Simplified Statement Extraction Using Machine Learning Techniques

The automatic generation of basic, factual questions from a single sentence of text is a problem in the field of natural language processing (NLP) that has received a considerable amount of attention in the past five years. Some studies have suggested splitting this problem into two parts: first, decomposing the source sentence into a set of smaller, simple sentences, and then transforming each of these sentences into a question. This paper outlines a novel method for the first part, combining two techniques recently developed for related NLP problems. Our method uses a trained classifier to determine which phrases of the source sentence are potential answers to questions, and then creates different compressions of the sentence for each one.