Open AccessHelpfulness Prediction of Product Assessments using Machine Learning
Author(s) -
Kagolanu Trishul,
S.R. Naidu
Publication year - 2019
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.b1068.0782s719
Subject(s) - helpfulness , random forest , metadata , computer science , product (mathematics) , machine learning , artificial intelligence , information retrieval , natural language processing , data science , world wide web , mathematics , psychology , social psychology , geometry
Paper Customers express their opinion on products through reviews. Since there will be a lot of reviews that will be posted, only those reviews which are helpful should be made accessible to the customer. Hence, helpfulness of review needs to be predicted. This work categorizes the features into reviewer, review text and review metadata. Machine Learning algorithms Linear Regression and Random Forests are used for prediction of helpfulness using these features. It is observed that rating of a review has the highest influence on predicting helpfulness followed by user average rating deviation, difficult words and positive words. This work defines the features such as stem sim length and lem sim length which are derived from the product description which have performed reasonably well. Using all the features with Random Forests algorithm for prediction gave the best performance in automatically predicting helpfulness.