Data Storytelling in Learning Analytics: AI-Driven Competence

Assessment

Ainhoa Álvarez

, Aitor Renobales-Irusta

and Mikel Villamañe

Department of Computer Languages and Systems University of the Basque Country UPV/EHU, Spain

Keywords: Learning Analytics, Generative Artificial Intelligence, Data Storytelling, Dashboards.

Abstract: Learning dashboards have become very popular, but the information shown on them is often difficult to

interpret by users. Different authors have worked to improve dashboards including narratives or data

storytelling techniques. However, creating these narratives is a complex process. Several studies have begun

to analyse the use of GenAI tools to generate these narratives in a scalable way, but this is still an area of

study that is at an early stage. In this paper, we present a proposal and a study aimed at generating narratives

using GenAI, extending previous work by aligning the generated narratives with the curriculum design of the

course. We first present a proposal for generating the narratives and then a study to evaluate their adequacy.

1 INTRODUCTION

A learning analytic dashboard is “a single display

that aggregates different indicators about learner(s),

learning process(es) and/or learning context(s) into

one or multiple visualizations” (Schwendimann

et al., 2017). Dashboards have gained popularity as a

tool to show analytical data regarding students

(Pozdniakov et al., 2025), and thus, providing

teachers with insights about the learning process of

their students (Fernandez-Nieto et al., 2024).

However, these dashboards are often challenging to

interpret by teachers and usually provide no guidance

on how to interpret them (Fernandez Nieto et al.,

2022). Several authors have worked in the inclusion

of explanatory features, for example, through Data

Storytelling (Fernandez-Nieto et al., 2024). Those

approaches are very powerful but generating the

narratives is a complex process that requires much

effort from creators (Li et al., 2024) and therefore,

some works have begun to explore the use of

Generative Artificial Intelligence (GenAI) to

automate and ease their generation in a scalable way

(Pinargote et al., 2024).

In a previous work (Villamañe, Mikel et al.,

2025), authors have begun to explore the use of

https://orcid.org/0000-0003-0735-5958

https://orcid.org/0000-0002-2148-9269

https://orcid.org/0000-0002-4450-1056

Generative AI to facilitate the comprehension of

dashboards and reduce the problem of data-literacy

lack that some teachers face when presented with

dashboards. Authors enhanced the dashboards

generated by the AdESMuS system (Alvarez et al.,

2020) with GenAI capabilities. As shown in Figure 1,

GenAI was introduced with three main objectives:

give general explanations about the chart, provide

interpretations about the data shown in the chart and

provide pedagogical insights and recommendations

in order to facilitate teachers taking remediation

actions when needed.

Figure 1: Dashboard with GenAI capabilities.

536

Álvarez, A., Renobales-Irusta, A., Villamañe and M.

Data Storytelling in Learning Analytics: AI-Driven Competence Assessment.

DOI: 10.5220/0013567300003967

In Proceedings of the 14th International Conference on Data Science, Technology and Applications (DATA 2025), pages 536-543

ISBN: 978-989-758-758-0; ISSN: 2184-285X

We also carried out a study with 15 teachers, and

one of its conclusions was that it could be useful to

relate assessment items to the contents of the course

in order to enhance the recommendations given by the

GenAI tool. This involves aligning pedagogical

intentions and course curriculum design with the

narratives about the data as other authors (Pinargote

et al., 2024) have also suggested.

In this paper, we address the alignment of

narratives with the curriculum design of the course

and present an initial approach. We present how we

propose to tackle the alignment and a study aimed at

answering these research questions:

 RQ1: How does providing course curriculum

design information to GenAI influence the

depth and utility of pedagogical insights?

 RQ2: To what extent do conclusions derived

from GenAI-driven analysis align with

pedagogical intentions and expectations of the

teachers?

This paper first presents some previous works

related to the study carried out. Next, the study design

is presented followed by the obtained results, a

discussion about them and finally some conclusions

and future work are presented.

