=Paper=
{{Paper
|id=Vol-3864/quasoq-2024-paper-02
|storemode=property
|title=A Report on Sentiment Analysis of Requirements Engineering Artifacts created in University Course
|pdfUrl=https://ceur-ws.org/Vol-3864/quasoq-2024-paper-02.pdf
|volume=Vol-3864
|authors=Takumi Katsuie,Shinpei Ogata,Kozo Okano,Yukako Iimura,Shinobu Saito
|dblpUrl=https://dblp.org/rec/conf/apsec/KatsuieOOIS24
}}
==A Report on Sentiment Analysis of Requirements Engineering Artifacts created in University Course==
https://ceur-ws.org/Vol-3864/quasoq-2024-paper-02.pdf
A Report on Sentiment Analysis of Requirements Engineering
Artifacts created in University Course
Takumi Katsuie1 , Shinpei Ogata1 , Kozo Okano1 , Yukako Iimura2 and Shinobu Saito2
1
Shinshu University Faculty of Engineering 4–17–1 Wakasato, Nagano-shi, Nagano, 380–8553 Japan
2
NTT Computer & Data Science Laboratories 3–9–11 Midori-Cho Musashino-shi, Tokyo, 180–8585 Japan
Abstract
This technical report introduces the results of sentiment analysis of artifacts in requirements engineering phase. These artifacts contain
descriptions of requirements and functions for the development target such as software product and solution. The descriptions of
requirements reflect user needs and problems are described based on the analysis of users’ dissatisfaction with the current situation and
their expectations. On the other hand, the description of functions describes the behavior of the system and the interaction between the
system and humans. Therefore, we apply sentiment analysis to requirements artifacts which are created in an exercise for university
students. We, then, investigate how the sentiment of the descriptions in the artifacts are changed. Sentiment analysis is performed
using Google Cloud’s Natural Language API on the descriptions included in the artifacts such as customer journey maps and user story
mappings. From the results of the application, we confirmed that the sentiment score of each artifact was different.
Keywords
Sentiment Analysis, Requirements Engineering Artifacts, Design Thinking
1. Introduction On the other hand, there are no reports of sentiment anal-
ysis on artifacts in requirements engineering phase. In the
Sentiment Analysis is a method for measuring and under- requirements engineering phase, the problem awareness,
standing the feelings of individuals from text data such as needs and expectations of stakeholders (users, operators,
reviews on the web, blog posts and SNS posts, and is used etc.) are analysed and the functions and performance that
in various situations such as understanding customer prod- satisfy these needs are defined. Goal models have tradi-
uct satisfaction and checking employee stress. Sentiment tionally been used to analyse problems and the consistency
Analysis determines whether an opinion is positive, nega- between problems and solutions. In initiatives that integrate
tive or neutral from text data including phrases, words and design thinking and requirements engineering, personas,
expressions contained in sentences. customer journey maps, etc., are created [6]. In these ar-
Sentiment analysis is also widely used in various research tifacts, it is recommended to describe the image of stake-
fields in software engineering. In the field of software repos- holders (users, operators, etc.) and realistic images of the
itory mining, efforts have been reported to apply sentiment system’s usage scenario. Therefore, it is conceivable that the
analysis to textual data extracted from developers’ discus- emotional tendencies measured in the deliverables created
sions (e.g. ticket comments, commit messages) in order during the requirements engineering phase may differ from
to predict defects in source code and interruptions in OSS the emotional tendencies of the SRS described above.
projects [1] . In addition, efforts to predict support ticket In this paper, we analyse the tendency of measured emo-
escalation by performing sentiment analysis on support tick- tions in the artifacts created in the requirements engineering
ets that represent issues raised by customers and combining phase (refer to Figure 1). At this time, the analysis approach
it with machine learning has been reported [2] [3]. In the that has been used for a long time is called the classical ap-
field of requirements engineering, efforts to acquire require- proach, while the analysis used in design thinking is called
ments by applying sentiment analysis to ratings and review the modern approach.
comments on products have also been reported [4]. In this paper, we set the following Research Question
The main data handled in software engineering can be (RQ).
