SILESIAN UNIVERSITY OF TECHNOLOGY PUBLISHING HOUSE

SCIENTIFIC PAPERS OF SILESIAN UNIVERSITY OF TECHNOLOGY 2023

ORGANIZATION AND MANAGEMENT SERIES NO. 177

ANALYSIS OF EMOTIONS IN IT PROJECTS

IMPLEMENTED IN THE OPEN SOURCE FORMULA
USING MACHINE LEARNING METHODS

Krzysztof S. TARGIEL
University of Economics in Katowice, Faculty of Informatics and Communication, Katowice;
[email protected], ORCID: 0000-0001-7815-1210

Purpose: The aim of this paper is to analyze possibilities of using automatic emotion analysis
in project management.
Design/methodology/approach: The approach adopted involves literature review, analysis of
data availability and availability of IT tools. Then, an attempt was made to adapt these elements
for use in project management.
Findings: The paper discusses three fundamental research questions that arise in the context of
using machine learning methods to analyze emotions in projects. The first of them concerned
what data can be used for analysis. It was established that electronic communication in projects
implemented in the open source formula is publicly available and susceptible to text analysis.
The second question concerned the methods that can be used in the analysis of emotions.
Here it was established that machine learning methods may be useful due to the problems
described in the literature with the use of dictionary methods. The third question concerned the
purposes for which the analysis of emotions can be useful. In response to this question, it was
established that recognizing particularly destructive emotions, such as anger, can be useful in
effective project management.
Research limitations/implications: The presented work is limited only to conceptual
digressions on the possibility and usefulness of using methods of automatic emotion detection
in project management. In future studies, these concepts should be verified on real data.
Originality/value: The novelty of paper is an attempt to define a framework for the use of
known methods of automatic emotion detection in project management.
Keywords: project management; emotion recognition; natural language processing.
Category of the paper: Conceptual paper.

http://dx.doi.org/10.29119/1641-3466.2023.177.37 http://managementpapers.polsl.pl/
646 K.S. Targiel

1. Introduction

Projects in the current economy face a unique set of challenges and opportunities.
Some of the key issues related to project management in the context of the problem considered
in this work are listed in the following paragraphs.
First one is Economic Uncertainty. The global economy is constantly evolving, and factors
such as inflation rates, interest rates, and market volatility can impact project planning and
execution. It's crucial to regularly monitor economic indicators and adjust project plans
accordingly.
Next one is Remote Work and Virtual Collaboration. The COVID-19 pandemic has
accelerated the adoption of remote work and virtual collaboration. Project teams may be
distributed across different locations, requiring the use of digital tools and platforms for
communication, project tracking, and collaboration.
Related with previous one is Technology and Digital Transformation. The rapid pace of
technological advancements offers opportunities for innovation and efficiency improvements.
Project managers should stay updated on emerging technologies relevant to their projects and
explore how they can leverage them to enhance project outcomes.
Most important key parameter is Stakeholder Engagement. Engaging stakeholders,
including customers, employees, investors, and regulators, is crucial for project success.
Clear communication, managing expectations, and addressing stakeholder concerns are
essential to maintain support and alignment throughout the project life cycle.
Taking these factors into account, we may try to use new elements in project management,
such as, for example, sentiment or emotion analysis in electronic communication and to propose
new tools for their use.
Electronic communication, based on emails, are especially susceptible to automatic analysis
based on Natural Language Processing. Many computer methods of natural language
processing (NLP) are currently being developed. They are methods of text and speech
processing (Speech recognition, Word segmentation), Morphological analysis (Lemmatization,
Stemming), Syntactic analysis (Parsing), Lexical semantics (Sentiment analysis with Emotion
recognition, Terminology extraction) and many others like Automatic summarization and
Machine translation. Sentiment analysis seems to be a particularly useful tool for analyzing
communication in a project. Having such tools at our disposal, we can analyze the impact of
external phenomena on IT projects implemented in the open-source format.
The aim of the study is to analyze possibilities of using automatic emotion analysis in
project management. On the one hand, it is an exploration of the models used in psychological
literature to describe human emotions as a complex phenomenon. On the other hand,
it is an examination of the availability of data for later research. On the third hand, it is also
an analysis of available IT tools that can be used in the automatic detection of emotions.
Analysis of emotions in IT projects... 647

