A Free and Open Dataset from a Prototypical Data-driven Study

Assistant in Higher Education

Johannes Schrumpf

1,* a

, Felix Weber

2,* b

, Katharina Schurz

2,* c

, Niklas Dettmer

and

Tobias Thelen

Institute of Cognitive Science, Osnabrück University, Wachsbleiche 27, Osnabrück, Germany

virtUOS, Osnabrück University, Heger-Tor-Wall 12, Osnabrück, Germany

tthelen@uni-osnabrueck.de

Keywords: Artificial Intelligence, Dataset, Digital Study Assistant, Higher Education, Educational Recommendation

Engines.

Abstract: Digital study assistants (DSAs) are an as of yet sparsely explored method to build bridges between classical,

on-campus higher education and novel digital education opportunities. The DSA we present in this paper

(SIDDATA) aims at supporting students to identify, reflect upon and follow their personal educational goals.

Over the course of 11 months, students interacted with a prototype version 2.0 of the software, generating

data about what features were interacted with, users' study-related data, and which features were deemed as

useful. In this data paper, we present a preprocessed version of the DSA database for research in the domain

of digital higher education. We present the data model design of the DSA and its relation to its’ features. We

further expand on the data extraction method used to generate the present dataset from the DSA’s database.

We discuss potential research paths that can be explored based on the dataset as well as its limitations

1 INTRODUCTION

Digital higher education remains an innovative and

fast changing field: An ever-increasing assortment of

digital education opportunities blends, but also

partially clashes, with classical education formats

such as lectures and seminars (

Castro, 2019)

. At the

same time, emerging technologies such as Artificial

Intelligence (AI) find their way onto campuses not

only as a subject but also as a tool to support and

enhance student education.

Project SIDDATA (Studienindividualisierung

durch Digitale, Datengestützte Assistenten, eng:

study individualization through digital, data-driven,

assistants) aims to aggregate these education

opportunities and streamline them into an easy-to-use

DSA for long term usage, oriented towards individual

educational goals. As a collaboration project between

https://orcid.org/0000-0002-0068-273X

https://orcid.org/0000-0002-7012-3378

https://orcid.org/0000-0003-3804-1134

https://orcid.org/0000-0001-9944-1671

https://orcid.org/0000-0002-3337-6093

* Authors contributed equally

the University of Osnabrück, the University of

Bremen, and the Leibniz University of Hannover, it

is part of the “Innovationspotentiale Digitaler

Hochschulbildung” (eng: Innovation Potentials in

Digital Higher Education) funding line, funded by

the BMBF (Bundesministerium für Bildung und

Forschung, eng: Federal Ministry of Education and

Research).

The development of the SIDDATA digital study

assistant software (referenced as SIDDATA) has

occurred iteratively with an agile development

approach allowing user feedback to be incorporated

into the latter two of the three annual releases.

Research about goals in higher education and

requirements for DSAs can be found in the related

project publications (Reinken & Greiff, 2021,

Schrumpf et al., 2021, Thelen et al., 2019; Vogelsang

et al., 2019; Weber & Le Foll, 2020; Weber et al.,

Schrumpf, J., Weber, F., Schurz, K., Dettmer, N. and Thelen, T.

A Free and Open Dataset from a Prototypical Data-driven Study Assistant in Higher Education.

DOI: 10.5220/0011038800003182

In Proceedings of the 14th International Conference on Computer Supported Education (CSEDU 2022) - Volume 2, pages 155-162

ISBN: 978-989-758-562-3; ISSN: 2184-5026

155

2019; Weber, 2019; Weber et al., 2021a, 2021b,

2021c; Weber & Thelen, 2021).

By collecting and processing educational resource

data on the one hand and user-system interaction, on

the other hand, the data collected by the DSA, version

P2, throughout its eleven months of operation offers

a unique opportunity to probe and analyze potential

opportunities and challenges for digital higher

education.

Thus, we here present data generated through the

interaction between users and SIDDATA. This

dataset (Weber et al., 2022) serves as a basis for free

exploration, offering insights into student interests,

users’ study-related data, the usefulness of features,

and their usage frequency. Subsets of this dataset

already have served as a basis for publications in the

domain of educational science (Schurz et al., 2021;

Weber et al., 2021c).