roughly classified into two categories: data obtained from RQ: Are the emotional expression measured from
the development and operation process and data obtained texts in the requirements engineering artifacts created
from the development artifacts (product). In addition to the using classical and modern approaches neutral?
application of sentiment analysis in software engineering In order to answer the above-mentioned research ques-
to the development and operational process data mentioned tion, we analyse and evaluate the artifacts created based on
above, there are also efforts targeting development artifact two approaches (classical and modern) as university exer-
data. For example, in the paper [5], they took the Software cises task of the requirements engineering phase in software
Requirements Specification (SRS), which is one of the final development.
products of the requirements definition process, and applied The composition of this paper is as follows: Section 2
sentiment analysis to the text data obtained from the SRS, describes the content of the artifacts to be analysed; Sec-
and found that They report that almost all sentences in the tion 3 describes the analysis methods and results; Section 4
SRS (about 96%) were neutral. considers the results of the analysis; and finally Section 5
provides a summary.
QuASoQ 2024: 12th International Workshop on Quantitative Approaches
to Software Quality, December 03, 2024, Chongqing China
$ 24w6020j@shinshu-u.ac.jp (T. Katsuie); ogata@cs.shinshu-u.ac.jp
(S. Ogata); okano@cs.shinshu-u.ac.jp (K. Okano);
yukako.iimura@ntt.com (Y. Iimura); shinobu.saito@ntt.com (S. Saito)
0000-0001-6996-3073 (S. Ogata); 0009-0006-9865-8362 (K. Okano);
0009-0006-3030-3442 (Y. Iimura); 0000-0002-6259-3521 (S. Saito)
© 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License
Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
12
Requirements definition Design
Software Requirements
Modern approach Specification
Persona
Customer Journey User Story Mapping
Map (CJM) (USM)
Service scenario Screen prototype
Users’ opinions
and requests Screen transition
diagram
Business flow
Goal model
Classical approach
Artifacts Creation flow Analysis target
Figure 1: Requirements engineering artifacts and creation clow
2. Artifacts to be analysed Table 1
List of artifacts
2.1. How to create the target data (how to process artifact approach analysis
proceed with the exercise) name target
Users’ opinions - ✓
In this paper, we target several artifacts created by the stu- and requests
dents in an exercise of a lecture on the upstream process Persona modern ✓
of software development (part of the requirements defini- requirements Service scenario modern ✓
tion and external design process) at a university (refer to definition Customer Jour- modern ✓
Table 1). The number of students taking the lecture was 54, ney Map (CJM)
and more than 90% of them were third-year undergraduate User Story Map- modern ✓
students in science and engineering. The students have al- ping (USM)
ready taken lectures on programming and modelling (UML, Goal model classical ✓
etc.) and have basic knowledge of software development. Business flow classical -
external Screen - -
In the exercises, after the teacher explained the contents of
design prototype
the artifacts, each student independently created all nine
artifacts in the order shown in Figure 1.In the first stage
of the requirements definition process, they assume users’
these 6 artifacts created in requirements definition process.
opinions and requests for the ideas of services provided by
the teacher, and describe them in writing. In the subsequent
exercises in the requirements definition process, they create 2.2. Contents of target data (6 artifacts)
artifacts based on the Classical Approach (2 types) and the
2.2.1. Users’ opinions and requests
Modern Approach (4 types). The creation of artifacts by sev-
eral people and third-party reviews of the created artifacts Users’ opinions and requests are created in order to verbalise
are not carried out. Therefore, a series of artifacts for 54 their opinions and requests for services. In this artifact, 2
students were created. In advance, we obtained permission opinions or requests such as ‘This kind of service would
to use the artifacts for this study from the students who be useful’ or ‘This kind of service is disappointing’ are de-
produced the analysed artifacts. scribed for each of the three services listed below.
We targeted artifacts that contained a certain amount of
natural language descriptions for sentiment analysis. Specif- • A service wanted to enrich your learning (lessons,
ically, there are 6 artifacts in total: users’ opinions and re- research, etc.) at university
quests, persona requirement, service scenario, customer • A service wanted for self-development during long
journey map and user story mapping, which are the arti- holidays (summer holidays, etc.)
facts created using the modern approach, and goal model, • A Service wanted to enjoy daily life (housework,
which is the artifact created using the classical approach. entertainment, meals, etc.)
We exclude the business flow, which is a typical artifact cre-
ated using classical approach from the sentiment analysis. 2.2.2. Persona
This is because the business flow also include natural lan-
guage descriptions in the labels of activities and flows, but A persona is a fictional character that represents a typical
the amount of it is small. Similarly, we excluded the screen user of the product or service to be developed. Details of
prototypes created in the external design process from the the character, such as its specific profile and requirements,
sentiment analysis. Similarly, we excluded the screen proto- are set. A Persona are used in the persona scenario method
types created during the external design process from the to devise and design services and systems that satisfy the
analysis. In the following we will explain the content of defined persona, as well as for the characters in the artifacts
13
to be created later. Setting a persona helps developers to We show an example of USM in Figure 3. A USM consists
develop user-centered services centered on the persona, of 5 elements: ‘Persona Problem’, ‘Service Value’, ‘Activity
rather than on the developer’s self-indulgent services. Overview’, ‘Narrative Flow’ and ‘User Stories’. In the ‘Per-
First, one service is selected from the services considered sona Problem’, describe the persona’s problem obtained
in ‘Users’ opinions and requests’. Then, a persona is deter- from CJM, and in the ‘Service Value’, describe how the ser-
mined, assuming the person who must be satisfied with the vice defined with the persona scenario method solves the
service. The persona is then made detailed by adding not persona’s problem. In the ‘Activity Overview’, describe the
only the basic profile (name, age, height and weight), but implementation overview of the service provided, and in
also the place of work, place of residence and hobbies and the ‘Narrative Flow’, describe the stories of the persona
preferences. After the detailed information of the persona using the services provided with reference to the CJM, in
is determined, what the persona wants for the selected ser- chronological order. In the ‘User Stories’, the user stories
vice (persona requirements) is described in 350 characters required to experience the elements of the ‘Narrative Flow’
or more. In this paper, we only analyse the description of are arranged in such a way that the essential services with
persona requirements among the persona. the highest priority are at the top, and the optional services
with the lowest priority as you move down. The user story
2.2.3. Service scenario is a requirement for the realisation of a service. The service
is composed of a set of user stories. It does not describe
The description of service scenarios is one of the processes about the system, but the requirements and goals of the
in the persona scenario method, and is created to assume persona to use the service. It is written as ‘The user wants
how the main persona will use the service in his/her daily to ∼ (so that ∼)’.
life. Specifically, it is described in a scenario format with 6
or more steps, when, how, in which situations, and what
2.2.6. Goal model
operations are performed by the main persona to realize the
service. A goal model is a representation in a tree structure of the
persona’s goals and the ways to achieve them in the system
2.2.4. Customer Journey Map (CJM) to be developed. Creating this can help developer organize
the requirements regarding the system so they can avoid
A customer journey map (hereafter CJM) is a visual rep- creating gaps between the user’s requirements and the sys-
resentation of the predicted actions and emotions that a tem.
pre-defined persona will take until using a service or prod- We show an example of a goal model in Figure 4. A
uct, arranged in chronological order. This is created to goal model is a tree structure, in which the higher goals
vividly imagine the user experience after determining the are the objectives of the lower goals. The top goal of the
target user profile and the key process to focus on when tree structure is the desired situation when the problem
considering the service. Creating this can help developer of the persona defined in the USM is solved. The top goal
visualize how persona feel so they can avoid potential issues is then decomposed and detailed to create subgoals. The
ahead of time, increase persona retention, and discover key subgoals are decomposed and detailed in the same way, and
information to make the best decisions for development. this process is repeated to finally derive the functional and
We show an example of a CJM in Figure 2. A CJM consists non-functional requirements.
of 6 elements: ‘Persona Requirements’, ‘Specific Scenes’,
‘Scenes Name’, ‘Persona Actions’, ‘Persona Emotions’ and
‘Insights (from persona’s actions and emotions)’. In the 3. Analysis Methods and Results
‘Specific Scenes’, write a concrete sentence that enables
the reader to imagine the situation in which the persona’s 3.1. Sentence extraction and Sentiment
requirements are generated. Then, in the ‘Scenes Name’, Analysis methods
describe the specific scene in chronological order by dividing
it into four or five scenes. In the ‘Persona Actions’, describe We extracted the only texts described by the students from
the actions the persona is likely to take in each scene, and in the 6 artifacts described in chapter 2, except for the elements
the ‘Persona Feelings’, describe the feelings and thoughts of names. Then, we split the extracted texts with symbols
the persona in each scene, including positive and negative such as punctuation marks, periods, exclamation points,
ones, in text form. Then, organise the actions and feelings and question marks, as well as with spaces and line breaks.
and describe in the ‘Insights’ why they act that way, why We obtained 2639 texts from all artifacts. We show the
they feel that way, the solutions, etc. number of extracted texts for each artifact type in Table 2.
We performed a sentiment analysis on these texts.
2.2.5. User Story Mapping (USM) In this paper, we use Google Cloud’s Natural Language
API [7] for sentiment analysis of text. Natural Language
A user story mapping (hereafter USM) is a visual representa- API is a service by Google Inc. that provides natural lan-
tion of the values (functions) that a service wishes to realize guage processing techniques such as sentiment analysis,
in chronological order and in order of priority, based on the entity analysis, entity sentiment analysis, content classifica-
actions of personas. After the persona and CJM have been tion, and syntactic analysis using machine learning, and is
created and the image of the service has been established, available for free for a certain number of times. In sentiment
a USM is created to concrete the image of the service. Cre- analysis, given a text, we obtain a score, which indicates the
ating this can help the entire development team organize polarity of the overall sentiment of the text, and a magni-
persona behavior and the value the service will bring so tude, which indicates the intensity of the sentiment, based
they can understand the value of the service, and determine on word meanings, etc. score indicates the emotion of the
development priorities. text and has values from -1.0 (negative) to 1.0 (positive).
14
Persona I wants to look good when I turn the camera ON, even in a first period non-face-to-face class on a very busy day in the
Requirements morning.
A day when I overslept and woke up 30 minutes before the start of class. It happens to be a day with a full morning of
Specific
classes, so I'd like to have a light breakfast. But I also want to put on some makeup in case the camera is turned on, and
Scenes
I don't want to be slammed into the computer right before first period.
Scenes Name Immediately after waking up Assess the situation once Hurry up and get dressed
As soon as I wake up, I look at the Check what day it is today. Do my make-up in a hurry.
clock as usual. Remind the class schedule. Change clothes, even if only the top
Persona
Seeing the time on the clock, I instantly half of clothing, in case I have to turn
Actions
wake up and jump out of bed. on the camera due to the content of the
class.
No way. Why do I oversleep only today! What shall I do! I want to eat breakfast. I would like to have a little time for
My tension has dropped. But I don't have time. breakfast.
Persona I want to change my clothes, at least I'll get dressed and do my makeup, but
Emotions my upper body, because sometimes it's in the house, can I manage that?
the camera will be on. What about hair and makeup?
Oh no, I don't have time! What shall I wear?
She wakes up and immediately can't By counting backwards in time, she If she has messy hair, she won't be
Insights
grasp if she overslept more than usual. might panic and falter. able to get her hair set in time for class.
She may become impatient by being Under what situations would she be
from
surprised and her heart beating very unsure of which clothes to wear?
persona's
fast. She doesn't want people to think she
actions and
She may waste time by worrying about always wears the same clothes.
emotions
what to do.
Figure 2: The example of a Customer Journey Map
She has no time before class because she oversleeps and gets so impatient. So, she feels that her computer starts up too
Persona Problem
slowly.
Talk to it like a smart speaker and it will automatically start your PC even when you are away from it.
Service Value
Being able to start up the PC quickly, so you can calmly participate in class even if you don't have much time before class.
Activity Overview Prepare to use the service. Automatically start up the PC earlier. Manage service usage records.
Narrative Flow Register own information. Attend morning classes calmly. Able to track recent morning activity.
