A User-centric System for Improving Human-Computer Interaction

through Fuzzy Logic-based Assistive Messages

Christos Troussas

, Akrivi Krouska

and Cleo Sgouropoulou

Department of Informatics and Computer Engineering, University of West Attica, Egaleo, Greece

Keywords: Assistive Messages, Computer Knowledge Level, COVID-19, e-Learning, Feedback, Fuzzy Logic,

Human-Computer Interaction, Personalized System, Recommender System, Rule-based System.

Abstract: The fast growth of the internet and communication technology in recent years has resulted in rendering

computers easily accessible to everyone. However, people have different knowledge and characteristics that

can affect their ability to use computers and at the same time create barriers to achieve an effective user

experience. The reason for this is to provide dynamic adaptability to users' individual needs. In view of this

compelling need, this paper presents a user-centric system that seeks mainly to improve the interaction of

users with the software they use. To achieve this, the system employs fuzzy logic to model the computer

knowledge of users and based on this classification, it delivers assistive messages, which are pertinent to the

interaction with the system. These messages are tailored to the user groups that have been created, as well as

the degree of detail which is more adequate for each group. As a testbed for our research, the presented

approach has been incorporated in a learning management system to support tutors towards having a better

experience while interacting with this software. The system has been evaluated by users during the COVID-

19 lockdown with promising results.

1 INTRODUCTION

The proliferation of the internet and communication

technology (ICT) in recent years has brought

significant changes in many parts of people's daily

lives. These changes rendered the technology an

indispensable tool to better organize their time,

communicate with others, learn, make purchases,

make bookings for places of entertainment or means

of travel, etc. As such, different software or web sites

are being used by many people around the world, who

are different in terms of their needs, preferences and

computer skills (Krouska et al., 2020 (a)). Thus,

adapting these systems to a diverse audience is a

significant notion that can improve human-computer

interaction (HCI) (Papakostas et al., 2021). In order

to provide a personalized interaction route for each

different user, current approaches involve improving

and modifying sotware content, feedback, novigation

support, etc. (Troussas et al., 2021).

https://orcid.org/0000-0002-9604-2015

https://orcid.org/0000-0002-8620-5255

https://orcid.org/0000-0001-8173-2622

Adaptive software has great interactivity

affordances, because it delivers a user-centered

experience increasing user engagement. The delivery

of assistive messages and feedback to users is an

integral component of it (Wang et al., 2019). For

more qualitative interaction, assistive messages can

be facilitated by sophisticated techniques towards

delivering appropriate assistance to users, when

needed (Zadeh Kashani and Hamidzadeh, 2020). The

more adaptive the messages are to the users’ needs,

the more effective and accurate they are. For instance,

when interacting with a software, if users has a high

level of computer skills and knowledge, it is useless

or even tiring for them to receive a huge volume of

messages; in this sense, their experience may be

unpleasant.

Based on reviews of the relevant scientific

literature (Bhanuse and Mal, 2021; Suhaim and Berri,

2021), it appears that the research on the delivery of

assistive messages to users is still in its early stages.

Several academics have investigated the delivery of

Troussas, C., Krouska, A. and Sgouropoulou, C.

A User-centric System for Improving Human-Computer Interaction through Fuzzy Logic-based Assistive Messages.

DOI: 10.5220/0010702800003058

In Proceedings of the 17th International Conference on Web Information Systems and Technologies (WEBIST 2021), pages 365-370

ISBN: 978-989-758-536-4; ISSN: 2184-3252

 2021 by SCITEPRESS – Science and Technology Publications, Lda. All rights reser ved

365

assistive messages in various environments,

including:

 e-learning (Soulef et al., 2021; Tahir et al.,

2021; Krouska et al., 2020 (b); Wan and Niu,

2020; Qomariyah and Fajar, 2019);

 e-commerce (Peng et al., 2020; Zhao, 2019;

Chauhan et al., 2019; Ouaftouh et al., 2019);

 e-travel and smart tourism sites (Fararni et al.,

2021; Chaudhari and Thakkar, 2020;

Hassannia et al., 2019);

 e-health systems (Ferretto et al., 2020; Tellería

et al., 2020; Gräßer et al., 2016);

 online movies systems (Singla et al., 2020;

Deldjoo et al., 2019; Troussas et al., 2018);

 search engines (Peña et al., 2020; Ahmedi and

Shabani, 2017).

