Principle-to-program: Neural methods for similar question retrieval in online communities

Abstract

Similar question retrieval is a challenge due to lexical gap between query and candidates in archive and is very different from traditional IR methods for duplicate detection, paraphrase identification and semantic equivalence. This tutorial covers recent deep learning techniques which overcome feature engineering issues with existing approaches based on translation models and latent topics. Hands-on proposal thus will introduce each concept from end user (e.g., question-answer pairs) and technique (e.g., attention) perspectives, present state of the art methods and a walkthrough of programs executed on Jupyter notebook using real-world datasets demonstrating principles introduced.