User wants to register his/her User wants his/her PC to start User wants to check the history of
1 information with the service so that 1 automatically at a set time 1 automatic startup of his/her PC in a
he/she can use the service. certain period of time in the past.
User wants to register his/her User wants to know from a remote In a certain period of time in the
phone information with the service location that his/her PC has started past, user wants to check whether
1 so that he/she can set the time 2 up without any problems. 2 or not his/her PC has actually
from his/her phone. attended an online class after
automatic startup.
User Stories
User wants to register a time with User wants to start up his/her PC at User wants to be informed of days
the service when his/her PC will a time other than a set time, even when automatic PC startup is not
1 2 3
automatically start up in the from a remote location. required, based on past PC startup
morning. times and class attendance.
User wants to register a mascot
with the service so that he/she
3
wants his/her PC to be started up
by his/her favorite mascot.
Figure 3: The example of a User Story Mapping
Spend time relaxing before
first-period classes begin.
Get time before a Get time before a
non-face-to-face class starts. face-to-face class starts.
Improve efficiency of
Improve efficiency of Improve efficiency of Improve efficiency of
equipment (computers) for
getting dressed getting dressed movement
non-face-to-face classes
Turn on the computer Connect to the campus
Set the startup time
( connect it to the network) portal
Figure 4: The example of a Goal model
15
magnitude indicates how much emotional content a text which means that texts with emotional expressions are more
contains, and has values from 0.0 to +inf. magnitude is frequent.
not normalized unlike the score, so the magnitude value
of a text increases each time emotions are expressed in the 3.2.2. Analysis of the range of emotions
text. In this paper, we use the score obtained from the senti-
ment analysis of each sentence, and analyze them in units We analyzed the range of emotions in artifacts by artifact
of artifact and artifact type. We show an example of texts type. First, texts with score greater than 0 were defined as
extracted and split from artifacts, and the score obtained by positive, and texts with other score were defined as negative.
sentiment analysis on the texts in the Table 3. Then, we calculated the maximum value from the positive
score and the minimum value from the negative score for
each artifact. Also, we calculated the average of the maxi-
Table 2
mum positive score and of the minimum negative score by
Number of sentences extracted by artifact type
artifact type. We show the result of this analysis in Figure
Aritifact name Number of extracted text 6. As shown in Figure 6, the range of emotions is larger
Users’ opinions and requests 237 for CJM and USM created using the modern approach, and
Persona requirements 205 smaller for the service scenario and the goal model created
Service scenario 242 using the classical approach.
Customer Journey Map (CJM) 1008
User Story Mapping (USM) 633
Goal model 314 4. Discussion
total 2639
The result of the analysis of the percentage of texts without
emotional expressions in the session 3.2.1 showed that emo-
tions were measured in about 30 % or more of the texts for
Table 3
all types of artifacts. In particular, artifacts created using
Example of sentiment analysis
modern approaches such as persona requirements and CJM
artifact texts extracted and split score were found to have emotional expressions in more than half
name of the texts on average.
Customer Immediately after waking up 0 Therefore, the answer to the Research Question RQ: Are
Journey Map the emotional expression measured from texts in the
Customer My tension has dropped. -0.7 requirements engineering artifacts created using clas-
Journey Map sical and modern approaches neutral? is that the emo-
User Story User wants to register his/her in- -0.2
tional expression measured from almost all texts in the arti-
Mapping formation with the service so that
fact created using the both approaches is not only neutral,
he/she can use the service
Goal model Spend time relaxing before first 0.6 but also negative and positive. Also, the artifacts created us-
period classes begin. ing the modern approach except for service scenarios tend
to have a higher percentage of texts with emotional expres-
sions than artifacts created using the classical approach.
This is different from the tendency, reported in the paper
3.2. Score analysis methods and results [5], of emotional expression measured from texts in the
We analyzed the scores obtained by sentiment analysis for SRS. We believe that the modern approach mainly requires
the texts by artifact type in terms of two aspects: the ratio to describe the persona’s expectations and dissatisfaction,
of texts without emotional expression and the range of the so that the sentences are more likely to have emotional
emotions. expressions in artifacts created using modern aproach. For
the service scenario, the functional descriptions such as the
operations performed by the persona to realize the service
3.2.1. Analysis of the percentage of texts without
and the accompanying system behavior are mainly required,
emotional expression
so the percentage of texts without emotional expressions
We analyze the percentage of texts without emotional ex- may have increased compared to the artifacts created by the
pression by artifact type. First, texts with absolute scores of other modern approaches.
