HELP TPARSE                                       Aaron Sloman, May 1986

    lib tparse;

This is a variant of LIB GRAMMAR which is somewhat more general in that
it will find all the parses of a list of words corresponding to a
non-terminal. In order to do this it makes use of the POP-11 process
mechanism. (See HELP * PROCESS).

See HELP * GRAMMAR for a brief summary of LIB * GRAMMAR, TEACH * GRAMMAR
gives more details.

LIB TPARSE assumes that a grammar and a lexicon have been set up, as for
LIB GRAMMAR.

E.g.

    vars grammar lexicon;
    [
        [s  [np vp]   ]
        [vp [v np] [v np prep np]    ]
        [np [pn] [det noun] [np and np] [np prep np]]
    ] -> grammar;

    [
        [noun   cat dog mouse man girl boy book tree]
        [pn     fred aaron steve john sharon]
        [v      liked killed thanked bought ate put]
        [prep   on over in at under]
        [det    each every the a some]
    ] -> lexicon;

The procedure setup makes these the grammar and lexicon for use by the
program:

    setup(grammar,lexicon);

Several procedures are available. showparses takes a non-terminal symbol
and a list of words, and prints out all possible parses of initial
segments of the list, and the corresponding unused words. E.g.

    showparses("s", [the man ate the dog ]);
    ** [s [np [det the] [noun man]] [vp [v ate] [np [det the] [noun dog]]]]
    ** [unused []]
    ** [no more parses]

    showparses("s", [the man ate the dog and the mouse]);
    ** [s [np [det the] [noun man]] [vp [v ate] [np [det the] [noun dog]]]]
    ** [unused [and the mouse]]
    ** [s [np [det the] [noun man]]
          [vp [v ate]
              [np [np [det the] [noun dog]] and [np [det the] [noun mouse]]]]]
    ** [unused []]
    ** [no more parses]

listallparses is similar, except that it makes a list of two element
lists, each containing a parse tree and a list of unused words, e.g.

    listallparses("s", [the man ate the dog and the mouse])==>

    ** [[[s [np [det the] [noun man]] [vp [v ate] [np [det the] [noun dog]]]]
         [and the mouse]]
        [[s [np [det the] [noun man]]
            [vp [v ate]
               [np [np [det the] [noun dog]] and [np [det the] [noun mouse]]]]]
         []]]

Whereas listparses makes a list of parse trees for the whole list of
words: e.g.

    listparses("s", [the man ate the dog and the mouse])==>

    ** [[s [np [det the] [noun man]]
           [vp [v ate]
               [np [np [det the] [noun dog]] and [np [det the] [noun mouse]]]]]]

Not that this list contains just one parse tree:
 (Warning - this may take a minute or two to complete)

    listparses("s", [the man put the cat on the book in the tree]) ==>
    ** [[s [np [det the] [noun man]]
           [vp [v put]
               [np [np [det the] [noun cat]]
                   [prep on]
                   [np [np [det the] [noun book]]
                       [prep in]
                       [np [det the] [noun tree]]]]]]
        [s [np [det the] [noun man]]
           [vp [v put]
               [np [np [np [det the] [noun cat]]
                       [prep on]
                       [np [det the] [noun book]]]
                   [prep in]
                   [np [det the] [noun tree]]]]]
        [s [np [det the] [noun man]]
           [vp [v put]
               [np [det the] [noun cat]]
               [prep on]
               [np [np [det the] [noun book]]
                   [prep in]
                   [np [det the] [noun tree]]]]]
        [s [np [det the] [noun man]]
           [vp [v put]
               [np [np [det the] [noun cat]]
                   [prep on]
                   [np [det the] [noun book]]]
               [prep in]
               [np [det the] [noun tree]]]]]

    listallparses("s", [the man put the cat on the book in the tree]) ==>
    ** [[[s [np [det the] [noun man]] [vp [v put] [np [det the] [noun cat]]]]
         [on the book in the tree]]
        [[s [np [det the] [noun man]]
            [vp [v put]
                [np [np [det the] [noun cat]]
                    [prep on]
                    [np [det the] [noun book]]]]]
         [in the tree]]
        [[s [np [det the] [noun man]]
            [vp [v put]
                [np [np [det the] [noun cat]]
                    [prep on]
                    [np [np [det the] [noun book]]
                        [prep in]
                        [np [det the] [noun tree]]]]]]
         []]
        [[s [np [det the] [noun man]]
            [vp [v put]
                [np [np [np [det the] [noun cat]]
                        [prep on]
                        [np [det the] [noun book]]]
                    [prep in]
                    [np [det the] [noun tree]]]]]
         []]
        [[s [np [det the] [noun man]]
            [vp [v put]
                [np [det the] [noun cat]]
                [prep on]
                [np [det the] [noun book]]]]
         [in the tree]]
        [[s [np [det the] [noun man]]
            [vp [v put]
                [np [det the] [noun cat]]
                [prep on]
                [np [np [det the] [noun book]]
                    [prep in]
                    [np [det the] [noun tree]]]]]
         []]
        [[s [np [det the] [noun man]]
            [vp [v put]
                [np [np [det the] [noun cat]]
                    [prep on]
                    [np [det the] [noun book]]]
                [prep in]
                [np [det the] [noun tree]]]]
         []]]


The basic procedure is called appparses:

    appparses(symbol, list of words, procedure);

i.e. for every initial segment of list which is an instance of symbol,
the procedure is applied to the parse-tree and the list of unused words.

Thus this could be used to define listparses as follows:

    define listparses(symbol,wordlist);
        [%
          appparses(symbol,
                    wordlist,
                    procedure(tree,wordlist);
                        if null(wordlist) then tree endif
                    endprocedure)
        %]
    enddefine;


--- C.all/help/tparse
--- Copyright University of Sussex 1996. All rights reserved.