Method, device, computer program and computer program product for processing linguistic data in accordance with a formalized natural language

• US 8,515,733 B2
• Filed: 10/17/2007
• Issued: 08/20/2013
• Est. Priority Date: 10/18/2006
• Status: Expired due to Fees

• Current Assignee: Calculemus BV
• Original Assignee: Calculemus BV
• Inventors: Jansen, Wilhelmus Johannes Josephus
• Primary Examiner(s): SPOONER, LAMONT M
• Application Number:
  US12/446,232
• Publication Number:
  US 20110010163A1
• Time in Patent Office: 2,134 Days
• Field of Search: 704/1, 704/2, 704 8- 10, 704/251, 704/257, 707706-708, 715/264
• US Class Current: 704/9
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars

Abstract

A method, device and computer program product for parsing linguistic input data by a computer system, in accordance with a grammar of a Formalized Natural Language. The grammar of the Formalized Natural language is a text grammar representing an infinite set of texts of type Text and is stored in electronic form in a computer readable medium constituting a text grammar device. This text grammar is defined by a set of four elements W, N, R and Text. W is a finite set of invariable words of type Word, to be used as terminal, elementary expressions of a text. N is a finite set of non-terminal help symbols, to be used for the derivation and the representation of texts. R is a finite set of inductive rules for the production of grammatical expressions of the Formalized Natural Language, and Text is an element of N and start-symbol for grammatical derivation of all texts of type Text of the Formalized Natural Language. Linguistic input data to be parsed are received from an input device acquired and parsed by the computer system in accordance with the Formalized Natural Language of the text grammar device. A physical representation of a syntactic and semantic structure of the parsed linguistic input data is provided by a data output device.

Claims

1. A computer controlled method for parsing linguistic input data in a computer, said method comprising the steps of:

storing, in a computer readable storage medium of said computer, data comprising a text grammar of a Formalized Natural Language, the text grammar comprising a finite set of elements that represent a set of texts of a human language, wherein said text grammar of said Formalized Natural Language defines a finite set W of words of type Word which is stored in a lexicon of type Lexicon, comprising a finite set of lexical words of type LexWord, in such a way that each lexical word is constructed as a tuple of one unique word form and a list of disjunctive grammar forms, and each grammar form is constructed as a tuple of a unique category and a list of disjunctive feature sequences, and every feature is constructed as a string referring to a set of disjunctive instances of a feature,

receiving by said computer said linguistic input data to be parsed,

parsing said received linguistic input data by said computer in accordance with said text grammar of said Formalized Natural Language, and

generating parsed linguistic output data that is consistent for every instance when said linguistic input data is provided to said computer.

2. The method according to claim 1, wherein said finite set of elements is a set of four elements W, N, R and Text, wherein:

W is a finite set of invariable words of type Word, to be used as terminal, elementary expressions of a text,

N is a finite set of non-terminal help symbols, to be used for the derivation and the representation of texts,

R is a finite set of inductive rules for the production of grammatical expressions of said Formalized Natural Language, and

Text is an element of N and start-symbol for grammatical derivation of all texts of type Text of said Formalized Natural Language.

3. The method according to claim 1, wherein said generating parsed linguistic output data comprises providing a gramprint of said linguistic input data, said gramprint comprising a physical representation of parsed syntactic and semantic relations between all words used in said linguistic input data parsed by said computer.

4. The method according to claim 1, wherein said generating parsed linguistic output data comprises providing a gramprint of said linguistic input data parsed by said computer, said gramprint comprising a physical representation of parsed syntactic and semantic relations between all phrases used in said linguistic input data parsed by said computer as an hierarchically ordered system of syntactically typed phrases wherein every phrase P is functionally transformed in a name referring to a set of phenomena realizing a property called “

to be P”

, in such a way that every syntactically determined phrase in particular, which is constructed with a determiner, is marked with a unique index indicating that said determined phrase is a name referring to a singleton set comprehending one phenomenon of which the unique identity can be distinguished and recognized by the properties named in the name of said singleton set, according to the grammar of said Formalized Natural Language of a text grammar device.