0.25 or less were considered neutral (neutral texts without On the other hand, the classical approach mainly re-
emotional expression), and texts with other scores were con- quires to describe the functional and non-functional re-
sidered emotional (texts with emotional expression). Then, quirements of the system. In the goal model, functional
we calculated the percentage of texts without emotional and non-functional requirements for the system are derived
expression and the percentage of texts with emotional ex- by detailing the goals from the higher-level goals to the
pression by artifact type for all 30 students. We show the lower-level goals. In the detailing process, the top goal de-
result of this analysis in Figure 5. As shown in Figure 5, the scribed the requirements for the persona, such as the desired
percentage of neutral tends to be higher in artifacts created situation when the persona’s problem is solved, so it is as-
later in the process, such as in CJM and USM, than in arti- sumed that emotions were measured from the sentences of
facts created earlier in the process. However, the percentage the some high-level goals.
of texts with emotional expressions in the artifacts created Now that we have confirmed that texts in artifacts in
using both classical and modern approaches is more than 30 requirements engineering phase often contain emotional
percent. Especially in artifacts such as users’ opinions and expressions, we will discuss the results of the section 3.2.2
requests, persona requirements, and CJM, the percentage of analysis of the range of emotions. The result of this analysis
texts with emotional expressions is more than 50 percent, confirmed that the range of emotions in CJM is particularly
16
Percentage of neutral (by artifact type)
100%
80%
60%
40%
20%
0%
User Story Mapping
Customer Journey
requirements
Goal model
Service scenario
Users' opinions and
Persona
requests
Map
neutral emotional
Figure 5: The percentage of texts without emotional expression by artifact type
Range of emotions (Average of positive and negative score)
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
Users' opinions
requirements
Goal model
scenario
Journey Map
User Story
Service
Mapping
and requests
Customer
Persona
negative positive
Figure 6: Average of maximum positive score and average of minimum negative score by artifact type
large. This suggests that many artifacts of the CJM tend to expressions or a small range of sentiment, we believe that
contain both strongly positive and strongly negative texts. the artifact may not have successfully acquired or extracted
This is because the CJM include a direct verbal description stakeholders’ demands. Therefore, we believe that perform-
of what the persona is feeling, such as I’m happy!,” “Good,” ing sentiment analysis on artifacts can be used to measure
“My tension is low” etc., in the “Persona Emotions” item, the degree to which artifacts are acquiring demands. Thus,
and thus it is easier to measure strong emotions from such sentiment analysis of artifacts will facilitate the extraction
descriptions, and we believe that we were able to measure of descriptions of stakeholder sentiments and functions, and
strong emotions from many of the CJMs. Thus, not only will measure the success of artifacts in extracting and ob-
the appearance frequency of text with emotional expres- taining stakeholder requirements, thereby supporting the
sions but also the range of emotions that emerge differs efficiency of system development, and so on.
depending on the type of artifact, and in particular, artifacts
that directly describe emotions and artifacts that describe
dissatisfaction and expectations are likely to have a large 5. Summary
range of emotions.
