TCS OPTRA

OPTRA is an opinion mining and tracking tool that enables creating actionable intelligence by mining unstructured and structured text available in online in blogs, discussion forums, websites, etc.

 

Challenges

Internet has facilitated rapid posting and consultation of feedback from consumers. Organizations have no choice but take proactive steps to gather business-critical intelligence from text generated across communication media such as, blogs, news posts, discussion forums, complaint forums and also initiate rapid customer feedback assimilation through online customer surveys. Given the humungous volumes of text, the need of the hour is for intelligent, automated techniques to derive insights from such unstructured text.
Consumer feedback is inherently noisy. From irregular sentence constructions to spelling errors, these may also contain information that is totally irrelevant to business. Automated feedback analysis can help decision makers identify causes for customer dissatisfaction, their desire and appreciation, which in turn can be converted to actionable intelligence. Segregating the relevant and irrelevant information as well as the possible impact of the extracted information are key challenges.  Since source data is constantly increasing, a near real-time analysis of high volumes of data is required, to be effective.

Solution Overview

TCS’ OPTRA employs goal-driven text-mining and integrated analytics to extract business-critical information from vast sources of noisy textual communication. The extracted knowledge can be presented to the end-user at different levels of abstraction. OPTRA is designed to perform integrated analytics over structured and unstructured data. The system employs a range of integrated analytical techniques that can lend organizations with a better understanding of their current business processes and performance.  Organizations can monitor customer reactions to their  products, services, policies and any other allied aspects through sentiment mining and trend analysis.
OPTRA is an in-house tool developed completely on JAVA which makes it run on almost any platform. It requires JRE 6.0 to run properly. It is a web-based application and requires a JAVA EE-based application server like Tomcat.Current Research

 

Benefits

• 360 degree visualization of data over different dimensions and granularity
• Goal-driven statistical and semantic analysis
• Ontology based report generation
• Business decision validation through trend analysis

The following are the research papers submitted on OPTRA:

• Lipika Dey, S.K. Mirajul Haque: Studying the effects of noisy text on text mining applications. AND 2009: 107-114
• Lipika Dey, S. K. Mirajul Haque: Opinion mining from noisy text data. IJDAR (2009): 205-226
• Lipika Dey, S. K. Mirajul Haque: Opinion mining from noisy text data. AND 2008: 83-90