In the aforementioned research efforts, the authors

employed, among others, semantic approaches,

collaborative filtering, content-based filtering,

genetic algorithms and machine learning to deliver

adaptive assistive messages to users.

In view of the above, this article presents a user-

centric system that aims to boost users’ engagement

with the software they use. Hence, the system

employs fuzzy logic to represent users' computer

expertise and, based on this classification, sends out

helpful messages that are relevant to their

engagement with the system. These messages are

customized to the user groups that have been formed,

as well as the level of detail that each group requires.

The offered method has been implemented into a

learning management system as a testbed for our

study in order to assist instructors in having a better

experience when engaging with this software.

The remainder of this paper is organized as

follows. Section 2 presents the fuzzy weights, being

used to model users’ computer knowledge. In Section

3, the rules for the delivery of assistive messages are

described. Section 4 presents the experimental results

and discussion. Finally, Section 5 continues with the

work's conclusions and recommendations for further

research activities.

2 FUZZY WEIGHTS FOR

MODELING COMPUTER

KNOWLEDGE

The improvement of HCI and the delivery of a

pleasant user experience involves mostly the

https://www.tests.com/practice/computer-skills-practice-

test

consideration of the users’ computer knowledge (Nie

et al., 2021; Khodr et al., 2020; Méndez et al., 2019;

Wynn and Hult, 2019). Nonetheless, determining the

users’ computer knowledge level is a difficult task,

riddled with ambiguity. For example, a user with a

score of 92/100 in a computer proficiency test cannot

be classified as exceptional or very good. Both states

include some truth. Fuzzy logic may be the answer in

such uncertain situations. In our case, the Computer

Tech Skills Practice Test was used

In this study, four fuzzy weights, namely Novice

(N), Intermediate (I), Good (G) and Expert (E), are

used to represent the computer knowledge level of

users of any kind of software. Trapezoidal

membership functions are used to express each fuzzy

weight. These functions are represented by four

boundary values: the degree of membership grows

from 0 to 1 between a1 and a2, flattens between a2

and a3, and then drops from 1 to 0 between a3 and a4.

Each category of computer knowledge level has an

interval when users' scores fully belong to the

category, and this is why trapezoidal membership

functions were employed. The following equations

illustrate the fuzzy weights membership functions,

where x is the student score; while Fig. 1 depicts their

scheme:

𝜇





𝑥





1𝑥≤30

1−

𝑥−30

30<𝑥<40

0𝑥≥40

𝜇





𝑥



⎩

⎪

⎨

⎪

⎧

𝑥−30

30<𝑥<40

1 40≤𝑥≤60

1−

𝑥−60

60<𝑥<70

0 𝑥≤30 𝑜𝑟 𝑥≥70

𝜇





𝑥



⎩

⎪

⎨

⎪

⎧

𝑥−60

60<𝑥<70

1 70≤𝑥≤80

1−

𝑥−80

80<𝑥<90

0 𝑥≤60 𝑜𝑟 𝑥≥90

𝜇





𝑥





𝑥−80

80<𝑥<90

1 90≤𝑥≤100

0𝑥≤80

WEBIST 2021 - 17th International Conference on Web Information Systems and Technologies

366

Figure 1: Scheme of fuzzy weights membership functions.

In light of the foregoing, the four fuzzy weights

are utilized to portray users’ computer knowledge

level. The values of these fuzzy weights, which range

from 0 to 1, are defined using the aforementioned

membership functions. Thereby, the equation

𝜇





𝑥





𝜇





𝑥



𝜇





𝑥



𝜇





𝑥



is correct.

Ten faculty members from Greek public

institutions defined the fuzzy weights and thresholds

of their membership functions. In more detail, the

faculty members were requested to describe the

computer knowledge levels of users during the whole

educational process, as well as the intervals of a user's

degree of success, characterizing each of these

computer knowledge levels. The faculty members

have more than 10 years of expertise in instructing

computing in university settings and can vouch for an

error-free portrayal of users’ knowledge levels.

3 RULES FOR ASSISTIVE

MESSAGES DELIVERY

This section describes the rules that are associated

with the four fuzzy weights explained earlier for

determining the quantity and detail of the assistive

messages that should be provided to each user. Each

user will be delivered these messages to be helped

while navigating in the software based on the fuzzy

weight s/he belongs to.

The detail and quantity of assistive messages

varies. These rules were established by the same 15

professors who specialize in teaching computing.

Particularly, they are given the seven derived

categories of membership in the above-described

fuzzy weights, and they are asked their opinion about

the number and detail of messages that are required

for helping users belonging to different computer

knowledge level categories. Then, the average of their

answers is taken into consideration for the formation

of the rules. The whole set of the rules are presented

in Table 1. Table 1 depicts the number and the

complexity degree of the learning activities for each

instance of the fuzzy weights, namely the number and

detail of assistive messages that a particular user will

receive based on his/her computer knowledge level.

Note that d is the detail degree of the message that can

take values from 1 (of low detail) to 3 (of high detail).

Table 1: Rules for assistive messages based on fuzzy

weights.

Fuzzy weights

Assistive messages

(with degrees of

detail and quantity)

3 of d=3

2 of d=2

0 of d=1

> μ

2 of d=3

3 of d=2

0 of d=1

=1 or μ

> μ

2 of d=3

2 of d=2

1 of d=1

> μ

1 of d=3

3 of d=2

1 of d=1

=1 or μ

> μ

1 of d=3

2 of d=2

2 of d=1

> μ

0 of d=3

2 of d=2

3 of d=1

0 of d=3

1 of d=2

4 of d=1

It needs to be underlined that the assistive

messages are pertinent to the screen the user sees. For

each screen, the messages are different and according

to the fuzzy weights. The software delivers five

messages per screen; this number is adequate so that

the user can be informed but not in a tedious way.

However, this number can be changed based on the

kind of software in which this approach is

incorporated. Following, several examples of

assistive messages of an educational software are

given based on users’ computer knowledge level

vector (

, μ

 A user with computer knowledge (1, 0, 0, 0)

who want to create an announcement will get

more detailed messages, such as “To upload an

announcement to students, firstly you should

write the title and the body text, and then select

if the students receive it in their emails. The

Date Ends field is to declare for how much time

the announcement will be visible.”. Whereas a

user with computer knowledge (0, 0, 0, 1) will

get less detailed messages on the same subject,

such as “To upload an announcement to

A User-centric System for Improving Human-Computer Interaction through Fuzzy Logic-based Assistive Messages

367

students, enter the title, body text, email

delivery and date ends option.”.

 A user with computer knowledge (0, 0.2, 0.8,

0) will get more detailed messages, such as:

“You can upload an xml file with your course

questions. Next, select ‘Introduction from IMS

QTI’ to enter these questions in your course

Test Bank. Then, you can modify them

according to your needs.”. Whereas a user with

computer knowledge (0, 0, 0, 1) will get less

detailed messages on the same subject, such as

“Upload the xml file with course questions,

select ‘Introduction from IMS QTI’ and modify

the questions if needed.”.

4 EXPERIMENTAL RESULTS

AND DISCUSSION

The software must always be evaluated to determine

its effectiveness and user acceptance. During the

COVID-19 pandemic, special emphasis has been

placed on the imposed asynchronous and online

education. As such, we have incorporated the

presented approach in a learning management system

that was used by faculty members of public

universities in the capital city of the country to

support the online education.

The evaluation process took place in two parallel

phases during the lockdown of the spring semester of

the academic year 2019-2020:

 Phase 1: 20 Faculty members used the learning

management system that provided static

assistive messages to them which are the same

for every user.

 Phase 2: 20 Faculty members used the learning

management system that provided the fuzzy

logic-based assistive messages (our presented

approach).

