Phan mem dich thuat truc tuyen Can quan tam den chat luong

Tags: Theo Rychtyckyj, Tay Ban Nha, Hang Google, Franz Josef Och, Vulgar Latin, Steve Richardson, Trinh soan thao van ban, sach huong dan su dung, co so du lieu, mot cach chinh xac, huong dan lap rap, nha nghien cuu, ngon ngu goc, cua cong ty, van ban goc, dich thuat, hang


Phan mem dich thuat truc tuyen Can quan tam den chat luong

Nhung cong cu dich thuat tu dong da xuat hien tu lau, va cac ky thuat moi dang giup day manh hieu suat hoat dong cua chung. Tuy nhien, chat luong van con la van de nan giai.

Ban vao trang Language Tools cua Google ( http://www.google.com/language_tools ) va thu lam nhu sau : Danh vao o Translate text cau hoi “Automated language translation – is it an idea whose time has come?” (Dich thuat tu dong – No co la mot y tuong ma thoi cua no da den khong ?) va cho chuong trinh dich cau hoi do sang tieng Phap. Tiep theo, dich ket qua tu tieng Phap sang tieng Duc. Va cuoi cung, dich cau tieng Duc co duoc tro lai tieng Anh. Ban se co ket qua la “Did language translation automate – is it an idea, from which the time came ?” (Dich thuat tu dong hoa – No co la mot y tuong ma thoi gian den tu do ?).

Lam lai giong nhu vay voi cau “Reboot your computer and try again.” (Khoi dong lai may tinh cua ban va thu lai lan nua) ban se co ket qua cuoi cung la “Their computer and attempt still again load.” (May tinh cua ho va no luc van con nhap vao lan nua.) Oi troi ! Dung la “tam sao that ban” !!! Co le ban se that su ngan ngam neu nhu nhung sach huong dan su dung thiet bi co nhieu ngon ngu duoc dich theo kieu nhu the nay.

Tuy vay, cong bang ma noi thi cac cong cu dich thuat cung co nhung ho tro tich cuc o mot so mat nao do. Va cac nha nghien cuu cho biet nhung cach tiep can moi se mang lai nhieu cai thien ve cach thuc hien cho nhung cong cu dich thuat.

Cap nhat co so du lieu

Hang xe hoi Ford da bat dau su dung phan mem dich thuat (machine translation) tu nam 1998, va den nay hang nay da dich duoc nam trieu cau huong dan lap rap tu tieng Anh sang tieng Tay Ban Nha, Duc, Bo Dao Nha va tieng Tay Ban Nha cua nguoi Mexico. Cac sach huong dan lap rap bang tieng Anh duoc cap nhat moi ngay, va nhung ban dich cua chung – khoang 5.000 trang mot ngay – duoc nhanh chong chuyen den cac nha may cua Ford tren khap the gioi.

Nestor Rychtyckyj, chuyen gia ve tri tue nhan tao (Artificial Intelligence – AI) cua hang Ford, noi : “Ban khong the lam toan bo cong viec nay mot cach thu cong duoc.”

Hang xe hoi nay su dung phan mem Enterprise Global Server cua Systran Software, mot cong ty chuyen cung cap cac phan mem dich thuat co tru so o San Diego, bang California. Ben canh do, cac ky su cua Ford cung tu viet mot chuong trinh AI de chuyen ngu nhung cau huong dan thuoc ve chuyen mon o cap do cao mot cach chinh xac va ro rang. Moi mot cau huong dan deu duoc luu tru vao co so du lieu de dung cho viec dich thuat.

Hang Ford cung phai phat trien nhung tu dien thuat ngu chuyen nganh xe hoi. Theo Rychtyckyj, duy tri nhung tu dien nhu the de dang hon viec tim kiem mot chuyen gia dich thuat thong thao cac thuat ngu. Ong noi : “Hau het nhung no luc cua chung toi la xay dung cac bang thuat ngu, va thuong xuyen thay doi chung. Ket qua dich thuat cua ban se tot hon nhieu neu ban co su chuan bi truoc.”

Phan mem cua cong ty Systran su dung mot ky thuat dang tin cay goi la dich theo quy tac (rules-based translation). Nhung he thong nhu the su dung nhung tu dien song ngu ket hop voi co so du lieu ve cach dung tu va cac quy tac van pham. Cac trinh dich thuat thuong mai thuong duoc bo sung bang cac bang thuat ngu, hoac nhung co so du lieu luu tru nhung cau hoac doan van mau da duoc dich san (translation memory). Nhung phan bo sung nay thuong do nguoi su dung lap ra trong qua trinh dich thuat. Nho vay, phan mem dich thuat co the dua ra nhung tu hoac cau dich chinh xac hoac gan dung nhat so voi tu hoac cau goc de nguoi dich de dang xem lai.

Huan luyen phan mem

Dich thuat theo kieu thong ke (statistical machine translation) la mot ky thuat moi nhung chua duoc su dung rong rai. Ky thuat nay su dung nhung van ban goc va ban dich cua chung de “huan luyen” phan mem. Theo thoi gian, nhung he thong nay se “hoc” duoc cach tao ra mot ban dich tot, roi dung phep tinh xac suat va thong ke de chon ra mot tu hoac cum tu gan dung nhat dua tren ngu canh.

Cac he thong thong ke doi hoi phai co rat nhieu van ban de huan luyen cac thuat toan, nhung chung khong can cac quy tac van pham, tu dien song ngu hoac translation memory. Chung tu phat trien nhung quy tac va tiep tuc tu dieu chinh cac quy tac do theo thoi gian.

