JP2010134730A - Language processor, language processing system and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a language processor which acquires base information as a criteria for determining whether a word used in a sentence in combination with a word representing a number functions as either or both a numerical classifier or/and a noun, on the basis of sentence group data. <P>SOLUTION: A word acquisition means (21) acquires a word used in a sentence in combination with a word representing a number. A statistical information acquisition means (22) acquires statistical information relating to how to use the word, on the basis of sentence group data. A base information generation means (23) generates base information as a criteria for determining whether the word functions as either or both a numerical classifier or/and a noun, on the basis of the statistical information. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a language processing device, a language processing system, and a program.

  A word used in a sentence in combination with a word representing a number may function as one or both of a classifier and a noun. For example, in the sentence “40 members”, “40” indicates the number of members. In this case, “person” is an assist in modifying “members” in which the number “40” is a noun. Functions as a classifier. On the other hand, in the sentence “40 cleaning activities”, “40” does not indicate the number of cleaning activities and qualifies “person”. The “person” in this case represents the object itself directly modified by the numeral “40” and functions as a noun. Thus, a “person” can function as both a classifier and a noun. Further, for example, when the word “individual” is used in a sentence in combination with a word representing a number, it does not function as a noun but functions as a classifier. Further, for example, the word “SDF” may be used in combination with a word representing a number, such as “Joint training of 3 Self-Defense Forces”, but it does not function as a classifier but functions as a noun.

  For example, when performing language processing such as parsing in consideration of whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or both, Basic information is required as a basis for determining whether the word functions as one or both of a classifier and a noun.

In relation to this point, Patent Document 1 discloses a technique in which a correspondence table describing the semantic role of a classifier is stored in advance, and a sentence is analyzed based on the correspondence table. Patent Document 2 discloses a technique in which a classifier selected by a user according to a noun is stored in dictionary data in association with the noun, and the subsequent translation processing is performed based on the dictionary data.
Patent 4039282 Japanese Patent Laid-Open No. 2000-90088

  The present invention acquires, based on sentence group data, basic information that serves as a basis for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both An object of the present invention is to provide a language processing device, a language processing system, and a program.

  The invention according to claim 1 is a language processing apparatus, which relates to a word acquisition means for acquiring a word used in a sentence in combination with a word representing a number, and how the word is used based on sentence group data. Statistical information acquisition means for acquiring statistical information, and basic information for generating basic information based on the statistical information as a basis for determining whether the word functions as either a classifier or a noun or both And an information generating means.

  Invention of Claim 2 is the language processing apparatus of Claim 1, Comprising: The said statistical information acquisition means WHEREIN: In the said sentence group data, the said word is a predetermined 1st aspect corresponding to a classifier. And means for obtaining the number of times the word is used in a predetermined second mode corresponding to a noun in the sentence group data. And

  The invention according to claim 3 is the language processing apparatus according to claim 1 or 2, wherein the statistical information acquisition means includes means for acquiring the number of modifiers that modify the word in the sentence group data. It is characterized by including.

  Invention of Claim 4 is the language processing apparatus in any one of Claim 1 thru | or 3, Comprising: The analysis object character string acquisition means which acquires an analysis object character string, The said word in the said analysis object character string Is included, based on the basic information of the word, it is determined whether the word functions as either a classifier or a noun or both, and based on the determination result, the character to be analyzed Analyzing means for analyzing the column.

  The invention according to claim 5 is the language processing apparatus according to claim 4, wherein the analysis means includes analysis rule determination means for determining an analysis rule for the analysis target character string based on the determination result. And the analysis target character string is analyzed using the analysis rule determined by the analysis rule determination means.

  The invention according to claim 6 is a language processing system, which relates to a word acquisition means for acquiring a word used in a sentence in combination with a word representing a number, and how the word is used based on sentence group data. Statistical information acquisition means for acquiring statistical information, and basic information for generating basic information based on the statistical information as a basis for determining whether the word functions as either a classifier or a noun or both And an information generation means.

