The strongExcitementstrong Open Platform for Textual Inferences.pdfVIP

The Excitement Open Platform for Textual Inferences ¨ Bernardo Magnini , Roberto Zanoli , Ido Dagan , Kathrin Eichler , Gunter Neumann , Tae-Gil Noh , Sebastian Pado , Asher Stern , Omer Levy FBK (magnini|zanoli@fbk.eu) Heidelberg, Stuttgart Univ. (pado|noh@cl.uni-heidelberg.de) DFKI (neumann|eichler@dfki.de) Bar Ilan University (dagan|sterna3|omerlevy@cs.biu.ac.il) Abstract and finally designed and implemented a rich envi- ronment to support open source distribution. This paper presents the Excitement Open The goal of the platform is to provide function- Platform (EOP), a generic architecture and ality for the automatic identification of entailment a comprehensive implementation for tex- relations among texts. The EOP is based on a modu- tual inference in multiple languages. The lar architecture with a particular focus on language- platform includes state-of-art algorithms, independent algorithms. It allows developers and a large number of knowledge resources, users to combine linguistic pipelines, entailment al- and facilities for experimenting and test- gorithms and linguistic resources within and across ing innovative approaches. The EOP is languages with as little effort as possible. For ex- distributed as an open source software. ample, different entailment decision approaches can share the same resources and the same sub- 1 Introduction