Social media sentiment analysis for local Kenyan products and services
It has become a common practice on the web for a consumer to learn how others like or dislike a product before buying, or for a manufacturer or service provider to keep track of customer opinions on its products so as to improve the user satisfaction. However, as the number of reviews available for any given product or service grows, it becomes harder and more difficult for people to understand and evaluate what the majority opinions about the product are. In this work, we attempt to resolve this problem for customer care service center in a Kenyan context by applying Sentiment analysis on people’s opinions expressed on social media. Sentiment Analysis or opinion mining attempts to resolve this problem by first presenting the user with an aggregate view of the entire data set, summarized by a label or a score, and secondly by segmenting the opinions/sentiments into three classes (positive, negative and neutral) that can be further explored as desired. We began by developing a module for facilitating searching, extraction and storage of opinions from social media. This was followed by development and training of a polarity classifier using the Python Natural Language Toolkit (NLTK) where we used the Naïve Bayes machine learning technique. The study went ahead to develop a web based application that integrates Facebook Graph API, Twitter API and the developed classifier to provide functionality for extracting, classifying and presenting classification results of data obtained from social media in a manner that can give a user with summarized as well as detailed view of customer opinion about a service or product.