Resolving ambiguities in the semantic interpretation of natural language questions

Our project is about an e-librarian service which is able to retrieve multimedia resources from a knowledge base in a more efficient way than by browsing through an index or by using a simple keyword search. The user can formulate a complete question in natural language and submit it to the semantic search engine. However, natural language is not a formal language and thus can cause ambiguities in the interpretation of the sentence. Normally, the correct interpretation can only be retrieved accurately by putting each word in the context of a complete question. In this paper we present an algorithm which is able to resolve ambiguities in the semantic interpretation of NL questions. As the required input, it takes a linguistic pre-processed question and translates it into a logical and unambiguous form, i.e. ALC terminology. The focus function resolves ambiguities in the question; it returns the best possible interpretation for a given word in the context of the complete user question. Finally, pertinent documents can be retrieved from the knowledge base. We report on a benchmark test with a prototype that confirms the reliability of our algorithm. From 229 different user questions, the system returned the right answer for 97% of the questions, and only one answer, i.e. the best one, for nearly half of the questions.