{"id":3725,"date":"2023-04-20T23:31:27","date_gmt":"2023-04-20T23:31:27","guid":{"rendered":"https:\/\/www.clc.hcmus.edu.vn\/?p=3725"},"modified":"2025-11-11T23:54:10","modified_gmt":"2025-11-11T16:54:10","slug":"ung-dung-tri-tue-nhan-tao-giai-quyet-bai-toan-dich-tu-dong-van-ban-chu-nom-sang-chu-quoc-ngu","status":"publish","type":"post","link":"https:\/\/www.clc.hcmus.edu.vn\/?p=3725","title":{"rendered":"\u1ee8ng d\u1ee5ng tr\u00ed tu\u1ec7 nh\u00e2n t\u1ea1o gi\u1ea3i quy\u1ebft &#8220;b\u00e0i to\u00e1n&#8221; d\u1ecbch t\u1ef1 \u0111\u1ed9ng v\u0103n b\u1ea3n ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef"},"content":{"rendered":"<p><em><strong>Website chuy\u1ec3n t\u1ef1 (d\u1ecbch) t\u1ef1 \u0111\u1ed9ng t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef g\u00f3p ph\u1ea7n b\u1ea3o t\u1ed3n v\u00e0 ph\u00e1t huy c\u00e1c gi\u00e1 tr\u1ecb v\u0103n h\u00f3a truy\u1ec1n th\u1ed1ng c\u1ee7a d\u00e2n t\u1ed9c, \u0111\u1eb7c bi\u1ec7t c\u00f3 th\u1ec3 \u0111\u01b0\u1ee3c \u1ee9ng d\u1ee5ng ngay \u0111\u1ec3 ph\u00e1t tri\u1ec3n ph\u1ea7n m\u1ec1m d\u1ecbch thu\u1eadt tr\u00ean thi\u1ebft b\u1ecb di \u0111\u1ed9ng nh\u1eb1m ph\u1ee5c v\u1ee5 du kh\u00e1ch khi tham quan, t\u00ecm hi\u1ec3u c\u00e1c \u0111\u1ecba \u0111i\u1ec3m c\u00f3 s\u1eed d\u1ee5ng ch\u1eef N\u00f4m.<\/strong><\/em><\/p>\n<p>C\u00f3 th\u1ec3 kh\u1eb3ng \u0111\u1ecbnh r\u1eb1ng, ch\u1eef N\u00f4m l\u00e0 th\u1ec3 lo\u1ea1i ch\u1eef vi\u1ebft tay \u0111\u1ea7u ti\u00ean c\u1ee7a ng\u01b0\u1eddi Vi\u1ec7t Nam do c\u00e1c b\u1eadc ti\u1ec1n nh\u00e2n x\u00e2y d\u1ef1ng d\u1ef1a tr\u00ean ch\u1ea5t li\u1ec7u c\u1ee7a ch\u1eef H\u00e1n, \u0111\u01b0\u1ee3c s\u1eed d\u1ee5ng trong g\u1ea7n 1.000 n\u0103m t\u1eeb th\u1ebf k\u1ef7 X \u0111\u1ebfn th\u1ebf k\u1ef7 th\u1ee9 XIX. Trong su\u1ed1t 10 th\u1ebf k\u1ef7 \u0111\u00f3, r\u1ea5t nhi\u1ec1u c\u00f4ng tr\u00ecnh v\u1ec1 l\u1ecbch s\u1eed, v\u0103n h\u1ecdc, y h\u1ecdc, n\u00f4ng nghi\u1ec7p, \u0111\u1ecba l\u00fd, \u2026 \u0111\u00e3 \u0111\u01b0\u1ee3c bi\u00ean so\u1ea1n, vi\u1ebft b\u1eb1ng ch\u1eef N\u00f4m v\u00e0 c\u00f2n \u0111\u01b0\u1ee3c l\u01b0u gi\u1eef cho \u0111\u1ebfn ng\u00e0y nay.<\/p>\n<p>&#8220;Tuy nhi\u00ean, ph\u1ea7n l\u1edbn t\u00e0i li\u1ec7u ch\u1eef N\u00f4m v\u1eabn ch\u01b0a \u0111\u01b0\u1ee3c d\u1ecbch (chuy\u1ec3n t\u1ef1) sang ch\u1eef Qu\u1ed1c ng\u1eef s\u1eed d\u1ee5ng con ch\u1eef Latin, v\u00e0 th\u1ef1c t\u1ebf l\u00e0 hi\u1ec7n kh\u00e1 \u00edt ng\u01b0\u1eddi c\u00f3 kh\u1ea3 n\u0103ng \u0111\u1ecdc \u0111\u01b0\u1ee3c ch\u1eef N\u00f4m \u0111\u1ec3 t\u00ecm hi\u1ec3u, khai th\u00e1c kho t\u00e0ng v\u0103n h\u00f3a, tri th\u1ee9c, t\u01b0 li\u1ec7u l\u1ecbch s\u1eed do ng\u01b0\u1eddi x\u01b0a \u0111\u1ec3 l\u1ea1i&#8221;, PGS.TS \u0110inh \u0110i\u1ec1n (Tr\u01b0\u1eddng \u0110\u1ea1i h\u1ecdc Khoa h\u1ecdc T\u1ef1 nhi\u00ean &#8211; \u0110HQG TP.HCM) cho bi\u1ebft.