In this work we are trying to build model that predicts the risk level of suicide based on users behavioural data.

• Task B: Risk Assessment for SW posters based on their SW postings and other Reddit postings

• Task_B_trainin.posts.csv contains 57016 records with these information[ post_id, user_id, subreddit]

• Crowd_train.csv This file contain the main risk level lables contains 994 records with 497 blank lables ?!

• 1. risk level a: 127 records (users).
• 2. risk level b: 50 records (users).
• 3. risk level c: 113 records (users).
• 4. risk level d: 206 records (users).

• Shared_task_posts.csv which contains 1048576 records [post_id,user_id,Timestamp, subreddit, post_title, post_body]

• Chen, Lushi, Abeer Aldayel, Nikolay Bogoychev, and Tao Gong. "Similar minds post alike: Assessment of suicide risk using a hybrid model." Proceedings of the Sixth Workshop on Computational Linguistics and Clinical Psychology. 2019.

• Mar 18 Preliminary system description paragraphs due (for our overview shared task paper)
• Mar 20-22 Test set downloadable. Output (and IRB approval or determination of Exempt status) due 4 days from your download timestamp. Submissions without documentation of IRB approval, exemption, or equivalent will not be evaluated.
• Mar 30 Final system description paragraphs due (no late submissions accepted)
• Mar 30 Short shared task papers + final system description paragraphs due (no late submissions accepted)

TimeFeatures.py: this script computes mean time intervals from any of the two posts in text from a certain topic, topics are defined by dictionaries we created. It also checks whether author mention sucidal methods

• Features.py: This file contains script to Construct the features based on the users set provided in the basic training file [Features extraction].

• empathFea.py: This script compute mean empath score on user level, we can select posts contain certain dictions and the empath variables we want to retain for these posts (e.g. we want to know the negative emotion scores when people mention mental health topics) The script also merge all the features genereted from empath by user level

• ExtractDictionsFromHTML.ipynb: this file extract dictions from webpages

• clpsych_model.py: This file contains script to train the basic ML models.

##Folders Description: dictionaries: dictionarities from different topics

data_sample_clpsy19: sample data

SampleShareTask: more data for feature observations (Posts from suicide watch in Jan, 2019 and all the posts of the authors from other subreddit)

• Empath: features extracted from empath and which empath feature might be useful for our classification task

• senti_value: This file read each user post and assign the sentiment value for each file to prepare it for next step of feature construction.

• reddit_scrape: This script read each user's postlist and provide the list of commenters.