Predicting Web User’s Tendency of Depression Using Negative Thought-Driven Depression Model

Abstract

　　Melancholia is problem in recent society People probably will pay attention to physical heath but mental health WHO said “DEPRESSION: A Global Crisis” in October 9 2012 Depression is a global public health concern Nowadays network already is a society of microcosm With mobile network become popular people prefer to post something on the internet including thinking mood feeling and trifles However we want to mine article on web and find who has depression tendency that we can remind the author and give immediately treatment 　　In this work we propose a Negative Thought-Driven Depression Model (NTDM) to predict depression tendency We find sign of depression and the goal is not diagnose depression Therefore we refer to diagnostic criteria and depression scale for finding features of depression We use negative thought negative emotion and symptom to predict depression tendency Furthermore we build three lexicons for NTDM there are negative thought lexicon negative emotion lexicon and symptom lexicon NTDM to predict depression tendency for single article We also propose NTDMlong to predict depression tendency for long term articles We collected some user’s articles from a well-known BBS station (PTT) to observe train and experiment These articles also were labeled depression tendency by experts 　　The experiment results showed the performance by using our model was better than using ENDE model and MPFA model We get conclusion that identifies the major negative thought to predict depression tendency is helpful even better than the model of major negative emotion

Date of Award	2015 Aug 5
Original language	English
Supervisor	Wen-Hsiang Lu (Supervisor)