Clustering in the Linked Open Data Cloud

 Organisatorisches

• Art der Arbeit:  Diplomarbeit / Masterarbeit
•      Status: ausgeschrieben
•      Institute:
  •             Computervisualistik         
  •             Informatik
•     Prüfungsamt-ID:
•     Interne BetreuerIn: Ansgar Scherp
•     Student:
•      Beginn:
•     Ende:

Beschreibung

The Linked Open Data cloud provides a rich source of background knowledge that can be used for searching and browsing different kind of information such as articles in DBpedia (a semantic version of Wikipedia), points of interests in Linked Geo Data, persons, events, and many others. However, the amount of data in this linked open data cloud exceeds today's support for an intelligent processing and management of the information. Clustering techniques are one possible means to alleviate this situation as they allow to aggregate data and possibly reduce the complexity of the information presented to the users.

In this thesis, different approaches for clustering and aggregation shall be applied on the Linked Open Data cloud. One approach for clustering the data is by means of distance metrics where the data contains, e.g., spatial information and/or time information. Other ways are based on linguistics, comparing the mere textual similarity of the name of two concepts, or semantics similarity that computes the (dis-)similarity of individuals based on their semantic annotation. The applicability of the approaches shall be evaluated and implemented in a generic library.