Bài giảng Xử lý ngôn ngữ tự nhiên: Chương 2 - Nguyễn Kiêm Hiếu (ĐH Bách khoa Hà Nội)

Bài giảng "Xử lý ngôn ngữ tự nhiên - Chương 2: Gán nhãn từ loại" cung cấp cho người học các kiến thức: PennTreebank, Hidden Markov model, conditional random fields, đánh giá. Mời các bạn cùng tham khảo nội dung chi tiết.
Nội dung trích xuất từ tài liệu:
Bài giảng Xử lý ngôn ngữ tự nhiên: Chương 2 - Nguyễn Kiêm Hiếu (ĐH Bách khoa Hà Nội)Chương 2 Gán nhãn từ loạiPennTreebankHidden Markov modelConditional Random FieldsIT4772 Xử lý ngôn ngữ tự nhiênViện CNTT-TT, ĐHBKHNĐánh giá2Chương 2 Xác định từ loạiPennTreebankINFORMATION EXTRACTIONNATURAL LANGUAGE UNDERSTANDINGEND-TO-ENDAPPLICATIONS●Created by University of Pennsylvania●Eight-years project: 1989 – 1996●7 millions words of POS tagged texts●POS tagset is based on Brown CorpusNATURAL LANGUAGE GENERATIONDATA + LINGUISTICS + MACHINE LEARNING34Penn POS tagset●CCHe bought a car and a house.●CDFive years later, autocar will be popular.●DTPierre Vinken will join the board.●EXThere is no asbestos in our product now.5●IN6●The percentage of lung cancer appears to be highest.Mr Vinken is chairman of Elsevier N.V.●●JJMDUS should regulate the class of asbestos.Rudolph Agnew was named an executive director.●JJS●JJRNNIt’s more than three times the expected number.The number of death was higher than expected●NNSPortfolio managers expect further declines in interest rates.78●NNP●It expects to obtain regulatory approval.Alexis Sanchez joined Manchester Unitedyesterday.●●NNPSPP$Shareholders approve its acquisition by Royal Trustco Ltd.… the Japan Automobile Dealers’ Association...●PRP●RB… depends heavily on creativityPOS●… at Monday’s auctionRBR… worked for the project for more than six years9●RBS10●… to return homethe most mundane aspect of its workers●VBTO●VBDthe executives joined Mayor WilliamHe decided to stay●VBG… before boarding the buses again●VBNA buffet breakfast was held in the museum1112●VBP●Plans that give advertisers disscount●WRBwhere employees are assigned lunch partnersVBZThe plan is not an attempt●WDTa project that did not include Seymor●WPwho couldn’t be reach for comment13corenlp.run14http://45.117.171.213/bknlptool/1516Chương 2 Xác định từ loạiHidden Markov model●Chương 2 Xác định từ loạiHidden Markov modelMarkov model●Categorical mixture model1718First-order HMMChương 2 Xác định từ loạiHidden Markov model●Transition probabilityPr(xt = NN | xt-1 = DT)●DTNNVBDINDTNNThecatsatonthematEmission probabitlityPr(ot = cat | xt = NN)1920