³ò 4ÒÇIc@sàddkZddklZlZlZlZddkTddklZde fd„ƒYZ de fd„ƒYZ d e fd „ƒYZ d e fd „ƒYZ d „Zd„Zd„Zd„Zedjo eƒndS(iÿÿÿÿN(tDependencyProductiontDependencyGrammartStatisticalDependencyGrammartparse_dependency_grammar(t*(tpformattDependencySpancBsVeZdZd„Zd„Zd„Zd„Zd„Zd„Zd„Z d„Z RS( sW A contiguous span over some part of the input string representing dependency (head -> modifier) relationships amongst words. An atomic span corresponds to only one word so it isn't a 'span' in the conventional sense, as its _start_index = _end_index = _head_index for concatenation purposes. All other spans are assumed to have arcs between all nodes within the start and end indexes of the span, and one head index corresponding to the head word for the entire span. This is the same as the root node if the dependency structure were depicted as a graph. cCsR||_||_||_||_t|||t|ƒfƒ|_||_dS(N(t _start_indext _end_indext _head_indext_arcsthashttuplet_hasht_tags(tselft start_indext end_indext head_indextarcsttags((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__init__ s     !cCs|iS(sn @return: An value indexing the head of the entire C{DependencySpan}. @rtype: C{int}. (R (R((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR(scCsd|i|i|ifS(so @return: A concise string representatino of the C{DependencySpan}. @rtype: C{string}. sSpan %d-%d; Head Index: %d(RRR (R((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__repr__/scCsed|i|i|if}xBtt|iƒƒD]+}|d||i||i|f7}q2W|S(so @return: A verbose string representation of the C{DependencySpan}. @rtype: C{string}. sSpan %d-%d; Head Index: %ds %d <- %d, %s(RRR trangetlenR R(Rtstrti((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__str__6s )cCs\t||iƒoI|i|ijo6|i|ijo#|i|ijo|i|ijS(sk @return: true if this C{DependencySpan} is equal to C{other}. @rtype: C{boolean}. (t isinstancet __class__RRR R (Rtother((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__eq__@s cCs ||j S(sk @return: false if this C{DependencySpan} is equal to C{other} @rtype: C{boolean} ((RR((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__ne__KscCsLt||iƒpdSnt|i|i|if|i|i|ifƒS(s” @return: -1 if args are of different class. Otherwise returns the cmp() of the two sets of spans. @rtype: C{int} iÿÿÿÿ(RRtcmpRR R(RR((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__cmp__RscCs|iS(sD @return: The hash value of this C{DependencySpan}. (R (R((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyt__hash__[s( t__name__t __module__t__doc__RRRRRR R"R#(((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRs      t ChartCellcBs2eZdZd„Zd„Zd„Zd„ZRS(så A cell from the parse chart formed when performing the CYK algorithm. Each cell keeps track of its x and y coordinates (though this will probably be discarded), and a list of spans serving as the cell's entries. cCs%||_||_tgƒ|_dS(s“ @param x: This cell's x coordinate. @type x: C{int}. @param y: This cell's y coordinate. @type y: C{int}. N(t_xt_ytsett_entries(Rtxty((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRks  cCs|ii|ƒdS(s Appends the given span to the list of spans representing the chart cell's entries. @param span: The span to add. @type span: C{DependencySpan}. N(R+tadd(Rtspan((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR.vscCsd|i|i|ifS(sk @return: A verbose string representation of this C{ChartCell}. @rtype: C{string}. s CC[%d,%d]: %s(R(R)R+(R((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR€scCsd|S(sk @return: A concise string representation of this C{ChartCell}. @rtype: C{string}. s%s((R((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR‡s(R$R%R&RR.RR(((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR'es  