The work is divided into the following parts. The first section presents literature survey on
topics of interest, namely open source projects, emotions models, automated emotion
recognition and IT tools for emotions analysis. Next section formulating research questions that
appear in the context of the problem under consideration. The third section contains discussions
of the questions raised and presents some findings. The whole paper ends with a summary.

2. Literature survey

2.1. Open source projects

Nowadays, software development is a very complicated undertaking in which many

specialist are involved. One of the most effective ways to develop software is the open-source
formula. Open-source software (OSS) is computer software that is released under a license in
which the copyright holder grants users the rights to use, study, change, and distribute the
software and its source code to anyone and for any purpose (Laurent, Andrew, 2008).
Some organizations which are follows open-source formula are the Linux Foundation,
the Eclipse Foundation, home of the Eclipse software development platform, the Debian
Project, creators of Debian GNU/Linux distribution; the Mozilla Foundation, home of the
Firefox web browser and finally the Apache Software Foundation.
Mailing lists are the core means of project communication in open source projects,
where they are used during software development and maintenance to discuss technical issues,
propose changes, report bugs, or ask how-to questions about configuration or any other parts of
the product (Obaidi, Klünder, 2021). The idea of using open source project mailing list
communication to analysis comes from Tourani et al., (2014).

2.2. Emotions models

Starting with the work of von Neumann and Morgenstern (1944), decision theory was based
on rationality. Later work by Simon (1955) and Bertalanffy (1968) merely modified the basic
assumption of rational decision-making. In the real world, the decision maker's emotional states
strongly influence the decisions made. Stressful situations they evoke strong emotions that lead
to wrong decisions.
Clarke (2010) was probably the first who notice that projects are emotional. Despite
previous work on the significance of conflict in projects (Chen, 2006), and recognizing that
conflict is a source of strong emotions (Barki, Hartwick, 2004), the topic of recognizing
emotions in the project has not been considered in the literature on the subject. In mentioned
paper Clark (2010) analyze how emotions can influence project manager behaviors and
decisions in order to better understand why projects can often take very different directions to
648 K.S. Targiel

those expected within the predominant rationalist paradigm. He interviewed PMs who took part
in Emotional Intelligence training, so they were aware of the important role of emotions in
decision making.
Virine et al. (2015) analyzed emotions in the context proposed by D. Goleman (Goleman,
2006) and adopted for project management by A. Mercino (Mercino, 2007) model of emotional
intelligence. Proposed model analyse five domains: Self-awareness, Self-management, Social
awareness, Relationship management and Team leadership. Virine et al. notice that not only
negative emotions may cause wrong decisions but also positive ones may lead to mistakes.
They propose some methods to deal with emotions but they first step is emotion recognition.
Human emotions are very complex phenomenon. For this reason, models of this
phenomenon are considered in psychology. A fairly wide overview of the models can be found
in (Nandwani, Verma, 2021). Most widely are used Ekman model (Ekman, 1992) and Plutchik
Wheel of Emotions (Plutchik, 1980). Ekman model is categorical one, with six defined
emotions: anger, disgust, fear, joy, sadness, surprise. Plutchik considered two types of
emotions. Basic ones which include Ekman six emotions supplemented by trust and anticipation
and mixed emotions which are made from the combination of basic emotions. Plutchik
represent his emotions on a colored wheel. Plutchik model is also categorical one. Opposite
type of models are Dimensional Emotion models (Nandwani, Verma, 2021). They are based on
three parameters: Valence (positive, neutral, negative), Activation or Arousal (excited, neutral,
calm) and Dominance or Power (weak, neutral, strong).

