Question Answering Chatbot for Troubleshooting Queries based on Transfer Learning
Zeeshan Haque Syed, Asma Trabelsi, Emmanuel Helbert, Vincent Bailleau, Christian Muths
Abstract
Open-Domain Question Answering (ODQA) is a technique for finding an answer to a given query from a large set of documents. In this paper, we present an experimentation study to compare ODQA candidate solutions in the context of troubleshooting documents. We mainly focus on a well known open-source framework which is called Haystack. This framework comprises two key components which are the Retriever and the Reader. The Haystack Framework comes with several Retriever-Reader combinations and the choice of the best one is still unanswered till now. In this paper, we conduct an experimentation study to compare different Retriever-Reader combinations. Our aim is to come up with the best combination of components in regard to the speed and the processing power within the context of troubleshooting queries.