5. The method according to claim 1, wherein said generating parsed linguistic output data comprises providing a hierarchically ordered sequence of Venn-diagrams in such a way that every text sentence of said linguistic input data parsed by said computer is represented by a Venn diagram that represents every syntactical phrase P of said text sentence as a set S of phenomena realizing a property called “

to be P”

, according to the grammar of said Formalized Natural Language of a text grammar device.

6. The method according to claim 1, wherein said generating parsed linguistic output data comprises providing a hierarchically ordered sequence of Venn-diagrams in such a way that every text sentence of said linguistic input data parsed by said computer is represented by a Venn diagram that represents every syntactical phrase P of said text sentence as a set S of phenomena realizing a property called “

to be P”

, and wherein a Venn-diagram of a text sentence represents in particular every syntactically determined phrase P of a current text sentence as an indexed singleton set comprehending a phenomenon of which an identity can be distinguished and recognized by properties named in names of cross sections of said singleton set, according to the grammar of said Formalized Natural Language of a text grammar device.

7. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a word of type Word as an invariable molecule of one word form of type WordForm and one grammar form of type GrammarForm, wherein said word form is an invariable string of type String of at least one character of a character alphabet used in said Formalized Natural Language, and said grammar form is a set of grammatical features, said set containing a category of type Category and a meaning of type Meaning.

8. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a category of type Category as a string of at least one character of type Character, which string refers to a position that a corresponding word form of a current word of said linguistic input data parsed by said computer can take up in syntactic phrases according to said text grammar of said Formalized Natural Language of a text grammar device.

9. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a meaning of type Meaning by comprising a list of features of type Feature, which list of features refers to a choice of grammatical conditions wherein a categorically typed word form of a current word of said linguistic input data parsed by said computer can get a reference to an intended set of phenomena sharing a same property, according to said text grammar of said Formalized Natural Language of a text grammar device.

10. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a feature of type Feature as a code of type String which code refers to a single, determined grammatical condition that can be met by a set of instances of said condition, according to said text grammar of said Formalized Natural Language of a text grammar device.

11. The method according to claim 1, wherein said text grammar of said Formalized Natural Language axiomatically defines an elementary phrase as a word that is syntactically selected from a lexicon of said Formalized Natural Language and wherein all phrases are inductively generated according to a set of production rules of said text grammar of said Formalized Natural Language of a text grammar device.

12. The method according to claim 1, wherein said computer generates syntactic phrases of said Formalized Natural Language according to a set of recursive, context-free production rules of the form XP→

UP . . . ZP, wherein XP is a category-type of a phrase to be generated by a function xp taking respectively as arguments an element of types UP, . . . , ZP, in such a way that a result of said function xp is of type (String, (XP, [Meaning]), wherein String represents a concatenation of word forms of selected elements of the respective types UP, . . . , ZP and where Meaning represents a type of a result of grammatical rules for unification of meanings of said selected elements of type UP, . . . , ZP, according to said text grammar of said Formalized Natural Language of a text grammar device.

13. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a Formalized Natural Language as a set of all and only syntactic phrases of type Text, wherein a text of type Text is a recursive function of a modal sentence of a data type ModalSentence and a dependent list of type Text, according to said text grammar of said Formalized Natural Language of a text grammar device, such that:

→

Text→

ModalSentence[Text].

14. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines an element of a data type ModalSentence by representing a constructor of type Constructor and an abstract sentence of type AbstractSentence, where a constructor of type Constructor defines a concrete, modal form in which a given abstract sentence can be transformed by said computer by a function text, according to said text grammar of said Formalized Natural Language of a text grammar device, such that:

data ModalSentence=Constructor AbstractSentence.

15. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines an abstract sentence of type AbstractSentence is defined as a tuple of five arguments of a function text, wherein said arguments are respectively of types Integer, Topic, Subject, InflexionData and DictumPredicativum, wherein an integer of type Integer indicates an order of a current modal sentence in a recursively produced text, a topic of type Topic is a function of an already given determined phrase, a subject of type Subject is a function of a determined phrase that must be marked for casus 1, a dictum predicativum of type DictumPredicativum is a function of a definite verb phrase, not yet marked for any casus, with its objects and adjuncts, and inflexion data of type InflexionData is a set of coded strings needed for a functional production of a concrete modal form of a text sentence, according to said text grammar of said Formalized Natural Language of a text grammar device, such that:

type AbstractSentence=(Integer, Topic, Subject, InflexionData, DictumPredicativum).

16. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a dictum predicativum of type DictumPredicativum as a function of a definite verb phrase with its objects and adjuncts, in such a way that sub-phrases of said dictum predicativum are ordered as they have to be modally applied on a grammatical subject of type Subject, according to a current constructor of said text grammar of said Formalized Natural Language of a text grammar device.

17. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a subject of type Subject as a function of a determined phrase that must be marked for casus 1 and to which the dictum predicativum has to be modally applied by a function text in order to generate a concrete modal sentence in accordance with a given constructor according to said text grammar of said Formalized Natural Language of a text grammar device, such that:

→

ModalSentence→

AbstractSentence ConcreteSentence.

18. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a topic of type Topic as an already given determined indexed phrase, performing as an element of an abstract sentence heading a text T and acting as a title of said text T, in such a way that every topic has to be explicitly present in said text T, and such that every topic of a dependent text must be borrowed from a previous concrete modal sentence higher up in a text tree according to said text grammar of said Formalized Natural Language of a text grammar device.

19. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a function sem by assigning a semantic value to every word w of said linguistic input data parsed by said computer, which word w has been syntactically selected as an elementary phrase of a text sentence, such that said word w becomes a name referring to a set of well known phenomena realizing a same property called “

to be w”

, which name is interpretable under given contextual conditions according to said text grammar of said Formalized Natural Language of a text grammar device, in such a way that said function sem projects a set of elementary phrases to a set called Universum Nominale comprising all elementary names, each name referring to a set of well known phenomena that are presumed to realize the property called “

to be w”

.

20. The method according to claim 1, wherein said text grammar of said Formalized Natural Language defines a function sem inducing a function sem* assigning a semantic value to every syntactic phrase of a text T, such that each phrase becomes a name referring to a set of phenomena identified within said text T and presumed to be realizing a property called with said name, which name is interpretable by said computer under given contextual conditions according to said text grammar of said Formalized Natural Language of a text grammar device, in such a way that said function sem* projects a set of all syntactic phrases to a set of all grammatical names, each name referring to a set of well known phenomena that are presumed to realize a property called with said name, such that said function sem* reveals an assumed set called Universum Reale comprising concrete and abstract, well known phenomena to which users of said Formalized Natural Language may refer with grammatical names, each name being an element of a sub-set of a power-set of the Universum Nominale.

21. The method according to claim 1, wherein an expression of said Formalized Natural Language is lazily parsed by said computer by producing successively grammatical sub-phrases of said expression using said text grammar of said Formalized Natural Language, said parsing providing as output data by a data output device one of a group comprising a failure message and at least one parse tree of said given expression.

22. The method according to claim 1, wherein a unique index-number is assigned to each identified phenomenon for which a syntactically determined name has been created in a text T being parsed by said computer in accordance with said text grammar of said Formalized Natural Language, in such a way that an indexed identity is identified by said name with said index-number and that different names for a same identity receive a same index-number within text T.

23. The method according to claim 1, wherein a grammar-based spelling check is executed during said parsing process by said computer in accordance with said text grammar of said Formalized Natural Language, in such a way that grammatical mistakes against one rule of a group comprising production-rules, unification rules and indexing rules of a text grammar device are reported and respective optional improvements are suggested by a data output device.

24. The method according to claim 1, wherein a lexicon of a Formalized Natural Language is extended during said parsing process by said computer in accordance with said text grammar of said Formalized Natural Language in case a string of characters is read but is not recognized as an element of said lexicon and said string should be recognized as a legal element of type Word of said Formalized Natural Language of a text grammar device.

25. The method according to claim 1, wherein a text T of a Formalized Natural Language is expressed as a computer program comprehending a hierarchically ordered domanium nominale DNT that is a proper sub-set of a power-set of a universum nominale UN and that refers unambiguously to a domanium reale DRT that is a presumed proper sub-set of an assumed universum reale UR, in such a way that there is a bijective relation between DNT and DRT.

26. The method according to claim 1, wherein spoken linguistic input data parsed by said computer in accordance with said text grammar of said Formalized Natural Language are outputted as written linguistic output data.

27. The method according to claim 1, wherein linguistic input data comprising a text of type Text of a first Formalized Natural Language are converted by said computer into an analogous text of a second Formalized Natural Language.

28. The method according to claim 1, wherein written linguistic input data parsed by said computer in accordance with said text grammar of said Formalized Natural Language of a text grammar device are outputted as spoken linguistic output data.

31. A computer program, comprising program code stored in a computer-readable storage medium that is not a signal, the program code executable by a computer to carry out the method according to claim 1.

32. A computer program, comprising program code stored in a computer-readable storage medium that is not a signal, the program code executable by a computer to carry out the method according to claim 1, the program code arranged for being integrated in or added to a computer application for joint execution of said computer program and said computer application by said computer.

33. A computer program product comprising program code stored on computer-readable storage medium that is not a signal, the program code readable by a computer, which operates to carry out the method according to claim 1, when said program code is executed by said computer.

34. A computer program product comprising program code stored on computer-readable storage medium that is not a signal, the program code readable by a computer, which is operable to carry out the method according to claim 1, and further comprising integrating said program code into a computer application for joint execution of said program code and said computer application by said computer.

29. A device comprising:

a computer readable storage medium storing data comprising a text grammar of a Formalized Natural language, the text grammar comprising a finite set of elements that represent a set of texts of a human language, wherein said text grammar of said Formalized Natural Language defines a finite set W of words of type Word which is stored in a lexicon of type Lexicon, comprising a finite set of lexical words of type LexWord, in such a way that each lexical word is constructed as a tuple of one unique word form and a list of disjunctive grammar forms, and each grammar form is constructed as a tuple of a unique category and a list of disjunctive feature sequences, and every feature is constructed as a string referring to a set of disjunctive instances of a feature,

an input device for receiving said linguistic input data to be parsed,

processing means for parsing said received linguistic input data in accordance with said text grammar of said Formalized Natural Language, and

a data output device for outputting parsed linguistic data, wherein said parsed linguistic data is consistent for every instance when said linguistic input data is provided to said input device.

30. A computer readable storage medium that is not a transient signal, the computer readable medium storing instructions for parsing linguistic input data, said instructions executable for:

receiving said linguistic input data to be parsed,

parsing said linguistic input data using a lexicon of a Formalized Natural Language, the lexicon accessible by a computer for parsing linguistic input data, the lexicon comprising a finite set of lexical words of type LexWord in such a way that one lexical word contains all the different words of type Word with one common word form, which word form is able to satisfy a given list of different disjunctive grammar forms, each grammar form given as a tuple of a unique category and a list of all its disjunctive meanings, every meaning constituting a feature sequence wherein each feature is constructed as a string indicating disjunctive instances of said feature, and outputting parsed linguistic data, wherein said parsed linguistic data is consistent for every instance when said linguistic input data is provided to said computer.