Investigation the Effect of Users’ Tagging Motivation on the Digital Educational Resources Metadata Descriptions

University of Piraeus Centre for Research and Technology – Hellas (CE.R.T.H.)
Department of Digital Systems Information Technologies Institute (I.T.I.)

Investigation the Effect of Users’ Tagging Motivation
on the Digital Educational Resources Metadata
Descriptions
Panagiotis Zervas1,2, Demetrios G Sampson1,2 and Maria
Aristeidou1
1
Department of Digital Systems, University of Piraeus

2
Information Technologies Institute, Centre for Research and Technology
Hellas

This work is licensed under the Creative Commons Attribution-NoDerivs-NonCommercial License. To view a copy of this license, visit
http://creativecommons.org/licenses/by-nd-nc/1.0 or send a letter to Creative Commons, 559 Nathan Abbott Way, Stanford, California
94305, USA.
P. Zervas, D. G. Sampson and M. Aristeidou 1/21 CELDA 2012, Madrid, Spain, October 2012


Presentation Outline
 Problem Definition
 Paper Contribution
 Related Studies
 Proposed Evaluation Methodology
 Applying the Proposed Evaluation Methodology
to an Existing Learning Object Repository (LOR)
 Conclusions – Future Work


Problem Definition (1/3)
 Several Open Educational Resources (OERs) initiatives have been emerged worldwide towards
the provision of open access to digital educational resources, in the form of Learning Objects
(LOs) [1].
 A key objective of OERs initiatives is to support the process of organizing, classifying and storing
digital educational resources and their associated metadata in web-based repositories which are
referred to as Learning Object Repositories (LORs). LORs are mainly developed to facilitate
search, retrieval and access to LOs through their metadata descriptions [2].
 Within this context, a popular way for describing digital educational resources is by using a
formal and centrally agreed classification system, such as the IEEE Learning Object Metadata
(LOM) [3].

[1] McGreal, R., 2008. A typology of learning object repositories. In: H.H. Adelsberger, Kinshuk, J. M. Pawlovski and D. Sampson, eds. International Handbook on
Information Technologies for Education and Training, pp. 5-18. 2nd Edition, Springer.
[2] Caswell, T., Henson, S., Jensen, M. and Wiley, D. (2008) ‘Open Educational Resources: Enabling universal education’, The International Review of Research in
Open and Distance Learning, Vol. 9 No.1, pp.1-11.
[3] IEEE Learning Technology Standards Committee (LTSC), 2002. Final Standard for Learning Object Metadata, [online] IEEE Learning Technology Standards
Committee. Available at: <http://ltsc.ieee.org/wg12/> [Accessed 28 August 2012].



 On the other hand, the emerging Web 2.0 applications allowed new ways of
characterizing digital educational resources, which moves from the expert-based
description based on formal classification systems to a less formal user-based tagging
[4].
 This new way of characterizing digital educational resources is referred to as Social
Tagging and is defined as the process of adding keywords, also known as tags, to any
type of digital resource by the users rather than the creators of the resources [5].
Moreover, the collection of tags created by the different users independently is
referred to as folksonomy [6].

[4] Derntl, M., Hampel, T., Motschnig-Pitrik, R. and Pitner, T., 2011. Inclusive social tagging and its support in Web 2.0 services. Computers in Human Behavior,
27(4), pp. 1460-1466.
[5] Bonino, S., 2009. Social Tagging as a Classification and Search Strategy. Germany: VDM.
[6] Bi, B., Shang, L. and Kao, B., 2009. Collaborative Resource Discovery in Social Tagging Systems. In: 18th ACM Conference on Information and Knowledge
Management, Hong – Kong, China, 2-6 November.