tProjectiveDependencyParsercBs)eZdZd„Zd„Zd„ZRS(s« A projective, rule-based, dependency parser. A ProjectiveDependencyParser is created with a DependencyGrammar, a set of productions specifying word-to-word dependency relations. The parse() method will then return the set of all parses, in tree representation, for a given input sequence of tokens. Each parse must meet the requirements of the both the grammar and the projectivity constraint which specifies that the branches of the dependency tree are not allowed to cross. Alternatively, this can be understood as stating that each parent node and its children in the parse tree form a continuous substring of the input sequence. cCs ||_dS(s Create a new ProjectiveDependencyParser, from a word-to-word dependency grammar C{DependencyGrammar}. @param dependency_grammar: A word-to-word relation dependencygrammar. @type dependency_grammar: A C{DependencyGrammar}. N(t_grammar(Rtdependency_grammar((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR¡scCsnt|ƒ|_g}x¼tdt|iƒdƒD]ž}|igƒxˆtdt|iƒdƒD]j}||it||ƒƒ||djo9|||it|d||ddgdgƒƒqbqbWq2WxÐtdt|iƒdƒD]²}x©t|dddƒD]‘}xˆt|d|dƒD]p}xg|||iD]T}xK|||iD]8}x/|i ||ƒD]}|||i|ƒqtWq[WqCWq+WqWqñWg} g} x´|t|iƒdiD]˜} d} x]tt|ƒƒD]I}| d|d||||ddd| i |ddddf 7} qíWt | ƒ} | i| ƒ| i| i ƒƒqÎW| S( sU Performs a projective dependency parse on the list of tokens using a chart-based, span-concatenation algorithm similar to Eisner (1996). @param tokens: The list of input tokens. @type tokens:a C{list} of L{String} @return: A list of parse trees. @rtype: a C{list} of L{tree} iiiÿÿÿÿtnullits %d %s %s %s %s %s %d %s %s %s t-( tlistt_tokensRRtappendR'R.RR+t concatenateR tDependencyGraphttree(RttokenstchartRtjtktspan1tspan2tnewspantgraphsttreestparset conll_formattdg((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRE«sF  A-G  cCsYg}|i|ijo dGHn|i|ijo|}|}|}n|i|i}|i|i}|ii|i|i|i|iƒoC|i||i|i<|it|i|i |i||ƒƒn|i|i}|ii|i|i|i|iƒoC|i||i|i<|it|i|i |i||ƒƒn|S(s Concatenates the two spans in whichever way possible. This includes rightward concatenation (from the leftmost word of the leftmost span to the rightmost word of the rightmost span) and leftward concatenation (vice-versa) between adjacent spans. Unlike Eisner's presentation of span concatenation, these spans do not share or pivot on a particular word/word-index. return: A list of new spans formed through concatenation. rtype: A C{list} of L{DependencySpan} s8Error: Mismatched spans - replace this with thrown error( RR RR1tcontainsR7R R8RR(RR@RAtspanst temp_spantnew_arcstnew_tags((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR9Ós"   *,*,(R$R%R&RRER9(((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR0”s  (t'ProbabilisticProjectiveDependencyParsercBs;eZdZd„Zd„Zd„Zd„Zd„ZRS(s A probabilistic, projective dependency parser. This parser returns the most probable projective parse derived from the probabilistic dependency grammar derived from the train() method. The probabilistic model is an implementation of Eisner's (1996) Model C, which conditions on head-word, head-tag, child-word, and child-tag. The decoding uses a bottom-up chart-based span concatenation algorithm that's identical to the one utilized by the rule-based projective parser. cCs dGHdS(sq Create a new probabilistic dependency parser. No additional operations are necessary. R4N((R((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRscCs)t|ƒ|_g}xtdt|iƒdƒD]ø}|igƒxâtdt|iƒdƒD]Ä}||it||ƒƒ||djo“|iii||dƒo\xn|ii||dD];}|||i t |d||ddg|gƒƒqÊWq&d||dGHgSqbqbWq2WxÐtdt|iƒdƒD]²}x©t|dddƒD]‘}xˆt|d|dƒD]p}xg|||i