Automating Knowledge Extraction from Unstructured Knowledge Articles

Customer Service Agents face many difficulties in finding suitable knowledge articles for resolution of customers' queries. This paper introduces Information Filtering System that helps customer service agents finding suitable knowledge articles as well as specific sections of text within knowledge articles for resolution of customers' queries through extraction of keywords from customers' interactions with customer service agents. This paper describes two approaches for resolution of queries, Coarse grain Approach that helps in finding Knowledge Article IDs and Fine grain approach helps in finding knowledge article ID and specific section with steps to be followed within that knowledge article using Domain Ontology Builder. This paper also describes two methods for annotation of Knowledge articles, dataset setup, experimental details of different algorithms with results, and list of benefits by using proposed system.