2 PREVIOUS WORKS

This section resumes a preliminary study conducted

by the authors and that sets the basis of the work

presented in this paper. It also depicts the definition

of the ontology for the domain and student data used

in this work.

2.1 Preliminary Study

Previous to this work, in (Villamañe, Mikel et al.,

2025) we presented the use of GenAI to enhance

dashboards in different aspects. One of those aspects

was related to the generation of pedagogical

conclusions about the data shown in the charts (see

Figure 1) and providing recommendations so that the

teacher could take remediation actions.

To this end, we used the prompt template shown

in Table 1, along with any of the charts displayed on

the dashboard showing the student's performance

across different assessment items of the course (see

Figure 1), and asked the GenAI to provide the teacher

with conclusions about the performance of the

student.

Table 1: Prompt template.

I am a university professor and I would like you to

indicate in a maximum of 250 words what

conclusions can be drawn from the data contained in

the attached file and indicate if you consider that I

should make any recommendation to the person to

improve their learning process. I don't need you to

describe the information, just give me the conclusions

and recommendations in a general way, focusing only

on those elements that are es

eciall

relevant.

This process was repeated with the charts of

several students and Table 2 shows some of the

sentences included in the conclusions and

recommendations generated by the GenAI.

Table 2: Sentences selected from the answers generated by

the GenAI.

This suggests a relatively solid knowledge in this topic

Here the student shows a deep understanding of the

topic

It would be useful for him to strengthen his knowledge

in analysis and design.

It is suggested to dedicate more time to studying and

practicing topics related to Elem3, considering the

possibility of requesting additional support or tutoring.

As can be seen in Table 2, the system introduces

sentences such as “

This suggests a relatively solid

knowledge in this topic”. However, the system does not

know the topics each assessment item is related to, so

it can not derive more insightful conclusions. For

example, what happens if there are two assessment

items related to the same topic and with very different

performance results?

Our thesis is that relating the curriculum design

that defines the course topics or competences to the

assessment items would create more comprehensive

and more useful information for the teacher.

2.2 Ontology for Domain and Student

Data

The objective of this paper is to analyse the

comparative impact of curriculum design-enriched

versus standard performance-based prompts on the

effectiveness of AI-generated learning

recommendations included in dashboards. We have

therefore defined an ontology to formalise the domain

model for the curriculum structure including its

competences, learning units, learning materials, and

other related elements, as well as the student model

with the student-related information.

Data Storytelling in Learning Analytics: AI-Driven Competence Assessment

537

The domain model represents “the skills,

knowledge and strategies of the topic being tutored”

(Sottilare et al., 2016). It defines a conceptual

framework to represent all the elements and

relationships within a course. There are many ways to

represent the domain model but it is important for the

model to be general enough to be able to integrate

data from different sources, such as Learning

Management Systems (LMS) or Intelligent Tutoring

Systems (ITS) (Samuelsen et al., 2019).

Figure

shows an excerpt of the domain model

ontology used, which is an extension of the ontology

defined in (Villamañe et al., 2018) that has been

successfully used in previous studies. It is also

founded on the Competence-based knowledge space

theory (CbKST) (Idrissi et al., 2017) and the domain

models of classical educational systems (Aleven

et al., 2023). This ontology provides a framework for

structuring educational content of courses with a

focus on competence-based learning, establishing

clear relationships between learning units, resources,

activities, and the competences they help develop. It

defines five core entities: Courses, Resources,

LearningUnits, AssessmentItems, and Competences,

connected through various relationships. This

structure creates a comprehensive framework that

links educational content, activities, and assessments

to the skills and knowledge learners should acquire.

This structure is general enough to accommodate

information coming from different educational

systems such as ITSs or LMSs as Moodle, and it can

incorporate data from different courses. That is, the

ontology is not dependent on any specific course or

educational system.

Figure

also shows the main elements of the

student model that are used in this study. Taking into

account that a key element of any student model is the

student performance data (Pelánek, 2022), we have

included the student general information together

with the representation of its relationship to courses,

assessment items, and competences.