2.3. Automated emotion recognition

Recent years have seen a strong development of computer natural language processing
methods. After the first periods of Symbolic NLP (1950s - early 1990s), and Statistical NLP
(1990s-2010s), present NLP methods have huge potential for implementation. Natural
Language Processing (NLP) refer to automated machine-driven algorithms for understanding
of human language and extracting information (Dinov, 2018). Common tasks for these methods
include text and speech processing, morphological analysis, syntactic analysis, lexical
semantics, relational semantics, and discourse (Natural language processing, 2021). Some new
applications includes: automatic summarization, machine translation, natural language
generation. Very interesting directions of NLP development in the context of project
management are the Sentiment Analysis (SA) and Emotion Analysis (EA) which leads to
emotion recognition.
Sentiment Analysis (SA) and Emotion Analysis (EA) are sometimes equated, but there are
important differences between them (Nandwani, Verma, 2021). Sentiment analysis is a means
of assessing emotions are positive, negative, or neutral. In contrast, Emotion detection is
a means of identifying complex human emotion like fear, love etc. As mention Yamini (2023),
Emotion Analysis has also outperformed Sentiment Analysis in some ways. First it reveals
Analysis of emotions in IT projects... 649

complex emotions. Then it provides deeply meaningful and relevant insights and at last it helps
to turn insight into action.
Automated emotion recognition use various methods like electroencephalogram (EEG),
facial, and speech signals, text analysis. Current overview of trends and future perspectives can
be found in paper (Maithri et al., 2022).
By the nature of things, emotion detection from text will be used to analyze emotions in
open source projects where almost all communication takes place via mailing lists.
We are talking about emotion detection here, narrowing down the area of emotion analysis.
Many papers have been written in recent years devoted to this topic. Worth mentioning are
(Shivhareand, Khethawat, 2012) where Word Ontology was used and (Minu, Ezhilarasi, 2012)
which describes an English emotion ontology based on WordNet. Batbaatar have used novel
neural network architecture, called SENN (Semantic-Emotion Neural Network) which can
utilize both semantic/syntactic and emotional information by adopting pre-trained word
representations (Batbaatar et al., 2019). Ho et al. (2020) have used to analyse other than English
language namely Vietnamese. Most recent surveys on emotion detection from text can be found
in paper (Nandwani, Verma, 2021) and also in two papers which present utilization of deep
learning models (Uymaz, Metin, 2022; Chen, 2022).

2.4. Tools for emotion recognition

There are two general ways for dealing with automatic emotion detection: knowledge-based
techniques and statistical methods (Emotion recognition, 2023). Knowledge-based techniques
are referred to as lexicon-based techniques but also contain rule-based systems. Statistical
methods are based on the use of different supervised machine learning algorithms.
Emotion lexicons are dictionaries that associate words or phrases with specific emotions.
Tools that can be used for emotion recognition are:
 NRC Word-Emotion Association Lexicon (Mohammad and Turney, 2010; Mohammad
and Turney, 2013).
 WordNet (Princeton University, 2010).
 EmotiNet (Balahur et al., 2012).
Example of rule based system is ANEW (Affective Norms for English Words) (Bradley,
Lang, 1999).
Statistical methods based on supervised machine learning algorithms use such architecture
as:
 Convolutional neural network (CNN) (Wang et al., 2016).
 Bidirectional Encoder Representations from Transformers (BERT) and Bi-directional
Long Short-Term Memory (BiLSTM) (Chen, 2022).
There are also effective hybrid methods, an overview of which can be found in (Alswaidan,
Menai, 2020).
650 K.S. Targiel

3. Research questions

When starting research in the selected area, one should be aware that the proposed models
will have to be verified based on data from the real world. The question arises about the
availability of such data, hence the first research question was formulated:
RQ1: What data from real projects can research be based?
Another problem that needs to be faced is the availability of methods and tools for detecting
emotions. Their spectrum is quite extensive as discussed in the previous section. However,
another research question arises in the form of:
RQ2: What methods can be used to analyze emotions in project management?
The last question that arises in the context of the conducted considerations is what this
knowledge can be used for in project management. If data for research are available, methods
and tools will be selected, whether the ability to automatically detect emotions can be used in
project management. Hence the next research question arise:
RQ3: For what purposes can emotion recognition be used in a project management?
A discussion of the presented research questions is presented in the next section.