For the evaluation, the faculty members of Phases

1 and 2 were asked three questions about their

experience with the system. After that, the results of

the questions posed to Faculty members of Phase 2

are presented, as well as t-test was employed to assess

the effectiveness of the assistive messages that is

achieved with the use of the presented system

(through fuzzy logic) in comparison to its

conventional version (static messages). The questions

are in a five-point Likert scale from -2 (absolutely

disagree) to 2 (absolutely agree) and are the

following:

 Were the assistive messages helpful? (Q1)

 Did the assistive messages improve your

interaction with the system? (Q2)

 Were the assistive messages adjusted to your

computer knowledge level? (Q3)

The results of the above questions for Phase 2

were aggregated and are shown in three pie charts

(Fig. 2-4).

Figure 2: Pie chart for Q1.

Figure 3: Pie chart for Q2.

Figure 4: Pie chart for Q3.

Concerning the first question (Q1), 14 faculty

members (a percentage of 70%) declared that the

assistive messages were helpful. Furthermore, 16

faculty members (a percentage of 80%) stated that the

assistive messages improved your interaction with the

system. Also, 15 faculty members (a percentage of

76%) confirmed that the assistive messages were

adjusted to your computer knowledge level. These

10%

20%

70%

Low Fair High

15%

80%

Low Fair High

10%

15%

75%

Low Fair High

WEBIST 2021 - 17th International Conference on Web Information Systems and Technologies

368

results were anticipated given that the assistive

messages using fuzzy logic improved the interaction

of people with the software, offering them an

optimized user experience.

Furthermore, t-test was employed on the results of

questions (Q1, Q2, Q3) posed to faculty members in

Phase 1 and 2, to compare the presented approach

(with assistive messages using fuzzy logic) with the

conventional approach (with static assistive

messages, same for all users). The alpha value, for the

experiment, was set to 0.05, and we discovered that

there is a statistically significant difference between

the means of the two trials for Q1, Q2, and Q3, by

evaluating the p-values (Table 2).

Table 2: T-test results.

Q1 Q2 Q3

Phase 1 Phase 2 Phase 1 Phase 2 Phase 1 Phase 2

Mean -0,3 1,45 -0,25 1,7 -0,55 1,55

Variance 0,536842 0,997368 1,25 0,536842 1,3131580,892105

Observations 20 20 20 20 20 20

Hypothesized

Mean Difference

0,767105 0,893421 1,102632

df 0 0 0

t Stat 38 38 38

P(T<=t) one-tail -6,31845 -6,52389 -6,32418

t Critical one-tail 1,04E-07 5,46E-08 1,02E-07

P(T<=t) two-tail 1,685954 1,685954 1,685954

t Critical two-tail 2,08E-07 1,09E-07 2,05E-07

Analyzing the results of the t-test, it can be

inferred that our presented approach for assistive

messages offered a better experience to the users. The

reason is because the assistive messages found to be

helpful for the users (Q1), improved their interaction

with the system (Q2) and were adjusted to their

computer knowledge level (Q3).

5 CONCLUSIONS

This paper presents a user-centric approach to

enhance the interaction between humans and

computers. To achieve this, it supports the delivery of

adaptive and assistive messages to users, when

needed. Since the computer knowledge plays a

significant role for HCI, in this paper, it is modelled

using fuzzy weights to figure out the level to which

each user belongs. Moreover, in this research, several

rules have been constructed, considering the fuzzy

weights (as created using fuzzy logic) and the detail

of the assistive messages (meant to be achieved);

these rules are employed by the system and are

delivered to users, when appropriate. As a testbed for

this research, the presented approach has been

incorporated in an e-learning software to help

instructors create a pedagogically sound learning

environment during the COVID-19 lockdown period.

The experimental results are very promising, showing

that the users are helped by the presented approach

and have a good experience during the interaction

with the software.

It needs to be noted that this approach can be

incorporated in any kind of software, e.g. e-

commerce, e-health systems etc., if the messages are

properly altered. Future research steps include the

incorporation of this approach into other kind of

software and its evaluation to measure the

effectiveness of the assistive messages in other

domains as well.

REFERENCES

Ahmedi, L., & Shabani, D. (2017, April). Search Engine

Query Recommendation-Using SNA over Query Logs

with User Profiles. In International Conference on Web

Information Systems and Technologies (Vol. 2, pp. 370-

375). SCITEPRESS.