3 STUDY DESIGN

To evaluate and validate our proposal, we conducted

an initial study involving two teachers from the

course Analysis and Design of Information Systems

and the assessment data of the 60 students enrolled in

the course. In this section, we present the design of

the study detailing the competences defined for the

course and the student data collection together with

the methodology used for the study.

The process and instruments used in the study

were approved by the Ethics Commission for

Research and Teaching (CEID/IIEB) of the

University of the Basque Country UPV/EHU with

code M10-2016-181 and informed consent was

obtained from all individual participants included in

the study. Following the recommendations of the

ethics commission, all direct identifiers and attributes

that could potentially be used to identify students

were supressed. Afterwards, a numerical id was

randomly assigned to each student.

3.1 Competence and Student Data

Definition

As mentioned before, the aim of this study is to

analyse whether it would be interesting to enhance the

system with course information in order to obtain

more insightful conclusions and recommendations.

To that end it is necessary to populate the prompt

template used in the preliminary work (see Table 1),

with information following the ontology shown in

Figure 2.

First, the teachers defined the main competences

of the course, as shown in Table 3. These

competences were then linked to the assessment items

that students should complete throughout the course.

Figure 2: Extract of the defined ontology.

DATA 2025 - 14th International Conference on Data Science, Technology and Applications

538

Table 3: Course Competences.

CODE COMPETENCE

C1 Interact with users to gather requirements for information systems, ensuring clear understanding of

stakeholder needs through effective communication techniques.

C2 Identify client requirements for software development projects, translating business needs into functional

ecifications usin

UML methodolo

ies.

C3 Create data models that represent information structures and relationships, using UML class diagrams and

entit

-relationshi

models for database desi

C4 Analyse software specifications to determine feasibility and technical approaches, evaluating requirements

against system constraints in information systems development.

C5 Create flowcharts and UML diagrams to visualize system processes, data flows, and interactions between

com

onents in software desi

C6 Plan software tests to verify functionality and validate that information systems meet requirements, developing

com

rehensive test cases and scenarios.

C7 Program computer systems by implementing designs into functional code, translating UML models into

working software components.

C8 Write project documentation that clearly explains system architecture, design decisions, and implementation

details for information system stakeholders.

Table 4 shows the relationship between the course

competences and the assessment items.

Table 4: Assessment items and their relation with the course

competences.

ASSESSMENT ITEM

COMPETENCES

(CODE)

Use Case Exa

C1, C2

Domain Model Exa

C2, C3

Anal

sis and Desi

n Exa

C4, C5

Extended Use Cases Practice C1, C2

Domain Model Practice C2, C3

Test Plan Practice C6

Communication Diagrams

Practice

C4, C5

Class Diagram and Relational

Schema Practice

C3, C4

uence Dia

rams Practice C4, C5

Final Practice and

Documentation

C3, C5, C8

lementation Practice C7

Test Im

lementation Practice C6

Next, the assessment data of the 60 students

enrolled in the course was collected from Moodle and

formatted according to the defined ontology.

Figure

3 shows an extract of the graphical

representation of some of the information related to

the student with id 006.

All the structured information was then used to

populate the prompt template before submitting it to

the GenAI. The data was structured into JSON-LD

format to assure its readability and scalability.

Figure 3: Extract of student 006 assessment data.

3.2 Methodology

To analyse the utility and alignment of the

conclusions and recommendations provided by the

GenAI with those of the educators, teachers

participating in the study were invited to assess the

reliability and appropriateness of the results for a

sample of the group.

Since the opinions of the teachers are subjective

and depend on their own perceptions, this analysis

was approached as a qualitative study. Therefore, the

students selected to be part of the sample were

purposefully chosen based on the variety they

contributed to the study, as recommended by the

literature (Abrams, 2010; Shaw & Holland, 2014).

To that end, and with the collaboration of the

course teachers, the different situations that could

arise among the students of the course were defined.

Data Storytelling in Learning Analytics: AI-Driven Competence Assessment

539

The situations described next were identified by the

teachers, who considered various potential scenarios

that could occur in the course:

1. No data at all.

2. Missing some data.

3. All competences acquired.

4. Not all competences acquired

Next, students were distributed among the

situations considering their assessment data. From

each situation, as many students as necessary were

randomly selected to maintain the percentage of

representativeness of each situation in the sample,

which size was decided to be of 15 students (25% of

the population) aligning with recommendations in

research literature for studies employing random

sampling techniques (Boddy, 2016).

Table 5 shows the percentage of students that met

the criteria to be assigned into each situation defined

by the teachers, the number of students to be selected

from those in each situation and the id of the

randomly selected students.

Table 5: Distribution of students among the different

situations and selected students for the sample.

Situation

Percentage

of students

Number of

students

selected

Selected

students

(id)

1 1.7% 1 014

2 15% 2 038 049

3 55% 8

003 020

004 023

006 043

016 052

4 28.3% 4

013 039

035 058

4 RESULTS

With the students for the sample selected, the prompt

template was used with the assessment data of each

of these students. Then, the teachers participating in

the study were asked to indicate on a Likert-scale of

1 to 5 their level of alignment with the narrative

generated by the GenAI. With 1 being the lowest

level of alignment and 5 indicating that they strongly

agreed with the recommendations and conclusions

obtained. They also had an open-ended question to

include other comments.

When evaluating the results for the students in

situation number 4 (not all competences acquired),

teachers stated that the narratives generated by the

GenAI were not adequate.

As shown in Table 6, the AI highlights some

inconsistencies such as the student achieving

significantly different results in assessment items that

assess the same competence.

Table 6: First version of generated conclusions and

recommendations.

Conclusions and Recommendations for Student

006

The student demonstrates a notably better

performance in practical activities (average of 7.5)

than in theoretical exams (average of 5.1),

suggesting a greater aptitude for practical

application of knowledge rather than theoretical

conceptualization.

The student particularly excels in competencies

related to software testing planning (8.9) and

technical diagram development (8.2-8.5), showing

strengths in technical and design aspects.

However, there is a significant weakness in the

"create data models" competency, evidenced by

the low score in the Domain Model Exam (3.2),

although paradoxically the student obtains good

results in the corresponding practical work (7.2).

Recommendations:

1. Strengthen theoretical understanding of

domain models and their conceptualization,

possibly through study techniques that better

connect theory with the practical skills already

mastered.

2. Work on the "identify client requirements"

competency, where the student shows

disparate results between exams and practical

work.

3. Maintain the good performance in design and

documentation competencies, which

constitute the student's strengths.

4. Implement study strategies that allow

transferring evident practical capability to the

theoretical domain, possibly through self-

assessment exercises or exam simulations.

Discussions with the teachers participating in the

study revealed that this inconsistency was due to the

fact that the assessment item “Domain Model Exam”

was carried out individually, while the “practical

work” was carried out in groups.

The distinction between whether an assessment

item is conducted individually or in groups is crucial

in this case. A student who scores very low on

individual components but very high on group

components may be benefiting from the work of their

peers, indicating that competency acquisition may not

be occurring adequately, despite student’s average

grades concealing this fact.

DATA 2025 - 14th International Conference on Data Science, Technology and Applications

540

Following the analysis of the initial test, the

ontology, and therefore, the JSON-LD were updated

to include information on the categorisation of each

assessment item as either individual or in-group and

a new analysis was requested to the GenAI using the

same prompt template as before.

Table 7 shows the main differences among the

conclusions and recommendations for the same

student including the new information.

Table 7: Extract from generated conclusions and

recommendations.

Conclusions and recommendations for student

006

The student shows contrasting performance

between individual and group assessments.

….

This difference suggests possible difficulties in

applying theoretical knowledge individually or

problems with exam pressure.

Recommendations:

1. Strengthen individual study ….

2. Practice more individual domain

modeling exercises ....

3. Develop strategies to transfer skills

demonstrated in group work to individual

assessment contexts.

…

The student has demonstrated potential in

practical environments but needs to consolidate

their autonomy in applying theoretical concepts to

improve their overall per

ormance.

In this occasion, the GenAI suggests that the

student should engage in additional individual study

and practical exercises to consolidate knowledge. The

GenAI also indicates the possibility that the student

may experience challenges in managing the pressure

of an exam.

As this GenAI generated narrative was considered

more adequate by the teachers, the study was

replicated and completed using the adjusted ontology

and JSON-LD to populate the prompt template for all

the students in the sample. Next, teachers evaluated

their alignment with the new narratives.

Table 8 shows, for each situation and for each

student selected, the average value of teachers’

alignment with the GenAI provided narrative.

Table 8: Alignment measure for the sample.

Situation

Student

(id)

Ali

nment

score

(average

among

teachers

)

Ali

nment

score

(average

for the

situation

)

1 014 5 5

038 2

2.25

049 2.5

003 5

4.38

004 3.5

006 4

016 3.5

020 4.5

023 5

043 4.5

052 5

013 5

4.38

035 4.5

039 4

058 4

5 DISCUSSION

The analysis of the generated narratives reveals that

the information generated by the AI references

specific competences such as 'create data models'

which helps in better understanding which are the

student’s strengths and weaknesses. This finding

directly addresses our first research question (RQ1).

When providing curriculum design information, the

results show that GenAI provides more insightful

conclusions and recommendations. This represents a

significant enhancement of the generated narratives

and facilitates teachers identifying the precise

competences in which students need specific support.

Addressing our second research question —

examining the extent to which GenAI-driven analysis

aligns with teachers' pedagogical intentions and

expectations— we analysed data in Table 8. The

results demonstrate high alignment across most

situations what positively answers RQ2. However,

for situation number 2 the agreement with the GenAI

generated narrative for all of its students (049 and

038) was considerably lower. This situation

represents cases in which some of the data is missing

because the students have not completed some of the

assessment items. In these cases, the GenAI

highlighted low scores in one of the exams but failed

to mention that the student had not completed the

other two exams, leaving out critical missing data.

Data Storytelling in Learning Analytics: AI-Driven Competence Assessment

541

As these omissions are important from an

educational perspective, the prompt should be refined

to include specific instructions for addressing

incomplete student assessment data.

Despite this limitation, the overall high alignment

suggests that, when provided with appropriate

curriculum design information, GenAI tools can

produce insights that closely correspond to what

experienced educators would identify as

pedagogically relevant. This alignment between

teachers’ expectations and generated narratives

affirmatively answers our second research question

(RQ2) whilst identifying specific areas for the

improvement.

6 CONCLUSIONS

Learning analytics dashboards are very popular but

present many problems to users. One of the main ones

is related to the difficulties users face to interpret data

shown on dashboards. To reduce this problem several

authors have proposed the enhancement of those

dashboards with narratives. This has shown positive

results but generating the narratives is a complex

process with a great workload. In this paper, we have

presented a proposal that enhances dashboards with

narratives automatically generated by GenAI tools.

The presented proposal has been validated in a course

with two participating teachers and the data of 60

students.

Populating the prompts with a context that defines

the course curriculum design to align the narratives

with them is very promising and significantly

enhances their relevance and usefulness for

educators. By incorporating course context into the

prompts, the generated conclusions are more detailed

and aligned with pedagogical goals, providing

actionable recommendations for teachers (RQ1).

The generated narratives were evaluated by

teachers and found to closely align with their own

interpretations of student performance, as evidenced

by a high average alignment score higher than 4 on a

1 to 5 Likert scale (RQ2).

Although the study was conducted with a

relatively small sample, the results are encouraging,

and the outcomes point to significant potential. This

emphasizes the value of including the curriculum

design of the course on the prompts. Therefore, we

plan to continue the proposal validation process with

a larger participant base, conducting further

experiments across different courses and areas of

knowledge to test its potential generalization.

The results have also shown some aspects that can

be improved that we next point as future work.