 In the field of Technology Enhanced Learning (TeL), a number of studies have been
reported aiming to evaluate (a) the anticipated added value of social tagging when
searching digital educational resources stored in LORs and compare it with the
traditional approach of searching based on expert-based formal description following
centrally agreed classification systems, such as IEEE LOM and (b) the enlargement of
educational resources possible description compared to the anticipated creatros’
descriptions [7], [8].
 Additionally, recent studies in the field of social tagging systems suggests that users’
tagging motivation has a direct influence on the properties of resulting tags and
folksonomies [9] but there are not existing studies for investigating this issue in the
field of TeL.
[7] Trant, J., 2009b. Tagging, Folksonomy and Art Museums: Results of steve.museum’s Research. [online] Archives & Museum Informatics. Available at:
<http://conference.archimuse.com/files/trantSteveResearchReport2008.pdf/> [Accessed 28 August 2012].
[8] Vuorikari, R. and Ayre, J., 2009. MELT Final Report of Phase II. [online] European Schoolnet. Available at: <http://info.melt-
project.eu/shared/data/melt/D5_5_PhaseIIreport_Final.pdf> [Accessed 28 August 2012].
[9] Korner, C., 2009. Understanding the motivation behind tagging. In: 20th ACM Conference on Hypertext and Hypermedia, Torino, Italy, 29 June - 1 July .



Paper Contribution
 In this paper we aim to investigate this issue and we propose a
methodology that aims to evaluate whether users’ tagging
behaviour can influence:
 the enlargement of metadata descriptions of digital educational
resources and
 the resulted folksonomy compared to the formal structured
vocabularies used by metadata experts or content providers for
characterizing digital educational resources.



Related Studies (1/2)
 An initial study has been conducted in the framework of the EU-funded project MELT (
http://info.melt-project.eu/). MELT project developed a LOR where LOs were characterized with
educational metadata (following IEEE LOM) by LOs’ authors, as well as with social tags (related
with LOs’ topic) by teachers. The main issues investigated were:
 Added value of social tags when searching for LOs: 23% of searches in MELT Repository
were performed based on social tags and 40% of these searches were found very useful by
the teachers. However, for accurate searches within MELT Repository, social tags were
considered not useful by the teachers.
 Added value of social tags for the enlargement of LOs description compared to authors’
descriptions: 25% of social tags contained additional to authors’ information, 26%
unnecessary information and 49% no new information. Moreover, social tags were found to
be more useful than formal vocabulary terms and most of the teachers wanted to change
the original metadata description to adopt some of the social tags as formal metadata
vocabulary terms.



Related Studies (2/2)
 Another study has been conducted in the framework of the US-funded project
“STEVE: The Museum Social Tagging Project” (http://www.steve.museum/). STEVE
project developed a repository with cultural heritage resources, which were
characterized with metadata by professional museum experts as well as with social
tags (related with resources’ topic) added by the teachers. The main issues
investigated were:
 Added value of social tags for the enlargement of cultural heritage resources
descriptions compared to museum experts’ descriptions: 86% of social tags
didn’t match with museum metadata added by professional museum experts
 Added value of social tags as formal museum metadata vocabulary terms: 88%
of social tags were evaluated and considered useful by the professional museum
experts



Proposed Evaluation Methodology (1/3)
 Step 1 – Identify different underlying behaviours for social taggers: we
adopt two types of social taggers motivations proposed by Korner [9]:
 Categorizers, who are motivated by categorization and use tags to
construct and maintain a navigational aid to the resources they annotate.
For this purpose, categorizers aim to establish a stable and bounded
vocabulary based on their personal preferences and motivation.
 Describers, who are motivated by description aim to describe the
resources they annotate accurately and precisely. As a result, their tag
vocabulary typically contains an open set of tags.



 In order to discriminate between categorizers and describers we adopt a set of measures proposed by Korner
et al. [10]:
 Tag/Resource Ratio: relates the vocabulary size of a user to the total number of digital educational Resources tagged by
this user. Describers, who use a variety of different tags for their resources, score higher values for this measure than
categorizers, who use fewer tags.
 Orphaned Tag Ratio: characterizes the degree to which users produce orphaned tags (that is tags assigned to few
resources only, and therefore are used infrequently). The orphaned tag ratio captures the percentage of tags in a user's
vocabulary that represent such orphaned tags. Categorizers vocabulary scores values closer to 0 because orphaned tags
would introduce noise to their personal vocabulary, whereas describers vocabulary scores values closer to 1 due to the fact
that describers tag resources in a more verbose and descriptive way.
 Overlap Factor: measures the phenomenon of an overlap produced by the assignment of more than one tag per resource
on average. Categorizers are interested in keeping this overlap relatively low. On the other hand, describers produce a high
overlap factor since they do not use tags for navigation but instead aim to best support later retrieval.
 Tag/Title Intersection Ratio: measures how likely users choose tags from the words of an educational resource title.
Categorizers use tags taken from the title and they score values closer to 1, whereas describers rarely use tags from the
title and they score values closer to 0.