In this paper, we reported the results of sentiment analysis
From these results, we confirmed that artifacts in require-
on artifacts in requirements engineering phase of software
ments engineering phase often contain texts with emotional
development, which were created using two approaches,
expressions, and that some types of artifacts tend to contain
classical and modern. Specifically, we conducted sentiment
strong emotions. We believe that sentiment analysis of the
analysis using Google Cloud’s Natural Language API on
texts in artifacts and extraction of texts with large score will
the descriptions in six artifacts, such as customer journey
facilitate understanding of the stakeholders’ dissatisfaction
map and goal model, and analyzed emotion scores obtained
and expectations, and the scenes in which these feelings are
by artifact type. The results showed that the percentage
held. On the other hand, We believe that by extracting neu-
of text with emotional expressions in all types of artifacts
tral (texts without emotilnal expressions), it will be possible
created using the two approaches was more than 30 per-
to extract descriptions of functional and non-functional re-
cent, and especially in the persona requests and customer
quirements for the system from the artifacts. In addition, if
journey maps created using the modern approach, the per-
the results of sentiment analysis of an artifact (e.g., CJM),
centage of text with emotional expressions was more than
which should reflect stakeholders’ expectations and dissat-
50 percent. From this, as an answer to the research question,
isfaction, show a small percentage of text with emotional
“Are the emotional expression measured from texts in the
17
requirements engineering artifacts created using classical Broy: “On Integrating Design Thinking for Human-
and modern approaches neutral?”, it was confirmed that Centered Requirements Engineering,” IEEE Software,
emotional expressions measured from texts in artifacts cre- vol. 37, no. 2, pp. 25-31, March-April 2020.
ated using both approaches were not only neutral, but also [7] https://cloud.google.com/natural-language?hl=ja
negative and positive. In addtion, it was confirmed that (2024/9/23 referred)
artifacts created using modern approach tended to have a
higher percentage of texts with emotional expressions than
artifacts created using classical approach. This may be due
to the fact that the modern approach is more likely than
the classical approach to describe requirements that per-
sona has. It was also confirmed that the range of emotions
differed depending on the type of artifact. This is because
the required descriptions differ depending on the artifact,
and the range of emotion is considered to be larger for arti-
facts that directly describe emotions and those that describe
dissatisfaction and expectations.
we believe that sentiment analysis of artifacts can be
used to measure the degree to which artifacts are acquiring
demands. Thus, sentiment analysis of artifacts will facilitate
the extraction of descriptions of stakeholder sentiments
and functions, and will measure the success of artifacts in
extracting and obtaining stakeholder requirements, thereby
supporting the efficiency of system development, and so on.
In this report, we analyzed the percentage of texts without
emotional expressions and the range of emotions in each
artifact type, but in the future we would like to conduct
more detailed analysis of the characteristics of emotions
in artifacts by analyzing artifact units and analyzing other
factors besides the range of emotions. We would also like
to investigate the relationship between the emotion of the
artifact and the quality of that artifact and the quality of
the artifacts (e.g., screen prototypes) that are created behind
the process. In addition, we would like to confirm whether
similar trends can be obtained using other artifact sets.
References
[1] P. Tourani and B. Adams: “The Impact of Human Dis-
cussions on Just-in-Time Quality Assurance: An Em-
pirical Study on OpenStack and Eclipse,” 2016 IEEE
23rd International Conference on Software Analysis,
Evolution, and Re-engineering (SANER), Osaka, Japan,
2016, pp. 189-200.
[2] C. Werner, G. Tapuc, L. Montgomery, D. Sharma, S. Do-
dos and D. Damian: “How Angry are Your Customers?
Sentiment Analysis of Support Tickets that Escalate,”
2018 1st International Workshop on Affective Comput-
ing for Requirements Engineering (AffectRE), Banff,
AB, Canada, 2018, pp. 1-8.
[3] C. Werner, Z. S. Li and D. Damian: “Can a Machine
Learn Through Customer Sentiment?: A Cost-Aware
Approach to Predict Support Ticket Escalations,” IEEE
Software, vol. 36, no. 5, pp. 38-45, Sept-Oct 2019.
[4] E. Guzman and W. Maalej: “How Do Users Like This
Feature? A Fine Grained Sentiment Analysis of App
Reviews,” 2014 IEEE 22nd International Requirements
Engineering Conference (RE), Karlskrona, Sweden,
2014, pp. 153-162.
[5] C. Werner, Z. S. Li and N. Ernst: “What Can the Senti-
ment of a Software Requirements Specification Docu-
ment Tell Us?” 2019 IEEE 27th International Require-
ments Engineering Conference Workshops (REW),
Jeju, Korea (South), 2019, pp. 106-107.
[6] J. Hehn, D. Mendez, F. Uebernickel, W. Brenner and M.
18