<\/p>\n<p>V\u00ec th\u1ebf, nh\u00f3m chuy\u00ean gia \u0111ang c\u00f4ng t\u00e1c t\u1ea1i Tr\u01b0\u1eddng \u0110\u1ea1i h\u1ecdc Khoa h\u1ecdc T\u1ef1 nhi\u00ean &#8211; \u0110HQG TP.HCM \u0111\u00e3 tri\u1ec3n khai nhi\u1ec7m v\u1ee5 khoa h\u1ecdc &#8211; c\u00f4ng ngh\u1ec7 &#8220;X\u00e2y d\u1ef1ng h\u1ec7 th\u1ed1ng chuy\u1ec3n t\u1ef1 t\u1ef1 \u0111\u1ed9ng v\u0103n b\u1ea3n ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef&#8221; v\u1edbi m\u1ee5c ti\u00eau then ch\u1ed1t l\u00e0 x\u00e2y d\u1ef1ng h\u1ec7 th\u1ed1ng c\u00f3 kh\u1ea3 n\u0103ng d\u1ecbch t\u1ef1 \u0111\u1ed9ng ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef.<\/p>\n<p>Theo l\u1eddi PGS.TS \u0110inh \u0110i\u1ec1n, vi\u1ec7c chuy\u1ec3n t\u1ef1 ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef r\u1ea5t ph\u1ee9c t\u1ea1p do hai h\u1ec7 ch\u1eef kh\u00e1c lo\u1ea1i h\u00ecnh ch\u1eef vi\u1ebft. Ch\u1eef N\u00f4m thu\u1ed9c lo\u1ea1i h\u00ecnh ch\u1eef ghi \u00fd (ideographic), c\u00f2n ch\u1eef Qu\u1ed1c ng\u1eef thu\u1ed9c lo\u1ea1i h\u00ecnh ch\u1eef ghi \u00e2m v\u1ecb (phonemic). C\u00f9ng m\u1ed9t ch\u1eef N\u00f4m c\u00f3 th\u1ec3 \u0111\u01b0\u1ee3c d\u1ecbch sang nhi\u1ec1u ch\u1eef Qu\u1ed1c ng\u1eef kh\u00e1c nhau t\u00f9y theo tri th\u1ee9c v\u0103n h\u00f3a, l\u1ecbch s\u1eed, \u0111\u1ecba l\u00fd, ti\u1ebfng Vi\u1ec7t c\u1ed5, ti\u1ebfng \u0111\u1ecba ph\u01b0\u01a1ng, t\u1eeb chuy\u00ean ng\u00e0nh,\u2026<\/p>\n<p>Hay n\u00f3i c\u00e1ch kh\u00e1c, vi\u1ec7c ch\u1ecdn ch\u1eef Qu\u1ed1c ng\u1eef n\u00e0o cho b\u1ea3n d\u1ecbch c\u1ea7n ph\u1ea3i suy \u0111o\u00e1n, v\u00e0 vi\u1ec7c suy \u0111o\u00e1n n\u00e0y ph\u1ea3i s\u1eed d\u1ee5ng \u0111\u1ebfn nhi\u1ec1u &#8220;tri th\u1ee9c&#8221; c\u1ea3 trong v\u00e0 ngo\u00e0i ng\u00f4n ng\u1eef (extra-linguistic).<\/p>\n<p><img decoding=\"async\" loading=\"lazy\" class=\"richtext-image full-width img-thumbnail\" src=\"https:\/\/dost.hochiminhcity.gov.vn\/media\/images\/NV-04-Hinh_1_-_Anh_Dai_dien_web_So.width-800.jpg\" alt=\"NV-04-Hinh 1 - Anh Dai dien (web So).jpg\" width=\"800\" height=\"471\" data-embedtype=\"image\" data-id=\"1750\" data-format=\"fullwidth\" data-alt=\"NV-04-Hinh 1 - Anh Dai dien (web So).jpg\" \/><\/p>\n<p><em>Giao di\u1ec7n website chuy\u1ec3n t\u1ef1 t\u1ef1 \u0111\u1ed9ng ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef l\u00e0 s\u1ea3n ph\u1ea9m c\u1ee7a nhi\u1ec7m v\u1ee5 khoa h\u1ecdc &#8211; c\u00f4ng ngh\u1ec7<\/em><\/p>\n<p>&#8220;Kh\u00f3 kh\u0103n l\u1edbn nh\u1ea5t trong vi\u1ec7c chuy\u1ec3n t\u1ef1 ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef ch\u00ednh l\u00e0 vi\u1ec7c ch\u1ecdn ch\u1eef Qu\u1ed1c ng\u1eef \u0111\u00fang trong s\u1ed1 c\u00e1c ch\u1eef Qu\u1ed1c ng\u1eef kh\u1ea3 d\u0129 c\u1ee7a ch\u1eef N\u00f4m&#8221;, \u0111\u1ea1i di\u1ec7n nh\u00f3m tri\u1ec3n khai nhi\u1ec7m v\u1ee5 th\u00f4ng tin, &#8220;Vi\u1ec7c ch\u1ecdn l\u1ef1a n\u00e0y ph\u1ee5 thu\u1ed9c v\u00e0o ng\u1eef c\u1ea3nh, th\u1ec3 lo\u1ea1i (v\u0103n xu\u00f4i\/v\u1ea7n), l\u0129nh v\u1ef1c (v\u0103n h\u1ecdc, y h\u1ecdc, t\u00f4n gi\u00e1o, \u2026) v\u00e0 c\u1ea3 v\u00e0o ni\u00ean \u0111\u1ea1i, v\u00f9ng mi\u1ec1n.&#8221;.<\/p>\n<p>Do \u0111\u00f3, v\u00ec nguy\u00ean t\u1eafc c\u1ee7a h\u1ecdc m\u00e1y l\u00e0 n\u1ebfu h\u1ec7 th\u1ed1ng\/con ng\u01b0\u1eddi hu\u1ea5n luy\u1ec7n cho m\u00e1y t\u00ednh ng\u1eef li\u1ec7u thu\u1ed9c th\u1ec3 lo\u1ea1i, l\u0129nh v\u1ef1c n\u00e0o th\u00ec m\u00e1y s\u1ebd d\u1ecbch t\u1ed1t h\u01a1n v\u1edbi nh\u1eefng v\u0103n b\u1ea3n thu\u1ed9c th\u1ec3 lo\u1ea1i\/l\u0129nh v\u1ef1c \u0111\u00f3.<\/p>\n<p>B\u00e1o c\u00e1o tr\u01b0\u1edbc H\u1ed9i \u0111\u1ed3ng t\u01b0 v\u1ea5n nghi\u1ec7m thu nhi\u1ec7m v\u1ee5 v\u1eeba \u0111\u01b0\u1ee3c S\u1edf KH&amp;CN TP.HCM t\u1ed5 ch\u1ee9c, PGS.TS \u0110inh \u0110i\u1ec1n kh\u1eb3ng \u0111\u1ecbnh: g\u1ea7n \u0111\u00e2y v\u1edbi s\u1ef1 ph\u00e1t tri\u1ec3n v\u01b0\u1ee3t b\u1eadc c\u1ee7a l\u0129nh v\u1ef1c tr\u00ed tu\u1ec7 nh\u00e2n t\u1ea1o (AI) c\u0169ng nh\u01b0 c\u00e1c c\u00f4ng ngh\u1ec7 h\u1ecdc m\u00e1y (machine learning) ti\u00ean ti\u1ebfn trong ng\u00e0nh khoa h\u1ecdc m\u00e1y t\u00ednh \u0111\u00e3 gi\u00fap b\u00e0i to\u00e1n chuy\u1ec3n t\u1ef1 t\u1ef1 \u0111\u1ed9ng hay chuy\u1ec3n t\u1ef1 m\u00e1y (machine transliteration) c\u00f3 th\u1ec3 th\u1ef1c hi\u1ec7n \u0111\u01b0\u1ee3c, d\u00f9 ch\u01b0a th\u1ec3 ch\u00ednh x\u00e1c ho\u00e0n to\u00e0n.<\/p>\n<p>V\u1edbi c\u00f4ng ngh\u1ec7 h\u1ecdc m\u00e1y, m\u00e1y c\u00f3 th\u1ec3 &#8220;t\u1ef1 h\u1ecdc&#8221; \u0111\u01b0\u1ee3c c\u00e1ch ch\u1ecdn (suy \u0111o\u00e1n) ch\u1eef Qu\u1ed1c ng\u1eef ph\u00f9 h\u1ee3p v\u1edbi t\u1eebng ch\u1eef N\u00f4m th\u00f4ng qua ng\u1eef c\u1ea3nh trong r\u1ea5t nhi\u1ec1u c\u00e1c b\u1ea3n d\u1ecbch N\u00f4m &#8211; Qu\u1ed1c ng\u1eef tr\u01b0\u1edbc \u0111\u00f3 c\u1ee7a con ng\u01b0\u1eddi. Do \u0111\u00f3, n\u1ebfu h\u1ec7 th\u1ed1ng &#8220;d\u1ea1y&#8221; cho m\u00e1y t\u00ednh b\u1eb1ng c\u00e1ch cung c\u1ea5p (\u0111\u01b0a v\u00e0o kho ng\u1eef li\u1ec7u hu\u1ea5n luy\u1ec7n) cho m\u00e1y c\u00e0ng nhi\u1ec1u b\u1ea3n d\u1ecbch N\u00f4m &#8211; Qu\u1ed1c ng\u1eef chu\u1ea9n, th\u00ec m\u00e1y s\u1ebd c\u00e0ng &#8220;th\u00f4ng minh&#8221; h\u01a1n v\u00e0 cho k\u1ebft qu\u1ea3 d\u1ecbch ch\u00ednh x\u00e1c h\u01a1n. Ngo\u00e0i ra, m\u00e1y c\u0169ng c\u00f3 kh\u1ea3 n\u0103ng t\u1ef1 h\u1ecdc \u0111\u1ec3 ho\u00e0n thi\u1ec7n h\u01a1n b\u1eb1ng c\u00e1ch r\u00fat kinh nghi\u1ec7m t\u1eeb c\u00e1c l\u1ed7i d\u1ecbch sai c\u1ee7a m\u00e1y sau khi con ng\u01b0\u1eddi hi\u1ec7u \u0111\u00ednh l\u1ea1i nh\u1eefng ch\u1ed7 d\u1ecbch sai \u0111\u00f3. Qu\u00e1 tr\u00ecnh n\u00e0y n\u1ebfu \u0111\u01b0\u1ee3c l\u1eb7p l\u1ea1i c\u00e0ng nhi\u1ec1u th\u00ec m\u00e1y s\u1ebd c\u00e0ng cho nh\u1eefng b\u1ea3n d\u1ecbch t\u1ed1t h\u01a1n sau n\u00e0y.<\/p>\n<p><img decoding=\"async\" loading=\"lazy\" class=\"richtext-image full-width img-thumbnail\" src=\"https:\/\/dost.hochiminhcity.gov.vn\/media\/images\/NV-01-Hinh_2_-_TS_Dinh_Dien.width-800.jpg\" alt=\"NV-01-Hinh 2 - TS Dinh Dien.jpg\" width=\"800\" height=\"600\" data-embedtype=\"image\" data-id=\"1748\" data-format=\"fullwidth\" data-alt=\"NV-01-Hinh 2 - TS Dinh Dien.jpg\" \/><\/p>\n<p><em>PGS.TS \u0110inh \u0110i\u1ec1n thuy\u1ebft minh v\u00e0 v\u1eadn h\u00e0nh th\u1ecb ph\u1ea1m t\u00ednh n\u0103ng chuy\u1ec3n t\u1ef1 t\u1ef1 \u0111\u1ed9ng v\u0103n b\u1ea3n ch\u1eef N\u00f4m sang v\u0103n b\u1ea3n ch\u1eef Qu\u1ed1c ng\u1eef t\u1ea1i H\u1ed9i \u0111\u1ed3ng t\u01b0 v\u1ea5n nghi\u1ec7m thu nhi\u1ec7m v\u1ee5 khoa h\u1ecdc &#8211; c\u00f4ng ngh\u1ec7 do S\u1edf KH&amp;CN TP.HCM t\u1ed5 ch\u1ee9c<\/em><\/p>\n<p><strong>K\u1ebft h\u1ee3p &#8220;h\u1ecdc m\u00e1y&#8221; v\u1edbi m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef<\/strong><\/p>\n<p>Tr\u00ean tinh th\u1ea7n \u0111\u00f3, trong ph\u1ea1m vi c\u1ee7a nhi\u1ec7m v\u1ee5, nh\u00f3m c\u00e1c nh\u00e0 khoa h\u1ecdc v\u1edbi s\u1ef1 d\u1eabn d\u1eaft c\u1ee7a PGS.TS \u0110inh \u0110i\u1ec1n \u0111\u00e3 \u0111\u1ec1 xu\u1ea5t v\u00e0 ho\u00e0n thi\u1ec7n vi\u1ec7c x\u00e2y d\u1ef1ng h\u1ec7 th\u1ed1ng chuy\u1ec3n t\u1ef1 t\u1ef1 \u0111\u1ed9ng t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef b\u1eb1ng c\u00f4ng ngh\u1ec7 h\u1ecdc m\u00e1y c\u00f3 b\u1ed5 sung th\u00eam tri th\u1ee9c t\u1eeb \u0111i\u1ec3n v\u00e0 m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef theo l\u0129nh v\u1ef1c.<\/p>\n<p>C\u1ee5 th\u1ec3, PGS.TS \u0110inh \u0110i\u1ec1n (ch\u1ee7 nhi\u1ec7m nhi\u1ec7m v\u1ee5) cho bi\u1ebft, t\u1eadn d\u1ee5ng c\u00f4ng ngh\u1ec7 h\u1ecdc s\u00e2u (deep learning), nh\u00f3m c\u00e1c nh\u00e0 khoa h\u1ecdc t\u1ea1i gia Tr\u01b0\u1eddng \u0110\u1ea1i h\u1ecdc Khoa h\u1ecdc T\u1ef1 nhi\u00ean &#8211; \u0110HQG TP.HCM \u0111\u00e3 t\u1eadp trung d\u1ecbch m\u1ed9t chi\u1ec1u t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef. C\u1ee5 th\u1ec3, nh\u00f3m x\u00e2y d\u1ef1ng T\u1ef1 \u0111i\u1ec3n ch\u1eef N\u00f4m &#8211; Qu\u1ed1c ng\u1eef (b\u1ea3n ch\u1ea5t l\u00e0 m\u1ed9t t\u1eadp h\u1ee3p c\u00f3 h\u1ec7 th\u1ed1ng c\u00e1c N\u00f4m t\u1ef1 \u0111\u01b0\u1ee3c gi\u1ea3i ngh\u0129a Qu\u1ed1c ng\u1eef) nh\u1eb1m t\u1eadp trung gi\u1ea3i ngh\u0129a c\u1ee7a t\u1eebng t\u1ef1, cung c\u1ea5p c\u00e1c th\u00f4ng tin s\u00e2u v\u1ec1 m\u1eb7t ng\u00f4n ng\u1eef h\u1ecdc.<\/p>\n<p>S\u1ed1 l\u01b0\u1ee3ng N\u00f4m t\u1ef1 r\u00fat tr\u00edch \u0111\u01b0\u1ee3c t\u1eeb kho ng\u1eef li\u1ec7u ch\u1eef N\u00f4m v\u00e0 c\u00e1c ngu\u1ed3n tham kh\u1ea3o \u0111\u1ea1t 22.264 m\u1ee5c t\u1ef1. Sau \u0111\u00f3, x\u00e2y d\u1ef1ng ti\u1ebfp T\u1eeb \u0111i\u1ec3n ch\u1eef N\u00f4m &#8211; Qu\u1ed1c ng\u1eef ch\u1ee9a 6.198 m\u1ee5c t\u1eeb. Nh\u00f3m c\u0169ng x\u00e2y d\u1ef1ng T\u1ef1 \u0111i\u1ec3n ch\u1eef H\u00e1n &#8211; Vi\u1ec7t ch\u1ee9a 26.330 m\u1ee5c t\u1ef1 v\u00e0 T\u1eeb \u0111i\u1ec3n ch\u1eef H\u00e1n &#8211; Vi\u1ec7t ch\u1ee9a 66.450 m\u1ee5c t\u1eeb.<\/p>\n<table border=\"1\" width=\"705\">\n<tbody>\n<tr>\n<td>&#8220;T\u1ef1&#8221; l\u00e0 \u0111\u01a1n v\u1ecb nh\u1ecf nh\u1ea5t trong qu\u00e1 tr\u00ecnh x\u1eed l\u00fd ng\u1eef li\u1ec7u \u0111\u1ec3 chu\u1ea9n b\u1ecb cho qu\u00e1 tr\u00ecnh chuy\u1ec3n t\u1ef1. C\u1ea5p \u0111\u1ed9 l\u1edbn h\u01a1n &#8220;t\u1ef1&#8221; l\u00e0 t\u1eeb v\u00e0 c\u1ee5m t\u1eeb. M\u1ed9t t\u1eeb hay c\u1ee5m t\u1eeb c\u00f3 th\u1ec3 g\u1ed3m hai, ba, \u0111\u1ebfn b\u1ed1n t\u1ef1. Cao h\u01a1n t\u1eeb v\u00e0 c\u1ee5m t\u1eeb l\u00e0 c\u1ea5p \u0111\u1ed9 c\u00e2u. C\u00e2u c\u00f3 th\u1ec3 bao g\u1ed3m c\u00e2u th\u01a1 ho\u1eb7c c\u00e2u v\u0103n xu\u00f4i v\u1edbi \u0111\u1ed9 d\u00e0i thay \u0111\u1ed5i t\u00f9y v\u0103n b\u1ea3n thu th\u1eadp \u0111\u01b0\u1ee3c.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>&nbsp;<\/p>\n<p>M\u00f4 h\u00ecnh \u0111\u01b0\u1ee3c nh\u00f3m tri\u1ec3n khai nhi\u1ec7m v\u1ee5 \u0111\u1ec1 xu\u1ea5t v\u00e0 hi\u1ec7n th\u1ef1c l\u00e0 d\u1ef1a tr\u00ean m\u00f4 h\u00ecnh hi\u1ec7n h\u1eefu c\u1ee7a gi\u1ea3i ph\u00e1p d\u1ecbch m\u00e1y th\u1ed1ng k\u00ea SMT b\u1eb1ng Moses c\u1ee7a gi\u1ea3i ph\u00e1p N\u00f4m Converter (www.chunom.org) m\u00e0 PGS.TS \u0110inh \u0110i\u1ec1n v\u00e0 c\u00e1c c\u1ed9ng s\u1ef1 t\u1eebng nghi\u00ean c\u1ee9u tr\u01b0\u1edbc \u0111\u00f3 v\u1edbi m\u1ed9t s\u1ed1 kh\u00e1c bi\u1ec7t v\u00e0 c\u1ea3i ti\u1ebfn d\u1ef1a tr\u00ean kinh nghi\u1ec7m h\u1ec7 d\u1ecbch Hoa &#8211; Vi\u1ec7t b\u1eb1ng th\u1ed1ng k\u00ea v\u00e0 Anh &#8211; Vi\u1ec7t b\u1eb1ng h\u1ecdc s\u00e2u c\u0169ng c\u1ee7a ch\u00ednh nh\u00f3m nghi\u00ean c\u1ee9u, \u0111\u00f3 l\u00e0 thay v\u00ec d\u1ecbch c\u1ea3 hai chi\u1ec1u (t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef v\u00e0 ng\u01b0\u1ee3c l\u1ea1i) nh\u01b0 h\u1ec7 th\u1ed1ng N\u00f4m Converter, th\u00ec nh\u00f3m nghi\u00ean c\u1ee9u ch\u1ec9 t\u1eadp trung d\u1ecbch m\u1ed9t chi\u1ec1u t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef v\u00ec th\u1ea5y th\u1ea5y chi\u1ec1u ng\u01b0\u1ee3c l\u1ea1i kh\u00f4ng c\u00f3 \u00fd ngh\u0129a th\u1ef1c ti\u1ec5n l\u1edbn.<\/p>\n<p>Ngo\u00e0i ra, vi\u1ec7c d\u1ecbch m\u1ed9t chi\u1ec1u t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef s\u1ebd gi\u00fap d\u1ec5 t\u1eadp trung c\u1ea3i ti\u1ebfn ch\u1ea5t l\u01b0\u1ee3ng \u0111\u1ea7u ra c\u1ee7a ch\u1eef Qu\u1ed1c ng\u1eef h\u01a1n b\u1eb1ng c\u00e1ch \u0111\u1ea7u t\u01b0 nhi\u1ec1u cho m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef (Language Model) c\u1ee7a ch\u1eef Qu\u1ed1c ng\u1eef.<\/p>\n<p>\u0110\u1ec3 kh\u1eafc ph\u1ee5c t\u00ecnh tr\u1ea1ng thi\u1ebfu ch\u1eef N\u00f4m trong t\u1eadp hu\u1ea5n luy\u1ec7n nh\u01b0 trong h\u1ec7 th\u1ed1ng N\u00f4m Converter, nh\u00f3m tri\u1ec3n khai nhi\u1ec7m v\u1ee5 b\u1ed5 sung t\u1ef1 \u0111i\u1ec3n H\u00e1n &#8211; Vi\u1ec7t v\u00e0o b\u1ea3ng d\u1ecbch (phrase table) c\u1ee7a h\u1ec7 th\u1ed1ng Moses. Ngo\u00e0i ra, nh\u00f3m c\u0169ng b\u1ed5 sung nhi\u1ec1u b\u1ea3n d\u1ecbch (th\u1ee7 c\u00f4ng) song song N\u00f4m &#8211; Qu\u1ed1c ng\u1eef kh\u00e1c (m\u00e0 h\u1ec7 th\u1ed1ng N\u00f4m Converter ch\u01b0a \u0111\u01b0a v\u00e0o hu\u1ea5n luy\u1ec7n) \u0111\u1ec3 n\u00e2ng cao ch\u1ea5t l\u01b0\u1ee3ng d\u1ecbch.<\/p>\n<p>&#8220;C\u1ea3i ti\u1ebfn ch\u00ednh c\u1ee7a nh\u00f3m nghi\u00ean c\u1ee9u l\u00e0 ph\u00e2n chia theo l\u0129nh v\u1ef1c cho ng\u1eef li\u1ec7u hu\u1ea5n luy\u1ec7n \u1edf \u0111\u1ea7u v\u00e0o v\u00e0 m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef ch\u1eef Qu\u1ed1c ng\u1eef \u1edf \u0111\u1ea7u ra&#8221;, PGS.TS \u0110inh \u0110i\u1ec1n nh\u1ea5n m\u1ea1nh.