ConQA is an extension of the international project QALL-ME that is funded by the EC in its 6th framework. Whereas the EU-project QALL-ME mainly consists of application-oriented research in the area of open-domain question-answering (QA), the ConQA project will mainly concentrate on the realization of the achieved research results in an economically relevant application. ConQA will focus on the development of a domain-specific QA-service for a restricted but scalable knowledge area: the answering of questions about cultural events in the Saarland and adjacent regions, e.g., Luxembourg or Nancy. The questions to answer will not only cover fact-based questions about concrete events ("Where will be Jazz concerts next month?"), but also questions about background information ("Who is the lead singer of that Jazz band?") and opinion-based questions ("What negative reviews did this band get after the last gig?").
ConQA will focus on the following main aspects. On the one hand side, the focus will be on the development of an easy-to-use robust formulation of user questions which do not require from the user any background information about the actual fact basis nor functional details about the answer-engine technology. We will combine the advantage of the simplicity of standard search engine technology with the advantages of a semantic analysis of natural language questions. On the other hand side, we will perform an automatic semantic analysis and classification of relevant web-pages, which supports the precise answering of the questions. We are assuming that such a semantic analysis is performed dynamically by ontology-based information extraction which automatically constructs a corresponding database. The above mentioned simple and robust access will then be realized by means of automatically acquired natural language patterns that are uniquely define a mapping between natural language questions and SQL-based database queries. The realization of such a controlled QA interface will be based on textual entailment, a novel powerful mechanism of robust semantic inferencing.