Deliverable 6.1 Infrastructure for Extractive Summarization

Due to the overabundance of textual information on-line, automatic text summarization [Saggion and Poibeau, 2013], the reduction of a text to its essential content, is fundamental for information systems dealing with textual content. In recent years text summarization research has intensified with well known evaluation programmes promoting the interest in the area. Comparison of different summarization approaches, which is usually done by relying on well stablished or widely used or accesible systems such as MEAD [Radev et al., 2004], is of paramount importance in text summarization. However, MEAD only provides few summarization functionalities or features such as a position-based feature, a controid-based feature, and a first-sentence similarity feature which could be limited for comparison purposes. Availability of customizable natural language processing systems, and not only ready-made applications, are very important. In this deliverable we describe the infrastructure for text summarization we are relying on in SKATER. It will allow us to create different summarization applications to be used in SKATER’s demonstrators.