International Joint Conference on Artificial Intelligence (IJCAI) Macau - 2019

"Concept to Code: Aspect Sentiment Classification with Deep Learning"
Proposal is available here
Presentation is available File1    File2

Title: Concept-to-code: Aspect Sentiment Classification with Deep Learning

Authors:
Muthusamy Chelliah (muthusamy.c@flipkart.com)
Asif Ekbal (asif@iitp.ac.in)
Mohit Gupta (mohit.gupta@flipkart.com)

Keywords: [Artificial Intelligence] Knowledge Representation and Reasoning
[Artificial Intelligence] Machine Learning Applications
[Artificial Intelligence] Natural Language Processing

Abstract:
Aspect sentiment classification (ASC) is more fine-grained than document- or sentence- level tasks in sentiment analysis. Neural networks alleviate feature engineering, and attention mechanism in particular addresses targeted-context detection problem. LSTM and memory networks are 2 models which incorporate attention in recent literature for ASC. This tutorial is an advanced survey equally of interest to academic researchers and industry practitioners - very timely with so much vibrant research in the NLP community over the past 5 years. We not only review relevant concepts from papers across multiple research groups but also present code fragments which illustrate such techniques and could be leveraged in due course in use cases from online marketplaces like Flipkart where product reviews influence user purchases.