In the following, we give a brief overview of the

SIDDATA study assistant software architecture.

Further, we introduce SIDDATA’s data model and

briefly highlight the data extraction process. We then

list descriptive statistics characterizing the dataset.

We conclude with a discussion of potential research

avenues based on the present dataset as well as its

limitations.

2 BACKGROUND

The present data were collected from user interaction

with version 2.0 of the SIDDATA study assistant

software. Students from three German universities

used the software, and the dataset contains data from

all three universities. Data aggregation occurred over

the course of 11 months, from December 2020 to

November 2021.

Because the dataset follows the student assistant

software’s database structure, for a better

understanding, we present the software architecture

underlying the DSA. We highlight the hierarchical

structure of database objects on a conceptual level

and outline user-database object interactions

facilitated by the software.

2.1 Software Architecture

The SIDDATA study assistant software is

implemented as a web service by utilizing the django

web framework. Figure 1 illustrates the software

architecture and its components. The study assistant

software consists of two interlocking parts that

communicate via a RESTful-API: the SIDDATA

plugin (“frontend”) and the SIDDATA backend.

The frontend serves as a graphical user interface

integrated into an existing learning management

system (LMS) to incorporate DSA functionalities

without the need to integrate them into the core of the

LMS directly. SIDDATA was developed and tested

with the “Stud.IP” (see https://www.studip.de/) LMS.

Figure 1: The components of SIDDATA’s web-based

software architecture.

On the one hand, the frontend provides data such

as course or user information to the backend. On the

other hand, it serves as an interface between users and

SIDDATA backend to request and receive

functionalities implemented in the backend logic.

The SIDDATA backend implements and serves

SIDDATA functionalities. It receives and processes

data from the LMS via the SIDDATA plugin. It also

handles user interaction by evaluating user inputs and

by creating appropriate answers, as implemented in

individual backend functions. A central component of

the SIDDATA backend is the SIDDATA database.

This PostgreSQL database stores, manages, and

retrieves information on individual SIDDATA

functions, user data, educational resource data, and

the state of interaction between users and the study

assistant software.

2.2 Hierarchical Data Structure

SIDDATA’s data model follows a strict hierarchy:

The uppermost hierarchy level is formed by so-called

Recommender modules. Attached to these modules

are Goal type objects. Activity type objects are

attached to goals and form the building blocks for

user-system interactions. Recommenders can have

multiple Goals, and Goals can have multiple

Activities attached. Figure 2 illustrates the

hierarchical structure of the SIDDATA data model.

SIDDATA’s functionalities are realized in

Recommender modules. Each recommender aims at

supporting student self-regulated learning through

offering a specific service. Recommenders vary in

terms of complexity: While some implement, for

example, a simple guide for applying and pursuing a

CSEDU 2022 - 14th International Conference on Computer Supported Education

156

Figure 2: Hierarchical structure of the SIDDATA data

model.

semester abroad, others offer more complex

functionalities such as a natural language based

semantic search engine for educational resources

(Schrumpf et al., 2021).

Attached to each recommender are Goals. These

objects represent abstract, long-term, or persistent

sub-functionalities of Recommenders. One example

of a Goal is a guide for navigating the administrative

landscape for a semester abroad. Figure 3 shows an

example of a Goal as displayed for the user.

Figure 3: Example of a Goal type object from the

“Academic Interests” recommender. The student has

entered “climate change mitigation” as one of their

interests. The Goal object is displayed as the dark blue strip

(Climate Change mitigation) above an attached Activity

object (Mein Thema[German: My Topic]). Goal type

objects can hold multiple Activity type objects.

Goal type objects can be discarded by clicking the

trash can symbol on the top right of the frame. This

may be useful if users decide not to pursue a goal

further and want to remove it from the interface.

Attached to Goals are Activities; these represent

atomic actions the user can perform within the study

assistant software. An example of an Activity is

visiting the foreign office of one’s home university to

sign a learning agreement for a semester abroad.

Activity objects are commonly displayed as a

dedicated interactive graphic with a bounding box

(see Figure 4).

Figure 4: Example of an Activity type object as displayed

in a learning management system.

Activities can be discarded and snoozed by users.

A snoozed Activity is removed from the user

interface and listed under the “paused” tab. Such

Activities can be resumed at a later point. Discarded

Activities are logged under the “discarded” tab.

Finally, Activities that were completed successfully

are listed under the “completed” tab.

2.3 Recommender Modules

Recommenders provide specific functionalities of the

study assistant software. They serve as the highest

attachment point for Goal objects. From a technical

perspective, Recommenders have their own

interaction logic and handle the processing of

attached Goals and Activity objects. From the user

perspective, recommenders are sub-functions that can

be enabled and disabled and provide specific services,

such as finding personally relevant learning resources

(Fachliche Interessen, eng: Academic Interests) or

connecting to other learners (Get-together, eng:

Academic Contacts

Figure 5: Example of the SIDDATA navigation menu

displaying activated Recommenders above and links to

paused, discarded, and completed Recommenders below.

A Free and Open Dataset from a Prototypical Data-driven Study Assistant in Higher Education

157

Recommenders offer students the service to

engage with and reflect upon individual educational

goals associated with the aspect of studying

associated with a recommender’s coverage: The

“Open Educational Resource” recommender, for

example, introduces learners to and highlights key

properties of, OERs and how they can assist with

learning subjects relevant to the user. Recommenders

thus are to be interpreted as independent interactive

features of the study assistant system. Once enabled,

recommender modules appear in the navigation menu

and are accessible via the graphical user interface (see

Figure 5).

Our dataset holds information about which users

have activated individual Recommender modules, as

well as which Goals and Activities are attached to

Recommenders. Table 1 shows the Recommenders

for which data is available in our dataset.

Table 1: Recommenders and their function.

Recommender title

(German title)

Description

Academic Contacts

(Get-together)

A tool for students to connect,

based on their personal

matching preferences

Academic Interests

(Fachliche

Interessen)

Semantic search engine for

Educational Resources based

on natural language processing

(see Schrumpf et al., 2021)

Data Ethics

(Souverän mit

meinen Daten)

Providing information about

data ethical aspects of the

digital age

Evaluation

(Evaluation)

Survey for users to give

detailed feedback

Funding

(Studienfinanzierung)

Providing information on how

to find funding for one’s

studies

Learning

Organization

(Lernorganisation)

Providing information about

methods and practices for self-

regulated learning

Open Educational

Resources

(Freie

Bildungsmaterialien)

Providing information about

Open Educational Resources

and a list of preselected

repositories

Personality Module

(Persönlichkeitsmodul)

Personality test for measuring

task-switching ability and

short/long-term memory

Scientific Career

(Wissenschaftliche

Karriere)

Providing general information

about a career in the scientific

field

Study Abroad

(Auslandssemester)

Providing general information

about study abroad

SIDDATA’s design philosophy is based on a

constructivist learning ideal where intrinsically

motivated learners take responsibility for their

personal and intellectual development. At the base of

this philosophy stand personal, educational goals,

which is why the substructure of recommenders is

called “goals” (see goal-related references for more

information). For some of the goal entities in the data

set it is the case that they correspond to a specific

personal goal (such as a specific personal interest in a

topic or a semester abroad), while some goal entities

in the data set were created by the developers to

structure the program flow and the appearance to the

user.

For instance, if students enter a text describing a

topic of personal interest in the Academic Interests

recommender, a goal appears with the interest as title,

and course recommendations appear as activities

within this goal. The goal here stands for an interest a

student wants to follow while studying. In contrast, a

goal within the Semester Abroad recommender is, for

example, to find a partner university with a fitting

study program. Here, a goal stands for a milestone in

the process of studying a semester abroad.

