Speaker: Maria Pontiki, researcher at Institute for Language and Speech Processing, Athens, Greece
Abstract: Aspect-based Sentiment Analysis (ABSA) methods analyze large amounts of unstructured texts and extract (coarse- or fine-grained) knowledge not explicitly stated in users’ text or included in the user ratings. The variety of the methods that have been proposed for ABSA are usually not directly comparable, since they adopt different representation schemes and focus on different domains. This talk will provide a thorough overview of ABSA based on the ABSA challenge experience in the context of the International Workshop on Semantic Evaluation (SemEval 2014, 2015, and 2016), providing for the first time a common framework for ABSA methods. The talk will focus on knowledge representation, annotation and evaluation issues and on the new ABSA framework that was designed and introduced in the context of the SemEval ABSA task. An overview of the current state of the art ABSA methods and techniques will also be presented. The talk will conclude with a snapshot of our work in progress that exploits the ABSA representation framework in a variety of applications and projects with a special focus on social and life sciences.