IRIT at INEX 2012: Tweet Contextualization

In this paper, we describe an approach for tweet contextualization developed in the context of the INEX 2012. The task was to provide a context up to 500 words to a tweet from the Wikipedia. As a baseline system, we used TF-IDF cosine similarity measure enriched by smoothing from local context, named entity recognition and part-of-speech weighting presented at INEX 2011. We modified this method by adding bigram similarity, anaphora resolution, hashtag processing and sentence reordering. Sentence ordering task was modeled as a sequential ordering problem, where vertices corresponded to sentences and sequential constraints were represented by sentence time stamps.