  The invention according to claim 7 is a program, the word acquisition means for acquiring a word used in a sentence in combination with a word representing a number, statistical information on how the word is used based on sentence group data Statistical information acquisition means for acquiring, and basic information generation for generating basic information based on the statistical information as a basis for determining whether the word functions as either a classifier or a noun or both A program for causing a computer to function as means.

  According to the first aspect of the present invention, the basic information serving as a basis for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both. Acquired based on sentence group data.

  According to the second aspect of the present invention, the basic information of a word used in a sentence in combination with a word representing a number is actually sentence group data in a predetermined first mode in which the word corresponds to a classifier. And the number of times actually used in the sentence group data in the predetermined second mode corresponding to the noun.

  According to the third aspect of the present invention, the basic information of the word used in the sentence in combination with the word representing the number is acquired in consideration of the number of modifiers that modify the word in the sentence group data.

  According to the fourth aspect of the invention, an analysis result in which whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or both can be obtained. .

  According to the invention described in claim 5, in accordance with an analysis rule that considers whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both An analysis result is obtained.

  According to the invention described in claim 6, basic information serving as a basis for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both. Acquired based on sentence group data.

  According to the seventh aspect of the present invention, basic information serving as a basis for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both. Acquired based on sentence group data.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating a configuration example of a language processing system 1 according to an embodiment of the present invention. A language processing system 1 shown in FIG. 1 includes a language processing device 10. The language processing apparatus 10 is, for example, a personal computer, and includes a control unit 11, a storage unit 12, an operation unit 13, and a display unit 14.

  The control unit 11 is, for example, a CPU or the like, and executes various types of information processing according to programs stored in the storage unit 12. The storage unit 12 includes a memory element such as a RAM and a ROM, a hard disk, and the like. The storage unit 12 holds programs executed by the control unit 11 and various data. In particular, in the present embodiment, the storage unit 12 stores dictionary and grammatical rule data used for morphological analysis and syntax analysis. The storage unit 12 also operates as a work memory for the control unit 11.

  The operation unit 13 is, for example, a keyboard or a mouse, and receives a user instruction operation, and outputs the content of the instruction operation to the control unit 11. The display unit 14 is, for example, a liquid crystal display, and displays an image in accordance with an instruction from the control unit 11.

  Hereinafter, functions realized by the language processing system 1 according to the present embodiment will be described. As shown in FIG. 2, the language processing system 1 functionally includes a word acquisition unit 21, a statistical information acquisition unit 22, a basic information generation unit 23, an analysis target character string acquisition unit 31, and an analysis unit 32. The analysis rule determination unit 33, the analysis result output unit 34, and the basic information storage unit 35 are included. The basic information storage unit 35 is realized by, for example, the storage unit 12, and the other functional blocks are realized by, for example, the control unit 11 executing a program stored in the storage unit 12. This program may be provided through communication means such as the Internet, or may be provided by being stored in various computer-readable information storage media such as CD-ROM and DVD-ROM.

  First, the word acquisition unit 21, the statistical information acquisition unit 22, and the basic information generation unit 23 will be described. These functional blocks are functions for generating basic information that serves as a basis for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both. It is a block.

Words used in sentences in combination with words that represent numbers are:
First group: one that can function as both a classifier and noun Second group: one that functions as a classifier Third group: one that functions as a noun Grouped into three groups. In the present embodiment, it is determined which of the first to third groups a word that can be used in a sentence in combination with a word representing a number, and information indicating the determination result is generated as the basic information.

  3 and 4 are flowcharts showing processing executed in the language processing system 1, and are flowcharts showing processing of the word acquisition unit 21, the statistical information acquisition unit 22, and the basic information generation unit 23. FIG. Hereinafter, these functional blocks will be described with reference to FIGS. 3 and 4.

  The word acquisition unit 21 acquires a word that can be used after a word representing a number.

