Linguistic rule-based ontology-driven chatbot system

Abstract

Chatbots are designed to change the way system and users interact. As the IT-support system interaction behaves like developer asks a query regarding a project from system, then system searches and results the best suitable answer. Chatbots usually roam around two components of a conversational system—intent and entity. In order to build a generic chatbot, it is important to employ natural language understanding system and process machine to understand language as does we human interpret. In this research work, linguistic rules, ontology, and similarity indexing are performed in a uniform manner to build a generic chatbot system. Initially, linguistic rules have been implied to understand context in correspondence to detect intent and entity in user query. Then, ontology is used to map intent and entity in varying parts of question–answering system. Even, ontology helps in finding similarity among relations used in query and relations in its ontological structure while applying syntactic ambiguity resolution. Syntactic ambiguity resolution is used to edit distance, cosine similarity, N-gram matching, and semantic feature similarity as a set of text preprocessing, information retrieval, and similarity algorithms. System provides a confidence score to consume resolved entity, and one can safely uses resolved entity if its score is above than defined threshold value.