Learning Management systems do not often

include the option to define the curriculum design and

relate assessment items to it using a systematic

approach. Therefore, in the near future we plan to

create a way to facilitate the definition of the

curriculum design of the course and to relate it to the

learning assessment items. We plan to start doing this

for Moodle as it is one of the most used Learning

Management System in higher education settings

(García-Murillo et al., 2020).

We also plan to improve the domain and student

data ontologies using educational standards and

semantic web techniques in order to ensure its

flexibility and applicability.

Some limitations were also identified. For

example, the GenAI occasionally omitted references

to assessment items not completed by students, which

impacted the comprehensiveness of its conclusions.

Addressing such omissions will require refining the

prompts to account for incomplete data scenarios.

Finally, it is important to address the ethical

concerns associated with the use of GenAI in

generating educational recommendations. Our

approach is designed to support, not replace, the

teachers’ role. The GenAI generated narratives are

intended to be supplementary, providing additional

insights based on the data available. To safeguard the

accuracy and integrity of the educational process, we,

as other authors (Chiu, 2024), propose

comprehensive training for the teachers. This training

will ensure teachers are aware of the limitations and

ethical considerations of using GenAI, thereby

maintaining their essential role in the decision-

making processes.

ACKNOWLEDGEMENTS

This work was partially funded by the Department of

Education, Universities and Research of the Basque

Government (ADIAN, IT-1437-22) and grant

RED2022-134284-T.

REFERENCES

Abrams, L. S. (2010). Sampling ‘Hard to Reach’

Populations in Qualitative Research: The Case of

Incarcerated Youth. Qualitative Social Work, 9(4), 536-

550. https://doi.org/10.1177/1473325010367821

Aleven, V., Rowe, J., Huang, Y., & Mitrovic, A. (2023).

Domain modeling for AIED systems with connections

DATA 2025 - 14th International Conference on Data Science, Technology and Applications

542

to modeling student knowledge: A review. En B. Du

Boulay, A. Mitrovic, & K. Yacef (Eds.), Handbook of

Artificial Intelligence in Education (pp. 127-169).

Edward Elgar Publishing. https://doi.org/10.4337/

b9781800375413.00017

Alvarez, A., Villamañe, M., & Ibargutxi, L. (2020).

Helping Teachers to Identify Students at Risk and

Analyze their Learning Process. Eighth International

Conference on Technological Ecosystems for

Enhancing Multiculturality, 747-751. https://doi.org/

10. 1145/3434780.3436645

Boddy, C. R. (2016). Sample size for qualitative research.

Qualitative Market Research: An International

Journal, 19(4), 426-432. https://doi.org/10.1108/QMR-

06-2016-0053

Chiu, T. K. F. (2024). The impact of Generative AI (GenAI)

on practices, policies and research direction in

education: A case of ChatGPT and Midjourney.

Interactive Learning Environments, 32(10), 6187-

6203. https://doi.org/10.1080/10494820.2023.2253861

Fernandez Nieto, G. M., Kitto, K., Buckingham Shum, S.,

& Martinez-Maldonado, R. (2022). Beyond the

Learning Analytics Dashboard: Alternative Ways to

Communicate Student Data Insights Combining

Visualisation, Narrative and Storytelling. LAK22: 12th

International Learning Analytics and Knowledge

Conference, 219-229. https://doi.org/10.114

5/3506860.3506895

Fernandez-Nieto, G. M., Martinez-Maldonado, R.,

Echeverria, V., Kitto, K., Gašević, D., & Buckingham

Shum, S. (2024). Data Storytelling Editor: A Teacher-

Centred Tool for Customising Learning Analytics

Dashboard Narratives. Proceedings of the 14th

Learning Analytics and Knowledge Conference, 678-

689. https://doi.org/10.1145/3636555.3636930

García-Murillo, G., Novoa-Hernández, P., & Rodríguez, R.

