xử lý ngôn ngữ tự nhiên,regina barzilay,ocw mit edu 6 891 Lecture 4 (September 20, 2005) Parsing and Syntax II CuuDuongThanCong com https //fb com/tailieudientucntt http //cuuduongthancong com?src=pdf[.]
6.891: Lecture (September 20, 2005) Parsing and Syntax II CuuDuongThanCong.com https://fb.com/tailieudientucntt Overview • Weaknesses of PCFGs • Heads in context-free rules • Dependency representations of parse trees • Two models making use of dependencies CuuDuongThanCong.com https://fb.com/tailieudientucntt Weaknesses of PCFGs • Lack of sensitivity to lexical information • Lack of sensitivity to structural frequencies CuuDuongThanCong.com https://fb.com/tailieudientucntt S VP NP NNP Vt NP IBM bought NNP Lotus PROB = P (S � NP VP | S) ×P (VP � V NP | VP) ×P (NP � NNP | NP) ×P (NP � NNP | NP) CuuDuongThanCong.com ×P (NNP � IBM | NNP) ×P (Vt � bought | Vt) ×P (NNP � Lotus | NNP) https://fb.com/tailieudientucntt Another Case of PP Attachment Ambiguity (a) S VP NP NNS PP VP workers VBD NP IN dumped NNS into sacks CuuDuongThanCong.com NP DT a NN bin https://fb.com/tailieudientucntt (b) S VP NP NNS workers NP VBD dumped PP NP NNS IN sacks into NP DT a CuuDuongThanCong.com NN bin https://fb.com/tailieudientucntt (a) Rules S � NP VP NP � NNS VP � VP PP VP � VBD NP NP � NNS PP � IN NP NP � DT NN NNS � workers VBD � dumped NNS � sacks IN � into DT � a NN � bin (b) Rules S � NP VP NP � NNS NP � NP PP VP � VBD NP NP � NNS PP � IN NP NP � DT NN NNS � workers VBD � dumped NNS � sacks IN � into DT � a NN � bin If P (NP � NP PP | NP) > P (VP � VP PP | VP) then (b) is more probable, else (a) is more probable Attachment decision is completely independent of the words CuuDuongThanCong.com https://fb.com/tailieudientucntt A Case of Coordination Ambiguity (a) NP NP PP NP NNS IN NP dogs in NNS CC NP and NNS cats houses CuuDuongThanCong.com https://fb.com/tailieudientucntt (b) NP PP NP NNS dogs NP IN in NP CC NNS and houses CuuDuongThanCong.com NP NNS cats https://fb.com/tailieudientucntt (a) Rules NP � NP CC NP NP � NP PP NP � NNS PP � IN NP NP � NNS NP � NNS NNS � dogs IN � in NNS � houses CC � and NNS � cats (b) Rules NP � NP CC NP NP � NP PP NP � NNS PP � IN NP NP � NNS NP � NNS NNS � dogs IN � in NNS � houses CC � and NNS � cats Here the two parses have identical rules, and therefore have identical probability under any assignment of PCFG rule probabilities CuuDuongThanCong.com https://fb.com/tailieudientucntt