[10] Korner, C., Benz, D., Strohmaier, M., Hotho, A. and Stumme, G., 2010. Stop thinking, start tagging - tag semantics emerge from collaborative verbosity. In:
19th International World Wide Web Conference (WWW 2010). Raleigh, NC 26-30 April. USA: ACM.



 Step 2 – Calculate similarity between social tags and educational metadata: we calculate the
similarity between social tags (offered by end-users, that is teachers) and educational metadata
(offered by metadata experts or content providers). The similarity is calculated for social tags
added by describers, as well as for social tags added by categorizers based on the users’
discrimination performed in step 1. At the end of this step, we expect to identify digital
educational resources enlarged with social tags offered by describers and/or categorizers that
are different by the formal metadata descriptions offered by metadata experts or content
providers.
 Step 3 - Compare folksonomy with formal vocabularies of educational metadata: we compare
the resulted folksonomy produced by the social tags with formal structured vocabularies of
educational metadata. The comparison is performed with the folksonomy produced by
describers, as well as with the folksonomy produced by categorizers following the users’
discrimination performed in step 1. At the end of this step, we would be able to identify new tags
offered by describers and/or categorizers that can enlarge the formal structured vocabularies of
educational metadata.


Applying Proposed Methodology
OpenScienceResources LOR
 We apply our proposed evaluation methodology to an existing
LOR, namely OpenScienceResources Repository (
http://www.osrportal.eu/).
 The OpenScienceResources Repository was developed in the
framework of an EU-funded project, referred to as
“OpenScienceResources: Towards the development of a Shared
Digital Repository for Formal and Informal Science Education”.
 It provides access to openly licensed (through Creative
Commons) science education resources, which can be used by
science teachers connecting formal science education in
schools with informal science education activities taken place
in European Science Centres and Museums [11]

[11] Sampson, D., Zervas, P. and Sotiriou, S., 2011. Science Education Resources Supported with Educational Metadata: The Case of the OpenScienceResources
Web Repository. Advanced Science Letters, Special Issue on Technology-Enhanced Science Education, 4(11/12), pp. 3353-3361 (9).



OSR LOR Tagging Approaches
Educational Metadata (according to IEEE LOM) Social Tags
Metadata Element Value Space Tags Categories Value Space
1.5 General.Keyword Free text describing the Free Tags Free text describing the
topic of the science topic and/or the subject
education resource domain of a science
9.2.2.2 Classification.Taxon Structured vocabulary education resource
Path.Taxon.Entry (when purpose describing the subject related with the science
metadata element has the value domain of the science curriculum
“discipline”) education resource related
with the science curriculum
9.2.2.2 Classification.Taxon Structured vocabulary Educational Structured vocabulary
Path.Taxon.Entry (when purpose describing the educational Objectives Tags describing the educational
metadata element has the value objectives that a science objectives that a science
“educational objective”) education resource intends education resource
to target (based on revised intends to target (based
Bloom’s Taxonomy) on revised Bloom’s
Taxonomy)



OSR LOR Dataset (March 2012)
Variables Value
Taggers 434
Tagged Science Education Resources 1877
Social Tags 14707
Free Tags 13117
Educational Objectives Tags 1590



Describers vs Categorizers (1/2)
 In order to cluster the Type of Taggers Value % per Total
Taggers
taggers of the
OpenScienceResources Categorizers 226 52,07 %

LOR we applied the set
Describers 208 47,92 %
of measures described in
our proposed Total 434 100,00 %

methodology in slide 10.



Describers vs Categorizers (2/2)
Type of Number of Tags Average Science Average Tagged Average Tags per
Taggers Taggers Contributed Tags per Education Science Education Science Education
Tagger Resources Resources per Resources
Tagged Tagger

Categorizers 226 1960 8,67 1852 8,19 1,05
Describers 208 12647 60,8 630 3,02 20,07

 Describers contributed the vast majority of social tags of the OpenScienceResources dataset but they
have tagged less science education resources than the categorizers.
 Describers mainly use a small set of digital educational resources, which they want to accurately and
precisely describe for later searching and retrieval.
 Categorizers contributed a small amount of social tags to a large set of digital resources aiming to
support future browsing to as many educational resources of the repository.