Specifically, as shown in FIG. 3, first, the word acquisition unit 21 acquires sentence group data including a plurality of sentences, and executes analysis of the sentence group related to the sentence group data (S101). For example, the sentence group data is stored in the storage unit 12. Hereinafter, as a specific example, a case will be described in which a sentence group G including the following eight sentences a to h is processed.
a: “There were only two Japanese.”
b: “He bought one of them.”
c: “Other people don't know.”
d: “The Air Self Defense Force at Atsugi Base was dispatched to the disaster area yesterday.”
e: “Joint training of 3 Self-Defense Forces was conducted.”
f: “If 5 are stacked, the height is just right.”
g: “I joined the Self-Defense Force.”
h: “Three people passed in class.”

  In S101, morphological analysis and syntax analysis are performed on the sentence group. Known methods are used for morphological analysis and syntax analysis.

  After executing the morpheme analysis and the syntax analysis, the word acquisition unit 21 extracts a word (morpheme) immediately after the morpheme (word) representing the number from the morphemes constituting the sentence group, and lists those words L is acquired (S102). For example, in the case of the sentence group G, “person” in the sentences a and h, “individual” in the sentences b and f, and “Self Defense Force” in the sentence e are extracted as words immediately after the word representing the number. Is done. For this reason, the word list L acquired in S102 is {person, individual, self-defense force}.

  The statistical information acquisition unit 22 acquires statistical information regarding how the word acquired by the word acquisition unit 21 is used in the sentence group. For example, the statistical information acquisition unit 22 acquires the number of times that the word acquired by the word acquisition unit 21 is used in a predetermined first mode corresponding to the classifier in the sentence group (see S105). In addition, for example, the statistical information acquisition unit 22 acquires the number of times that the word acquired by the word acquisition unit 21 is used in a predetermined second mode corresponding to the noun in the sentence group (see S106).

Specifically, as shown in FIG. 3, the statistical information acquisition unit 22 initializes a variable i to 0 (S103). Then, the statistical information acquisition unit 22 counts the number P _i of times that the word L _i that is the i-th element of the word list L is used in the sentence group (S104). That is, the number P _{i of} times that the word L _i appears in the sentence group is counted. For example, when the variable i is 0, the number of times P ₀ in which the word “person” that is the 0th element (that is, the first element) of the word list L is used is counted. Since “person” is used once in the sentences a, c, and h, “3” is acquired as the value of P ₀ .

In addition, the statistical information acquisition unit 22 counts the number of times Q _{i in} which the word L _i is used after the word representing the number in the sentence group (S105). In other words, the number of times the word L _i in the sentence group is to appear immediately after the word representing the number Q _i is counted. For example, in the case of the sentence group G, “2” is acquired as the value of Q ₀ because “person” is immediately after the word representing the number in the sentences a and h.

When a word that functions as a classifier is used as a classifier, it is often used after a word that represents a number, compared to a word that does not function as a classifier. Therefore, a word that functions as a classifier has a larger Q _i value than a word that does not function as a classifier. For this reason, the value of Q _i is a statistical value for determining whether a word functions as a classifier.

The statistical information acquisition unit 22, the word L _i in sentence group is at a position other than immediately following the word representing the number, and counts the number of times R _i which is used to modify the predicate (S106). For example, in the case of the sentence group G, “person” in the sentence c is not immediately after the word representing the number, and the predicate is modified, so “1” is acquired as the value of R ₀ .

Here, when a word that can be used after a word that represents a number is used at a position other than immediately after the word that represents a number and is used to modify a predicate, the word functions as a noun. it seems to do. Therefore, the value of R _i is greater for words that function as nouns than for words that do not function as nouns. For this reason, the value of R _i is a statistical value for determining whether or not a word functions as a noun.

The statistical information acquisition unit 22 counts the number _{S i} of modifier that modifies the word _{L i} in sentence group (S107). For example, in the case of the sentence group G, there modifier that modifies the "person" one in the sentence c, and, since the other statements no modifier that modifies the "human", S ₀ “1” is acquired as the value of.

