An Expressive phrases identification supported with feature prediction consuming unstructured data collection

Abstract

It’s evident that, the rate of unstructured data is increasing from different sources of social media. In these days those data are being used by many researchers for their producing research results. The sentiment analysis is one of the key research perspective predicted by analysing unstructured data, but these analyses are majorly used for business intelligence. Hence, the huge unstructured data related to medical intelligence is not used properly. In this paper, the domain is introduced in sentiment analysis in the field of medical intelligence. Herewith the Major Depression Disorder (MDD) phrases are predicted by positive and negative polarity calculation. The phrases are framed by generating the UN-gram classification methodology, which is continuously splits the sentences to identify the exact emotional phrases. The experimental results are statistically proven by the probability distribution of n-gram classifier, which is compared with the sentiment tree bank generated with the value of perplexity and polarity distribution.