DMIR

FolkPeers - Folksonomies in P2P systems

Project Coordinator: H. Sofia Pinto

Project Duration: From 25/11/2007 to 25/11/2009

Objectives: The new developments on the Semantic Web and in particular the new trend on Web 2.0 applications have a strong emphasis on user driven publishing and managing content platforms. Examples of such systems are Flickr (to store, search, sort and share photos), del.icio.us (to keep, share and discover favorite links), digg (to keep, classify, and share favorite news),etc. All of them are user driven social content websites that have a huge number of users sharing resources. Classification of shared resources is user-driven and lead to a new topic: Folksonomies.

The goal of this project is to develop an application for user driven social content management in the ontology area. In particular we aim at developing an ontology library based in P2P technology, that allows users to store, search, sort, share, discover and classify ontologies. Although there are a few Ontology Libraries, such as the Ontolingua Server, DAML Ontology Library, Schema Web, or Swoogle, none has all the necessary features and functionalities for effective, widespread and easy usage: (1) some are simple repositories, (2) others have ontologies represented in deprecated knowledge representation languages, and they all (3) lack essential functionality, such as ontology summarization, rating of ontologies, online graphical browsing, multiple-ontology search, ontology mapping and alignment among others. Moreover, there are several problems associated to centralized content management systems which recently have started to be substituted by distributed content management systems usually implemented on top of P2P systems. This current trend emerged first in Eduttela and was applied in Bibster and Oyster.

In the system to be developed in this project, the Peers make their local ontologies available to others and the whole library is composed by the sum of these local libraries. The local libraries are maintained by their owners which can add or remove ontologies, edit their meta-information, classify them. Their changes are propagated to the global system immediately. In what concerns P2P systems we will contribute to this area by looking for novel solutions - in particular we will investigate in resource and query propagation algorithms. Moreover other important problems also to be investigated in this project are:

• how can a social tagging system be implemented in a distributed system, since current systems are centralized? We will investigate and implement a solution for sharing information about users/resources/tags in a distributed system
• finding and analysis of relations between users/resources/tags? We will investigate solutions for existing systems, such as Flickr, del.icio.us and bibsonomy and then implement them in our system. For that we will use analysis and clustering algorithms to the data collected from these systems and try to identify evidences of emergent semantics.
• how can we have contextualized search? In social tagging systems, Flickr for instance, one looses the context when performing search in the resources. We will investigate solutions that will allow users to search for resources (through tags) without loosing its context and improve their navigation.
• suggestion of tags to resources. We will investigate information retrieval techniques to suggest tags to users. The solutions sought are to be as general as possible. For that we will develop them in more text intensive social tagging systems like digg and apply them to our system.

Prime Contractor: Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa (INESC-ID/INESC/IST/UTL)

Financed by: FCT