{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/surfacing-contextual-hate-speech-words-within","title":"Surfacing contextual hate speech words within social media","arxiv_id":"1711.10093","date":"2017-11-28","proceeding":null,"authors":["Jherez Taylor","Melvyn Peignon","Yi-Shin Chen"],"abstract":"Social media platforms have recently seen an increase in the occurrence of\nhate speech discourse which has led to calls for improved detection methods.\nMost of these rely on annotated data, keywords, and a classification technique.\nWhile this approach provides good coverage, it can fall short when dealing with\nnew terms produced by online extremist communities which act as original\nsources of words which have alternate hate speech meanings. These code words\n(which can be both created and adopted words) are designed to evade automatic\ndetection and often have benign meanings in regular discourse. As an example,\n\"skypes\", \"googles\", and \"yahoos\" are all instances of words which have an\nalternate meaning that can be used for hate speech. This overlap introduces\nadditional challenges when relying on keywords for both the collection of data\nthat is specific to hate speech, and downstream classification. In this work,\nwe develop a community detection approach for finding extremist hate speech\ncommunities and collecting data from their members. We also develop a word\nembedding model that learns the alternate hate speech meaning of words and\ndemonstrate the candidacy of our code words with several annotation\nexperiments, designed to determine if it is possible to recognize a word as\nbeing used for hate speech without knowing its alternate meaning. We report an\ninter-annotator agreement rate of K=0.871, and K=0.676 for data drawn from our\nextremist community and the keyword approach respectively, supporting our claim\nthat hate speech detection is a contextual task and does not depend on a fixed\nlist of keywords. Our goal is to advance the domain by providing a high quality\nhate speech dataset in addition to learned code words that can be fed into\nexisting classification approaches, thus improving the accuracy of automated\ndetection.","url_abs":"http://arxiv.org/abs/1711.10093v1","url_pdf":"http://arxiv.org/pdf/1711.10093v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"surfacing-contextual-hate-speech-words-within","repo_url":"https://github.com/JherezTaylor/hatespeech_codewords","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"community-detection","task_name":"Community Detection"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"hate-speech-detection","task_name":"Hate Speech Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.10093","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}