4. Discussion and findings

4.1. Data selection

Due to the specificity of projects in which teams are focused on achieving goals,
it is difficult to expect that during their implementation someone will have time to answer
surveys or undergo EEG tests. In this case, there is also a question about the relevance of the
results of the research conducted in this way. This is especially important in the context of the
use of surveys that need to be developed. We also resign from the use of facial analysis methods.
It seems that at the initial stage of research it will be beneficial to focus on the analysis of texts
and communication carried out in this form.
Communication in the project takes place in defined communication channels and is usually
confidential. Fortunately, in IT projects implemented in the open source formula, the projects
communication is based on mailing lists and by nature are publicly available.
The mailing list of the Apache OpenOffice project, implemented in the open-line formula,
was selected in this work. For this, project communication is publicly available at
"https://openoffice.apache.org/mailing-lists.html". Mailing list is maintained since 2011 till
today. Every month, several hundred messages appear on all sub lists together.
Analysis of emotions in IT projects... 651

OpenOffice is an open-source office suite. It was an open-sourced version of the earlier

StarOffice, which Sun Microsystems acquired in 1999 for internal use. Sun open-sourced the
OpenOffice suite in July 2000 as a competitor to Microsoft Office. In 2011, Oracle Corporation,
then the owner of Sun, announced that it would no longer offer a commercial version of the
suite and donated the project to the Apache Foundation. Apache renamed the software to
Apache OpenOffice. Today the most actively developed successor projects is LibreOffice
(OpenOffice.org, 2022).
Apache OpenOffice is an office productivity suite providing six productivity applications
(Writer, Calc, Impress, Draw, Math, Base) based around the OpenDocument Format (ODF).
OpenOffice is released on Windows, OS X, Linux. It is available in 41 languages.

4.2. Methods selection

It seems that methods based on dictionaries and deep neural networks are particularly
interesting. While the use of dictionary methods can be problematic, due to the problem
mentioned by Tourani (Tourani et al., 2017), involving the use of specific IT slang in open
source projects, the use of deep neural networks may be interesting. Another research problem
arises as to whether these tools will be able to capture emotions from e-mail communication.
It is also a question about the selectivity of these methods in detecting emotions.

4.3. The use of emotion recognition in project management

In the context of the latest views on the success of a project, according to which success is
considered to be meeting the expectations of and stakeholders, their emotions may be important
in the correct assessment of the project. We know that some negative emotions (such as anger)
as well as positive ones (such as love) can affect the way we view the situation and the decisions
we make. Hence, the ability to catch particularly bad emotions can be helpful in effectively
managing stakeholder engagement. As a consequence, it can also have an impact on achieving
project success more effectively.
For the above reasons, it seems justified to focus primarily on the detection of bad emotions,
such as anger.

5. Summary

The presented work deals with the topic of using automatic emotion recognition in projects.
On the one hand, there are effective tools for emotional analysis. On the other hand, there is
a need to manage stakeholder engagement.
652 K.S. Targiel

The available data for the analysis of emotions in projects was reviewed. It turned out that
it would be beneficial to use publicly available communication in open source projects.
It also implied the use of methods of emotional analysis in texts. In the course of the conducted
analyses, effective methods of analyzing emotions from the text were also reviewed.
Finally, an analysis of the possibility of using emotion recognition in project management was
made. The most beneficial seems to be the use of the proposed methods to detect bad emotions
that disturb the delay in project goals.
The presented work presents the concept of using emotion recognition in project
management. The initial considerations carried out led to many further questions that need to
be verified in empirical research. This will be the subject of further research.

