The aim of this project is to develop a search function based on textual inference, which can be used to find, for an incoming support request, relevant documents in an existing customer support system. Unlike state-of-the-art search technology, the system will analyze and compare the text of the support request and the documents in the system on the content rather than the keyword level.
Within the project, we will investigate how to analyze and represent texts on a semantic level to improve the search result. We will also develop a system that integrates external language resources and linguistic tools into the search process. In particular, we aim to integrate "textual entailment" components, which determine, whether the truth of one statement follows from the truth of another. These components will be developed as part of the ongoing EU project Excitement.
Software AG, Darmstadt