Activities are the smallest units in the data

structure and can be understood as atomic user

interactions with the system. Examples for Activities

are recommendations generated and displayed to the

user or questions which ask the users for input. The

term Activity was chosen because the system is

intended to elicit actions of students towards their

individual goals. All activity type objects have a title

and a description displayed directly in the UI of the

SIDDATA frontend. Additionally, Activities can be

paused or discarded. If Activities are successfully

interacted with, they are moved to the finished tab and

automatically set to inactive. Activities are sorted into

different types, each one implementing a unique

mode of interaction:

To-do type Activities act as an item on a checklist,

displaying information or reminding the user of a task

to be fulfilled to progress in reaching a certain goal.

Question type Activities require the user to make a

choice from a pre-selected list of answers or query the

user for an answer in natural language.

Resource type Activities present online resources in

the form of a weblink. This data type is intended to

represent a broad range of educational resources, such

as Open Educational Resources (OER), Massive

Open Online Courses (MOOCs), or goal-relevant

online resources, such as for instance a web-page of

an institution offering scholarships that the digital

assistant may identify with being of interest for the

personal goal to study abroad.

CSEDU 2022 - 14th International Conference on Computer Supported Education

158

Course type Activities are a recommendation for

enrollment into a course from the LMS of the local

and partner universities.

Additionally, users can rate any Activity on a

scale of 0 to 4. This feedback function provides hints

for the quality of the recommended items.

2.4 Data Extraction & Processing

In the SIDDATA database, long and cryptic IDs are

used to identify activities, students, recommenders,

courses, degrees, and goals for performance and

security reasons. To increase usability for researchers

working with the dataset and to further anonymize the

data to make it even less person-relatable, these IDs

were incrementally replaced by plain integer values.

In the original database, some personal or

personal-relatable data is present, which had to be

removed from the dataset. This includes names,

telephone numbers, and email addresses exchanged

between students in the Academic Contacts

recommender. In some cases, other personal

information such as health-related information was

present. Information of these kinds was replaced by

placeholders, such as for instance <emailadress>.

This procedure substitutes personal data in a way that

allows researchers to understand and analyze the data

while effectively protecting person-relatable

information.

3 DATASET PROPERTIES

The core objective of data processing from the

original data, stored in a PostgreSQL relational

database to one single CSV file, was optimal usability

for reuse by the scientific community. While the

database model was designed to avoid redundancies,

the preprocessed data set repeats information and at

the same time allows filtering, aggregation, and

custom analyses with all kinds of data processing

tools capable of reading CSV files, such as

LibreOffice Calc (or a commercial alternative, such

as Excel) or data analysis tools, such as R or PSPP (or

a commercial alternative, such as SPSS). The size of

the resulting comma-separated-value file (.csv) is

20.1 MB. The file format allows the comma or

semicolon as a string separator. In this case, the

semicolon was chosen because included texts do

contain commata, which leads to formatting issues

when working with the data. For scientists working

with the data, it is important to choose a semicolon

and not the comma as a string separator when

importing or reading the csv file.

3.1 Dataset Structure and Statistics

To make the dataset more readable, this section

introduces the data structure.

Inheriting from the hierarchical data structure of

the SIDDATA database, each activity belongs to

exactly one Goal, which belongs to exactly one

Recommender. Additionally, we have associated

each activity with a corresponding student interacting

with this activity. Hence, we aggregate all activity,

goal, and student information in one row of the

dataset. This enhances readability with the tradeoff of

increased repetition.

Students possess additional attributes such as field

of study, semester of enrollment, and home

university. Students can also study in multiple study

programs and, therefore, different semesters of

enrollment for each of their study programs. We have

aggregated student information within squared

brackets for dataset columns with potentially multiple

entries meaning that each indexed item within the

squared brackets corresponds to the data at the same

index in another column with multiple entries. For

example, if a student studies in two fields of study,

[mathematics, philosophy], the corresponding

semester of enrollment is stored as [3,5], where 3

corresponds to “mathematics” and 5 corresponds to

“philosophy”, in the semester of enrollment column.

Filtering functions allow to “zoom” into the data

to inspect, for instance, all activities of a user, all

users of a university, all students enrolled into a

specific study program, etc. Table 2 gives a detailed

explanation of variables and a range of possible

values for each column of the dataset.

Table 2: Summary of columns present in the dataset with a

more in-depth explanation of column meaning.

