1101_sent_parsng

1101_sent_parsng - 9.591 Parsing algorithms sentence...

Info iconThis preview shows pages 1–25. Sign up to view the full content.

View Full Document Right Arrow Icon
9.591 - Parsing algorithms, sentence complexity Ted Gibson November 1, 2004 (with thanks to Florian Wolf for these slides) Gibson lab, MIT
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
1. What is parsing? 2. Parsing strategies 1. Top-down 2. Bottom-up 3. Left-corner 4. Chart parsing Gibson lab, MIT
Background image of page 2
What is parsing? (1) The dog bites the man. (2) The man bites the dog. (3) *The dog bites man the. (1) = boring, (2) = interesting, (3) = not English Gibson lab, MIT
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
What is parsing? (1), (2), and (3) have the same words BUT: structure different Î different meaning for (1) and (2) Not every sentence structure is possible: (3) A grammar tells you what are possible sentence structures in a language: –S NP VP –N P Det N –e t c . Gibson lab, MIT
Background image of page 4
Why is parsing hard? Infinite number of possible sentences We can understand and say sentences we never heard before. Therefore representations of sentences’ meanings cannot be stored in and retrieved from memory. Ambiguity The man saw the woman on the hill with the telescope . Word-level ambiguity: saw Phrase-level ambiguity: PP-attachments Gibson lab, MIT
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
What is parsing? Parsing: discover how the words in a sentence can combine, using the rules in a grammar. Gibson lab, MIT
Background image of page 6
What is parsing? Parser sentence Æ representation of meaning Generator representation of meaning Æ sentence Gibson lab, MIT
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Parsing strategies - intro Our grammar: S Æ NP VP NP Æ Det Noun VP Æ Verb NP Det Æ the Noun Æ man Noun Æ woman Verb Æ likes Verb Æ meets Gibson lab, MIT
Background image of page 8
Parsing strategies - intro Our sentence: The man likes the woman Gibson lab, MIT
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Parsing strategies - intro Our grammar is unambiguous: not the case in real life –V P Æ Verb I walk Æ Verb NP I eat the apple Æ VP PP I see you with the telescope Gibson lab, MIT
Background image of page 10
Parsing strategies - intro How to deal with ambiguity in grammar: Serial approach: try one rule at a time, then backtrack if necessary need to specify with which rule to start Parallel approach: work on all alternative rules need data structure that can contain set of parse trees Gibson lab, MIT
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Parsing strategies - intro Top-down parsing: Start by looking at rules in grammar See what of the input is compatible with the rules Bottom-up parsing: Start by looking at input See which rules in grammar apply to input Combination of top-down and bottom-up: Only look at rules that are compatible with input Gibson lab, MIT
Background image of page 12
Top-down - intro Assume that input will eventually form a sentence Invoke S-node and all its possible extensions Keep expanding nodes until you find matching input Stack: keep track of what you still need to find in order to get grammatical sentence Gibson lab, MIT
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Top-down - intro If a LHS has more than one RHS: non-deterministic = parser does not specify expansion order deterministic = parser specifies expansion order Gibson lab, MIT
Background image of page 14
Top-down - example S Gibson lab, MIT
Background image of page 15

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Top-down - example S NP VP Gibson lab, MIT
Background image of page 16
Top-down - example S Det Noun NP VP Gibson lab, MIT
Background image of page 17

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Top-down - example S Det Noun NP VP The Gibson lab, MIT
Background image of page 18
Top-down - example S Det Noun NP VP The man Gibson lab, MIT
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Top-down - example S NP VP Det Noun The man Verb NP Gibson lab, MIT
Background image of page 20
Top-down - example S Det Noun Verb NP NP VP The man likes Gibson lab, MIT
Background image of page 21

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Top-down - example S Det Noun Verb Det Noun NP NP VP The man likes Gibson lab, MIT
Background image of page 22
Top-down - example S Det Noun Verb Det Noun NP NP VP The man likes the Gibson lab, MIT
Background image of page 23

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Top-down - example S Det Noun Verb Det Noun NP NP VP The man likes the woman Gibson lab, MIT
Background image of page 24
Image of page 25
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page1 / 105

1101_sent_parsng - 9.591 Parsing algorithms sentence...

This preview shows document pages 1 - 25. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online