【李宏毅机器学习课件】24.结构化预测：序列标记.pdf

Sequence Labeling Problem Sequence Labeling f : X  Y Sequence Sequence x: x1 x2 x3 …… xL (sequence) y: y1 y2 y3 …… yL (sequence) Application •Name entity recognition •Identifying names of people, places, organizations, etc. from a sentence •Harry Potter is a student of Hogwarts and lived on Privet Drive. • people, organizations, places, not a name entity Can be difficult … 楊公再興之神 馮氏埋香之塚 Ref: Ref: https://www.ptt.cc/bbs/JinYong/M. https://www.ptt.cc/bbs/JinYong/M. 1258625573.A.DC4.html 1195128035.A.31A.html Example Task •POS tagging • Annotate each word in a sentence with a part-of-speech. John saw the saw. PN V D N • Useful for subsequent syntactic parsing and word sense disambiguation, etc. Example Task •POS tagging John saw the saw. The problem cannot be solved PN V D N without considering the sequences.  “saw” is more likely to be a verb V rather than a noun N  However, the second “saw” is a noun N because a noun N is more likely to follow a determiner. Outline Hidden Markov Model (HMM) Conditional Random Field (CRF) Structured Perceptron/SVM Outline Hidden Markov Model (HMM) Conditional Random Field (CRF) Structured Perceptron/SVM HMM Just the assumption •How you generate a sentence? of HMM Step 1 • Generate a PO