MS Project Defense: Zenghui Yan, "Exploring Deep Learning Method for Question Retrieval in Community Question Answering"

Date: 
Monday, March 20, 2017 - 9:00am
Location: 
1132 Harold Frank Hall
Title: 
Exploring Deep Learning Method for Question Retrieval in Community Question Answering
Speaker: 
Zenghui Yan
Committee: 
Xifeng Yan (Chair), and William Wang

Abstract

Community question answering such as Stack Overflow and AskUbuntu has become a popular online service recent years. However, with the increase of the CQA archive, it accumulates massive duplicated questions on the CQA websites.  Therefore, retrieving similar questions is become more and more important.

Nowadays, deep learning has become popular in everywhere. Convolutional Neural Network(CNN) and Recurrent Neural Network(RNN) are two most popular deep learning models, while Long Short Term Memory(LSTM) is the most popular Recurrent Neural Network model.

This project involves survey and comparison of CNN and LSTM model in question retrieval task. In this project, CNN and LSTM models are implemented using tensorflow, and AskUbuntu dataset is used to evaluate these two models. 

Everyone welcome!