Applications of Text mining in G-Square’s Bigdator

G-Square Solutions applies Text Analytics in its Flagship product ‘Bigdator’. Bigdator uses Latent Dirichlet Allocation Model for Topic Modelling, Latent Semantic Analysis Model for summarizing a long article or comments and Support Vector Machine (SVM) model for sentiment analysis. Bigdator is built on Python 2.7, database used is MongoDB.

Bigdator consists of three modules namely:

  1. Newsrator: Newsrator module does Text Mining on the news articles based on current market affairs as well as historical news. It provides sentiment score on each news articles and collectively, highlights the overall Rating, Positive count, Negative Count, Sentiment Score, etc for the selected news articles as shown in the below screenshot. The business users can filter their search, based on Company, Company key people, Regulators, etc. and can also save their reports for future analysis.There is a functionality called as ‘EWS Report’ which helps the business users with Early Warning Signals related to its stakeholders and investors, which helps them to make better decisions and avoid risks. It also has a wordcloud and elaborative graphs for broader understanding of the analysis for the end users.2. Sociometor – Sociometor module does Text Mining on the unstructured data flowing from social media websites like Facebook and Twitter. The UI consist of graphical representation of the sentiment analysis of people who have posted or commented on various banks as shown in the below screenshot. On clicking the details buttons, the user can get a detailed analysis of a particular bank or any other company which includes graphs, pie charts, wordcloud, topic mining, etc. We’re also working on a functionality which will help the business users to track the ID’s of people who’ve complained or complimented the bank for its services as well as track the future prospects for the bank. This functionality will help the bank to reach to each of its customer and interact with them.

3. Textrator – Textrator module is used for analyzing and summarizing an unstructured text document. The text or csv document can have any kind of unstructured data within it, like, comments of people availing the services of the bank, sourced from Facebook or Twitter, or, a random news article. Textrator Analyzer can analyze the intent and sentiment of the comment and the summarizer can summarize bullet points of a long article or a set of comments as shown in the below screenshot.

Follow

Leave a Reply

Your email address will not be published. Required fields are marked *