{"id":"https://openalex.org/W4385284112","doi":"https://doi.org/10.1145/3610611","title":"SelfSeg: A Self-supervised Sub-word Segmentation Method for Neural Machine Translation","display_name":"SelfSeg: A Self-supervised Sub-word Segmentation Method for Neural Machine Translation","publication_year":2023,"publication_date":"2023-07-26","ids":{"openalex":"https://openalex.org/W4385284112","doi":"https://doi.org/10.1145/3610611"},"language":"en","primary_location":{"id":"doi:10.1145/3610611","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3610611","pdf_url":null,"source":{"id":"https://openalex.org/S4306421405","display_name":"ACM Transactions on Asian and Low-Resource Language Information Processing","issn_l":"2375-4699","issn":["2375-4699","2375-4702"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Asian and Low-Resource Language Information Processing","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"http://hdl.handle.net/2433/286918","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5033608663","display_name":"Haiyue Song","orcid":"https://orcid.org/0000-0003-1159-0918"},"institutions":[{"id":"https://openalex.org/I22299242","display_name":"Kyoto University","ror":"https://ror.org/02kpeqv85","country_code":"JP","type":"education","lineage":["https://openalex.org/I22299242"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Haiyue Song","raw_affiliation_strings":["Kyoto University, Japan"],"raw_orcid":"https://orcid.org/0000-0003-1159-0918","affiliations":[{"raw_affiliation_string":"Kyoto University, Japan","institution_ids":["https://openalex.org/I22299242"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077362985","display_name":"Raj Dabre","orcid":"https://orcid.org/0000-0003-0664-3421"},"institutions":[{"id":"https://openalex.org/I90023481","display_name":"National Institute of Information and Communications Technology","ror":"https://ror.org/016bgq349","country_code":"JP","type":"facility","lineage":["https://openalex.org/I90023481"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Raj Dabre","raw_affiliation_strings":["National Institute of Information and Communications Technology, Japan"],"raw_orcid":"https://orcid.org/0000-0003-0664-3421","affiliations":[{"raw_affiliation_string":"National Institute of Information and Communications Technology, Japan","institution_ids":["https://openalex.org/I90023481"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102757632","display_name":"Chenhui Chu","orcid":"https://orcid.org/0000-0001-9848-6384"},"institutions":[{"id":"https://openalex.org/I22299242","display_name":"Kyoto University","ror":"https://ror.org/02kpeqv85","country_code":"JP","type":"education","lineage":["https://openalex.org/I22299242"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Chenhui Chu","raw_affiliation_strings":["Kyoto University, Japan"],"raw_orcid":"https://orcid.org/0000-0001-9848-6384","affiliations":[{"raw_affiliation_string":"Kyoto University, Japan","institution_ids":["https://openalex.org/I22299242"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028836340","display_name":"Sadao Kurohashi","orcid":"https://orcid.org/0000-0001-5398-8399"},"institutions":[{"id":"https://openalex.org/I22299242","display_name":"Kyoto University","ror":"https://ror.org/02kpeqv85","country_code":"JP","type":"education","lineage":["https://openalex.org/I22299242"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Sadao Kurohashi","raw_affiliation_strings":["Kyoto University, Japan"],"raw_orcid":"https://orcid.org/0000-0001-5398-8399","affiliations":[{"raw_affiliation_string":"Kyoto University, Japan","institution_ids":["https://openalex.org/I22299242"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5033318800","display_name":"Eiichiro Sumita","orcid":"https://orcid.org/0000-0002-1028-4399"},"institutions":[{"id":"https://openalex.org/I90023481","display_name":"National Institute of Information and Communications Technology","ror":"https://ror.org/016bgq349","country_code":"JP","type":"facility","lineage":["https://openalex.org/I90023481"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Eiichiro Sumita","raw_affiliation_strings":["National Institute of Information and Communications Technology, Japan"],"raw_orcid":"https://orcid.org/0000-0002-1028-4399","affiliations":[{"raw_affiliation_string":"National