Here, when a word functions as a noun, it is often modified by a modifier. Thus, words that functions as a noun, as compared to a word that does not function as a noun, the value of S _i increases. Therefore, the value of S _i is a statistical value for the word to determine whether it functions as a noun.

  The basic information generation unit 23 obtains statistical information as basic information for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun or as both. Generated based on the statistical information acquired by the unit 22.

Specifically, as shown in FIG. 4, basic information generating section 23, the word _{L i} statistics _{(P i, Q i, R} i, S i) an evaluation value _{E i} of the word _{L i} based on the Calculate (S108). The evaluation value E _i is calculated by the following calculation formula (1).

E _i = (P _i * Q _i ) / (1 + R _i + S _i ) (1)

As described above, a word that functions as a classifier has a larger Q _i value than a word that does not function as a classifier. On the other hand, the value of R _i and S _i is greater for words that function as nouns than for words that do not function as nouns. For this reason, words belonging to the second group (functioning as a classifier) have a larger evaluation value E _i than words belonging to other groups, and words belonging to the third group (functioning as nouns) The evaluation value E _i is smaller than the words belonging to.

Further, as described above, since the statistical value (P ₀ , Q ₀ , R ₀ , S ₀ ) of “person” is ( ₃ , 2, ₁ , ₁ ), the evaluation value E ₀ of “person” is “ 2 "is acquired.

Thereafter, basic information generating unit 23 compares the evaluation value _{E i} of the word _{L i} with a reference value J (S109). As described above, since the evaluation value E _{i of the} words belonging to the third group is smaller than that of the words belonging to other groups, in this step, whether or not the evaluation value E _i of the word L _i is smaller than the reference value J. by determining, it is determined whether or not to classify the word L _i to a third group. That is, it is determined whether or not the word L _i should be classified as a noun.

  The reference value J is determined as follows, for example. That is, first, representative words that are known to function as nouns and not to function as classifiers when used in sentences in combination with words representing numbers are selected. Then, the evaluation value E of the word is calculated by executing the processing of S104 to S108 with the selected word as a processing target. Then, a reference value J is set based on the calculated evaluation value E. For example, a value near the calculated evaluation value E is set as the reference value J. In the following description, it is assumed that the reference value J is “1”.

If the evaluation value _{E i} of the word _{L i} is smaller than the reference value J (S109: Y), basic information generating unit 23 classifies the word _{L i} in the third group (S110). That is, the word L _i is classified as a noun.

On the other hand, if the evaluation value _{E i} of the word _{L i} is greater than or equal to the reference value J (S109: N), basic information generating unit 23 compares the evaluation value _{E i} of the word _{L i} and the reference value K (S111). As described above, since a word belonging to the second group has a higher evaluation value E _i than a word belonging to another group, in this step, whether or not the evaluation value E _i of the word L _i is larger than the reference value K. It is determined whether or not the word L _i should be classified into the second group. That is, it is determined whether or not the word L _i should be classified as a classifier.

  The reference value K is determined as follows, for example. That is, first, representative words that function as a classifier when used in a sentence in combination with a word representing a number and are known not to function as a noun are selected. Then, the evaluation value E of the word is calculated by executing the processing of S104 to S108 with the selected word as a processing target. Then, a reference value K is set based on the calculated evaluation value E. For example, a value near the calculated evaluation value E is set as the reference value J. In the following description, it is assumed that the reference value K is “3”.

If the evaluation value _{E i} of the word _{L i} is larger than the reference value K (S111: Y), basic information generating unit 23 classifies the word _{L i} in the second group (S112). That is, the word L _i is classified as a classifier.

On the other hand, if the evaluation value _{E i} of the word _{L i} is equal to or less than the reference value K (S111: N), basic information generating unit 23 classifies the word _{L i} in the first group (S113). That is, in this case, the word L _i is classified as both a classifier and a noun.