Column Explanation

activity_id unique identifier for activities,

uential numberin

activity_title

title of the activity

activity_description

description text

activity_status status of interactions (possible

values: “new”, “active”,

”snoozed”, ”done”, “discarded”,

”active”

)

activity_type type of activity (possible values:

“todo”, ”question”, ”resource”,

”course”)

feedback_scale_size Number of choices the user has to

rate an Activity. Ranges from 2 to

5, 0 stands for no feedback

tion.

A Free and Open Dataset from a Prototypical Data-driven Study Assistant in Higher Education

159

Table 2: Summary of columns present in the dataset with a

more in-depth explanation of column meaning (cont.).

Column Explanation

Given_feedback Feedback value for an Activity if it

was rated b

the use

timestam

time of creation

question_id unique identifier for questions,

sequential numbering (optional,

only for questions)

question question text (optional, only for

uestions

)

answers set of possible answers for multiple

choice questions (optional)

given_answers answers given by the student

(

tional

)

course_id unique identifier for courses,

sequential numbering (optional,

only for courses)

resource_id unique identifier for resources,

sequential numbering (optional,

only for recommended resources)

resource_title title of resource (optional, only for

resources)

resource_url url of a resource (optional, only for

resources

)

course_title title of a recommended course

(

tional, onl

for courses

)

goal_id unique identifier for goal,

sequential numberings

goal_title short title of a goal

goal_description detailed description of a goal

(

tional

)

goal_order order in which the goal is displayed

in the recommende

recommender name of the recommender

student_id unique identifier for students,

sequential numbering

student_university home university of the student

student_institute institute of the student (optional)

subject* study subject of the student

(optional)

subject_id unique identifier for subjects,

uential numberin

degree_id unique identifier for degrees,

uential numberin

degree* target degree for a subject of the

student

semester* semester for each combination or

subject and degree for a specific

student

*Subject, degree and semester together state an enrollment

of a student into a study program. One student can be

enrolled in more than one study program. In this dataset,

subject, degree, and semester contain ordered lists. The

order clarifies which subject, degree, and semester together

constitute one enrollment. The number of known

enrollments per student range from 0 to 5.

Table 3 shows the numbers for each entity type

represented in the data.

Table 3: The dataset holds information about 735 students

from three universities and the following other entity types

and quantities.

Entity Type n

Universities 3

Stud

Sub

ects 165

rees 19

Students 735

Recommende

Goals 6081

Activities 41520

Courses 1365

Questions 899

Semesters 15

4 DISCUSSION

With the SIDDATA prototype 2 dataset, we present a

source of data for investigations in the domain of

study assistant engineering, digital education,

educational psychology, recommendation engine

evaluation and the strategic considerations for

learning in the digital age.

The present dataset already served as a basis to

perform an effectiveness analysis of the SIDDATA

assistant software (see Schurz et al., 2021). However,

with this first publication, the depths of the

SIDDATA prototype 2 dataset are far from being

used to exhaustion.

4.1 Potential Research Avenues

We have identified the following paths as of yet

unexplored research opportunities for future

scientific endeavors:

So far, we have not performed analysis into

student-system interaction based on the field of study.

Aggregating statistical information about such usage

could produce novel insights into the student behavior

in terms of technology openness, the effectiveness of

software features, or how well the software design fits

the needs of students from different domains of

expertise. General statistical information about the

number of enrolled students per field of study is

available at each partner university's webpage (see

Appendix) and, in conjunction with the present

dataset, reveals new information.

Further, we have not analyzed the feedback from

students on single Activity objects, aside from a

general trend which indicates which Activities from

CSEDU 2022 - 14th International Conference on Computer Supported Education

160

which Recommender was generally perceived to be

useful or less useful, interactions between the

semester of enrolment, the field of study, and

likelihood to rate certain Activities with certain

values may be discoverable.

Additionally, an analysis of when which

recommenders were used may also give insights into

the needs and interests of students over the course of

two semesters.

Regarding individual recommenders, there are

multiple open questions that, as of yet, have not been

examined in detail:

The Academic Interests recommender (Fachliche

Interessen) logs interests that were entered by

students. These interests have not been analyzed for

their content or correlation with the field of study,

semester of enrollment, or other student data. Here,

an analysis of feedback for given recommendations

could yield insights into how fitting

recommendations were to requests.