Al Fararni, K., Nafis, F., Aghoutane, B., Yahyaouy, A.,

Riffi, J., & Sabri, A. (2021). Hybrid recommender

system for tourism based on big data and AI: A

conceptual framework. Big Data Mining and Analytics,

4(1), 47-55.

Bhanuse, R., & Mal, S. (2021, March). A Systematic

Review: Deep Learning based E-Learning

Recommendation System. In 2021 International

Conference on Artificial Intelligence and Smart

Systems (ICAIS) (pp. 190-197). IEEE.

Chaudhari, K., & Thakkar, A. (2019). A comprehensive

survey on travel recommender systems. Archives of

Computational Methods in Engineering, 1-27.

Chauhan, G., Mishra, D. V., & Begam, M. F. (2019,

October). Customer-Aware Recommender System for

Push Notifications in an e-commerce Environment. In

2019 Global Conference for Advancement in

Technology (GCAT) (pp. 1-7). IEEE.

Deldjoo, Y., Schedl, M., & Elahi, M. (2019, September).

Movie genome recommender: A novel recommender

system based on multimedia content. In 2019

International Conference on Content-Based

Multimedia Indexing (CBMI) (pp. 1-4). IEEE.

Ferretto, L. R., Bellei, E. A., Biduski, D., Bin, L. C. P.,

Moro, M. M., Cervi, C. R., & De Marchi, A. C. B.

(2020). A physical activity recommender system for

patients with arterial hypertension. IEEE Access, 8,

61656-61664.

Gräßer, F., Malberg, H., Zaunseder, S., Beckert, S.,

Schmitt, J., & Klinik, S. A. (2016, September).

Application of recommender system methods for

A User-centric System for Improving Human-Computer Interaction through Fuzzy Logic-based Assistive Messages

369

therapy decision support. In 2016 IEEE 18th

International Conference on e-Health Networking,

Applications and Services (Healthcom) (pp. 1-6). IEEE.

Hassannia, R., Vatankhah Barenji, A., Li, Z., & Alipour, H.

(2019). Web-based recommendation system for smart

tourism: Multiagent technology. Sustainability, 11(2),

323.

Kashani, S. M. Z., & Hamidzadeh, J. (2020, December).

Improvement of non-negative matrix-factorization-

based and Trust-based approach to collaborative

filtering for recommender systems. In 2020 6th Iranian

Conference on Signal Processing and Intelligent

Systems (ICSPIS) (pp. 1-7). IEEE.

Khodr, H., Kianzad, S., Johal, W., Kothiyal, A., Bruno, B.,

& Dillenbourg, P. (2020). AlloHaptic: Robot-Mediated

Haptic Collaboration for Learning Linear Functions. In

2020 29th IEEE International Conference on Robot and

Human Interactive Communication (RO-MAN) (pp. 27-

34). IEEE.

Krouska, A., Troussas, C., & Sgouropoulou, C. (2020,

October). A Personalized Brain-Based Quiz Game for

Improving Students’ Cognitive Functions. In

International Conference on Brain Function

Assessment in Learning (pp. 102-106). Springer, Cham.

(a)

Krouska, A., Troussas, C., & Sgouropoulou, C. (2020,

June). Applying Genetic Algorithms for

Recommending Adequate Competitors in Mobile

Game-Based Learning Environments. In International

Conference on Intelligent Tutoring Systems (pp. 196-

204). Springer, Cham. (b)

Méndez, J. I., Ponce, P., Meier, A., Peffer, T., Mata, O., &

Molina, A. (2019, November). Framework for

promoting social interaction and physical activity in

elderly people using gamification and fuzzy logic

strategy. In 2019 IEEE Global Conference on Signal

and Information Processing (GlobalSIP) (pp. 1-5).

IEEE.

Nie, G., Ullal, A., Zheng, Z., Swanson, A. R., Weitlauf, A.

S., Warren, Z. E., & Sarkar, N. (2021). An Immersive

Computer-Mediated Caregiver-Child Interaction

System for Young Children with Autism Spectrum

Disorder. IEEE Transactions on Neural Systems and

Rehabilitation Engineering, 29, 884-893.