As described above, the evaluation value E ₀ of “person” is “2”, which is larger than the reference value J (1) and not more than the reference value K (3). It is classified into one group (see S113). That is, “person” is classified as both a classifier and a noun.

After classification of the word L _i is determined, 1 is added to the value of the variable i (S114), the value of the variable i is equal to or a number of elements less than M word list L is determined (S115). When the value of the variable i is less than the number M of elements in the word list L, the processes of S104 to S115 are executed.

For example, when the value of the variable i is 1, the processing of S104 to S113 is executed for the word “pieces” in the word list L. Since “individual” is used once in the sentences b and f in the sentence group G, “2” is acquired as the value of P ₁ (S104). In the sentences b and f, since “number” is after the word representing the number, “2” is acquired as the value of Q ₁ (S105). Further, in the sentences b and f, since “number” is after the word representing the number, “0” is acquired as the value of R ₁ (S106). In addition, since there is no modifier that modifies “pieces” in the sentences b and f, “0” is acquired as the value of S ₁ (S107). As a result, “4” is acquired as the evaluation value E ₁ of “pieces” (S108). Since the “individual” evaluation value E ₁ is larger than the reference value J (1) and larger than the reference value K (3), the “individual” is classified into the second group (S112). That is, “individual” is classified as a classifier.

For example, when the value of the variable i is 2, the processing of S104 to S113 is executed for the word “Self Defense Force” in the word list L. In the sentence group G, “SDF” is used once in the sentences d, e, and g, so “3” is acquired as the value of P ₂ (S104). In addition, since the word “SDF” follows the word representing the number in the sentence e, “1” is acquired as the value of Q ₂ (S105). Furthermore, in the sentences d and g, “Self Defense Force” is not after the word representing the number and the predicate is modified, so “2” is acquired as the value of R ₂ (S106). Further, since there is one modifier that modifies “SDF” in the sentence d, “1” is acquired as the value of S ₂ (S 107). As a result, "0.75" is obtained as the evaluation value _{E 2} of the "SDF" (S108). Since "SDF" evaluation value E ₂ of equal to or less than the reference value J (1), "SDF" are classified into a third group (S110). That is, “Self Defense Force” is classified as a noun.

  In the process of S115, when it is determined that the value of the variable i is equal to or greater than the number M of elements in the word list L, the process of S104 to S113 is completed for each word included in the word list L. This is a case where the classification of each word included in L is completed.

  The basic information generation unit 23 stores the classification result of each word acquired as described above in the basic information storage unit 35.

  Next, the analysis target character string acquisition unit 31, the analysis unit 32, the analysis rule determination unit 33, the analysis result output unit 34, and the basic information storage unit 35 will be described. These functional blocks are functional blocks for analyzing the structure of the character string to be analyzed.

  The analysis target character string acquisition unit 31 acquires an analysis target character string to be analyzed. The analysis target character string is a sentence written in a natural language, and may be input by the user using the operation unit 13 or may be stored in advance in the storage unit 12 or the like. Here, it is assumed that the character string to be analyzed is a Japanese sentence.

In the following, as a specific example
X: “Three students”
Y: “3 apples”
Z: “3 Self-Defense Forces training”
A case where three analysis target character strings X, Y, and Z are analyzed will be described.

  The analysis unit 32 analyzes the analysis target character string. First, the analysis unit 32 performs morphological analysis of the character string to be analyzed. For example, in the case of the character string X to be analyzed, morphemes “3”, “people”, “no”, “student” are obtained, and the part of speech of each morpheme “3”, “people”, “no”, “student” is obtained. Are analyzed as numbers, classifiers, unionized particles, and nouns, respectively.

  Next, the analysis rule determination unit 33 determines whether the morpheme immediately after the morpheme representing the number in the analysis target character string functions as one or both of the classifier and the noun. As described above, the basic information storage unit 35 stores information indicating which of the first to third groups a word that can be used after a word representing a number belongs. For this reason, the analysis rule determination unit 33 refers to the stored contents of the basic information storage unit 35.

