Outline

Providing insights into the state, coverage and scope of available educational Linked Data

The lack of up-to-date and precise descriptive information has made this challenging.

The classification of datasets in the LD Cloud is highly specific to the resource types one is looking at.

Providing a systematic assessment of educational Linked Data which consider both, represented topics as well as resource types and their correlations.

Research Questions

Q1: Which types and topics are covered by existing educational Linked Data?

Q2: What are the central topics covered for particular types (e.g. Open Educational Resources metadata)?

Q3: Are certain topics underrepresented for certain types, or vice versa?

Methodology

This research is focused on datasets of the LinkedUp Catalog for which a topic profile is available

A topic profile connects a dataset with topics by applying elaboration algorithm to extracted resource samples. In this sense, a topic profile provides a comprehensive overview of the topic coverage of individual datasets.

Network Analysis metrics and approaches

Fetahu, B., Dietze, S., Nunes, B. P., Taibi, D., Casanova, M. A., Generating structured Profiles of Linked Data Graphs, 12th International Semantic Web Conference (ISWC2013), Sydney, Australia, (2013).

Fetahu, B., Dietze, S., Nunes, B. P., Casanova, M. A., Taibi, D., Nejdl, W., A Scalable Approach for Efficiently Generating Structured Dataset Topic Profiles, 11th Extended Semantic Web Conference (ESWC2014), Heraklion, Crete, Greece, (2014)

Networks

Dataset and Categories network

Dataset and Resource type network

Resource types and Categories network

Dataset and categories network

Dataset and categories network

Dataset and categories network

The effect of resource types on topic connections between datasets

Dataset and Resource type network

Dataset and Resource type network

Dataset and Resource type network

Resource type mapping

The effect of Mapping on resource types connections between datasets

The effect of Mapping on resource types connections between datasets

Resource types and categories network

Resource types and categories network

Interactive Explorer for educational Linked Data

http://data-observatory.org/led-explorer/

Taibi, D., Dietze, S., Fetahu, B., Fulantelli, G., Exploring type-specific topic profiles of datasets: a demo for educational linked data, in Poster & System Demonstration Proceedings of 13th International Semantic Web Conference (ISWC2014), Riva Del Garda, Italy, October 2014

Conclusions

Key findings of our study include: F1. Educational datasets can best be characterised

(profiled) by a combined representation of resource types and categories as part of dataset profiles

F2. The nature of categories differs significantly depending on the resource types they are associated with.

F3. Educational resource types can be characterised by their inherent topic distribution

F4. Educational resources, i.e. instances which represent some form of educational documents, currently are not equally spread across all disciplines. A topic bias exists towards fields in the area of Computer Science and the Life Sciences.

Ongoing works

Studying the relationships emerging from the inherent relatedness of DBpedia categories. For instance, if the dataset D1 refers to category Cx and dataset D2 refers to category Cy, the path between Cx and Cy in the DBpedia category graph (e.g. Cx might be a subcategory of Cy) might also hint at additional connections.

Investigating the possibilities to exploit the DBpedia category graph tied to specific resources as part of automated type and schema alignment methods. The intuition is that similar categories are likely to be tied to instances of similar resource types.

Thank you

Davide Taibi

National Research Council of Italy

Institute for Educational Technology

davide.taibi@itd.cnr.it