{"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/linguistically-informed-self-attention-for","title":"Linguistically-Informed Self-Attention for Semantic Role Labeling","arxiv_id":"1804.08199","date":"2018-04-23","proceeding":"EMNLP 2018 10","authors":["Emma Strubell","Patrick Verga","Daniel Andor","David Weiss","Andrew McCallum"],"abstract":"Current state-of-the-art semantic role labeling (SRL) uses a deep neural\nnetwork with no explicit linguistic features. However, prior work has shown\nthat gold syntax trees can dramatically improve SRL decoding, suggesting the\npossibility of increased accuracy from explicit modeling of syntax. In this\nwork, we present linguistically-informed self-attention (LISA): a neural\nnetwork model that combines multi-head self-attention with multi-task learning\nacross dependency parsing, part-of-speech tagging, predicate detection and SRL.\nUnlike previous models which require significant pre-processing to prepare\nlinguistic features, LISA can incorporate syntax using merely raw tokens as\ninput, encoding the sequence only once to simultaneously perform parsing,\npredicate detection and role labeling for all predicates. Syntax is\nincorporated by training one attention head to attend to syntactic parents for\neach token. Moreover, if a high-quality syntactic parse is already available,\nit can be beneficially injected at test time without re-training our SRL model.\nIn experiments on CoNLL-2005 SRL, LISA achieves new state-of-the-art\nperformance for a model using predicted predicates and standard word\nembeddings, attaining 2.5 F1 absolute higher than the previous state-of-the-art\non newswire and more than 3.5 F1 on out-of-domain data, nearly 10% reduction in\nerror. On ConLL-2012 English SRL we also show an improvement of more than 2.5\nF1. LISA also out-performs the state-of-the-art with contextually-encoded\n(ELMo) word representations, by nearly 1.0 F1 on news and more than 2.0 F1 on\nout-of-domain text.","url_abs":"http://arxiv.org/abs/1804.08199v3","url_pdf":"http://arxiv.org/pdf/1804.08199v3.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":"linguistically-informed-self-attention-for","repo_url":"https://github.com/strubell/LISA","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"dependency-parsing","task_name":"Dependency Parsing"},{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"},{"task_slug":"part-of-speech-tagging","task_name":"Part-Of-Speech Tagging"},{"task_slug":"predicate-detection","task_name":"Predicate Detection"},{"task_slug":"semantic-role-labeling","task_name":"Semantic Role Labeling"},{"task_slug":"semantic-role-labeling-predicted-predicates","task_name":"Semantic Role Labeling (predicted predicates)"},{"task_slug":"word-embeddings","task_name":"Word Embeddings"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/predicate-detection-on-conll-2005","task":"Predicate Detection","dataset":"CoNLL 2005","model":"LISA","rank_in_archive_order":1,"of":2,"metrics":{"F1":"98.4"},"uses_additional_data":false},{"leaderboard":"/sota/predicate-detection-on-conll-2012","task":"Predicate Detection","dataset":"CoNLL 2012","model":"LISA","rank_in_archive_order":1,"of":1,"metrics":{"F1":"97.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-role-labeling-on-conll-2005","task":"Semantic Role Labeling","dataset":"CoNLL 2005","model":"LISA","rank_in_archive_order":12,"of":15,"metrics":{"F1":"86.04"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-role-labeling-predicted-predicates","task":"Semantic Role Labeling (predicted predicates)","dataset":"CoNLL 2005","model":"LISA + ELMo","rank_in_archive_order":1,"of":5,"metrics":{"F1":"86.90"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-role-labeling-predicted-predicates","task":"Semantic Role Labeling (predicted predicates)","dataset":"CoNLL 2005","model":"LISA","rank_in_archive_order":4,"of":5,"metrics":{"F1":"84.99"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-role-labeling-predicted-predicates-1","task":"Semantic Role Labeling (predicted predicates)","dataset":"CoNLL 2012","model":"LISA + ELMo","rank_in_archive_order":3,"of":7,"metrics":{"F1":"83.38"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-role-labeling-predicted-predicates-1","task":"Semantic Role Labeling (predicted predicates)","dataset":"CoNLL 2012","model":"LISA","rank_in_archive_order":6,"of":7,"metrics":{"F1":"82.33"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1804.08199","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1804.08199"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/strubell/LISA","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"unverified":14},"by_repo_kind":{"official":{"samples":14,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"e17ea0fc4873cd3e","entry":"accuracy_np","repo":"strubell/LISA","repo_kind":"official","path":"src/evaluation_fns_np.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/evaluation_fns_np.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"e17ea0fc4873cd3e"}},{"code_sha256_prefix":"a8b322b3e51e24de","entry":"accuracy_tf","repo":"strubell/LISA","repo_kind":"official","path":"src/evaluation_fns.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/evaluation_fns.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a8b322b3e51e24de"}},{"code_sha256_prefix":"9d94a9256be463e0","entry":"convert_bilou","repo":"strubell/LISA","repo_kind":"official","path":"src/evaluation_fns_np.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/evaluation_fns_np.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9d94a9256be463e0"}},{"code_sha256_prefix":"03c2cf6d3b1daf1b","entry":"convert_conll","repo":"strubell/LISA","repo_kind":"official","path":"src/evaluation_fns_np.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/evaluation_fns_np.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"03c2cf6d3b1daf1b"}},{"code_sha256_prefix":"d5c0e80895ba13a2","entry":"create_metric_variable","repo":"strubell/LISA","repo_kind":"official","path":"src/evaluation_fns.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/evaluation_fns.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"d5c0e80895ba13a2"}},{"code_sha256_prefix":"c2f375306505868b","entry":"get_default","repo":"strubell/LISA","repo_kind":"official","path":"src/constants.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/constants.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c2f375306505868b"}},{"code_sha256_prefix":"0e84978dc1139b65","entry":"get_params","repo":"strubell/LISA","repo_kind":"official","path":"src/attention_fns.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/attention_fns.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"0e84978dc1139b65"}},{"code_sha256_prefix":"caca64fca95608c5","entry":"int_to_str_lookup_table","repo":"strubell/LISA","repo_kind":"official","path":"src/nn_utils.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/nn_utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"caca64fca95608c5"}},{"code_sha256_prefix":"ea3b43b22b209a25","entry":"leaky_relu","repo":"strubell/LISA","repo_kind":"official","path":"src/nn_utils.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/nn_utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ea3b43b22b209a25"}},{"code_sha256_prefix":"413d686580cb441b","entry":"lowercase_converter","repo":"strubell/LISA","repo_kind":"official","path":"src/data_converters.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/data_converters.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"413d686580cb441b"}},{"code_sha256_prefix":"c7484be80663fcc6","entry":"map_strings_to_ints","repo":"strubell/LISA","repo_kind":"official","path":"src/dataset.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/dataset.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c7484be80663fcc6"}},{"code_sha256_prefix":"deaa9d2fd33c5349","entry":"parse_roots_self_loop_converter","repo":"strubell/LISA","repo_kind":"official","path":"src/data_converters.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/data_converters.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"deaa9d2fd33c5349"}},{"code_sha256_prefix":"bbf19f387d296223","entry":"set_vars_to_moving_average","repo":"strubell/LISA","repo_kind":"official","path":"src/nn_utils.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/nn_utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"bbf19f387d296223"}},{"code_sha256_prefix":"15dbbde62d34c21c","entry":"strip_conll12_domain_converter","repo":"strubell/LISA","repo_kind":"official","path":"src/data_converters.py","file_url":"https://github.com/strubell/LISA/blob/HEAD/src/data_converters.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"15dbbde62d34c21c"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}