Efficient Tool for Diabetes Tracking through Layered Bigram Approach
7 Pages Posted: 29 May 2018
Date Written: February 7, 2018
Abstract
Today social media is playing a vital role on the users of internet. In twitter, social media, public posts tweets which include status, opinion sharing, discussion etc. These tweets may be related to different domains and health care is one of the prominent domain in which health related tweets will be shared by the users. The Domain chosen for this paper is “Health” to do opinion gathering about a particular Disease for a period of two years. Though strategic analysis of tweets data, can do predictions about the future and having the opinion to take preventive measures useful for public, Health Professionals, health Behavior Researchers (Health Education). This paper main focus is to make use of social media for Health Behavior Research like tracking the information of one particular disease “Diabetes” in detail using Bigram Text Classification. In this paper tweets are considered with the search word “Diabetes” contains fields like “User’s Opinion in text message”, source of the URL, Retweet count, Location source of the tweet etc.
The Tweets (text messages) are undergone Two-Level classification i.e. Bayesian Classification at the first level based on the unigram “diabetes” & Text Classification through Feature Extraction at the second level using “Term Tweet Classification Algorithm”. In this paper, some of the predictions regarding Diabetes are “People with Symptoms of Diabetes”, “People suffering from Type1 level Diabetes”, “People suffering from Type2 level Diabetes”, People Death information due to Diabetes” & “People using Insulin for Diabetes”.
Keywords: Twitter, Health Care, Diabetes, Bigram Text Classification, Health Data
Suggested Citation: Suggested Citation