  For example, in the case of the analysis target character string X, it is determined to which of the first to third groups “person” immediately after “3” belongs. As described above, since “person” belongs to the first group, it is determined that “person” belongs to the first group. That is, it is determined that “person” functions as both a classifier and a noun.

  Further, for example, in the case of the analysis target character string Y, it is determined which “group” immediately after “3” belongs to the first to third groups. As described above, since “pieces” belong to the second group, it is determined that “pieces” belong to the second group. That is, it is determined that “individual” functions as a classifier and does not function as a noun.

  Further, for example, in the case of the analysis target character string Z, it is determined to which of the first to third groups the “Self Defense Force” immediately after “3” belongs. As described above, since the “SDF” belongs to the third group, it is determined that the “SDF” belongs to the third group. That is, it is determined that “SDF” functions as a noun and does not function as a classifier.

  Then, the analysis rule determination unit 33 selects one or a plurality of analysis rules as the analysis rules used for the syntax analysis based on the determination result as to which of the first to third groups the morpheme immediately after the morpheme representing the number belongs. decide.

  For example, in the case of the analysis target character string X, since “person” is classified into the first group (functioning as both a classifier and a noun), an analysis rule that assumes that “person” is a classifier, and “ Two analysis rules are selected, the analysis rule assuming that "person" is a noun.

  FIG. 5 shows an example of an analysis rule on the assumption that “person” is a classifier. FIG. 5A shows grammar rules. Since all of the analysis target character strings X to Z are noun phrases, the grammatical rules for the noun phrase (Nadj) are shown in FIG. 5A, and the rules for the word order to be satisfied by the morphemes included in the noun phrase It is shown. Here, the symbol “|” means “OR”, which means that one of the elements surrounded by curly brackets and separated by the symbol “|” is selected alternatively. “PP”, “N”, “NUMBER”, and “CL” indicate the vocabulary type to which the morpheme belongs. FIG. 5B shows a dictionary that defines the correspondence between morphemes and vocabulary types, and shows to which vocabulary types the morphemes belong. The analysis rule shown in FIG. 5 is an analysis rule on the assumption that “person” is a classifier. Therefore, in FIG. 5B, “person” belongs to the vocabulary type “CL” corresponding to the classifier. ing.

  FIG. 6 shows an example of an analysis rule on the assumption that “person” is a noun. FIG. 6A shows the grammatical rules of the noun phrase (Nadj), and shows the word order rules that the morphemes included in the noun phrase should satisfy. FIG. 6B shows a dictionary that defines the correspondence between morphemes and vocabulary types, and shows to which vocabulary types the morphemes belong. The analysis rule shown in FIG. 6 is an analysis rule that assumes that “person” is a noun, and therefore, in FIG. 6B, “person” belongs to the vocabulary type “N” corresponding to the noun. ing. In this respect, FIG. 6B is different from FIG.

  Further, for example, in the case of the analysis target character string Y, since “persons” are classified into the second group (functions as classifiers), one analysis rule that assumes that “persons” are classifiers is selected. . In this analysis rule, “individual” is defined as belonging to the vocabulary type “CL” corresponding to the classifier.

  Also, for example, in the case of the analysis target character string Z, since “Self Defense Force” is classified into the third group (functioning as a noun), one analysis rule that assumes that “Self Defense Force” is a noun is selected. . In this analysis rule, it is defined that “SDF” belongs to the vocabulary type “N” corresponding to the noun.

  The analysis unit 32 executes syntax analysis in accordance with the analysis rule determined by the analysis rule determination unit 33. For example, when one analysis rule is selected by the analysis rule determination unit 33, syntax analysis is executed in accordance with the analysis rule, and one analysis result is obtained. On the other hand, when a plurality of analysis rules are selected by the analysis rule determination unit 33, syntax analysis is executed in accordance with each analysis rule, and a plurality of analysis results are obtained.

