lecture parsing

Upload: jitendra-kumar

Post on 05-Apr-2018

225 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Lecture Parsing

    1/40

    PARSING WITHCONTEXT-FREE GRAMMARS

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    2/40

    PARSING

    Parsing is the process of recognizing and

    assigning STRUCTURE Parsing a string with a CFG:

    Finding a derivation of the string consistent withthe grammar

    The derivation gives us a PARSE TREE

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    3/40

    EXAMPLE (CFR LAST WEEK)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    4/40

  • 7/31/2019 Lecture Parsing

    5/40

    TOP-DOWN AND BOTTOM-UP

    SEARCH STRATEGIES

    The search has to be guided by the INPUT

    and the GRAMMAR TOP-DOWN search: the parse tree has to

    be rooted in the start symbol S

    EXPECTATION-DRIVEN parsing

    BOTTOM-UP search: the parse tree must bean analysis of the input

    DATA-DRIVEN parsing

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    6/40

    AN EXAMPLE OF TOP-DOWN SEARCH

    (IN PARALLEL)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    7/40

    AN EXAMPLE OF BOTTOM-UP

    SEARCH

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    8/40

  • 7/31/2019 Lecture Parsing

    9/40

    TOP-DOWN, DEPTH-FIRST,

    LEFT-TO-RIGHT

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    10/40

  • 7/31/2019 Lecture Parsing

    11/40

    TOP-DOWN, DEPTH-FIRST,

    LEFT-TO-RIGHT (III)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    12/40

    TOP-DOWN, DEPTH-FIRST,

    LEFT-TO-RIGHT (IV)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    13/40

    A T-D, D-F, L-R PARSER

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    14/40

    TOP-DOWN vs BOTTOM-UP

    TOP-DOWN: Only search among grammatical answers

    BUT: suggests hypotheses that may not beconsistent with data

    Problem: left-recursion

    BOTTOM-UP: Only forms hypotheses consistent with data

    BUT: may suggest hypotheses that make nosense globally

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    15/40

    LEFT-RECURSION

    A LEFT-RECURSIVE grammar may cause a

    T-D, D-F, L-R parser to never return Examples of left-recursive rules:

    NP NP PP

    S S and S

    But also: NP Det Nom

    DetNPs

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    16/40

  • 7/31/2019 Lecture Parsing

    17/40

    LEFT-RECURSION: POOR SOLUTIONS

    Rewrite the grammar to a weakly equivalent

    one Problem: may not get correct parse tree

    Limit the depth during search

    Problem: limit is arbitrary

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    18/40

    LEFT-CORNER PARSING

    A hybrid of top-down and bottom-up parsing

    Strategy: dont consider any expansionunless the current input can serve as theLEFT-CORNER of that expansion

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    19/40

  • 7/31/2019 Lecture Parsing

    20/40

    COMMON STRUCTURAL AMBIGUITIES

    COORDINATION ambiguity

    OLD (MEN AND WOMEN) vs(OLD MEN) AND WOMEN

    ATTACHMENT ambiguity:

    Gerundive VP attachment ambiguity

    I saw the Eiffel Tower flying to Paris PP attachment ambiguity

    I shot an elephant in my pajamas

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    21/40

    PP ATTACHMENT AMBIGUITY

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    22/40

  • 7/31/2019 Lecture Parsing

    23/40

  • 7/31/2019 Lecture Parsing

    24/40

    INVARIANTS AND TOP-DOWN

    PARSING

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    25/40

  • 7/31/2019 Lecture Parsing

    26/40

    DYNAMIC PROGRAMMING

    A standard T-D parser would reanalyze AFLIGHT 4 times, always in the same way

    A DYNAMIC PROGRAMMING algorithmuses a table (the CHART) to avoid repeatingwork

    The Earley algorithm also Does not suffer from the left-recursion problem

    Solves an exponential problem in O(n3)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    27/40

    THE CHART

    The Earley algorithm uses a table (the CHART) ofsize N+1, where N is the length of the input

    Table entries sit in the `gaps between words

    Each entry in the chart is a list of

    Completed constituents

    In-progress constituents

    Predicted constituents

    All three types of objects are represented in thesame way as STATES

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    28/40

    THE CHART:

    GRAPHICAL REPRESENTATION

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    29/40

    STATES

    A state encodes two types of information: How much of a certain rule has been encountered

    in the input Which positions are covered

    A, [X,Y]

    DOTTED RULES VP V NP

    NP Det Nominal

    S VP

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    30/40

    EXAMPLES

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    31/40

    SUCCESS

    The parser has succeeded if entry N+1 of thechart contains the state

    S, [0,N]

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    32/40

  • 7/31/2019 Lecture Parsing

    33/40

  • 7/31/2019 Lecture Parsing

    34/40

  • 7/31/2019 Lecture Parsing

    35/40

  • 7/31/2019 Lecture Parsing

    36/40

    EXAMPLE:

    BOOK THAT FLIGHT

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    37/40

    EXAMPLE:

    BOOK THAT FLIGHT (II)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    38/40

  • 7/31/2019 Lecture Parsing

    39/40

    EXAMPLE:

    BOOK THAT FLIGHT (IV)

    - by FaaDoOEngineers.com

  • 7/31/2019 Lecture Parsing

    40/40