Institute of Information and Communications Technology, Japan","institution_ids":["https://openalex.org/I90023481"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.7931,"has_fulltext":true,"cited_by_count":5,"citation_normalized_percentile":{"value":0.76272342,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":"22","issue":"8","first_page":"1","last_page":"24"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":1.0,"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/T10181","display_name":"Natural Language Processing Techniques","score":1.0,"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":0.9998999834060669,"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/T12016","display_name":"Web Data Mining and Analysis","score":0.9868999719619751,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/artificial-intelligence","display_name":"Artificial intelligence","score":0.6806396245956421},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.673143208026886},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6570535898208618},{"id":"https://openalex.org/keywords/machine-translation","display_name":"Machine translation","score":0.6516638398170471},{"id":"https://openalex.org/keywords/translation","display_name":"Translation (biology)","score":0.6489720344543457},{"id":"https://openalex.org/keywords/word","display_name":"Word (group theory)","score":0.6070836782455444},{"id":"https://openalex.org/keywords/text-segmentation","display_name":"Text segmentation","score":0.541946530342102},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.5099886655807495},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.34678536653518677},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3304167687892914},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.15000340342521667},{"id":"https://openalex.org/keywords/chemistry","display_name":"Chemistry","score":0.07285124063491821}],"concepts":[{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6806396245956421},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.673143208026886},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6570535898208618},{"id":"https://openalex.org/C203005215","wikidata":"https://www.wikidata.org/wiki/Q79798","display_name":"Machine translation","level":2,"score":0.6516638398170471},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.6489720344543457},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.6070836782455444},{"id":"https://openalex.org/C98501671","wikidata":"https://www.wikidata.org/wiki/Q1948408","display_name":"Text segmentation","level":3,"score":0.541946530342102},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.5099886655807495},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.34678536653518677},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3304167687892914},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.15000340342521667},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.07285124063491821},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C105580179","wikidata":"https://www.wikidata.org/wiki/Q188928","display_name":"Messenger RNA","level":3,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3610611","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3610611","pdf_url":null,"source":{"id":"https://openalex.org/S4306421405","display_name":"ACM Transactions on Asian and Low-Resource Language Information Processing","issn_l":"2375-4699","issn":["2375-4699","2375-4702"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Asian and Low-Resource Language Information Processing","raw_type":"journal-article"},{"id":"pmh:oai:irdb.nii.ac.jp:01221:0006026428","is_oa":true,"landing_page_url":"http://hdl.handle.net/2433/286918","pdf_url":null,"source":{"id":"https://openalex.org/S7407056385","display_name":"Institutional Repositories DataBase (IRDB)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I184597095","host_organization_name":"National Institute of Informatics","host_organization_lineage":["https://openalex.org/I184597095"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"ACM Transactions on Asian and Low-Resource Language Information Processing","raw_type":"journal article"},{"id":"pmh:oai:arXiv.org:2307.16400","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2307.16400","pdf_url":"https://arxiv.org/pdf/2307.16400","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"pmh:oai:irdb.nii.ac.jp:01221:0006026428","is_oa":true,"landing_page_url":"http://hdl.handle.net/2433/286918","pdf_url":null,"source":{"id":"https://openalex.org/S7407056385","display_name":"Institutional Repositories DataBase (IRDB)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I184597095","host_organization_name":"National Institute of Informatics","host_organization_lineage":["https://openalex.org/I184597095"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"ACM Transactions on Asian and Low-Resource Language Information Processing","raw_type":"journal article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G317285154","display_name":"Multilingual corpus construction and domain adaptation for low-resource machine translation","funder_award_id":"22KJ1724","funder_id":"https://openalex.org/F4320334764","funder_display_name":"Japan Society for the Promotion of Science"},{"id":"https://openalex.org/G5742032516","display_name":"Neural Machine Translation Based on Bilingual Resources Extracted from Multimodal Data","funder_award_id":"19K20343","funder_id":"https://openalex.org/F4320334764","funder_display_name":"Japan Society for the Promotion of Science"},{"id":"https://openalex.org/G832091749","display_name":"\u610f\u56f3\u3092\u7684\u78ba\u306b\u4f1d\u3048\u308b\u97f3\u58f0\u5bfe\u8a71\u7ffb\u8a33\u306e\u57fa\u76e4\u6280\u8853\u306e\u5275\u51fa","funder_award_id":"23K28144","funder_id":"https://openalex.org/F4320334764","funder_display_name":"Japan Society for the Promotion of Science"}],"funders":[{"id":"https://openalex.org/F4320324891","display_name":"Iran Telecommunication Research Center","ror":"https://ror.org/01a3g2z22"},{"id":"https://openalex.org/F4320334764","display_name":"Japan Society for the Promotion of Science","ror":"https://ror.org/00hhkn466"},{"id":"https://openalex.org/F4320335839","display_name":"National Institute of Information and Communications Technology","ror":"https://ror.org/016bgq349"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":73,"referenced_works":["https://openalex.org/W46679369","https://openalex.org/W222053410","https://openalex.org/W343636949","https://openalex.org/W1516184288","https://openalex.org/W1522301498","https://openalex.org/W1524281572","https://openalex.org/W1603598191","https://openalex.org/W1753482797","https://openalex.org/W2016856586","https://openalex.org/W2025768430","https://openalex.org/W2100495367","https://openalex.org/W2100664567","https://openalex.org/W2118434577","https://openalex.org/W2121879602","https://openalex.org/W2123301721","https://openalex.org/W2123442489","https://openalex.org/W2124807415","https://openalex.org/W2130942839","https://openalex.org/W2133564696","https://openalex.org/W2154099718","https://openalex.org/W2326925005","https://openalex.org/W2440757793","https://openalex.org/W2487442924","https://openalex.org/W2559655401","https://openalex.org/W2573761686","https://openalex.org/W2592647456","https://openalex.org/W2595715041","https://openalex.org/W2600463316","https://openalex.org/W2612690371","https://openalex.org/W2613904329","https://openalex.org/W2757041753","https://openalex.org/W2799087757","https://openalex.org/W2891475925","https://openalex.org/W2891546148","https://openalex.org/W2933138175","https://openalex.org/W2951559648","https://openalex.org/W2952125979","https://openalex.org/W2952288254","https://openalex.org/W2952468927","https://openalex.org/W2952487930","https://openalex.org/W2962732637","https://openalex.org/W2962784628","https://openalex.org/W2962801832","https://openalex.org/W2963250244","https://openalex.org/W2963324947","https://openalex.org/W2963341956","https://openalex.org/W2963420272","https://openalex.org/W2963426332","https://openalex.org/W2963532001","https://openalex.org/W2963887123","https://openalex.org/W2963979492","https://openalex.org/W2965373594","https://openalex.org/W2971081948","https://openalex.org/W2998353611","https://openalex.org/W3001434439","https://openalex.org/W3034789084","https://openalex.org/W3034999214","https://openalex.org/W3035207248","https://openalex.org/W3035252911","https://openalex.org/W3098654368","https://openalex.org/W3126910220","https://openalex.org/W3156299562","https://openalex.org/W3172246349","https://openalex.org/W3175212568","https://openalex.org/W3197737114","https://openalex.org/W4226271314","https://openalex.org/W4288025088","https://openalex.org/W4288089799","https://openalex.org/W4289440907","https://openalex.org/W4292779060","https://openalex.org/W4298340572","https://openalex.org/W4385245566","https://openalex.org/W4394666973"],"related_works":["https://openalex.org/W2883671469","https://openalex.org/W2978383222","https://openalex.org/W2172629291","https://openalex.org/W2380773642","https://openalex.org/W2384559435","https://openalex.org/W2337707338","https://openalex.org/W2393940967","https://openalex.org/W2058548953","https://openalex.org/W2785359773","https://openalex.org/W2550455130"],"abstract_inverted_index":{"Sub-word":[0],"segmentation":[1,50,90,168],"is":[2,53,97,266,294],"an":[3],"essential":[4],"pre-processing":[5],"step":[6],"for":[7,140],"Neural":[8],"Machine":[9],"Translation":[10],"(NMT).":[11],"Existing":[12],"work":[13],"has":[14],"shown":[15],"that":[16,52,132],"neural":[17,48],"sub-word":[18,49],"segmenters":[19],"are":[20,28],"better":[21],"than":[22,184],"Byte-Pair":[23],"Encoding":[24,202],"(BPE),":[25],"however,":[26],"they":[27,31],"inefficient,":[29],"as":[30,69],"require":[32],"parallel":[33,65],"corpora,":[34],"days":[35],"to":[36,40,56,121,136,273,283,296],"train,":[37],"and":[38,58,87,113,157,192,194,204,225,248,252,259,271],"hours":[39],"decode.":[41],"This":[42],"article":[43],"introduces":[44],"SelfSeg,":[45],"a":[46,71,77,100,129,195,218,226,279],"self-supervised":[47],"method":[51,181,215,265],"much":[54],"faster":[55,268,275],"train/decode":[57],"requires":[59],"only":[60],"monolingual":[61,291],"dictionaries":[62],"instead":[63],"of":[64,76,107,147,165,236],"corpora.":[66],"SelfSeg":[67,108],"takes":[68],"input":[70],"word":[72,84,111,117],"in":[73,154,278],"the":[74,83,89,92,123,134,145,163,176,233,256],"form":[75],"partially":[78],"masked":[79],"character":[80],"sequence,":[81],"optimizes":[82],"generation":[85],"probability,":[86,95],"generates":[88],"with":[91,190],"maximum":[93],"posterior":[94],"which":[96],"calculated":[98],"using":[99,166],"dynamic":[101],"programming":[102],"algorithm.":[103],"The":[104,170,213],"training":[105,124,270],"time":[106],"depends":[109],"on":[110,175,211,243,255],"frequencies,":[112],"we":[114,127,150,161],"explore":[115],"several":[116],"frequency":[118],"normalization":[119],"strategies":[120],"accelerate":[122],"phase.":[125],"Additionally,":[126],"propose":[128],"regularization":[130,214],"mechanism":[131],"allows":[133],"segmenter":[135],"generate":[137],"various":[138],"segmentations":[139],"one":[141],"word.":[142],"To":[143],"show":[144],"effectiveness":[146],"our":[148,180,264],"approach,":[149],"conduct":[151],"MT":[152],"experiments":[153],"low-,":[155],"middle-,":[156],"high-resource":[158,280],"scenarios,":[159],"where":[160],"compare":[162],"performance":[164],"different":[167],"methods.":[169],"experimental":[171],"results":[172,254],"demonstrate":[173],"that,":[174],"low-resource":[177],"ALT":[178],"dataset,":[179],"achieves":[182,216],"more":[183],"1.2":[185,227],"BLEU":[186,220,228],"score":[187,197,221,229],"improvement":[188,198,222,230],"compared":[189,282],"BPE":[191,224],"SentencePiece,":[193],"1.1":[196],"over":[199,223,231],"Dynamic":[200],"Programming":[201],"(DPE)":[203],"Vocabulary":[205],"Learning":[206],"via":[207],"Optimal":[208],"Transport":[209],"(VOLT),":[210],"average.":[212],"approximately":[217],"4.3":[219],"BPE-dropout,":[232],"regularized":[234],"version":[235],"BPE.":[237],"We":[238,285],"also":[239],"observed":[240],"significant":[241],"improvements":[242],"IWSLT15":[244],"Vi\u2192En,":[245],"WMT16":[246],"Ro\u2192En,":[247],"WMT15":[249],"Fi\u2192En":[250],"datasets":[251],"competitive":[253],"WMT14":[257,260],"De\u2192En":[258],"Fr\u2192En":[261],"datasets.":[262],"Furthermore,":[263],"17.8\u00d7":[267],"during":[269,276],"up":[272],"36.8\u00d7":[274],"decoding":[277],"scenario":[281],"DPE.":[284],"provide":[286],"extensive":[287],"analysis,":[288],"including":[289],"why":[290],"word-level":[292],"data":[293],"enough":[295],"train":[297],"SelfSeg.":[298]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":2}],"updated_date":"2026-05-21T06:26:12.895304","created_date":"2025-10-10T00:00:00"}