<\/p>\n<p>&#8220;Kh\u00f3 kh\u0103n l\u1edbn nh\u1ea5t trong vi\u1ec7c chuy\u1ec3n t\u1ef1 ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef ch\u00ednh l\u00e0 vi\u1ec7c ch\u1ecdn ch\u1eef Qu\u1ed1c ng\u1eef \u0111\u00fang trong s\u1ed1 c\u00e1c ch\u1eef Qu\u1ed1c ng\u1eef kh\u1ea3 d\u0129 c\u1ee7a ch\u1eef N\u00f4m \u0111\u00f3. Vi\u1ec7c ch\u1ecdn l\u1ef1a n\u00e0y ph\u1ee5 thu\u1ed9c v\u00e0o ng\u1eef c\u1ea3nh, th\u1ec3 lo\u1ea1i, l\u0129nh v\u1ef1c v\u00e0 c\u1ea3 v\u00e0o ni\u00ean \u0111\u1ea1i. H\u1ec7 th\u1ed1ng hi\u1ec7n h\u1eefu ch\u1ec9 m\u1edbi ch\u1ecdn ch\u1eef Qu\u1ed1c ng\u1eef theo ng\u1eef c\u1ea3nh c\u00f3 trong t\u1eadp ng\u1eef li\u1ec7u hu\u1ea5n luy\u1ec7n m\u00e0 t\u1eadp hu\u1ea5n luy\u1ec7n n\u00e0y l\u1ea1i \u0111\u01b0\u1ee3c hu\u1ea5n luy\u1ec7n chung (l\u1eabn l\u1ed9n th\u1ec3 lo\u1ea1i, l\u0129nh v\u1ef1c, ni\u00ean \u0111\u1ea1i)&#8221;, PGS.TS \u0110inh \u0110i\u1ec1n cho bi\u1ebft th\u00eam, &#8220;V\u00ec v\u1eady, trong m\u00f4 h\u00ecnh \u0111\u1ec1 xu\u1ea5t, ch\u00fang t\u00f4i ph\u00e2n chia t\u1eadp hu\u1ea5n luy\u1ec7n, c\u0169ng nh\u01b0 m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef theo th\u1ec3 lo\u1ea1i v\u00e0 l\u0129nh v\u1ef1c.&#8221;.<\/p>\n<p>PGS.TS \u0110inh \u0110i\u1ec1n cho bi\u1ebft, m\u1ee5c \u0111\u00edch quan tr\u1ecdng nh\u1ea5t l\u00e0 tra c\u1ee9u ngh\u0129a Qu\u1ed1c ng\u1eef t\u01b0\u01a1ng \u1ee9ng c\u1ee7a c\u00e1c N\u00f4m t\u1ef1. \u0110\u1ec3 \u0111\u1ea1t \u0111\u01b0\u1ee3c \u0111i\u1ec1u n\u00e0y, x\u00e9t v\u1ec1 c\u1ea5u tr\u00fac vi m\u00f4, t\u1ef1 \u0111i\u1ec3n ch\u1eef N\u00f4m ph\u1ea3i \u0111\u01b0\u1ee3c x\u00e2y d\u1ef1ng m\u1ed9t c\u00e1ch c\u00f3 h\u1ec7 th\u1ed1ng th\u00e0nh c\u00e1c tr\u01b0\u1eddng thu\u1ed9c t\u00ednh c\u1ee5 th\u1ec3. \u0110\u00f3 l\u00e0, tr\u01b0\u1eddng ngh\u0129a Qu\u1ed1c ng\u1eef, l\u00e0 tr\u01b0\u1eddng c\u0103n b\u1ea3n lu\u00f4n c\u00f3, gi\u1ea3i ngh\u0129a Qu\u1ed1c ng\u1eef cho m\u1ee5c N\u00f4m t\u1ef1; tr\u01b0\u1eddng t\u1ea7n su\u1ea5t: cho bi\u1ebft m\u1ee9c \u0111\u1ed9 ph\u1ed5 bi\u1ebfn c\u1ee7a N\u00f4m t\u1ef1; tr\u01b0\u1eddng l\u0129nh v\u1ef1c: cho bi\u1ebft ng\u1eef c\u1ea3nh c\u1ee5 th\u1ec3 c\u1ee7a N\u00f4m t\u1ef1; v\u00e0 tr\u01b0\u1eddng th\u1ec3 lo\u1ea1i: cung c\u1ea5p th\u00eam th\u00f4ng tin v\u1ec1 ngu\u1ed3n g\u1ed1c N\u00f4m t\u1ef1. Vi\u1ec7c x\u00e2y d\u1ef1ng t\u1ef1 \u0111i\u1ec3n ch\u1eef N\u00f4m &#8211; Qu\u1ed1c ng\u1eef bao g\u1ed3m giai \u0111o\u1ea1n thu th\u1eadp, x\u00e2y d\u1ef1ng kho ng\u1eef li\u1ec7u ch\u1eef N\u00f4m, t\u1eeb \u0111\u00f3 tr\u00edch xu\u1ea5t danh s\u00e1ch c\u00e1c N\u00f4m t\u1ef1 v\u00e0 ho\u00e0n thi\u1ec7n c\u1ea5u tr\u00fac v\u0129 m\u00f4, vi m\u00f4 c\u1ee7a t\u1ef1 \u0111i\u1ec3n.<\/p>\n<p>\u0110\u1ec3 hu\u1ea5n luy\u1ec7n m\u00e1y h\u1ecdc v\u00e0 x\u00e2y d\u1ef1ng m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef, trong c\u00f4ng tr\u00ecnh nghi\u00ean c\u1ee9u v\u1eeba \u0111\u01b0\u1ee3c nghi\u1ec7m thu n\u00e0y, PGS.TS \u0110inh \u0110i\u1ec1n v\u00e0 c\u1ed9ng s\u1ef1 \u0111\u00e3 x\u00e2y d\u1ef1ng ng\u1eef li\u1ec7u cho l\u0129nh v\u1ef1c v\u0103n h\u1ecdc, \u0111\u1eddi s\u1ed1ng v\u00e0 t\u00f4n gi\u00e1o.<\/p>\n<p>&#8220;M\u1ed7i l\u0129nh v\u1ef1c c\u00f3 nh\u1eefng v\u1ed1n t\u1eeb kh\u00e1c nhau, gi\u00fap ch\u00fang ta gi\u1edbi h\u1ea1n l\u1ea1i mi\u1ec1n\/l\u0129nh v\u1ef1c l\u1ef1a ch\u1ecdn ch\u1eef Qu\u1ed1c ng\u1eef (trong tr\u01b0\u1eddng h\u1ee3p ch\u1eef N\u00f4m \u0111a tr\u1ecb) \u0111\u1ec3 n\u00e2ng cao kh\u1ea3 n\u0103ng ch\u1ecdn \u0111\u00fang \u0111\u01b0\u1ee3c ch\u1eef Qu\u1ed1c ng\u1eef t\u01b0\u01a1ng \u1ee9ng&#8221;, PGS.TS \u0110inh \u0110i\u1ec1n ph\u00e2n t\u00edch, &#8220;Cu\u1ed1i c\u00f9ng, thay v\u00ec ch\u1ec9 l\u1ea5y ng\u1eef li\u1ec7u ch\u1eef Qu\u1ed1c ng\u1eef trong t\u1eadp hu\u1ea5n luy\u1ec7n \u0111\u1ec3 hu\u1ea5n luy\u1ec7n cho m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef (qu\u00e1 \u00edt, ch\u1ec9 v\u00e0i ng\u00e0n c\u00e2u), nh\u00f3m nghi\u00ean c\u1ee9u ch\u1ee7 \u0111\u1ed9ng s\u1eed d\u1ee5ng th\u00eam ng\u1eef li\u1ec7u ch\u1eef Qu\u1ed1c ng\u1eef \u1edf b\u00ean ngo\u00e0i (r\u1ea5t l\u1edbn, h\u00e0ng tri\u1ec7u c\u00e2u) v\u00e0 \u0111\u00e3 \u0111\u01b0\u1ee3c ph\u00e2n chia theo th\u1ec3 lo\u1ea1i v\u00e0 l\u0129nh v\u1ef1c n\u00f3i tr\u00ean \u0111\u1ec3 hu\u1ea5n luy\u1ec7n cho m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef N-gram c\u1ee7a ch\u1eef Qu\u1ed1c ng\u1eef \u1edf \u0111\u1ea7u ra nh\u1eb1m n\u00e2ng cao kh\u1ea3 n\u0103ng ch\u1ecdn \u0111\u00fang ch\u1eef Qu\u1ed1c ng\u1eef theo t\u00ednh t\u1ef1 nhi\u00ean nh\u1ea5t c\u1ee7a ng\u00f4n ng\u1eef. \u0110\u01b0\u1ee3c bi\u1ebft, kho ng\u1eef li\u1ec7u \u0111\u01a1n ng\u1eef ch\u1eef Qu\u1ed1c ng\u1eef \u0111\u01b0\u1ee3c nh\u00f3m nghi\u00ean c\u1ee9u &#8220;n\u1ea1p d\u1ea1y&#8221; cho h\u1ec7 th\u1ed1ng hi\u1ec7n \u1edf m\u1ee9c 823.533 c\u00e2u v\u00e0 13.024.774 t\u1eeb.<\/p>\n<p>V\u00ec nguy\u00ean t\u1eafc c\u1ee7a h\u1ecdc m\u00e1y l\u00e0 n\u1ebfu ch\u00fang ta hu\u1ea5n luy\u1ec7n cho m\u00e1y t\u00ednh hi\u1ec3u \u0111\u01b0\u1ee3c\/bi\u1ebft \u0111\u01b0\u1ee3c ng\u1eef li\u1ec7u thu\u1ed9c th\u1ec3 lo\u1ea1i, l\u0129nh v\u1ef1c n\u00e0o th\u00ec m\u00e1y s\u1ebd d\u1ecbch t\u1ed1t h\u01a1n v\u1edbi nh\u1eefng v\u0103n b\u1ea3n thu\u1ed9c th\u1ec3 lo\u1ea1i hay l\u0129nh v\u1ef1c \u0111\u00f3.<\/p>\n<p>Tr\u00ean tinh th\u1ea7n n\u00e0y, khi tri\u1ec3n khai\/\u00e1p d\u1ee5ng \u0111i v\u00e0o th\u1ef1c t\u1ebf, ng\u01b0\u1eddi s\u1eed d\u1ee5ng mu\u1ed1n d\u1ecbch v\u0103n b\u1ea3n th\u1ec3 lo\u1ea1i hay l\u0129nh v\u1ef1c n\u00e0o, ch\u1ec9 c\u1ea7n ch\u1ecdn (tr\u00ean tr\u00ecnh \u0111\u01a1n) th\u1ec3 lo\u1ea1i hay l\u0129nh v\u1ef1c mu\u1ed1n d\u1ecbch \u0111\u1ec3 m\u00e1y t\u00ednh l\u1ef1a ch\u1ecdn ki\u1ebfn th\u1ee9c \u0111\u00e3 h\u1ecdc ph\u00f9 h\u1ee3p v\u1edbi th\u1ec3 lo\u1ea1i hay l\u0129nh v\u1ef1c m\u00e0 m\u00e1y \u0111\u00e3 \u0111\u01b0\u1ee3c hu\u1ea5n luy\u1ec7n. Trong tr\u01b0\u1eddng h\u1ee3p ng\u01b0\u1eddi s\u1eed d\u1ee5ng kh\u00f4ng x\u00e1c \u0111\u1ecbnh \u0111\u01b0\u1ee3c th\u1ec3 lo\u1ea1i hay l\u0129nh v\u1ef1c c\u1ee7a v\u0103n b\u1ea3n ch\u1eef N\u00f4m c\u1ea7n d\u1ecbch, th\u00ec m\u00e1y t\u00ednh v\u1eabn c\u00f3 th\u1ec3 t\u1ef1 x\u00e1c \u0111\u1ecbnh (ch\u1ebf \u0111\u1ed9 t\u1ef1 \u0111\u1ed9ng &#8211; auto) th\u1ec3 lo\u1ea1i hay l\u0129nh v\u1ef1c c\u1ee7a v\u0103n b\u1ea3n \u0111\u00f3 tr\u01b0\u1edbc khi d\u1ecbch (t\u1ef1a nh\u01b0 ch\u1ee9c n\u0103ng ch\u1ecdn ng\u00f4n ng\u1eef t\u1ef1 \u0111\u1ed9ng c\u1ee7a c\u00f4ng c\u1ee5 Google Translator).