1aPh.D. Scholar, School of Management, RK University
2Professor and Head of Computer Department, Indian Institute of Social Welfare and Business Management, Kolkata, West Bengal, India
*Corresponding author email id: subhasis@praxis.ac.in
Online published on 16 September, 2015.
The purpose of this research study is to find the applicability of text mining concepts in marketing area to analyse consumer reviews from review sites. An approach has been tried to see if perceptual maps can be created from these reviews. Apart from this, the study tries to explain which features are important for making a model more liked by consumers.
In this study, total 4710 reviews are collected from 22 different smart phones using the web crawling technique. Afterwards, text mining procedures are applied on those reviews to convert the texts to structured data. After data preparation, sentiment analyses are done to extract sentiment scores for six different features which people talked about very frequently. These sentiment scores are used for developing perceptual maps and for classification modelling.
Based on the perceptual maps, it is seen how different models are positioned on the basis of three distinct dimensions extracted. Galaxy Note is found distinctly ahead of other phones in audio but in video quality, Xperia P is found very much preferred. From classification modelling, it is seen that battery backup is very important to improve positive perception about the phone and RAM is least important.
The models extracted could not be verified with structured questionnaire because getting above 4000 reviews for 22 different models is very costly. Another limitation is the way of extracting sentiment scores. The method is not applicable for ‘’hinglish’ (mixture of Hindi and English). Moreover, for complicated English as well, this method will not give reliable output and hence, it failed to judge iphone properly.
Text mining, Factor analysis, Sentiment score, Logistic regression, Perceptual Mapping
