Semantic Search and Question-Answering Systems

Abstract

Keyword Search Systems have been in wide use since the availability of computers themselves. Keyword search returns the relevant results which contain the exact keywords used in a search query. But in today’s world of extensive information, a simple keyword search is bound to miss a lot of other relevant results which are phrased or written differently. This leads us to develop Semantic Search systems, which can also consider the context and meaning behind a search query, and return results which may not have the exact keywords, but still are relevant because of the intended meaning. This can be taken one-step further by introducing (Extractive) Question-Answering Systems, which if given a question-like query, can directly extract the short answer from the massive wealth of existing information. We explore the development and implementation of such a Semantic Search and Question-Answering system in the domain of finance for a financial product, AlgoFabric, at AlgoAnalytics Pvt. Ltd.