{"id":"https://openalex.org/W3168767448","doi":"https://doi.org/10.1145/3447548.3467307","title":"Why Attentions May Not Be Interpretable?","display_name":"Why Attentions May Not Be Interpretable?","publication_year":2021,"publication_date":"2021-08-13","ids":{"openalex":"https://openalex.org/W3168767448","doi":"https://doi.org/10.1145/3447548.3467307","mag":"3168767448"},"language":"en","primary_location":{"id":"doi:10.1145/3447548.3467307","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3447548.3467307","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5090022501","display_name":"Bing Bai","orcid":"https://orcid.org/0000-0002-6953-1948"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Bing Bai","raw_affiliation_strings":["Tencent Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Tencent Inc., Beijing, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112582835","display_name":"Jian Liang","orcid":"https://orcid.org/0000-0003-3890-1894"},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jian Liang","raw_affiliation_strings":["Alibaba Group, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Alibaba Group, Beijing, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100438476","display_name":"Guanhua Zhang","orcid":"https://orcid.org/0000-0003-1445-1817"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guanhua Zhang","raw_affiliation_strings":["Tencent Inc., Guangzhou, China"],"affiliations":[{"raw_affiliation_string":"Tencent Inc., Guangzhou, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100348617","display_name":"Hao Li","orcid":"https://orcid.org/0000-0002-6294-6761"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hao Li","raw_affiliation_strings":["Tencent Inc., Beijing, China"],"affiliations":[{"raw_affiliation_string":"Tencent Inc., Beijing, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102906188","display_name":"Kun Bai","orcid":"https://orcid.org/0000-0002-3773-5364"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kun Bai","raw_affiliation_strings":["Tencent Inc., Guangzhou, China"],"affiliations":[{"raw_affiliation_string":"Tencent Inc., Guangzhou, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5115695181","display_name":"Fei Wang","orcid":"https://orcid.org/0000-0001-5768-7323"},"institutions":[{"id":"https://openalex.org/I4387153466","display_name":"Weill Cornell Medicine","ror":"https://ror.org/02r109517","country_code":"US","type":"education","lineage":["https://openalex.org/I205783295","https://openalex.org/I4387153466"]},{"id":"https://openalex.org/I205783295","display_name":"Cornell University","ror":"https://ror.org/05bnh6r87","country_code":"US","type":"education","lineage":["https://openalex.org/I205783295"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Fei Wang","raw_affiliation_strings":["Weill Cornell Medicine, New York, NY, USA"],"affiliations":[{"raw_affiliation_string":"Weill Cornell Medicine, New York, NY, USA","institution_ids":["https://openalex.org/I205783295","https://openalex.org/I4387153466"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5090022501"],"corresponding_institution_ids":["https://openalex.org/I2250653659"],"apc_list":null,"apc_paid":null,"fwci":3.6711,"has_fulltext":false,"cited_by_count":50,"citation_normalized_percentile":{"value":0.94107042,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":97,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"25","last_page":"34"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9998000264167786,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9998000264167786,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9991000294685364,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11273","display_name":"Advanced Graph Neural Networks","score":0.996399998664856,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/interpretability","display_name":"Interpretability","score":0.9453320503234863},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7182202935218811},{"id":"https://openalex.org/keywords/uncorrelated","display_name":"Uncorrelated","score":0.594097375869751},{"id":"https://openalex.org/keywords/interpretation","display_name":"Interpretation (philosophy)","score":0.5936952829360962},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.5553325414657593},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4948822855949402},{"id":"https://openalex.org/keywords/phenomenon","display_name":"Phenomenon","score":0.47828325629234314},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.4416255056858063},{"id":"https://openalex.org/keywords/root","display_name":"Root (linguistics)","score":0.4314863979816437},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.1638321876525879},{"id":"https://openalex.org/keywords/epistemology","display_name":"Epistemology","score":0.1337474286556244}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.9453320503234863},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7182202935218811},{"id":"https://openalex.org/C169345407","wikidata":"https://www.wikidata.org/wiki/Q8216221","display_name":"Uncorrelated","level":2,"score":0.594097375869751},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.5936952829360962},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.5553325414657593},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4948822855949402},{"id":"https://openalex.org/C50335755","wikidata":"https://www.wikidata.org/wiki/Q483247","display_name":"Phenomenon","level":2,"score":0.47828325629234314},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4416255056858063},{"id":"https://openalex.org/C171078966","wikidata":"https://www.wikidata.org/wiki/Q111029","display_name":"Root (linguistics)","level":2,"score":0.4314863979816437},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.1638321876525879},{"id":"https://openalex.org/C111472728","wikidata":"https://www.wikidata.org/wiki/Q9471","display_name":"Epistemology","level":1,"score":0.1337474286556244},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3447548.3467307","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3447548.3467307","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.7099999785423279,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":41,"referenced_works":["https://openalex.org/W1967188199","https://openalex.org/W2032536435","https://openalex.org/W2068144165","https://openalex.org/W2112796928","https://openalex.org/W2113459411","https://openalex.org/W2132917208","https://openalex.org/W2144996578","https://openalex.org/W2150291618","https://openalex.org/W2155195660","https://openalex.org/W2170240176","https://openalex.org/W2250539671","https://openalex.org/W2253795368","https://openalex.org/W2265846598","https://openalex.org/W2282821441","https://openalex.org/W2562607067","https://openalex.org/W2564898401","https://openalex.org/W2626778328","https://openalex.org/W2750779823","https://openalex.org/W2885318751","https://openalex.org/W2906152891","https://openalex.org/W2950178297","https://openalex.org/W2950768109","https://openalex.org/W2951025380","https://openalex.org/W2951568144","https://openalex.org/W2962790223","https://openalex.org/W2963069010","https://openalex.org/W2963271116","https://openalex.org/W2963365341","https://openalex.org/W2963393688","https://openalex.org/W2963875806","https://openalex.org/W2964308564","https://openalex.org/W2966869984","https://openalex.org/W2970155250","https://openalex.org/W2970670424","https://openalex.org/W2970726176","https://openalex.org/W2972324944","https://openalex.org/W3034224415","https://openalex.org/W3035625410","https://openalex.org/W3081430124","https://openalex.org/W3102937163","https://openalex.org/W3138819813"],"related_works":["https://openalex.org/W2905433371","https://openalex.org/W2888392564","https://openalex.org/W4310278675","https://openalex.org/W4388422664","https://openalex.org/W4390569940","https://openalex.org/W4361193272","https://openalex.org/W2963326959","https://openalex.org/W4388685194","https://openalex.org/W4312407344","https://openalex.org/W2894289927"],"abstract_inverted_index":{"Attention-based":[0],"methods":[1,154,173],"have":[2],"played":[3],"important":[4],"roles":[5],"in":[6,25,102],"model":[7],"interpretations,":[8],"where":[9],"the":[10,18,96,105,108,129,171,177],"calculated":[11],"attention":[12,44,73,109,124,130,180],"weights":[13,45,110,131],"are":[14,57,132],"expected":[15],"to":[16,104,147,155],"highlight":[17,47],"critical":[19],"parts":[20],"of":[21,92,179],"inputs":[22],"(e.g.,":[23],"keywords":[24],"sentences).":[26],"However,":[27],"recent":[28,69],"research":[29],"found":[30],"that":[31,88,116,170],"attention-as-importance":[32],"interpretations":[33],"often":[34],"do":[35],"not":[36,78],"work":[37],"as":[38],"we":[39,86],"expected.":[40],"For":[41],"example,":[42],"learned":[43],"sometimes":[46],"less":[48],"meaningful":[49],"tokens":[50],"like":[51,65],"\"[SEP]\",":[52],"\",\",":[53],"and":[54,56,151],"\".\",":[55],"frequently":[58],"uncorrelated":[59],"with":[60],"other":[61],"feature":[62],"importance":[63,136],"indicators":[64],"gradient-based":[66],"measures.":[67],"A":[68],"debate":[70],"over":[71],"whether":[72],"is":[74,95],"an":[75],"explanation":[76],"or":[77],"has":[79],"drawn":[80],"considerable":[81],"interest.":[82],"In":[83],"this":[84,93,157],"paper,":[85],"demonstrate":[87],"one":[89,144],"root":[90],"cause":[91],"phenomenon":[94],"combinatorial":[97,141],"shortcuts,":[98,142],"which":[99],"means":[100],"that,":[101],"addition":[103],"highlighted":[106],"parts,":[107],"themselves":[111],"may":[112],"carry":[113],"extra":[114],"information":[115],"could":[117],"be":[118],"utilized":[119],"by":[120],"downstream":[121],"models":[122],"after":[123],"layers.":[125],"As":[126],"a":[127],"result,":[128],"no":[133],"longer":[134],"pure":[135],"indicators.":[137],"We":[138,159],"theoretically":[139],"analyze":[140],"design":[143],"intuitive":[145],"experiment":[146],"show":[148,169],"their":[149],"existence,":[150],"propose":[152],"two":[153],"mitigate":[156],"issue.":[158],"conduct":[160],"empirical":[161],"studies":[162],"on":[163],"attention-based":[164],"interpretation":[165],"models.":[166],"The":[167],"results":[168],"proposed":[172],"can":[174],"effectively":[175],"improve":[176],"interpretability":[178],"mechanisms.":[181]},"counts_by_year":[{"year":2025,"cited_by_count":23},{"year":2024,"cited_by_count":12},{"year":2023,"cited_by_count":10},{"year":2022,"cited_by_count":5}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}