Web3.0 and Language Resources. Marta Sabou. Knowledge Media Institute (
KMi). The Open University, Milton Keynes, United Kingdom. R.M.Sabou@open.
ac.
Web3.0 and Language Resources Marta Sabou Knowledge Media Institute (KMi) The Open University, Milton Keynes, United Kingdom
[email protected]
The Web is constantly evolving to meet the requirements of its increasingly important role in our society. The first, largely textual generation of the Web has evolved into the so-called social Web, or Web2.0, where content is primarily contributed by users in the form of wiki pages (most notably Wikipedia), blogposts and various tag-annotated multimedia resources (images, URL’s, video) in social sharing sites (e.g., Flickr). Technology experts predict another transformation into Web3.0. – a social Web that leverages from the benefits of large-scale, semantic-annotations provided by the Semantic Web (SW). Indeed, in parallel to the Web2.0 movement, the intensifying SW initiative is generating a large body of semantic data (ontologies and annotations), which is now available online and easily accessible through Google-like semantic Web gateways. The importance of the Web for language technology as a large-scale, heterogeneous and up-to-date collection of data has been recognized early on. The large body of research in using Web1.0. for the benefit of LRs has already been extended with research focusing on Web2.0 data. For example, Wikipedia and folksonomy tagspaces are used to estimate relatedness between terms [1, 2]. We believe that besides deepening research on the frontier of Web2.0 and LRs, the next important wave is in exploring Web3.0. resources. In our lab, we are carrying out pioneering work on exploring and combining resources specific to Web3.0., namely social and semantic web data. Specifically, we exploit SW data in the form of online ontologies to derive semantic relations between terms. This technology has been successfully applied in a variety of SW related tasks that traditionally would rely on established LRs such as WordNet: ontology matching (finding relations between the terms of two ontologies) and ontology evolution (extending a given ontology with new terms). In both tasks we obtained high precision values (above 70%), but coverage is still limited in some topic domains [3]. We have also applied this research to folksonomy tagspaces, with the aim to transform semantically weak, tag-based annotations of media resources into a semantically richer structure. We accomplish this both by relying on traditional LRs such as WordNet but also by employing SW ontologies. Specifically, our algorithms (FLOR and Scalet) associate tags with ontological entities and link them to one another through a variety of semantic relations. The result is a richer annotation for media resources, which can support more powerful forms of search and browsing. An initial evaluation of such enrichment process has yielded high precision but a relatively low recall, due primarily to the aforementioned relative sparseness of semantic web data. Initial user based evaluation of the enhanced resource search functionalities has produced positive feedback [4].
In conclusion, it is our view that the results from these initial experiments on using Web3.0 resources as novel LRs encourage further research in this direction. In particular, we need to explore how to combine these various resources with more traditional LRs in order to obtain optimal results. For our particular focus on exploiting online ontologies this could lead to increasing recall values. Additionally, we need to identify methods for exploring these novel LRs, possibly by adapting methods used for more traditional LRs (e.g., WordNet). [1] M. Strube, S.P. Ponzetto, Simone P. WikiRelate! Computing Semantic Relatedness Using Wikipedia. In Proc. of the 21st National Conference on Artificial Intelligence, 2006. [2] G. Stumme , D. Benz , C. Cattuto , A. Hotho. Semantic Grounding of Tag Relatedness in Social Bookmarking Systems. In Proc. of ISWC, 2008. [3] M. Sabou, M. d’Aquin, E. Motta. Exploring the Semantic Web as Background Knowledge for Ontology Matching. Journal on Data Semantics, XI, 2008. [4] S. Angeletou, M. Sabou, E. Motta. Improving folksonomy search with FLOR. Submitted for peer review, 2009.