JPH10187724A - Method and device for assisting document preparation - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect an error in numerical expression by performing a document analysis, specially, an analysis of units of counting way for words from meaning relation between words. SOLUTION: A word division part 104 divides a sentence inputted from an input part 101 into words and a word relation analysis part 106 analyzes modification relations among the divided words. A numerical relation extraction part 108 extracts a numerical expression having a relation of counting a number from the modification relations of the respective words. A numerical expression error detection part 109 refers to counting way information (including units or ways of counting and conditions of numerals when the units are used) registered in a counting way storage part 107 corresponding to words counting numbers in the extracted numeric expression to detect an auxiliary numeral when units of ways of counting match each other or a numeral limit error unless the represented numerical meets numeral conditions although the unit of counting way match with each other.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document creation support method and apparatus for supporting document creation, and more particularly to a document creation support method and apparatus suitable for detecting an error in the expression of a word count.

[0002]

2. Description of the Related Art In recent years, with the spread of Japanese document creation software (so-called Japanese word processing software) that can be read and executed by a document input device such as a Japanese word processor or a personal computer, the creation of documents has become difficult.
The use of electronic means instead of conventional handwriting is increasing. However, electronic means are mainly provided for inputting sentences, and at present, human beings still consider the sentences to be input.

[0003] In recent years, word processors and similar document input devices (or word processing software) have been provided with information useful for humans to think about sentences, and for helping created sentences to be more accurate. Some are provided with a function to perform Analyze the input text,
The function to find the error is one of them.

[0004]

However, in the above-mentioned prior art, since information obtained from a specific word is used, it is difficult to detect a use error of a word that cannot be understood by one word alone. There was a problem. In particular, there is a problem in the case where the counting method (unit) differs for each word as in Japanese sentences.

The present invention has been made in consideration of the above circumstances, and has as its object to analyze a sentence based on the semantic relation between words and words, in particular, to analyze units of counting of words, and to make an error in the numerical expression thereof. It is an object of the present invention to provide a document creation support method and device for detecting a document.

[0006]

SUMMARY OF THE INVENTION According to the present invention, for each word that can be a numerical expression, information on how to count is registered in a storage means (counting storage means), and an error in a numerical expression in an input sentence is stored. When detecting the input sentence, the input sentence is divided into words, and the relation of the dependency of each of the divided words is analyzed to extract a numerical expression having a relation of counting the number,
For each of the extracted numerical expressions, an erroneous numerical expression is detected by referring to the counting information in the storage means corresponding to the word in the numerical expression.

[0007] In the present invention, the information of the counting method is registered in advance in the storage means in correspondence with each word which can be a target of numerical expression, that is, each word (noun) whose number can be counted, so that the input can be performed. A sentence is analyzed to extract a numerical expression having a relationship that counts the number, and by simply referring to the counting information about the word (the word that can be counted) in the numerical expression, an expression of the counting method using the word is obtained. Can be easily determined whether the expression is out of the counting method indicated by the counting information, that is, whether the extracted numerical expression is incorrect.

As described above, in the present invention, a numerical expression having a word (a word that can be counted) that is counted from the input sentence differently from the counting method indicated by the counting information registered in the storage means. Can be detected as an error.

Here, if the counting information has a unit (numerical number) for counting the corresponding word, it is possible to detect an error in the numerically expressed portion where the number is counted in the wrong unit.

In addition, if the counting information has a unit for counting the corresponding word and a numerical condition when the unit is used, a numerical expression having a relation of counting the number from the input sentence is extracted. In this case, by referring to the counting information for the words (words that can be counted) in the numerical expression, it is possible to determine whether or not the counting unit is incorrect,
Although the unit of counting is not wrong, it can be determined whether or not the expressed numerical value is an erroneous numerical expression that does not satisfy numerical conditions (for example, the size and number of digits of the numerical value).

As described above, in the present invention, attention is paid to the fact that, depending on the number of words that can be counted, there are cases where the size and digits of numerical values used for the words are limited, and such restrictions are set as conditions (numerical conditions). By giving the word counting information, even if the correct unit is used, a numerical expression portion using a numerical value that does not match the condition can be detected as an error (numerical limit error).