<\/p>\n<p>L\u00e0 m\u1ed9t ph\u1ea7n c\u1ee7a nhi\u1ec7m v\u1ee5, PGS.TS \u0110inh \u0110i\u1ec1n v\u00e0 nh\u00f3m c\u1ed9ng s\u1ef1 \u0111\u00e3 x\u00e2y d\u1ef1ng th\u00e0nh c\u00f4ng m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef v\u00e0 m\u00f4 h\u00ecnh d\u1ecbch tr\u00ean nh\u1eefng ng\u1eef li\u1ec7u thu th\u1eadp \u0111\u01b0\u1ee3c, \u0111\u1ed3ng th\u1eddi ho\u00e0n thi\u1ec7n th\u1eed nghi\u1ec7m website h\u1ed7 tr\u1ee3 chuy\u1ec3n t\u1ef1 (d\u1ecbch) t\u1ef1 \u0111\u1ed9ng t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef (t\u1ea1m th\u1eddi \u0111\u01b0\u1ee3c \u0111\u1eb7t t\u1ea1i \u0111\u1ecba ch\u1ec9 http:\/\/clcnom.kimtudien.com.vn\/), v\u00e0 b\u1ed9 c\u00f4ng c\u1ee5 &#8220;d\u1ecbch m\u00e1y&#8221; n\u00e0y c\u0169ng \u0111\u00e3 \u0111\u01b0\u1ee3c nh\u00f3m tri\u1ec3n khai nhi\u1ec7m v\u1ee5 tr\u00ecnh di\u1ec5n, b\u00e1o c\u00e1o tr\u01b0\u1edbc H\u1ed9i \u0111\u1ed3ng t\u01b0 v\u1ea5n nghi\u1ec7m thu nhi\u1ec7m v\u1ee5 khoa h\u1ecdc &#8211; c\u00f4ng ngh\u1ec7 do S\u1edf KH&amp;CN TP.HCM t\u1ed5 ch\u1ee9c.<\/p>\n<p>\u0110\u01b0\u1ee3c bi\u1ebft, ngay trong \u0111\u1ea7u th\u00e1ng 4\/2023 n\u00e0y, Tr\u01b0\u1eddng \u0110\u1ea1i h\u1ecdc Khoa h\u1ecdc T\u1ef1 nhi\u00ean v\u00e0 nh\u00f3m nghi\u00ean c\u1ee9u \u0111\u00e3 \u0111\u01b0a h\u1ec7 th\u1ed1ng chuy\u1ec3n t\u1ef1 ch\u1eef N\u00f4m n\u00f3i tr\u00ean l\u00ean website ch\u00ednh th\u1ee9c c\u1ee7a tr\u01b0\u1eddng (https:\/\/tools.clc.hcmus.edu.vn\/) nh\u1eb1m ph\u1ee5c v\u1ee5 nhu c\u1ea7u tra c\u1ee9u c\u1ee7a \u0111\u00f4ng \u0111\u1ea3o ng\u01b0\u1eddi d\u00e2n, c\u00e1c nh\u00e0 khoa h\u1ecdc v\u00e0 c\u00e1c t\u1ed5 ch\u1ee9c, doanh nghi\u1ec7p.<\/p>\n<p>Ngu\u1ed3n: https:\/\/dost.hochiminhcity.gov.vn\/hoat-dong-so-khcn\/ung-dung-tri-tue-nhan-tao-giai-quyet-bai-an-dich-tu-dong-van-ban-chu-nom-sang-chu-quoc-ngu\/<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Website chuy\u1ec3n t\u1ef1 (d\u1ecbch) t\u1ef1 \u0111\u1ed9ng t\u1eeb ch\u1eef N\u00f4m sang ch\u1eef Qu\u1ed1c ng\u1eef g\u00f3p ph\u1ea7n b\u1ea3o t\u1ed3n v\u00e0 ph\u00e1t huy [&hellip;]<\/p>\n","protected":false},"author":17,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[126],"tags":[],"_links":{"self":[{"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=\/wp\/v2\/posts\/3725"}],"collection":[{"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=\/wp\/v2\/users\/17"}],"replies":[{"embeddable":true,"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=3725"}],"version-history":[{"count":2,"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=\/wp\/v2\/posts\/3725\/revisions"}],"predecessor-version":[{"id":3734,"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=\/wp\/v2\/posts\/3725\/revisions\/3734"}],"wp:attachment":[{"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3725"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3725"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.clc.hcmus.edu.vn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3725"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}