Interpretation of Product Quality by Rule based and Machine Learning Approaches Using Opinion Mining

EasyChair Preprint 6145

Abstract

It is no longer important to have direct interaction with people to know their opinion regarding the product they are using.  Everything is available online in commercial and social networking platforms. To analyze the emotion of the customer or a public individual based on the tweet or review to find out how he thinks the product is a difficult task. In this work, we detect the emotion /opinion of a consumer with respect to a particular product, by using different methodologies and algorithms and analyze which is the most optimal solution for unstructured data. The main aim of this paper is 1) To test the traditional topic models like LDA and LSA by VADER algorithm which is a Rule Based approach and 2) To test the Machine Learning techniques like CNN-LSTM and Bi-LSTM and obtain the best suitable model in terms of loss, accuracy and also to find a model with less training and testing time i.e., operating time for the given unstructured data.

Keyphrases: Bi-LSTM, CNN with LSTM, Opinion Mining, Topic Modelling Techniques, VADER

@booklet{EasyChair:6145,
  author    = {Madhapur Vineelaswathi and Chuan-Ming Liu},
  title     = {Interpretation of Product Quality by Rule based and Machine Learning Approaches Using Opinion Mining},
  howpublished = {EasyChair Preprint 6145},
  year      = {EasyChair, 2021}}