{"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/identification-interpretability-and-bayesian","title":"Identification, Interpretability, and Bayesian Word Embeddings","arxiv_id":"1904.01628","date":"2019-04-02","proceeding":null,"authors":["Adam M. Lauretig"],"abstract":"Social scientists have recently turned to analyzing text using tools from\nnatural language processing like word embeddings to measure concepts like\nideology, bias, and affinity. However, word embeddings are difficult to use in\nthe regression framework familiar to social scientists: embeddings are are\nneither identified, nor directly interpretable. I offer two advances on\nstandard embedding models to remedy these problems. First, I develop Bayesian\nWord Embeddings with Automatic Relevance Determination priors, relaxing the\nassumption that all embedding dimensions have equal weight. Second, I apply\nwork identifying latent variable models to anchor the dimensions of the\nresulting embeddings, identifying them, and making them interpretable and\nusable in a regression. I then apply this model and anchoring approach to two\ncases, the shift in internationalist rhetoric in the American presidents'\ninaugural addresses, and the relationship between bellicosity in American\nforeign policy decision-makers' deliberations. I find that inaugural addresses\nbecame less internationalist after 1945, which goes against the conventional\nwisdom, and that an increase in bellicosity is associated with an increase in\nhostile actions by the United States, showing that elite deliberations are not\ncheap talk, and helping confirm the validity of the model.","url_abs":"http://arxiv.org/abs/1904.01628v1","url_pdf":"http://arxiv.org/pdf/1904.01628v1.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":"identification-interpretability-and-bayesian","repo_url":"https://github.com/adamlauretig/bwe","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null},{"paper_slug":"identification-interpretability-and-bayesian","repo_url":"https://github.com/adamlauretig/bwe_application_naacl_2019","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"word-embeddings","task_name":"Word Embeddings"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[{"method_slug":null,"method_name":"American"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}