Hang Google su dung phan mem dich theo quy tac cua cong ty Systran, nhung ho cung phat trien nhung he thong dua tren thong ke cua chinh ho de dich tu tieng Anh sang tieng A-Rap, Nga, Hoa, Han, Nhat, va nguoc lai. Theo chuyen gia nghien cuu Franz Josef Och cua Google, nhung ngon ngu nay rat kho cho cac phan mem dich thuat vi chung khac voi cac ngon ngu Tay phuong co goc la tieng Latin binh dan (Vulgar Latin).

Trong nhieu nam qua, hang Microsoft da dua mot phan mem phan tich cu phap dua tren quy tac van pham vao trinh soan thao van ban Word cua ho. Gan day, hang nay lai ket hop cac translation memory, cac phan mem dich thuat dua tren quy tac va thong ke voi su ho tro cua con nguoi de dich cac van ban cho co so du lieu ve thong tin ho tro khach hang cua ho. Steve Richardson, mot nha nghien cuu cao cap cua Microsoft, noi : “Khuynh huong moi trong cong dong cac nha nghien cuu la xem con nguoi co the ket hop nhung ky thuat thuan tuy ve thong ke nay voi mot so kien thuc ve ngon ngu ra sao.”

Khi duoc hoi la se co nhung dot pha trong viec dich thuat khong, ong tra loi : “Nhung dot pha da xuat hien trong qua trinh nghien cuu. Tren phuong dien thuc hanh, se co nhung dot pha trong viec tao ra nhung he thong duoc tich hop vao qua trinh hoat dong cua doanh nghiep.”

Xu huong trong tuong lai la ket hop cac co so du lieu luu tru nhung cau hoac doan van mau da duoc dich san voi cac phan mem dich thuat dua tren quy tac hoac thong ke, hoac ca hai. Nhung he thong lai tao nhu the se tro nen tinh vi va phuc tap hon.

Tuy nhien, nhung he thong dich thuat hien nay chua the dua ra nhung ban dich co chat luong cao nhu la nhung dich gia “sieu dang”, dac biet doi voi nhung van ban su dung ngon ngu binh dan. Theo Rychtyckyj cua hang Ford, nhung he thong lai tao tro nen thong minh nhu the nao khong la van de gi ca, chat luong dich thuat phu thuoc vao viec xay dung ngon ngu goc mot cach chuan xac.

Quy trinh dich thuat noi chung co hai cong doan chinh :

- Phan tich nguyen ban trong ngon ngu goc de “hieu” that ro tac gia “muon noi” gi.

- Tai lap y nghia ma tac gia muon chuyen tai trong ngon ngu dich de co duoc mot van ban tuong duong voi nguyen ban.

Thoat nhin thi quy trinh nay co ve don gian nhung dang sau no la ca mot qua trinh nhan thuc phuc tap. De “giai ma” y nghia cua van ban goc mot cach chinh xac, dich gia phai phan tich va dien giai tat ca cac dac tinh cua no. Day la mot qua trinh doi hoi mot kien thuc sau rong ve van pham, ngu nghia, cu phap, thanh ngu… cua ngon ngu goc, cung nhu van hoa cua tac gia van ban do. Dong thoi, dich gia cung phai co mot kien thuc tuong tu nhu the ve ngon ngu ma ho se chuyen doi.

Vi vay, thach thuc doi voi cac phan mem dich thuat la lap trinh nhu the nao de may tinh co the “hieu” mot van ban nhu la con nguoi.

(Theo TBVTSG)

Viet Bao
Thích và chia sẻ bài viết này trên facebook Theo dõi vietbao.vn trên facebook

Tìm hiểu: Theo Rychtyckyj, Tay Ban Nha, Hang Google, Franz Josef Och, Vulgar Latin, Steve Richardson, Trinh soan thao van ban, sach huong dan su dung, co so du lieu, mot cach chinh xac, huong dan lap rap, nha nghien cuu, ngon ngu goc, cua cong ty, van ban goc, dich thuat, hang

Xem Bản Tiếng Việt Có Dấu: Phần mềm dịch thuật trực tuyến: Cần quan tâm đến chất lượng

Ban co the doc ban tieng Viet co dau cua bai viet Phần mềm dịch thuật trực tuyến: Cần quan tâm đến chất lượng bang cach nhan chuot vao duong dan tren.

Lien he ve tin Phan mem dich thuat truc tuyen Can quan tam den chat luong

Nhan xet, hay lien he ve tin Phan mem dich thuat truc tuyen Can quan tam den chat luong co the gui bang duong dan o duoi. Ban nen gui kem tieu de bai viet Phan mem dich thuat truc tuyen Can quan tam den chat luong de lam tham khao. Bai viet nay thuoc chuyen de Kinh nghiem trong chuyen muc Cong nghe.

Cac tin khac:

Xem tiep ... Kinh nghiem
Kenh Tin RSS Kenh Tin RSS | Du Bao Thoi Tiet | Lich Xem TV | Lien He - Contact | Quang Cao | Viec Lam | Dieu Kien Su Dung | Bao Ve Tinh Rieng Tu | Sitemap

Copyright©2008 VietBao.vn, Thu Vien Thong Tin Tong Hop Viet Nam va The Gioi, phien ban thu nghiem phan mem Viet Bao Viet Nam, beta 1.0