DAIICT at TREC RTS 2016: Live Push Notification and Email Digest

This paper describes the participation of Information Retrieval Lab(IRLAB) at DA-IICT Gandhinagar,India in Real-Time Summarization track TREC 2016. This year TREC RTS offered two tasks. In the first task, that is scenario A, our system will be monitoring continuous posts from Twitter public stream and push the relevant tweet for each interest profile to RTS evaluation broker. For the same, we have expanded interest profile using Word2vec training model with past 30 days tweets. We have calculated relevance score between tweets and expanded interest profile using Okapi BM25 model. For Scenario B, Email digest, we anticipated summarization problem as a clustering problem. In scenario A, we reported result in terms of Expected Gain EG-1(primary metric)=0.1708 and in scenario B we have achieved primary metric nDCG-1 = 0.1972.