a Parallel Treebank,
Journal of Information and Library Science INFOtheca, vol. XIII, pp. 19-32,
Serbian Academic Library Association, ISSN: 1450-9687 (2012)
Colhon, M., Simionescu, R. Deriving a Statistical Syntactic Parsing from a
Treebank, Proceedings of International Conference on Web Intelligence, Mi-
ning and Semantics WIMS 12, ISBN: 978-1-4503-0915-8 (2012)
Colhon, M. Language Engineering for Syntactic Knowledge Transfer, Com-
puter Science and Information Systems Journal (ComSIS), 9(3), ISSN 1820-
0214, pp. 1231-1247 (2012)
Colhon, M. Syntactic Translation Patterns from a Parallel Treebank, Procee-
dings of the First Workshop on Computational Linguistics of Balkan Langu-
ages (CLoBL 2012), The 5th Balkan Conference in Informatics - BCI 2012,
ISBN 978-86-7031-200-5, pp. 85-88 (2012)
Colhon, M. An Architecture Model for English-Romanian Noun Phrase Alig-
nments, In JKCCT (Journal of Knowledge, Communications and Computing
Technologies), ISSN: 2067-095, 3(1), pp. 23-34 (2011)
Colhon, M. XML Semantic Schema Annotation for Dependency Relation-
ships, Annals of the University of Craiova, Mathematics and Computer
Science Series, ISSN: 1223-6934, vol. 37(3), pp. 64-70 (2010)
Colhon, M., Hristea, F. Syntactic Relations in the Context of Performing
Word Sense Disambiguation, Journal of Knowledge, Communications and
Computing Technologies (JKCCT), ISSN: 2067-0958, II(2), pp. 1-9 (2010)
Colhon, M. A Contrastive Study of Syntactic Constituents in English and
Romanian Texts, In Proceedings of the Workshop Language Resources and
Tools with Industrial Applications, Adrian Iftene, Diana-Maria Trandabat
(eds.), pag. 11-20, ISSN 1843-911X (2011)
Colhon, M. Parallel Treebank From Word-Aligned Bilingual Corpus. Langu-
age Engineering for Phrasal Alignments, In Proceedings of 15th International
Conference on System Theory, Control and Computing, Mihail Voicu (ed.),
pag. 172-177, ISSN 2068-0465, (2011)
Colhon, M., Danciulescu, D. Semantic Schemas for Natural Language Gene-
ration in Multilingual Systems, Journal of Knowledge, Communications and
Computing Technologies (JKCCT), ISSN: 2067-0958, II(1), pp. 10-17 (2010)
Colhon, M., Tandareanu, N. A Semantic Schema - based Approach for Na-
tural Language Translation, WSEAS Journal Transactions on Computers,
ISSN: 1109-2750, 9(11), pp. 1307 - 1317 (2010)
Colhon, M., Tandareanu, N. An Approach for Contextual Translation based
on Semantic Schemas, Latest Trends on Computers, vol. I, ISSN: 1792-4251,
ISBN: 978-960-474-201-1, pp. 294-299 (2010)
Ghindeanu, M., Poprlan, C. A Natural Language Processing System using
8 CUPRINS

Java-Prolog Technology, Proceedings of The 5th International Conference on


Artificial Intelligence and Digital Communications - AIDC 2005, vol. 105,
pp. 63-68, ISBN 973-671-055-6 (2005)

Reprezentari de cunostinte cu frameuri


Ghindeanu, M., Poprlan, C. A Natural Language Processing System using
Java-Prolog Technology, Proceedings of The 5th International Conference on
Artificial Intelligence and Digital Communications - AIDC 2005, vol. 105,
pp. 63-68, ISBN 973-671-055-6 (2005)

Reprezentari de cunostinte conditionale:


Tandareanu, N., Colhon, M., Zamfir, C. Embedding Conditional Knowledge
Bases into Question Answering Systems and Java Implementation, Interna-
tional Journal of Computers, 4(4), ISSN: 1998-4308, pp. 169-176 (2010)
Tandareanu, N., Colhon, M., Zamfir, C. A Spoken Question Answering Sys-
tem Based on Conditional Knowledge Latest Trends on Computers, vol. I,
ISSN: 1792-4251, pp. 220-225 (2010)
Tandareanu, N., Colhon, M., Zamfir, C. XML Schema Language Specifica-
tions for Conditional Knowledge, Advances in Communications, Computers,
Systems, Circuits and Devices, European Conference of Computer Science
(ECCS 10), pp. 149-154, ISSN: 1792-6637 (2010)
Colhon, M., Tandareanu, N. The Inference Mechanism in Conditional Sche-
mas, Annals of the University of Craiova, Mathematics and Computer Science
Series, 37(1), ISSN: 1223-6934, pp. 55-70 (2010)
Tandareanu, N., Ghindeanu, M. Conditional graphs generated by conditio-
nal schemas, Annals of University of Craiova, Mathematics and Computer
Science Series, ISSN: 1223-6934, 36(1), pp. 1-11 (2009)
Tandareanu, N., Ghindeanu, M. Towards a Mathematical Modelling of Con-
ditional Knowledge, Research Notes in Artificial Intelligence and Digital Co-
mmunications, vol.103, ISBN 973-8419-71-9, Craiova, p.5-15 (2003)

Reprezentarea cunostintelor incerte


Iancu, I., Constantinescu, N., Colhon M., Fingerprints Identification using a
Fuzzy Logic System, International Journal of Computers Communications &
Control, ISSN 1841- 9836, vol.5(4), pp. 525-531 (2010)
Iancu, I., Colhon, M., Dupac M., A Fuzzy Controller with Various T-norms
Applied in Robot Navigation, WSEAS Tranaactions on Systems and Control,
ISSN: 1991-8763, 5(11), pp. 882-891 (2010)
Iancu, I., Colhon, M., Dupac, M., A Takagi-Sugeno Type Controller for Mo-
bile Robot Navigation, Proceedings of WSEAS International Conf. on Com-
putational Intelligence, pp. 29-34 (2010)
CUPRINS 9

Iancu, I., Colhon, M., Mamdani FLC with various implications, Proceedings
of 11th International Symposium on Symbolic and Numeric Algorithms for
Scientific Computing - SYNASC 2009, published by IEEE Computer Soci-
etys Conference Publishing Service (CSP)
Ghindeanu, M., Fuzzy Controller for Spatial Objects Recognition, Proceedings
of the 6th International Conference on Artificial Intelligence and Digital Co-
mmunications - AIDC 2006, vol. 106, pp. 118-125 (2006)

Grafuri stratificate etichetate


Danciulescu, D., Colhon, M. Splitting the structured paths in stratified gra-
phs. Application in Natural Language Generation, An. St. Univ. Ovidius
Constanta, 22(2) pp. 59-69 (2014) Ghindeanu, M.Fuzzy Controller for Spa-
tial Objects Recognition, Research Notes in Artificial Intelligence and Digital
Commnunications, vol. 106,ISBN 973-742-413-1 978-973-742-413-6, pp.118-
125 (2006)
Tandareanu, N., Ghindeanu, M.Hierarchical Reasoning Based on Stratified
Graphs. Application in Image Synthesis, Proceedings of 15th International
Workshop on Database and Expert Systems Applications, Proceedings of
DEXA2004, IEEE Computer Society, Zaragoza, Spania, Los Alamitos Cali-
fornia, p.498-502 (2004)
Tandareanu, N., Ghindeanu, M. Image Synthesis from Natural Language
Description, Research Notes in Artificial Intelligence and Digital Communi-
cations, vol.103,ISBN 973-8419-71-9, Craiova, p.82-96 (2003)