Ouaftouh, S., Sassi, I., Zellou, A., & Anter, S. (2019,

October). Flat and hierarchical user profile clustering in

an e-commerce recommender system. In 2019 1st

International Conference on Smart Systems and Data

Science (ICSSD) (pp. 1-5). IEEE.

Papakostas, C., Troussas, C., Krouska, A., &

Sgouropoulou, C. (2021). Measuring User Experience,

Usability and Interactivity of a Personalized Mobile

Augmented Reality Training System. Sensors, 21(11),

3888.

Peña, P., Lado, R. T., Del Hoyo, R., del Carmen Rodríguez-

Hernández, M., & Abadía-Gallego, D. (2020).

Ontology-quality Evaluation Methodology for

Enhancing Semantic Searches and Recommendations:

A Case Study. In WEBIST (pp. 277-284).

Peng, M., Zhang, J., Wen, S., & Liu, C. H. (2020,

December). Comparing Recommendation Algorithms

in Session-based E-commerce Sites. In 2020

International Conference on Intelligent Computing,

Automation and Systems (ICICAS) (pp. 377-380).

IEEE.

Qomariyah, N. N., & Fajar, A. N. (2019, December).

Recommender system for e-learning based on personal

learning style. In 2019 International Seminar on

Research of Information Technology and Intelligent

Systems (ISRITI) (pp. 563-567). IEEE.

Singla, R., Gupta, S., Gupta, A., & Vishwakarma, D. K.

(2020, June). FLEX: A Content Based Movie

Recommender. In 2020 International Conference for

Emerging Technology (INCET) (pp. 1-4). IEEE.

Soulef, B., Abdesselam, B., & Raja, C. (2021, June). An

association rule-based recommender system for

learning materials recommendation. In 2021

International e-Engineering Education Services

Conference (e-Engineering) (pp. 49-54). IEEE.

Suhaim, A. B., & Berri, J. (2021). Context-Aware

Recommender Systems for Social Networks: Review,

Challenges and Opportunities. IEEE Access, 9, 57440-

57463.

Tahir, A., Ijaz, A., & Javed, F. (2021, May). Classy AA-

NECTAR: Personalized Ubiquitous E-Learning

Recommender System with Ontology and Data Science

Techniques. In 2021 International Conference on

Digital Futures and Transformative Technologies

(ICoDT2) (pp. 1-8). IEEE.

Tellería, C., Ilarri, S., & Sánchez, C. (2020). Text Mining

of Medical Documents in Spanish: Semantic

Annotation and Detection of Recommendations. In

WEBIST (pp. 197-208).

Troussas, C., Krouska, A., & Sgouropoulou, C. (2021).

Improving Learner-Computer Interaction through

Intelligent Learning Material Delivery Using

Instructional Design Modeling. Entropy, 23(6), 668.

Troussas, C., Krouska, A., & Virvou, M. (2018,

November). Multi-algorithmic techniques and a hybrid

model for increasing the efficiency of recommender

systems. In 2018 IEEE 30th International Conference

on Tools with Artificial Intelligence (ICTAI) (pp. 184-

188). IEEE.

Wan, S., & Niu, Z. (2019). A hybrid e-learning

recommendation approach based on learners’ influence

propagation. IEEE Transactions on Knowledge and

Data Engineering, 32(5), 827-840.

Wang, W., Zhang, G., & Lu, J. (2017). Hierarchy

visualization for group recommender systems. IEEE

Transactions on Systems, Man, and Cybernetics:

Systems, 49(6), 1152-1163.

Wynn, E., & Vallo Hult, H. (2019). Qualitative and critical

research in information systems and human-computer

interaction: Divergent and convergent paths. Now

Foundations and Trends.

Zhao, X. (2019, April). A study on e-commerce

recommender system based on big data. In 2019 IEEE

4th International Conference on Cloud Computing and

Big Data Analysis (ICCCBDA) (pp. 222-226). IEEE.

WEBIST 2021 - 17th International Conference on Web Information Systems and Technologies

370