References

1. Alswaidan, N., Menai, M.E.B. (2020). A survey of state-of-the-art approaches for emotion
recognition in text. Knowledge and Information Systems, Vol. 62, Iss. 8, pp. 2937-2987.
doi:10.1007/s10115-020-01449-0.
2. Balahur, A., Hermida, J.M., Montoyo, A. (2012). Detecting implicit expressions of emotion
in text: A comparative analysis. Decision Support Systems. Vol. 53, Iss. 4, pp. 742-753.
doi:10.1016/j.dss.2012.05.024.
3. Barki, H., Hartwick, J. (2004). Conceptualising the construct of interpersonal conflict.
International Journal of Conflict Management, Vol. 15, No. 3, pp. 216-44.
4. Batbaatar, E., Li, M., Ryu, K.H. (2019). Semantic-Emotion Neural Network for Emotion
Recognition From Text. IEEE Access, Vol. 7, pp. 111866-111878. doi: 10.1109/
ACCESS.2019.2934529
5. Bertalanffy von, L. (1968). General System Theory: Foundations, Development. New York:
George Braziller.
6. Bradley, M.M., Lang, P.J. (1999). Affective norms for English words (ANEW): Instruction
manual and affective ratings. Technical Report C-1. University of Florida, The Center for
Research in Psychophysiology.
7. Chen, M. (2022). Emotion Analysis Based on Deep Learning With Application to Research
on Development of Western Culture. Frontiers in Psychology, Vol. 13, doi:
10.3389/fpsyg.2022.911686.
8. Chen, M.H. (2006). Understanding the benefits and detriments of conflict on team creativity
process, Creativity and Innovation, Vol. 15, No. 1, pp. 105-16.
9. Clarke, N. (2010). Projects are emotional: How project managers’ emotional awareness
can influence decisions and behaviours in projects, Vol 3, Iss 4, pp. 604-624,
doi:10.1108/17538371011076073.
Analysis of emotions in IT projects... 653

10. Dinov, I.D. (2018). Natural Language Processing/Text Mining. In: I.D. Dinov (ed.), Data
Science and Predictive Analytics: Biomedical and Health Applications Using R (pp. 659-
95). Cham: Springer International Publishing.
11. Ekman, P. (1992). An argument for basic emotions. Cognition and Emotion, Vol. 6,
Iss. 3-4, pp.169-200. doi:10.1080/02699939208411068.
12. Emotion recognition (2023). Wikipedia. Retrieved from: https://en.wikipedia.org/w/
index.php?title=Emotion_recognition&oldid=1153310432, 31.05.2023.
13. Goleman, D. (2006). Emotional Intelligence. Why It Can Matter More Than IQ.
10th Anniversary Edition. New York: Random House Publishing Group.
14. Ho, V.A., Nguyen, D.H.-C., Nguyen, D.H., Pham, L.T.-V., Nguyen, D.-V., Nguyen, K.V.,
Nguyen, N.L.-T. (2020). Emotion Recognition for Vietnamese Social Media Text.
In: L.-M. Nguyen, X.-H. Phan, K. Hasida, S. Tojo (Eds.), Computational Linguistics
(pp. 319-333). Springer. doi: 10.1007/978-981-15-6168-9_27.
15. Laurent, St., Andrew, M. (2008). Understanding Open Source and Free Software Licensing.
O'Reilly Media.
16. Maithri, M., Raghavendra, U., Gudigar, A., Samanth, J., Prabal Datta Barua, Murugappan,
M., Chakole, Y., Acharya, U.R. (2022). Automated emotion recognition: Current trends and
future perspectives. Computer Methods and Programs in Biomedicine, Vol. 215,
pp. 106646. doi: 10.1016/j.cmpb.2022.106646.
17. Mersino, A.C. (2007). Emotional intelligence for project managers: The people skills you
need to achieve outstanding results. New York: American Management Association.
18. Minu, R.I., Ezhilarasi, R. (2012). Automatic Emotion Recognition and Classification.
Procedia Engineering, Vol. 38, doi: 10.1016/j.proeng.2012.06.004.
19. Mohammad, S.M., Turney, P.D. (2010). Emotions Evoked by Common Words and Phrases:
Using Mechanical Turk to Create an Emotion Lexicon. Proceedings of the NAACL-HLT
2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in
Text. Los Angeles, pp. 26-34.
20. Mohammad, S., Turney, P. (2013). Crowdsourcing a Word-Emotion Association Lexicon.
Computational Intelligence, Vol. 29, Iss. 3, pp. 436-465.
21. Nandwani, P., Verma, R. (2021). A review on sentiment analysis and emotion detection
from text. Social Network Analysis and Mining, Vol. 11, Iss. 1, p. 81. doi:10.1007/s13278-
021-00776-6.
22. Natural language processing. Available online https://en.wikipedia.org/wiki/Natural_
language_processing, 31.05.2023.
23. Neumann von, J., Morgenstern, O. (1944). Theory of games and economic behavior.
Princeton: Princeton University Press.
24. Obaidi, M., Klünder, J. (2021) Development and Application of Sentiment Analysis Tools
in Software Engineering: A Systematic Literature Review. Evaluation and Assessment in
Software Engineering, pp. 80-89.
654 K.S. Targiel

