Machine learning techniques for effective text analysis of social network E-health data

The World Wide Web has evolved very drastically and the recent advent in social networking and media rich websites has necessitated analysis of the social networks and opinions expressed by various users in media like blogs, tweets, and website pages alike. While a lot of previous researches have been done on product / CRM domain, relatively few research has been done on the network analysis of social media with a focus on extraction and opinion of e-Health data i.e. data expressed in various blogs or website pages by users regarding various aspects of their health. While marketing and medical companies can leverage this data to augment customer reach and thus further their profits, sentiment analysis and network analysis on the data can help various organizations understand health patterns, address people's concerns or predict outbreak patterns in case of contagious diseases. This paper outlines the machine learning techniques which are helpful in the analysis of medical domain data from Social networks. In this paper we compare the existing techniques of machine learning, discuss the advantages and challenges encompassing the perspectives involving the use of text mining methods for applications in E-health and medicine. We evaluate medical domain data classification efficiency using various metrics like ROC, AUC implemented via R language packages. We need to infer which machine learning technique is more relevant for processing data from social networks having medical terms in it.