Scheme semantice
Colhon, M. XML Semantic Schema Annotation for Dependency Relation-
ships, Annals of the University of Craiova, Mathematics and Computer
Science Series, ISSN: 1223-6934, vol. 37(3), pp. 64-70 (2010)
Colhon, M., Danciulescu, D. Semantic Schemas for Natural Language Gene-
ration in Multilingual Systems, Journal of Knowledge, Communications and
Computing Technologies (JKCCT), ISSN: 2067-0958, II(1), pp. 10-17 (2010)
Colhon, M., Tandareanu, N. A Semantic Schema - based Approach for Na-
tural Language Translation, WSEAS Journal Transactions on Computers,
ISSN: 1109-2750, 9(11), pp. 1307 - 1317 (2010)
Colhon, M., Tandareanu, N. An Approach for Contextual Translation based
on Semantic Schemas, Latest Trends on Computers, vol. I, ISSN: 1792-4251,
ISBN: 978-960-474-201-1, pp. 294-299 (2010)
Colhon, M. Knowledge Transfer and Path-based Reasoning in Hyper-schemas,
Journal of Knowledge, Communications and Computing Technologies, ISSN:
2067-0958, I(1)/2009, pp. 47-58 (2009)
Colhon, M. Hierarchical Distributed Systems Based on Semantic Schemas,
10 CUPRINS

Annals of University of Craiova, Mathematics and Computer Science Series,


ISSN: 1223-6934, 36(2), pp. 47-52 (2009)
Tandareanu, N., Ghindeanu, M., Nicolescu, S.-A. Hierarchical Distributed
Reasoning System for Geometric Image Generation International Journal of
Computers Communicationa & Control, ISSN 1841- 9836, 4(2), pp. 167-177
(2009)
Tandareanu, N., Ghindeanu, M., Path-based Reasoning in Semantic Sche-
mas, Annals of University Of Craiova, Mathematics and Computer Science
Series, 2008
Ghindeanu, M., A Spatial Reasoning HDR System, Proceedings of the 8th
International Conference on Artificial Intelligence and Digital Communica-
tions - AIDC 2008, vol. 108, pp. 88-102, ISBN 978-973-671-161-9 (2008)
M. Ghindeanu, Constructing Architectures for an Hierarchical Distributed
Reasoning System Based on its Inputs, Proceedings of the 5th Internatio-
nal Conference on Cybernetics and Information Technologies, Systems and
Applications (CITSA 2008), Orlando, Florida, USA , 2008
Tandareanu, N., Ghindeanu, M. Hierarchical Distributed Reasoning System
for Geometric Image Generation, International Journal of Computers, Com-
munication and Control, 2008
Tandareanu, N., Ghindeanu, M. Hierarchical Semantic Structures Applied in
Automatic Image Generation, Proceedings of the 11th IASTED Internatio-
nal Conference on Intelligent Systems and Control (ISC 2008), 2008
Ghindeanu, M. The Knowledge Domain of an Hierarchical Distributed Sys-
tem Determines its Architecture, Annals of University Of Craiova, Mathema-
tics and Computer Science Series, vol. 34, pag. 107-114, ISSN: 1223-6934,
2007
Ghindeanu, M. Joining Semantic Schemas in Vision of a Distributed System
Reasoning, Research Notes in Artificial Intelligence and Digital Commnuni-
cations, vol. 107, ISBN 978-973-671-135-0, Craiova, pag. 123-132, 2007
Tandareanu, N., Ghindeanu, M. Properties of Derivations in a Semantic
Schema, Annals of University of Craiova, Mathematics and Computer Science
Series, ISSN 1223-6934, vol.33, pp. 147-155 (2006)
Tandareanu, N., Ghindeanu, M. A Three-Level Distributed Knowledge Sys-
tem Based on Semantic Schemas, Proceedings of 16th International Work-
shop on Database and Expert Systems Applications, Proceedings of DEXA05
(TAKMA2005), IEEE Computer Society, Copenhagen, p.423-427, 2005.

S-ar putea să vă placă și