Advance Language Model-based Sentence Retrieval in Question Answering Systems
Speaker: Saeedeh Momtazi
Institution: Saarland University
Abstract:
Information Retrieval (IR) engine is one of the essential components in Question Answering (QA) systems. Since in the QA systems people are searching for small nuggets of data, the necessity of reducing the retrieved documents to smaller parts such as sentences is obvious. In this research we propose some ideas for improving the performance of sentence retrieval in QA systems. For this purpose we are suggesting some language model-based techniques. Our methods contain higher level n-gram models, word relationship and Maximum Entropy (ME).