Another recommender whose investigation could

hold valuable insight is the Academic Contacts (Get-

together) recommender: Here, students are able to get

matched to other students based on their common

interests. Analyzing the common interests students

have entered in order to find matches could give a

perspective on what kind of extracurricular

workshops or events could be offered by student

bodies or by other university institutions to

effectively facilitate student interest networks.

The present dataset may also support studies that

evaluate digital study assistant systems in general or

that seek to assess the state of digital higher education

in general.

4.2 Limitations

The dataset presented here offers a variety of data for

exploring the effectiveness of a digital study assistant

system. However, some limitations have to be noted,

which may impact future findings from the present

data: Before using the study assistant software, users

were asked whether they were willing to share their

data for research purposes. The data in the dataset

presented here, therefore, is only composed of

interactions logged from users who affirmed the

usage of their data. Therefore, a confounding effect

between willingness to share data and the willingness

to explore and interact with the student assistant

software caused by general user technology

acceptance cannot be ruled out.

Another limitation of the dataset is the quality of

data: Even though general data quality can be

considered high, some data such as field of study or

the title of recommended courses are drawn from the

learning management system of participating

universities. This means that the data quality of these

items may vary, as learning management system

settings for one university (Bremen) allowed students

to edit their field of study, while the other two

universities (Hannover, Osnabrück) only gave

officials the writing permissions for these data fields.

For courses, some course titles may only give

information in context and for a specific field of

study. This holds, for example, for acronyms that

carry little meaning outside a field. Further, semesters

of enrollment and users enrolled in a specific field of

study are not evenly distributed. This means that

investigations relying on semesters of enrollment and

field of study may not yield representative results and

may need to be pre-selected for a sufficient number

of samples before they can be used in a study.

4.3 Outlook

Since November 2021, the third and final version of

the software prototype has been released and is

running productively at the three partner universities.

This new version includes more and refined features

such as the inclusion recommendations for MOOCs

as motivated by Vogelsang et al. (2019), an

interactive tool for hierarchical goal setting as per

Weber et al. (2021b). The feature set has been

extended and improved, but the general database

structure has been maintained and refined. In the first

half of 2022, a new dataset with a compatible

structure will be published, which allows the reuse of

data analyses scripts and methods developed for the

dataset published along with this paper on new data.

Because the features the SIDDATA study

assistant software is offering are realized in the

SIDDATA backend, data from new LMS systems are

easy to integrate into the existing software. This

allows to expand SIDDATA’s reach beyond the three

partner universities and, potentially, to extend the

digital portfolio of many universities with low cost

and low maintenance requirements.

ACKNOWLEDGEMENTS

The authors acknowledge the financial support by the

Federal Ministry of Education and Research of

Germany for Siddata (project number 16DHB2124).

We thank all students who donated their data.

A Free and Open Dataset from a Prototypical Data-driven Study Assistant in Higher Education

161

REFERENCES

Castro, R. (2019), “Blended learning in higher education:

Trends and capabilities”, Education and Information

Technologies, Education and Information

Technologies, Vol. 24 No. 4, pp. 2523–2546.

Reinken, C., & Greiff, P. (2021). AIS Electronic Library

(AISeL) A Multi-Perspective Research for the

Investigation of Software Requirements for a Digital

Study Assistant

Schrumpf, J., Weber, F., & Thelen, T. (2021). A Neural

Natural Language Processing System for Educational

Resource Knowledge Domain Classification. In:

Kienle, A., Harrer, A., Haake, J. M. & Lingnau, A.

(Hrsg.), DELFI 2021. Bonn: Gesellschaft für

Informatik e.V. (S. 283-288).