Further, in the counting information, in addition to a unit for counting the corresponding word and a numerical condition when the unit is used, a recommendation for generating an expression recommended when the condition is not satisfied. If there is an expression rule, if the unit of counting is not wrong, but a numerical expression using a numerical value that does not meet the conditions is detected as an error, the recommendation in the counting information used for the error detection Recommended numerical expressions can be generated according to expression rules. Here, the recommended expression rule may include a unit to be used (a classifier) and a shift number of a numerical digit (necessary to satisfy a numerical condition when the unit is used).

[0013]

Embodiments of the present invention will be described below with reference to the drawings. [First Embodiment] FIG. 1 is a block diagram showing a functional configuration of a document creation support apparatus according to a first embodiment of the present invention. The document creation support apparatus shown in FIG.
2, output unit 103, word division unit 104, dictionary storage unit 10
5. It comprises a word relation analysis unit 106, a counting method storage unit 107, a numerical relation extraction unit 108, and a numerical expression error detection unit 109.

The input unit 101 inputs a character string to be supported for document creation, more specifically, a character string (or a document composed of character strings) to be detected as an error in how to count words. Used for The input unit 101 is, for example, a keyboard, a magnetic disk device, a communication line, or the like.

The control unit 102 controls the entire apparatus.
The output unit 103 is used, for example, to output a character string input by the input unit 101 and a detection result of an error in how to count words. The output unit 103 is, for example, a liquid crystal display, C
It is a display device such as an RT display.

The word dividing unit 104 analyzes the character string input by the input unit 101 and divides the character string into words. The analysis dictionary storage unit 105 stores an analysis dictionary 105a used for word analysis by the word division unit 104 in advance. The analysis dictionary 105a has, for each word, its heading (notation), part of speech, inflected form, and the like.

The word relation analysis unit 106 includes the word division unit 10
The relationship between the word divided by step 4 and the dependency of the word is analyzed. The counting method storage unit 107 registers a set of counting information including a word (noun) that can be a target (numerable) of a numerical expression and a unit (numerical number) of the counting method for various counting words. Counting information table)
107a is stored in advance. This counting information can include the character type of the used numeral determined by the classifier (for example, whether to use arithmetic numerals or Chinese numerals). Depending on the classifier, the counting information includes a condition (additional condition) of a numerical value (a numerical range) when the classifier is used. This counting storage unit 1
07 (registered contents of counting information table 107a)
An example is shown in FIG.

The numerical relationship extracting unit 108 extracts an expression (numerical expression) having a number counting relationship from the analysis result of the word relationship analyzing unit (dependency relationship between words). The numerical expression error detecting unit 109 determines whether the numerical expression extracted by the numerical relationship extracting unit 108 is correct based on the contents of the counting storage unit 107 (registered contents of the counting information table 107a).

Next, the operation of the document creation support apparatus having the configuration shown in FIG. 1 will be described from its outline. First, a character string serving as an input sentence is input from the input unit 101. This input statement is
It is passed to the word division unit 104 via the control unit 102. The word division unit 104 analyzes (morphologically analyzes) the input sentence (input character string) according to the analysis dictionary 105a in the analysis dictionary storage unit 105, and divides the sentence into the most natural word units. At this time, the word division unit 104 adds information such as the part of speech and the inflected form of the word obtained from the analysis dictionary 105a to the divided word. FIG. 3 shows the result of word division performed on the input character string (input sentence) by the word division unit 104, taking as an example the case where the input character string is "She bought two books." FIG.
The symbol “/” in the inside indicates the division position into word units.

The result of the word division by the word division unit 104 is passed to the word relation analysis unit 106. Word relation analysis unit 106
Receives the result of the word division and analyzes the dependency relationship between the words of each divided word. FIG. 4 shows the result of analyzing the dependency relationship between words with respect to the word segmentation result of FIG.

