Project

Excitement

EXploring Customer Interactions through Textual EntailMENT

EXploring Customer Interactions through Textual EntailMENT

  • Duration:

Identifying semantic inference relations between texts is a major underlying language processing task, needed in practically all text understanding applications. For example, Question Answering and Information Extraction systems should verify that extracted answers and relations are indeed inferred from the text passages. While such apparently similar inferences are broadly needed, there are currently no generic semantic "engines" or platforms for broad textual inference. Rather, annotation tools exist for narrow semantic tasks (i.e. they consider one phenomenon at a time and one single fragment of text at time), and systems have to independently assemble and augment them to obtain a complete inference process. Our primary scientific motivation is to change this ineffective setting and offer encompassing textual inference capabilities.

Our second, industrial-oriented, motivation is within the text analytics market. We focus on the customer interaction domain, which today spans multiple channels such as speech, email and social media. This growing market shows increasing demand for automatically analyzing customer inputs to harness their value. A major stumbling block, however, is the current inability to perform effective inferences over complete customer statements, rather than just keyword-based or topical analysis.

Accordingly, we set dual goals for our project. The first is to develop a generic multi-lingual platform for textual inference, based on the successful textual entailment paradigm, and make it available to the scientific and technological communities. This will enable diverse applications to leverage the open platform for their inference needs, while sharing around it the development of core semantic technology. Our second goal is to leverage the inference platform to develop a new generation of unsupervised text exploration technology for customer interactions, enabling business users to better grasp their diverse and often unpredicted content.

Partners

  • NICE Systems (coordinator), Israel
  • Bar Ilan University, Israel
  • Fondazione Bruno Kessler - FBK, Italy
  • Deutsches Forschungszentrum für Künstliche Intelligenz GmbH - DFKI, Germany
  • University of Heidelberg, Germany
  • Almawave S.r.l,m Italy
  • OMQ GmbH, Germany

Publications about the project

Tae-Gil Noh, Sebastian Padó, Vered Shwartz, Ido Dagan, Vivi Nastase, Lili Kotlerman, Meni Adler,

In: Proceedings of *SEM 2015. Joint Conference on Lexical and Computational Semantics (*SEM-15) located at NAACL 2015 June 4-5 Denver Colorado United States ACL 2015.

To the publication
Bernardo Magnini, Ido Dagan, Sebastian Pado, Günter Neumann

In: 17th International Conference on Text, Speech and Dialogue. International Conference on Text, Speech and Dialogue (TSD-2014) 17th September 8-11 Brno Czech Republic Pages 11-18 Lecture Notes in Computer Science (LNCS) 8655 ISBN 978-3-319-10815-5 (Print) 978-3-319-10816-2 (Online) Springer 9/2014.

To the publication
Günter Neumann, Aleksandra Gabryszak

In: Proceedings of the Third Joint Conference on Lexical and Computational Semantics. Joint Conference on Lexical and Computational Semantics (*SEM-14) located at COLING August 23-24 Dublin Ireland COLING 8/2014.

To the publication

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz