Challenge : For experience brands wanting to manage reputation managing and learning from data on review sites that is viewed by millions making it very important. However, collecting external data, cleaning it and making meaning out of the same is a highly difficult and expensive exercise.
External data offers critical insights into brand perception and competitive scenarios.
External data offers critical insights into brand perception and competitive scenarios.
Solution : We scrapped over 10 million customer reviews for more than 100000 restaurants across India. We focused on the following questions for analysing the reviews; • What are the keywords & topics of discussion across the comments? • What elements of the restaurant would they want improved? – Service, staff behaviour, ambience etc. • What is the overall mood of the customers visiting the restaurant? This project involved a beautiful concoction of text and structured data, which enriched the data manifold. We used advanced techniques of text analytics like, machine learning for topics classification, sentiment analysis, keyword parsing and relations extraction. We also created a module where restaurant collected reviews could also be analysed and compared against this thus leading to richer data and insights for decisions
Data Source : Restaurant listing and review sites in India
Tools used: Python and other open source tools
sign in at www.lunchbox.subcortext.com for more