  For example, in the case of the analysis target character string X, the syntax analysis is executed in accordance with an analysis rule (see FIG. 5) that “person” belongs to the vocabulary type “CL”. FIG. 7 shows the results of this analysis. Separately, syntax analysis is executed in accordance with an analysis rule (see FIG. 6) that “person” belongs to the vocabulary type “N”. FIG. 8 shows the result of this analysis. By the way, when an analysis result as shown in FIG. 8 is obtained, the character string “three students” is interpreted as meaning, for example, “a common student of three teachers”. Become. As described above, in the case of the analysis target character string X, two analysis results are obtained.

  Further, for example, in the case of the analysis target character string Y, only the syntax analysis is executed in accordance with the analysis rule that “piece” belongs to the vocabulary type “CL” corresponding to the classifier. FIG. 9 shows the results of this analysis. Thus, in the case of the analysis target character string Y, only one analysis result is obtained.

  Further, for example, in the case of the analysis target character string Z, only the syntax analysis is executed in accordance with the analysis rule that “Self Defense Force” belongs to the vocabulary type “N” corresponding to the noun. FIG. 10 shows the result of this analysis. Thus, in the case of the analysis target character string Z, only one analysis result is obtained.

  The analysis result output unit 34 outputs the analysis result of the analysis unit 32. For example, the analysis result output unit 34 displays the analysis result of the analysis unit 32 on the display unit 14. When a plurality of analysis results are obtained, the analysis result output unit 34 outputs the plurality of analysis results.

  As described above, when the morpheme immediately after the morpheme representing the number in the analysis target character string is classified into the first group (which can function as both a classifier and a noun), a plurality of analysis results are obtained. It is done. In such a case, the analysis result may be narrowed down by executing a known ambiguity resolution process such as machine learning so as to obtain an analysis result judged to be more appropriate. For example, Hiroki Yoshimura and three other authors, “Selection of f-structure based on Support Vector Machine” (Natural Language Processing Study Group Report, Information Processing Society of Japan, November 2003, Vol. 2003, No. .108, p.75-80).

  According to the present embodiment described above, basic information serving as a basis for determining whether a word used in a sentence in combination with a word representing a number functions as either a classifier or a noun, or as both, is a sentence. Acquired based on group data.

  The present invention is not limited to the embodiment described above.

  For example, all the processes described above are examples, and may be realized by a method different from the method described above.

Further, for example, the statistical information acquisition unit 22 may acquire statistical information other than the statistical information (S _i , P _i , Q _i , R _i ) described above. For example, the calculation formula of the evaluation value E _i is not limited to the above calculation formula (1), but may be calculated by other calculation formulas, or may be calculated using other statistical information. Also good.

Further, for example, basic information generating unit 23 the evaluation value E _i of the word L _i may also be stored in the basic information storage unit 35 as the basic information. In this case, the analysis rule determination unit 33 uses the evaluation value E of the morpheme (word) stored in the basic information storage unit 35 when determining which of the first to third groups the morpheme belongs to. That is, the analysis rule determination unit 33 determines which of the first to third groups the morpheme belongs by executing the same processing as the processing of S109 to S113. In this case, the reference values J and K may also be stored in the basic information storage unit 35 as part of the basic information.

Further, for example, basic information generating section 23 statistics word _{L i (S i, P i} , Q i, R i) a may also be stored in the basic information storage unit 35 as the basic information. In this case, when the analysis rule determination unit 33 determines which of the first to third groups the morpheme belongs to, statistical information (S _i , P _i ) on the morpheme (word) stored in the basic information storage unit 35. , Q _i , R _i ). That is, the analysis rule determination unit 33 determines which of the first to third groups the morpheme belongs by executing the same processing as the processing of S108 to S113. In this case, the reference values J and K may also be stored in the basic information storage unit 35 as part of the basic information.