Schurz, K., Schrumpf, J., Weber F., Lübcke M., Seyfeli, F.,

and Wannemacher, K., (2021). Towards a User-

Focused Development of A Digital Study Assistant

Through a Mixed Methods Design, in Sampson, D.G.,

Ifenthaler, D., Pedro, I., Mascia, M.L. (Eds.), 18th

International Conference on Cognition and Exploratory

Learning in the Digital Age (CELDA 2021). IADIS

Press

Thelen, T., König, C., Wannemacher, K., Wollersheim, H.-

W., Köhler, T., Igel, C., … Riedel, J. (2019). Digitale

Werkzeuge für Studienindividualisierung und

personalisierte Kompetenzentwicklung. In J. Hafer, M.

Martina, & M. Schumann (Eds.), Medien in der

Wissenschaft, Band 75, Teilhabe in der digitalen

Bildungswelt (pp. 258–262). Münster, New York:

Waxmann Verlag.

Vogelsang, K., Greiff, P., Tenspolde, C. and Hoppe, U.

(2019), “Agile by technique – The role of technology

enhanced learning in higher education ”, Beiträge Zur

Hochschulforschung, 41. Jahrgang, No. 3, pp. 28–46.

Weber, F., (2019). Goal Trees as Structuring Element in a

Digital Data-Driven Study Assistant, in Sampson, D.G.,

Ifenthaler, D., Pedro, I., Mascia, M.L. (Eds.), 16th

International Conference on Cognition and Exploratory

Learning in the Digital Age (CELDA 2019). IADIS

Press, Cagliari, Italy, pp. 413–416

Weber, F., Kernos, J., Grenz, M., and Lee, J., (2021a).

Towards a Web-Based Hierarchical Goal Setting

Intervention for Higher Education, in Sampson, D.G.,

Ifenthaler, D., Pedro, I., Mascia, M.L. (Eds.), 18th

International Conference on Cognition and Exploratory

Learning in the Digital Age (CELDA 2021). IADIS

Press.

Weber, F., & Le Foll, E. (2020). A Tagset for University

Students’ Educational Goals. In 17th International

Conference on Cognition and Exploratory Learning in

the Digital Age, CELDA 2020.

https://doi.org/10.33965/celda2020_202014l004

Weber, F., Osada, S., & Thelen, T. (2019). Towards a

Comprehensive Taxonomy of Study Goals of

University Students. Poster Session Presented at the

EuroCogSci 2019, 2125.

Weber, F., Schrumpf, J., and Thelen, T., (2021b).

Development of a Digital Goal Setting Companion for

Higher Education, Andrea Kienle et al. (Hrsg.): Die 19.

Fachtagung Bildungstechnologien (DELFI), Lecture

Notes in Informatics (LNI), Gesellschaft für

Informatik, Bonn 2021.

Weber, F., Schurz, K., Schrumpf, J., Seyfeli, F.,

Wannemacher, K., & Thelen, T. (2021c). Digitale

Studienassistenzsysteme Von der Idee zur Umsetzung

im Projekt SIDDATA. In H.-W. Wollersheim, M.

Karapanos, & N. Pengel (Eds.), Medien in der

Wissenschaft, Band 78, Bildung in der digitalen

Transformation (pp. 239–244). Münster, New York:

Waxmann Verlag.

Weber, F., and Thelen, T., (2021), Characterizing Personal

Educational Goals -Inter-rater Agreement on A Tagset

Reveals Domain-Specific Limitations of the External

Perspective, Ifenthaler, D., Sampson, D. G., Isaias, P.

(Eds.). (Forthcoming 2021). Balancing the Tension

between Digital Technologies and Learning Sciences.

New York, NY: Springer.

Weber, F., Schrumpf, J., Dettmer, N., Schurz, K., Thelen,

.T, (2022), "Siddata Study Assistant Dataset from

Prototpye 2.0", https://doi.org/10.26249/FK2/YYID

KE, osnaData, V1

APPENDIX

Statistics on student enrollment can be found for each

involved university by following the provided links:

Leibniz University of Hannover:

https://www.uni-hannover.de/de/universitaet/profil/

zahlen/studierendenstatistik/

University of Bremen:

https://www.finanzcontrolling.uni-bremen.de/daten/

index.htm

University of Osnabrück:

https://www.uni-osnabrueck.de/universitaet/zahlen

datenfakten/studierendenstatistiken/

CSEDU 2022 - 14th International Conference on Computer Supported Education

162