Project: Twitter Data Research Public Opinion Mining• Developed sentiment analysis module to classify tweets as positive, negative or neutral.User Location Analysis• Developed a gazetteer based search system to identify location of the users.• The developed system takes a free form string given in the location field of twitter user profiles and outputs a best match city using heuristics along with a confidence score.Tweet Classification• Used naïve bayes and SVM classifiers to classify tweets as relevant and irrelevant w.r.t. a particular context.• Used bag of words features to train the classifiers.Topic Analysis• Performed n-gram analysis of the sets of tweets.• Used part of speech tagging and aggregated n-gram frequencies with exponential decay to generate hourly, daily and weekly trending topics.Project: Lead Generation from Web Automated Data Collection and Organization• Developed robust and reliable system to collect relevant data from a variety of web sources and social networks like Twitter.• Developed modules to organize data in structured and usable form from the unstructured raw data collected from web sources.• Designed database to efficiently store and retrieve required information.Lead Scoring• Developed a heuristic based scoring model for generating ranked list of leads.• Developed robust fuzzy matching algorithms to match accounts gathered from different sources and to match with client’s current account list.• Used n-gram analysis to identify important keywords/phrases for lead scoring.