D]T}xK|||i D]8}x/|i ||ƒD]} |||i | ƒqÎWqµWqWq…WqhWqKWg} g} t } d} x|t|iƒdi D]ç}d}d}x”tt|ƒƒD]€}|d||d|i|ddf7}|d |d|||||i||i|d|i|ddd d f 7}qYWt|ƒ}|i|ƒ}|| jo|iƒ} |} q4q4W| | gS( sZ Parses the list of tokens subject to the projectivity constraint and the productions in the parser's grammar. This uses a method similar to the span-concatenation algorithm defined in Eisner (1996). It returns the most probable parse derived from the parser's probabilistic dependency grammar. iiiÿÿÿÿs7No tag found for input token '%s', parse is impossible.iR4s %s %s %d %s R3s %d %s %s %s %s %s %d %s %s %s R5(R6R7RRR8R'R1Rthas_keyR.RR+R9tNoneR R:t compute_probR;(RR<R=RR>ttagR?R@RARBRCRDt max_parset max_scoreRERFt malt_formatRGtscore((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRE s\ =-)U   cCsig}|i|ijo dGHn|i|ijo|}|}|}n|i|i}|i|i}|ii|i|i|i|iƒoC|i||i|i<|it|i|i |i||ƒƒn|i|i}|i|i}|ii|i|i|i|iƒoC|i||i|i<|it|i|i |i||ƒƒn|S(s Concatenates the two spans in whichever way possible. This includes rightward concatenation (from the leftmost word of the leftmost span to the rightmost word of the rightmost span) and leftward concatenation (vice-versa) between adjacent spans. Unlike Eisner's presentation of span concatenation, these spans do not share or pivot on a particular word/word-index. return: A list of new spans formed through concatenation. rtype: A C{list} of L{DependencySpan} s8Error: Mismatched spans - replace this with thrown error( RR RR1RHR7R R8RR(RR@RARIRJRKRL((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyR9;s$   *,*,c CsÊg}h}h}xœ|D]”}x‹tdt|iƒƒD]q}|i|d}|i|ƒ}|i|ƒ} || } x/td|d| dƒD]} |i|d} |i|d} |i| ƒo|| i| ƒnt| gƒ|| | |}|djo.|i||d}|i||d}n| d jo6|i||dd}|i||dd}n|djo|it | |gƒƒnd |||| | f}d || | f}|i|ƒo||cd7 'cats' | 'walls' 'walls' -> 'the' 'cats' -> 'the' tthetcatstscratchtwallsN(RR0RE(tgrammartpdpRDR;((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRsäs  c CsâHdGHdGHdGHtdƒ}|GHHdGHdGHt|ƒ}|iddd dd d gƒ}x|D] }|GHq`WHd GHd GHdGHdGHtdƒ}|GHHdGHt|ƒ}|iddd dd d gƒ}x|D] }|GHqÏWdS(s A demonstration showing the creation of a C{DependencyGrammar} in which a specific number of modifiers is listed for a given head. This can further constrain the number of possible parses created by a C{ProjectiveDependencyParser}. s>A grammar with no arity constraints. Each DependencyProductions;specifies a relationship between one head word and only onesmodifier word.:sj 'fell' -> 'price' | 'stock' 'price' -> 'of' | 'the' 'of' -> 'stock' 'stock' -> 'the' s<For the sentence 'The price of the stock fell', this grammars(will produce the following three parses:Rvtpricetoftstocktfells0Comparatively, the following grammar contains a s5C{DependencyProduction} that specifies a relationships4between the one head word, 'price', and two modifierswords, 'of' and 'the'.sh 'fell' -> 'price' | 'stock' 'price' -> 'of' 'the' 'of' -> 'stock' 'stock' -> 'the' s:This constrains the number of possible parses to just one:N(RR0RE(RzR{RDR;((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pytarity_parse_demoõs<  !   !cCs¥g}tidƒD]}|o|t|ƒqq~}tƒ}dGH|i|ƒddddddg}d Gd i|ƒGd GH|i|ƒ}d GH|d GHdS(sU A demo showing the training and use of a projective dependency parser. s s6Training Probabilistic Projective Dependency Parser...tCathytzagthentwildtzwaaient.s Parsing 't s'...sParse:iN(t conll_data2tsplitR:RMRntjoinRE(t_[1]tentryRCtppdptsentRE((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pyRt$s4  t__main__(tmathtnltkRRRRtdependencygraphtpprintRtobjectRR'R0RMRuRsR€RtR$(((s;/p/zhu/06/nlp/nltk/nltk/parse/projectivedependencyparser.pys s " P/gã   /