LASER: a Linguistic enriched And Scalable Event attribute extRaction System

With the fast growth of web data and HUMINT reports, intelligent analysts need the capability to rapidly monitor and analyze event information in those massive amounts of unstructured textual data, in order to achieve and maintain persistent Situational Awareness (SA). Intelligent Automation, Inc., along with our collaborators, proposes to develop a Linguistic enriched And Scalable Event attribute extRaction System: LASER. There are three major innovations in LASER, firstly, it integrates even richer and more specialized features into each of the four classification models for the four event attributes, namely, modality, polarity, genericity and tense; secondly, it adopts robust classification models that can handle unbalanced class problem (which is the case in event attribute extraction); thirdly, it incorporates three post-correction approaches which are expected to bring more performance gains to event attribute extraction. LASER also leverages a state-of-the-art event extraction system which extracts relative high quality events such that event attributes can be further extracted. Furthermore, LASER uses powerful cloud computing architecture for information management and algorithmic computation.