The result of the analysis of the dependency relationship between words by the word relationship analysis unit 106 is passed to the numerical relationship extraction unit 108. The numerical relationship extracting unit 108 extracts expressions having a counting relationship (in the example of FIG. 4, book and book 2, book 2) from among the words analyzed by the word relationship analyzing unit 106. I do.

The expression having the relation counting the number extracted by the numerical relation extracting unit 108 is the numerical expression error detecting unit 10.
9 passed. Upon receiving the expression having the relationship of counting the numbers, that is, the expression of the counting (numeric expression), the numerical expression error detecting unit 109 converts the expression of the counting into the counting information table in the counting storage unit 107. By sequentially collating with the counting method information registered in 107a, it is determined whether or not the expression of the counting method is erroneous, and an error is detected.

The error detection result for the expression of counting detected by the numerical expression error detecting unit 109 is passed to the control unit 102. The control unit 102 outputs this error detection result to the output unit 103 and presents it to the user by, for example, displaying it on a screen.

Here, the details of the numerical expression error detection processing by the numerical expression error detection unit 109 will be described with reference to the flowchart of FIG. The numerical expression error detecting unit 109 performs the following process on each numerical expression extracted by the numerical relationship extracting unit 108 among the relationships between words analyzed by the word relationship analyzing unit 106.

First, the counting information table 107a in the counting storage unit 107 is searched, and counting information for the countable words in the numerical expression extracted by the numerical relationship extracting unit 108 (counting in the extracted numerical expression). It is checked whether or not there is counting information corresponding to a word that matches the word to be obtained (step 51).

If there is no corresponding counting information,
It is determined that it is not an error as it is (step 57). On the other hand, if there is counting information, whether the classifier in the numerical expression extracted by the numerical relationship extraction unit 108 matches the classifier in the corresponding counting information, that is, whether the classifier is correct It is checked whether or not there is (step 52).

If the classifier is incorrect, an error is detected as a "classifier error" (step 53). On the other hand, if the classifier is correct, it is checked whether there is a numerical condition (additional condition) in the corresponding counting information (step 54), and if there is no numerical condition, it is determined that there is no error (step 57). .

On the other hand, if there is a numerical condition, it is checked whether or not the relationship between the classifier and the numerical value in the numerical expression extracted by the numerical relation extracting unit 108 matches the numerical condition (step 55). That is, as in the numerical condition, as in the example of FIG. 2, the size of the numerical value when using the classifier and the restriction on the number of digits are described, so that the numerical value of the numerical expression of the extracted input sentence matches it? Determine whether or not.

If they match, it is determined that there is no error (step 57), and if they do not match, it is determined that "numerical limit error" has occurred (step 56). Next, FIG.
With reference to FIG. 6, a description will be given of a specific example of inspection of a numerical expression (unit of counting) for an input sentence (input character string) by the document creation support device of the present invention.

First, “2” is input from the input unit 101 as an input sentence.
000 megabytes of memory "
04, word division is performed as shown in FIG. Further, it is assumed that a numerical relationship as shown in FIG. 6B has been extracted by the word relationship analyzing unit 106 and the numeric relationship extracting unit 108.

The numerical expression error detecting unit 109 refers to the registration information (see FIG. 2) of the counting information table 107a in the counting storage unit 107, and finds that the word "memory" is "byte", "kilobyte", "megabyte". Recognize how to count "individual". In this example, the use of the classifier is correct because "megabyte" is used for the classifier.

However, the counting information table 107 shown in FIG.
As is clear from the registration information of a, since the numerical range when using the classifier "megabyte" has a numerical condition of up to 3 digits, the 4-digit "2000" in the input sentence "2000 megabyte memory" is used. The figures do not meet the conditions. Therefore, the numerical expression error detecting unit 109 calculates
The numerical expression "2000 megabytes" including the digit number "2000" is detected as "numerical limit error" (step 56).