  Further, for example, the language processing system 1 may be configured to include a plurality of language processing devices 10. FIG. 11 shows an example of a language processing system 1 configured to include a plurality of language processing devices 10. The language processing system 1 shown in FIG. 11 includes a first language processing device 10a and a second language processing device 10b. The first language processing device 10a and the second language processing device 10b can exchange data with each other via communication means. In this case, as shown in FIG. 12, the word acquisition unit 21, the statistical information acquisition unit 22, and the basic information generation unit 23 are realized by the first language processing device 10 a, and the analysis target character string acquisition unit 31, the analysis unit 32, The analysis rule determination unit 33, the analysis result output unit 34, and the basic information storage unit 35 may be realized by the second language processing device 10b. In this case, the basic information generation unit 23 stores the basic information in the storage unit 12 of the first language processing device 10a. Moreover, the basic information generation part 23 supplies the said basic information to the 2nd language processing apparatus 10b via a communication means, for example according to the request | requirement from the 2nd language processing apparatus 10b.

It is a figure which shows the structural example of the language processing system which concerns on embodiment of this invention. It is a functional block diagram which shows an example of the function which the language processing system which concerns on embodiment of this invention implement | achieves. It is a flowchart which shows the process performed with a language processing system. It is a flowchart which shows the process performed with a language processing system. It is a figure which shows an example of an analysis rule. It is a figure which shows another example of an analysis rule. It is a figure which shows an example of an analysis result. It is a figure which shows another example of an analysis result. It is a figure which shows another example of an analysis result. It is a figure which shows another example of an analysis result. It is a figure which shows the structural example of the language processing system which concerns on other embodiment of this invention. It is a functional block diagram which shows an example of the function which the language processing system which concerns on other embodiment of this invention implement | achieves.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 language processing system, 10 language processing apparatus, 10a 1st language processing apparatus, 10b 2nd language processing apparatus, 11 control part, 12 memory | storage part, 13 operation part, 14 display part, 21 word acquisition part, 22 statistical information acquisition part , 23 Basic information generation part, 31 Analysis object character string acquisition part, 32 Analysis part, 33 Analysis rule determination part, 34 Analysis result output part, 35 Basic information storage part.

Claims (7)

Word acquisition means for acquiring a word used in a sentence in combination with a word representing a number;
Statistical information acquisition means for acquiring statistical information on how the word is used based on sentence group data;
Basic information generating means for generating basic information based on the statistical information as a basis for determining whether the word functions as one or both of a classifier and a noun;
A language processing apparatus comprising: The statistical information acquisition means includes
Means for acquiring the number of times the word is used in a predetermined first mode corresponding to a classifier in the sentence group data;
Means for acquiring the number of times the word is used in a predetermined second mode corresponding to a noun in the sentence group data;
The language processing apparatus according to claim 1.   The language processing apparatus according to claim 1, wherein the statistical information acquisition unit includes a unit that acquires the number of modifiers that modify the word in the sentence group data. An analysis target character string acquisition means for acquiring the analysis target character string;
When the word is included in the analysis target character string, based on the basic information of the word, it is determined whether the word functions as either a classifier or a noun or both, and the determination result Based on the analysis means for analyzing the analysis target character string,
4. The language processing apparatus according to claim 1, comprising:   The analysis means includes analysis rule determination means for determining an analysis rule for the analysis target character string based on the determination result, and the analysis target character string is determined using the analysis rule determined by the analysis rule determination means. The language processing device according to claim 4, wherein the language processing device is analyzed. Word acquisition means for acquiring a word used in a sentence in combination with a word representing a number;
Statistical information acquisition means for acquiring statistical information on how the word is used based on sentence group data;
Basic information generating means for generating basic information based on the statistical information as a basis for determining whether the word functions as one or both of a classifier and a noun;
A language processing system comprising: Word acquisition means for acquiring a word used in a sentence in combination with a word representing a number;
Statistical information acquisition means for acquiring statistical information on how the word is used based on sentence group data; and
Basic information generating means for generating basic information based on the statistical information, which is a basis for determining whether the word functions as one of a classifier and a noun or both.
As a program to make the computer function.

Images