{"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/revisiting-the-importance-of-encoding-logic","title":"Revisiting the Importance of Encoding Logic Rules in Sentiment Classification","arxiv_id":"1808.07733","date":"2018-08-23","proceeding":"EMNLP 2018 10","authors":["Kalpesh Krishna","Preethi Jyothi","Mohit Iyyer"],"abstract":"We analyze the performance of different sentiment classification models on\nsyntactically complex inputs like A-but-B sentences. The first contribution of\nthis analysis addresses reproducible research: to meaningfully compare\ndifferent models, their accuracies must be averaged over far more random seeds\nthan what has traditionally been reported. With proper averaging in place, we\nnotice that the distillation model described in arXiv:1603.06318v4 [cs.LG],\nwhich incorporates explicit logic rules for sentiment classification, is\nineffective. In contrast, using contextualized ELMo embeddings\n(arXiv:1802.05365v2 [cs.CL]) instead of logic rules yields significantly better\nperformance. Additionally, we provide analysis and visualizations that\ndemonstrate ELMo's ability to implicitly learn logic rules. Finally, a\ncrowdsourced analysis reveals how ELMo outperforms baseline models even on\nsentences with ambiguous sentiment labels.","url_abs":"http://arxiv.org/abs/1808.07733v1","url_pdf":"http://arxiv.org/pdf/1808.07733v1.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":"revisiting-the-importance-of-encoding-logic","repo_url":"https://github.com/martiansideofthemoon/logic-rules-sentiment","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"sentiment-analysis","task_name":"Sentiment Analysis"},{"task_slug":"sentiment-classification","task_name":"Sentiment Classification"}],"methods":[{"method_slug":"bilstm","method_name":"BiLSTM"},{"method_slug":"elmo","method_name":"ELMo"},{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}