FIG. 6C similarly shows that the input sentence is "2000".
(Word relation analysis unit 106,
The numerical relationship (extracted by the numerical relationship extracting unit 108) is shown. In this example, since the classifier is correct and also satisfies the numerical condition, the result is "no error" (step 57). Further, FIG. 6D shows a numerical relationship when the input sentence is "2000 memories". Since this uses a different classifier, the result is "Error in classifier" (step 53). [Second Embodiment] Next, a second embodiment of the present invention will be described with reference to the drawings.

FIG. 7 is a block diagram showing a functional configuration of a document creation support device according to the second embodiment of the present invention.
The same parts as those in FIG. FIG.
1 is different from the document creation support device of FIG. 1 in that a counting method information table 707a in place of the counting information table 107a in the counting storage unit 107 and a numeric value in place of the numerical expression error detection unit 109 Expression error detector 7
09, and a new recommended expression generation unit 710 that generates a recommended expression based on the expression determined to be “numeric restriction error” (by the numerical expression error detection unit 709).

The counting information table 707a registers a group of counting information including a word (noun) which can be an object of numerical expression (can be counted) and a unit of counting of the word (numerical number) for various counting words. Table
It is the same as the counting information table 107a in FIG. 1 in that, depending on the classifier, the condition (additional condition) of a numerical value (numerical range) when the classifier is used is registered in association with the classifier. The counting information table 707a is different from the counting information table 1107a in that the counting information to which a numerical condition is added has a numerical expression that does not match the numerical condition, that is, a numerical expression that exceeds the numerical range. That is, information indicating how appropriate the expression is, that is, information indicating a recommended expression (recommended expression rule) is further added. FIG. 8 shows an example of the registered contents of the counting information table 707a.

The numerical expression error detecting section 709 counts whether or not the numerical expression extracted by the numerical relation extracting section 108 is correct.
This is the same as the numerical expression error detection unit 109 in FIG. The difference between the numerical expression error detection unit 709 and the numerical expression error detection unit 109 is that when the erroneous expression is “numeric limit error” (instead of “numeric classifier error”), the correspondence in the counting method information table 707a Is passed to the recommended expression generation unit 710 to request generation of a recommended expression (instead of the expression).

Next, the operation of the document creation support device having the configuration shown in FIG. 7 will be described from its outline. First, the input unit 10
1, the input sentence is input, and then the input sentence is divided into words by the word division unit 104, and the inter-word relations are determined by the word relation analysis unit 106 based on the result of the word division. The operation up to the point where the numerical relationship extracting unit 108 extracts the expression having the relationship of counting the number of the analyzed relationships between the words,
The operation is the same as that in the document creation support device having the configuration shown in FIG.

When the numerical relationship extracting unit 108 extracts an expression having a number counting relationship, that is, a numerical expression (numerical expression), the numerical expression error detecting unit 709 counts the counted expression. By sequentially checking the counting method information registered in the counting information table 707a on the method storage unit 107, it is determined whether or not the expression of the counting method is incorrect, and an error is detected. Here, when an error is detected and the numerical expression error is a “numeric restriction error” that does not satisfy the numerical condition, the numerical expression error detection unit 709 instructs the recommended expression generation unit 710 to count the “numeric restriction error”. Counting information table 7 corresponding to the expression (a numerical value and a classifier) and the numerical condition used for the determination of the “numeric limit error”
The recommended expression rule registered in 07a is passed to the recommended expression generation unit 710, and a request is made to the recommended expression generation unit 710 to generate an appropriate expression that replaces the expression of the numbering that has become “numerical limit error”.

In response to this, the recommended expression generation unit 710
A recommended expression is generated by changing a numerical value and a classifier that represent the expression of “numerical restriction error” in accordance with the recommended expression rule passed from the numerical expression error detection unit 709, and passed to the control unit 102.

The control unit 102 outputs the recommended expressions generated by the recommended expression generation unit 710 to the output unit 103 and presents them to the user by, for example, displaying them on a screen. Here, the details of the numerical expression error detection processing by the numerical expression error detection unit 709 and the post-processing when an error is detected in the detection processing will be described with reference to the flowchart of FIG.

