Predicting Mental Health Problems with Automatic Identification of Metaphors

Algorithm 1

Framework of feature set generation for MSM.

	Input: The target text.
	Output: Metaphor-Sentiment feature set.
(1)	Identifying the metaphoricity of each word in the text, count the frequency, and generate metaphorical statistical features
(2)	Using Sentistrength to obtain the score of positive and negative emotions, and generate the statistical characteristics and sentiment uctuation value on sentence level
(3)	Determining the metaphorical words in the sentence by the sentiment information of the sentence to obtain the sentiment characteristics of the metaphorical words
(4)	Using SenticNet to get the word-level emotional scores of five dimensions, and calculating the average value to get the sentiment features of the text
(5)	Integrating the above characteristics, return Metaphor-Sentiment feature set