Traversing the Linking Open Data Cloud to Create News from Tweets

摘要

We propose a two-fold approach that is able to both consume and exploit semantics encoded in the Linking Open Data (LOD) cloud, and create news that document events reported in micro-blogging posts that correspond to documentary tweets. A documentary tweet is similar to a newspaper headline and reports an incident or event. Knowledge extracted from documentary tweets are used to develop a story line which will be augmented with RDF facts consumed from the LOD cloud. The resulting news content is represented in RDF using the rNews Ontology, facilitating news generation and retrieval. We study effectiveness of our approach with respect to a gold standard of manually tagged tweets. Initial experimental results suggest that our techniques are able to generate content that reflects up to 76.38% of the manually tagged terms.