S. (2020). Technological Satisfaction About Moodle in

Higher Education—A Meta-Analysis. IEEE Revista

Iberoamericana de Tecnologias del Aprendizaje, 15(4),

281-290. IEEE Revista Iberoamericana de Tecnologias

del Aprendizaje. https://doi.org/10.1109/RITA.20

20.3033201

Idrissi, M. K., Hnida, M., & Bennani, S. (2017).

Competency-Based Assessment: From Conceptual

Model to Operational Tool. En E. Cano & G. Ion (Eds.),

Innovative Practices for Higher Education Assessment

and Measurement. https://doi.org/10.4018/978-1-5225-

0531-0.ch004

Li, H., Wang, Y., & Qu, H. (2024). Where Are We So Far?

Understanding Data Storytelling Tools from the

Perspective of Human-AI Collaboration. Proceedings

of the 2024 CHI Conference on Human Factors in

Computing Systems, 1-19.

https://doi.org/10.1145/3613904.3642726

Pelánek, R. (2022). Adaptive, Intelligent, and Personalized:

Navigating the Terminological Maze Behind

Educational Technology. International Journal of

Artificial Intelligence in Education, 32(1), 151-173.

https://doi.org/10.1007/s40593-021-00251-5

Pinargote, A., Calderón, E., Cevallos, K., Carrillo, G.,

Chiluiza, K., & Echeverria, V. (2024). Automating data

narratives in Learning Analytics Dashboards using

GenAI. 2024 Joint of International Conference on

Learning Analytics and Knowledge Workshops, 150-

161. https://research.monash.edu/en/publications/

automating-data-narratives-in-learning-analytics-

dashboards-using

Pozdniakov, S., Martinez-Maldonado, R., Tsai, Y.-S.,

Echeverria, V., Swiecki, Z., & Gasevic, D. (2025).

Investigating the Effect of Visualization Literacy and

Guidance on Teachers’ Dashboard Interpretation.

Journal of Learning Analytics, 1-25. https://doi.org/

10.18608/jla.2024.8471

Samuelsen, J., Chen, W., & Wasson, B. (2019). Integrating

multiple data sources for learning analytics—Review of

literature. Research and Practice in Technology

Enhanced Learning, 14(1), 11. https://doi.org /10.1186

/s41039-019-0105-4

Schwendimann, B. A., Rodríguez-Triana, M. J., Vozniuk,

A., Prieto, L. P., Boroujeni, M. S., Holzer, A., Gillet,

D., & Dillenbourg, P. (2017). Perceiving Learning at a

Glance: A Systematic Literature Review of Learning

Dashboard Research. IEEE Transactions on Learning

Technologies, 10(1), 30-41. IEEE Transactions on

Learning Technologies. https://doi.org/10.1109/TLT

.2016.2599522

Shaw, I., & Holland, S. (2014). Doing Qualitative Research

in Social Work. SAGE Publications, Ltd.

https://doi.org/10.4135/9781473906006

Sottilare, R. A., Graesser, A., Hu, X., Olney, A., Nye, B.

D., & Sinatra, A. (2016). Design Recommendations for

Intelligent Tutoring Systems: Volume 4-Domain

Modeling (Vol. 4). US Army Research Laboratory.

https://www.semanticscholar.org/paper/Design-Reco

mmendations-for-Intelligent-Tutoring-Sottilare-Grae

sser/745d7f28e1718ed095c9d934aa4342a4be6fc3e3

Villamañe, M., Alvarez, A., & Larranaga, M. (2018, abril).

Supporting competence-based learning with visual

learning analytics and recommendations. 2018 IEEE

Global Engineering Education Conference

(EDUCON). https://doi.org/10.1109/educon.2018

.8363421

Villamañe, Mikel, Renobales-Irusta, Aitor, & Alvarez,

Ainhoa. (2025). A Preliminary Study on the Use of

Generative Artificial Intelligence to Enrich

Dashboards. Proceedings of Csedu 2025. 17th

International Conference on Computer Supported

Education, Porto, Portugal.

Data Storytelling in Learning Analytics: AI-Driven Competence Assessment

543