Natural Language Processing for Mental Disorders: An Overview

Abstract

least one of the keywords in our keywords list, which we discuss next.Criteria (1) and (2) above are automatic and implemented in software, i.e., one can directly download our code and run it to reproduce the output of criteria (1-2) using the same keywords we use.We now discuss each criterion in more detail. ACL Anthology (criterion 1)We use an authoritative index to search for the NLP publications to include in this survey.As stated in the ACL public website, the Association for Computational Linguistics "is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing (NLP)" 3 .The ACL Anthology indexes 62, 486 papers across journals, conferences, workshops and shared tasks, and is arguably the most important index for research publications in NLP. Keyword matching (criterion 2)We use exact substring match between a keyword and the title and abstract, and a paper is considered a match if a keyword is contained in either of the two.We choose keywords to identify papers dealing with mental disorders that have a high estimated burden and are the most epidemiologically prevalent (Whiteford et al., 2015).We use the keywords: "autism", "ASD" 4 , "dyslexia", "depression", "obsessive compulsive disorder", "obsessive-compulsive disorder", "OCD" 5 , "attention deficit and hyperactivity disorder", "ADHD" 6 , "schizophrenia", "down syndrome", "substance abuse", "tobacco abuse", "tabagism", "eating disorder", "anorexia", "bulimia", "mental health" and "developmental disorder". Manual validationNext, we manually validate all papers by reading their titles and abstracts and filtering out any incorrectly retrieved papers (false positives), leaving a total of 3 https://www.aclweb.org/portal/what-is-cl.Last access on February 1 st , 2021. 4 Abbreviation for Autism