OPINION MINING ON ADVERSE DRUG REACTIONS IN SOCIAL MEDIA USING MACHINE LEARNING TECHNIQUES

Abstract

Social media attracts millions of users to share their interests and opinions with other users in the flexible way. Medical organizations started to share their information about newly developed drugs through social media. It provides flexible platform for millions users to share their opinions about the newly developed drugs. With that information various peoples who suffered from disease might get an opinion about them. Analyzing various kind of drugs information posted by the millions of users is a most difficult task which is researched by various authors. In the existing work, two-step analysis framework is implemented which focuses on positive and negative sentiment extraction. It is done for the purpose of ascertaining user opinion of cancer treatment. It is done by using a self-organizing map to analyze word frequency data derived from user’s forum posts. However in the existing system is a static model where only the older posts that are posted by the users online previously are considered which might lead to less accurate detection of consumer opinions. The existing lacks in accurate prediction of user opinions. This is resolved in the proposed research method by introducing Dynamic Drug Data Analysis using Hybrid Fuzzy C Means with Transductive Support Vector Machine (DDDA-HFCM-TSVM) by focusing on user opinions from the user reviews to refine the product based on user opinions. This is implemented in the social media framework where more number of consumers post their reviews based on the drug product. In the proposed research twitter social network is considered for retrieving the user posted opinions. The main goal of the proposed research work is to support the dynamic retrieval of data from the twitter social web site from which drugs related information would be identified. The overall evaluation of the proposed research work is done in the matlab simulation environment from which it is proved that the proposed research methods leads to provide better result than the existing research methods.