Cognitive natural language search using calibrated quantum mesh

Abstract

This paper describes the application of a search system for helping users find the most relevant answers to their questions from a set of documents. The system is developed based on a new algorithm for Natural Language Understanding (NLU) called Calibrated Quantum Mesh (CQM). CQM finds the right answers instead of documents. It also has the potential to resolve confusing and ambiguous cases by mimicking the way a human brain functions. The method has been evaluated on a set of queries provided by users. The relevant answers given by the Coseer search system have been judged by three human judges as well as compared to the answers given by a reliable answering system called AskCFPB. Coseer performed better in 57.0% of cases, and worse in 16.5% cases, while the results were comparable to AskCFPB in 26.6% of cases. The usefulness of a cognitive computing system over a Microsoft-powered key-word based search system is discussed. This is a small step toward enabling artificial intelligence to interact with users in a natural manner like in an intelligent chatbot.