NEPOMUK D5.2 - Metadata Sharing and Recommendation Application - First Prototype

D5.2	Metadata Sharing and Recommendation Application - First Prototype
Publication date	17/12/2007
Dissemination level	Public
File	D5.2_v10_NEPOMUK_Metadata_Sharing_and_Recommendations.pdf

Executive summary

WP5000 investigates and exploits the knowledge exchange in social networks and provides tools and services for community identification and analysis as well as for supporting knowledge exchange in these semantic social networks. The goal of the second deliverable is to provide the basic functionality for metadata sharing and recommendation. The provided functionality is packaged in three components (as they have been identified in the Nepomuk architecture): i) Metadata Sharing, ii) Metadata Recommender, and iii) Expert Recommender. These components can be adopted and support several scenarios focusing on social interactions among desktops. The components? functionality is provided to other Nepomuk components or applications via well defined APIs, which have been tested by prototype applications.

Shortly, our work resulted in the following:

• The Metadata Sharing (1) component, being responsible for building a semantic overlay infrastructure based on the P2P (2) infrastructure provided by WP4000;
• The Metadata Recommender (3) component, which provides the user with recommendations built using the metadata in the network of Nepomuk peers. By analyzing the overlap between metadata and resources, it provides recommendation for: i) metadata enrichment, ii) metadata-based recommendation of new resources, and iii) recommendation for tags;
• The Expert Recommender (4) component, which provides the users with a list of persons who are experts on a given topic.

The main task of the Metadata Sharing component is to provide a semantic overlay infrastructure on top of a P2P overlay network which separates the logical layer - where data, schemas and schema mappings are managed - from the physical layer consisting of the structured P2P network. It fosters semantic interoperability through pair-wise schema mappings and query reformulation. Thus, heterogeneous but semantically related information sources can all be queried transparently using iterative query reformulation.

For the realization of the Metadata Sharing component:

• GridVine, the first metadata sharing system addressing simultaneously both scalability and semantic heterogeneity is described;
• A general overview of our infrastructure is provided with details on both the P2P access structure used for indexing information and routing messages and the higher semantic layer responsible for managing structured content;
• The decentralized information integration based on pair-wise schema mappings between heterogeneous but semantically related schemas is given;
• Several distributed query resolution mechanisms are discussed and we illustrate the performance of our infrastructure.

To accomplish the recommendation tasks provided by the Metadata Recommender and by the Expert Recommender:

• We describe how we can provide recommendations to enrich the metadata description of the current desktop resources using the shared metadata in the Nepomuk network of peers, and how we can provide recommendations of new resources similar to a given resource;
• We present a model for finding experts in a given topic using the resources on the desktop; our model also allows to retrieve both people and document together;
• We describe how the folksonomy structure of the Nepomuk network of peers can be used to find suitable tag recommendations for resources the user wants to annotate.

The implemented Metadata Recommender and Expert Recommender components allow the recommendation of new resources, persons, and of additional metadata using the Nepomuk P2P network.

Summarizing, we have developed the necessary functionalities for the semantic social networking and knowledge exchange. We have designed, analyzed and implemented methods for sharing of metadata and for recommending resources, persons, and metadata. In the rest of the project the focus will be put on extending the available functionalities, integration and deeper evaluation.

Together with Deliverable 5.2 we deliver a report on "Specification of Exchange Formats & Workflows for Metadata Sharing and Recommendations" which was written in the first year of the project. In the description of work it was not mentioned to be part of a deliverable but an intermediate report. We hereby hand this report in later, as promised to the reviewers. Since it has been written at the very beginning of the project, the terminology partly has changed. The report is delivered as an appendix to this document (cf. Appendix B).

Footnotes

(1) Located on the Nepomuk SVN Server https://dev.nepomuk.semanticdesktop.org in the directory /repos/trunk/java/org.semanticdesktop.nepomuk.comp.metadatasharing

(2) Abbreviations are listed in the Appendix A.

(3) Located on the Nepomuk SVN Server https://dev.nepomuk.semanticdesktop.org in the directory /repos/trunk/java/org.semanticdesktop.nepomuk.comp.metadatarecommender

(4) Located on the Nepomuk SVN Server https://dev.nepomuk.semanticdesktop.org in the directory /repos/trunk/java/org.semanticdesktop.nepomuk.comp.expertrecommender