Numerical expression error detecting section 709 determines the relation between words analyzed by word relation analyzing section 106.
The following processing is performed on each numerical expression extracted by the numerical relationship extracting unit 108.

First, the counting information table 707a on the counting storage unit 107 is searched to determine whether there is counting information for the countable word in the numerical expression extracted by the numerical relationship extracting unit 108 (step). 91).

If there is no corresponding counting information,
It is determined that it is not an error as it is (step 97). On the other hand, if there is counting information, whether the classifier in the numerical expression extracted by the numerical relationship extraction unit 108 matches the classifier in the corresponding counting information, that is, whether the classifier is correct It is checked whether or not it is (step 92).

If the classifier is incorrect, an error is detected as a "classifier error" (step 93). On the other hand, if the classifier is correct, it is checked whether a numerical condition is added to the corresponding counting information (step 9).
4) If there is no numerical condition, it is determined that there is no error (step 97).

On the other hand, when the numerical condition is added, it is checked whether or not the relation between the classifier and the numerical value in the numerical expression extracted by the numerical relation extracting unit 108 matches the numerical condition (step 95). .

If they match, it is determined that there is no error (step 97), and if they do not match, it is determined that "numerical limit error" has occurred (step 96). Numerical expression error detection unit 70
9 detects a numerical expression error as a “numerical restriction error”, extracts a recommended expression rule added to the corresponding counting information in the counting information table 707a together with a numerical condition, and It is passed to the recommended expression generation unit 710 together with the numerical value and the classifier forming the determined numerical expression,
Request generation of recommended expressions.

Then, the recommended expression generating unit 710 shifts the numerical value and the numerical value of the classifier received from the numerical expression error detecting unit 709 by the number of digits specified by the recommended expression rule received from the numerical expression error detecting unit 709. By changing the classifier to the classifier specified by the recommended expression rule, a recommended expression is generated (acquired) in place of the numerical expression determined to be “numerical limit error” (step 98). Then, the recommended expression generation unit 710 sets the generated recommended expression as information to be presented to the user and passes it to the control unit 102 (step 99).

On the other hand, when the numerical expression error detecting section 709 detects "numeric classifier error" (step 93), the numerical expression error detecting section 709 sets the "numeric classifier error" as information to be presented to the user, and sets the information to the control section. 102 (step 99).

The control unit 102 includes a numerical expression error detection unit 70
9 or the user presentation information passed from the recommended expression generation unit 710 is output to the output unit 103 and displayed on the screen. Next, a specific example of the above-described inspection of a numerical expression (unit of counting) for an input sentence (input character string) by the document creation support device of FIG. 7 will be described with reference to FIG.

First, “2” is input from the input unit 101 as an input sentence.
000 megabytes of memory "
04, word division is performed as shown in FIG. Further, it is assumed that a numerical relationship as shown in FIG. 6B has been extracted by the word relationship analyzing unit 106 and the numeric relationship extracting unit 108.

The numerical expression error detecting section 109 refers to the registration information (see FIG. 8) of the counting information table 707a on the counting storage section 107, and finds that the word "memory" is "byte", "kilobyte", "megabyte". Recognize how to count "individual". In this example, the use of the classifier is correct because "megabyte" is used for the classifier.

However, the counting information table 707 shown in FIG.
As is clear from the registration information (counting information) of a, when the classifier “megabyte” is used, the numerical range has a numerical condition of up to three digits, and when it exceeds three digits (instead of “megabyte”) Since there is a recommended expression rule of using “gigabyte” and shifting the numerical value by three digits (when “megabyte” is used), the numerical expression error detecting unit 709 uses the above-mentioned input sentence “2000 megabyte memory” The numerical expression including the four-digit numerical value “2000” of “2000 megabytes” therein is detected as “numeric limit error” (step 96).