Educational Metadata vs Social Tags (1/2)
 We calculated:
 the similarity of the metadata values added by content providers to the element
Nr. 1.6 Keyword of the IEEE LOM standard with the free tags category of the social
tags added by describers and categorizers
 the similarity of the metadata values added by content providers to the element
Nr. 9.2.2.2 Classification.TaxonPath.Taxon.Entry (when purpose metadata
element has the value “educational objective”) of the IEEE LOM standard with the
educational objectives tags category of the social tags added by describers and
categorizers.
 The similarity threshold was selected by considering relevant thresholds used in other
social tagging evaluation studies from the literature [8].


Educational Metadata vs Social Tags (2/2)
Type of Keyword vs. Classification.Taxon Science Education % per total tagged
Taggers Free Tags Path.Taxon.Entry vs. Resources with low science education
Educational Objectives similarity score (<0,5) resources
Tags
Categorizers 0,79 0,71 184 out of 1852 9,93%
Describers 0,76 0,45 307 out of 630 48,73%

 Describers added social tags (mainly to the educational objectives tags category) that were significantly
different from the educational metadata added by the content providers. They significantly contributed
to the enlargement of the metadata descriptions of 307 science education resources.
 Categorizers added tags that were quite similar with the educational metadata added by the content
providers.
 This distinction between categorizers and describers is very important because it facilitates capturing
enlarged metadata descriptions of digital educational resources.



Folksonomy vs Formal Structured Vocabualry (1/2)
 We compared the resulted folksonomy of free tags category of social tags added by describers and
categorizers with the structured formal vocabulary used by content providers to characterize science
education resources with terms related with science curriculum:
 we calculated the similarity of the describers folksonomy and categorizers folksonomy with the
formal structured vocabulary and we kept those tags that they achieved similarity score zero. The
number of these tags was 202 (from describers) and 94 (from categorizers).
 We excluded the semantic noise from these tags, that is synonyms and subjective tags and we
concluded to 46 (describers) and 10 (categorizers) possible new terms that could enlarge the
structured formal vocabulary used at the element Nr. 9.2.2.2 Classification.TaxonPath.Taxon.Entry
(when purpose metadata element has the value “discipline”) of the IEEE LOM standard.
 We investigated whether there were new terms contributed by describers and/or categorizers.



Folksonomy vs Formal Structured Vocabualry (2/2)
Type of Taggers New terms Contributed % per Total Terms
Only by Describers 46 82,14 %
Only by Categorizers 10 17,85 %
By Categorizers & Describers 0 0,00 %

 Describers outperformed categorizers by contributing 36 more new terms.
 There were not common terms contributed by both describers and categorizers.
 These results provided us evidence that describers have a stronger influence on the enlargement of
formal structured vocabularies, whereas categorizers’ contribution was limited.
 These findings could be relevant in LORs for indexing digital educational resources with describers’ tags
towards facilitating search and retrieval of digital educational resources.



Conclusions – Future Work
 In this paper, we investigated the influence of different tagging styles, namely
describers and categorizers on enlarging the metadata descriptions of digital
educational resources.
 The results of our study performed with the dataset of an existing LOR, namely
OpenScienceResources Repository showed that describers produce tags that are
significantly different from formal metadata, whereas categorizers mainly follow the
formal metadata generated by metadata experts or content providers.
 Considering tagging motivation during folksonomies analysis could facilitate capturing
enlarged metadata descriptions of digital educational resources.
 Future work includes deeper analysis to the results of our study, so as to identify the
effect of describers’ and categorizers’ social tags to the enlargement of metadata
descriptions for digital educational resources with different granularity levels and
different formats.

Investigation the Effect of Users’ Tagging Motivation on the Digital Educational Resources Metadata Descriptions

More Related Content

What's hot (20)

Similar to Investigation the Effect of Users’ Tagging Motivation on the Digital Educational Resources Metadata Descriptions (20)

Recently uploaded (20)

Investigation the Effect of Users’ Tagging Motivation on the Digital Educational Resources Metadata Descriptions