SenSe Seminar: LeanBigData (31-01-2017)

A presentation by  Dr. MoulosKnowledge, Media and Distributed Systems Lab of the National Technical University of Athens

Tackling the challenges posed by Social Networking content and addressing its casual nature, n-gram graphs technique provides a language-independent supervised approach for text mining and sentiment analysis. One of the fields that the sentiment analysis was developed for, was to accommodate the need for evaluating the sentiment of public opinion on a subject that exists in the big unstructured data available on the internet. Some methods of sentiment analysis include statistical approaches using machine learning algorithms. The training of machine learning algorithms using features extracted from the texts provides a model for predicting the polarity of new texts.

Taking advantage of LeanBigData ( which targets at building an ultra-scalable and ultra-efficient integrated big data platform, a real-time sentiment analysis system was introduced which could be applied in cases like : elections (US Presidential Elections), brand monitoring, public awareness (case of earthquake, unclassified events/event detection), security awareness –disclose new threats (case of protests : Real time map where tourists or citizens could know where is it safe to go based on people sentiments).

Moreover, an anonymity storage framework for the tweets is used in order to comply with EU legislation.