In this case, the numerical expression error detecting section 709
The numerical value "2000" and the classifier "megabyte" which form the numerical expression "2000 megabytes" which became "numerical limit error"
And a recommended expression rule for shifting the numerical value when using “megabyte” by three digits and replacing it with the recommended expression using “gigabyte”, to the recommended expression generation unit 710. In response, as shown in FIG. 10, the recommended expression generation unit 710 shifts the numerical value “2000” in “2000 megabytes” by three digits to “2” and changes the classifier “megabytes” to “gigabytes”. By doing, the expression of "2000 megabytes" is changed to "2
Gigabyte ".

The processing function of each part of the document creation support device having the configuration shown in FIG. 1 is a recording medium such as a floppy disk on which a program for realizing the processing function by a computer, for example, a personal computer 1100 shown in FIG. (FD) 1110, and the floppy disk 1110 is connected to the personal computer 11
00, the program recorded on the floppy disk 1110 is stored in the personal computer 11.
This is realized by reading and executing at 00. More specifically, the floppy disk 1110
In addition, as shown in FIG. 12A, a word division routine 1211 for functioning as the word division unit 104, a word relation analysis routine 1212 for functioning as the word relation analysis unit 106, and a numerical relation extraction unit 108 are functioned. A program 1210 having a numerical relationship extraction routine 1213 and a numerical expression error detection routine 1214 for functioning as the numerical expression error detection unit 109 may be recorded, and the program 1210 may be read and executed by the personal computer 1100. .

Similarly, in order to realize the processing function of each part of the document creation support device having the configuration shown in FIG.
As shown in FIG. 12 (b), the word
4, a word segmentation routine 1211 to function as
A word relation analysis routine 1212 for functioning as the word relation analysis unit 106, a numerical relation extraction routine 1213 for functioning as the numerical relation extraction unit 108, a numerical expression error detection routine 1224 for functioning as the numerical expression error detection unit 709, In addition, a program 1220 having a recommended expression generation routine 1225 for functioning as the recommended expression generation unit 710 may be recorded, and the program 1220 may be read and executed by the personal computer 1100.

It should be noted that the dictionary storage unit 105 and the counting storage unit 107 may be secured in a ROM generally included in the personal computer 1100 or a non-volatile storage device such as a magnetic disk device (hard disk device).

Further, in the first and second embodiments described above, a document creation supporting apparatus having an error detection function of a numerical expression for a Japanese input sentence has been described. The present invention can also be applied to error detection of numerical expressions in other languages such as Chinese and Thai, as long as the language has a counting method other than the above.

Further, in the above embodiment, an error in the numerical expression in the input sentence is detected and only the error or what kind of expression is recommended is presented to the user. By automatically replacing the incorrect numerical expressions in the language with the recommended expressions, it is possible to perform more appropriate numerical expressions, for example, in a machine translation system or a natural language understanding system, and to generate the target language and knowledge generated. Data and the like can be made more natural and easy to handle.

[0059]

As described above in detail, according to the present invention, the input sentence is analyzed by registering the counting method information in advance in the storage means in correspondence with each word whose number can be counted. It is possible to easily determine whether or not the numerical expression is erroneous simply by extracting a numerical expression having a relationship of counting the numbers and referring to the counting method information on the words in the numerical expression.

Further, according to the present invention, by using the counting information having the unit of word counting, it is possible to detect a numerical expression in which the unit of counting is wrong. According to the present invention,
By using the counting information that has the unit of the word count and the numerical condition when using the unit, the numerical unit is not wrong, but the expressed numerical value does not satisfy the numerical condition. Can be detected.

Further, according to the present invention, in addition to a unit for counting words and a numerical condition when using the unit, a recommended expression rule for generating an expression recommended when the condition is not satisfied is defined. By using the counting information that has, the unit of counting is not wrong, but if a numerical value expressed that does not satisfy the numerical condition is detected, a valid numerical expression without numerical limit error is generated be able to.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a functional configuration of a document creation support device according to a first embodiment of the present invention.

