This demo paper introduces LODr, a service providing semanticenrichment features for existing tagged content from various Web 2.0 services, based on the MOAT and Linked Data principles. A proposal for semantically-enhanced tagging While tagging is widely deployed on Web 2.0 websites, it raises various issues which have been largely studied and mainly consist in tags ambiguity and heterogeneity, as well as the lack of organisation between them [3]. While it may not be a problem regarding personal tagging, it becomes relevant when trying to discover and retrieve content that have been tagged by others. Our approach to solve these issues consists in letting people give meaning to their tags using URIs of Semantic Web resources, especially reference URIs from the Linking Open Data initiative [1]. This means modeling facts as: "When I tag this picture 'apple', I mean http://dbpedia.org/resource/Apple Records, i.e. the record label, not the fruit". Such vision of semantically-enhanced tagging has been recently published through MOAT [5], which consists in (1) an ontology to represent relationships between tags and resources URIs, extending the Tag Ontology [4] and (2) an open-source and collaborative framework to de ne and share those relationships within a community and help people to bridge the gap between tagging and semantic indexing, without directly facing RDF modeling1. This way of tagging content with URIs o er various advantages, as solving ambiguity and heterogenity issues by dealing with machine-understandable URIs rather than words. Most important, it makes tagged data enters the Semantic Web, being interlinked with other resources (DBpedia concepts, FOAF URIs ...), that can be used to retrieve and browse related content.
While our rst experiments with MOAT have been done in a corporate context2, we decided to extend the approach and implemented LODr { http://lodr.info {, a personal open-source application that allows one to re-tag his existing Web 2.0 content and weave it into the Semantic Web thanks to the previous principles.
LODr is thus not yet another tagging service, but a system that provide users a way to semantically enrich existing tagged data that have been created thanks to their favourite tools. The system is based on a set of wrappers (currently available for 5 di erent services including Flickr and Slideshare), that parse the RSS feeds of user's data, extract items and related tags and translate it to RDF using SIOC [2] and the Tagging Ontology. The data is then stored into a local triple-store and for each tagged item, the user can browse it and give meaning to its tags, using relationships that have been de ned by the community, as depicted in Fig. 1. To ease the process of chosing the right meaning, human-readable labels can be displayed instead of URIs. When no URI have been previously de ned or when existing ones do not correspond to the meaning of the tag in the current context, a new URI can be added, directly or using the Sindice search widget3.
As LODr is based on the MOAT principles, it requires interaction with a dedicated tag server that stores the relationships between tags and URIs for the community that uses it. While a default public server is available, a community can use the tool with its own tag server which might be useful, for instance, in a company. LODr is completely Semantic-Web based, and its RDF backend is powered by ARC24. It features tagcloud and conceptcloud interfaces, di erent ways to browse items (all items, re-tagged items ...) and use Exhibit5 for faceted browsing. The whole template uses RDFa6, so that content can be easily discovered and crawled by dedicated semantic search engine. Moreover, as we wanted to o er value-added services to end-users, we wrote a dedicated Ubiquity7 command to discover tagged items when browsing the Web, e.g. to nd all items linked to a DBpedia URI when browsing the related Wikipedia page.
Finally, as various augmented-tagging applications have been recently
published, we think the originality of LODr resides in: (