25. OpenOffice.org. Retrieved from: https://en.wikipedia.org/wiki/OpenOffice.org,

31.05.2023.
26. Pennington, J., Socher, R., Manning, Ch.D. (2014). GloVe: Global Vectors for Word
Representation.
27. Plutchik, R. (1980). General Psychoevolutionary Theory of Emotion. In: R. Plutchik,
H. Kellerman (Eds.), Theories of Emotion (pp. 3-33). Academic Press. doi:10.1016/B978-
0-12-558701-3.50007-7.
28. Princeton University (2010). About WordNet. Princeton University. Retrieved from:
https://wordnet.princeton.edu/, 31.05.2023.
29. R Development Core Team (2022). R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing. Vienna, Austria, http://www.R-
project.org/.
30. Shivhare, S.N., Khethawat, S. (2012). Emotion Detection from Text (arXiv:1205.4944).
ArXiv. Retrieved from: https://doi.org/10.48550/arXiv.1205.4944, 17.05.2023.
31. Simon, H.A. (1955). A behavioral model of rational choice. Quartely Journal of Economics,
Vol. 69, pp 99-118.
32. Tourani, P., Yiang, Y., Adams, B. (2017). Monitoring sentiment in open source mailing
lists: exploratory study on the apache ecosystem. Proceedings of 24th Annual International
Conference on Computer Science and Software Engineering, CASCON’14, pp. 34-44.
33. Uymaz, A.H., Metin, K.S. (2022). Vector based sentiment and emotion analysis from text:
A survey. Engineering Applications of Artificial Intelligence, Vol. 113, pp. 104922. doi:
10.1016/j.engappai.2022.104922.
34. Virine, L., Trumper, M., Virine, E. (2015). Emotions in Project Management. PM World
Journal, Vol. 4, Iss. 8, pp. 1-8.
35. Wang, Y., Feng, S., Wang, D., Yu, G., Zhang, Y. (2016). Multi-label chinese microblog
emotion classification via convolutional neural network. In: F. Li, K. Shim, K. Zheng,
G. Liu (Eds.), Web technologies and applications: APWeb 2016, vol. 9931. Lecture notes
in computer science (pp. 567-580). Cham: Springer.
36. Yamini (2023). Emotion and Sentiment Analysis: What are the differences? Analytics Steps.
Retrieved from: https://www.analyticssteps.com/blogs/emotion-and-sentiment-analysis-
what-are-differences, 26.05.2023.