FIG. 2 is a view showing an example of registered contents of a counting information table 107a in FIG. 1;

FIG. 3 is a view showing an example of a word division result for an input sentence by a word division unit 104 in FIG. 1;

FIG. 4 is a view showing a result of a word relationship analysis performed by the word relationship analysis unit 106 in FIG. 1 on the result of the word division in FIG. 3;

FIG. 5 is a flowchart for explaining a numerical expression error detection process by a numerical expression error detection unit 109 in FIG. 1;

FIG. 6 is a diagram illustrating an input with a “numeric limit error” in order to explain a specific example of checking a numerical expression (unit of counting) for an input sentence (input character string) in the first embodiment. Sentence, "error-free" input sentence, and "classifier error"
FIG. 6 is a diagram showing an input sentence with a symbol and a dependency relationship between words in the input sentence.

FIG. 7 is a block diagram showing a functional configuration of a document creation support device according to a second embodiment of the present invention.

FIG. 8 is a view showing an example of registered contents of a counting information table 707a in FIG. 7;

FIG. 9 is a flowchart for explaining a numerical expression error detection process by a numerical expression error detection unit 709 in FIG. 7;

FIG. 10 shows a “numerical value limit error” in the second embodiment.
FIG. 7 is a diagram showing a comparison between a numerical expression example detected and a recommended expression example generated based on the expression.

FIG. 11 is a diagram showing the appearance of a personal computer on which a floppy disk in which a program for realizing the function of each unit of the document creation support device having the configuration shown in FIG. 1 or 7 is recorded is mounted.

12 shows a main routine included in a program for realizing the function of each unit of the document creation support device having the configuration of FIG. 1 and a program for realizing the function of each unit of the document creation support device having the configuration of FIG. 7; The figure which shows the main routine included.

[Explanation of symbols]

 101 input unit, 102 control unit, 103 output unit, 104 word division unit, 105 dictionary storage unit, 105a analysis dictionary, 106 word relation analysis unit, 107 counting storage unit, 107a, 707a ... Counting information table, 108: Numerical relation extracting unit, 109, 709: Numerical expression error detecting unit, 710: Recommended expression generating unit.

Claims (8)

[Claims] 1. A document creation support method for dividing an input sentence into words, extracting related words, and detecting an error in a sentence based on the relationship. The counting method is registered in a storage unit, and the input sentence is divided into words in order to detect an error in the numerical expression in the input sentence, and the dependency of each divided word is determined. By analyzing the relationship, a numerical expression having a relationship counting the number is extracted, and for each of the extracted numerical expressions, refer to the counting method information in the storage unit corresponding to the word in the numerical expression. A document creation support method, wherein an erroneous numerical expression is detected. 2. An input unit for inputting a sentence, a word dividing unit for dividing the sentence input by the input unit into words, and a dependency relationship between the words divided by the word dividing unit is analyzed. Word relation analysis means; and numerical relation extraction means for extracting a numerical expression having a relation of counting the number from the dependency relation between words analyzed by the word relation analysis means; For each word, a counting method storage means for storing information on the counting method, and for each numerical expression extracted by the numerical relationship extracting means,
A document expression supporting apparatus comprising: a numerical expression error detecting means for detecting an erroneous numerical expression by referring to the counting information in the storage means corresponding to a word in the numerical expression. . 3. A document creation support method for dividing an input sentence into words, extracting related words, and detecting an error in a sentence based on the relationship. The counting information including the counting unit is registered in the storage unit, and the input sentence is divided into word units in order to detect an error in the numerical expression in the input sentence, and the divided sentence is divided. By analyzing the dependency relation of each word, a numerical expression having a relation counting the number is extracted, and for each of the extracted numerical expressions, the numerical value in the storage means corresponding to the word in the numerical expression is extracted. A document creation support method characterized by detecting a numerical expression in which a unit of counting is wrong by referring to counting information. 4. An input unit for inputting a sentence, a word dividing unit for dividing the sentence input by the input unit into words, and a dependency relationship between the words divided by the word dividing unit is analyzed. Word relation analysis means; and numerical relation extraction means for extracting a numerical expression having a relation of counting the number from the dependency relation between words analyzed by the word relation analysis means; For each word, counting method storage means for storing counting information including the counting unit, for each numerical expression extracted by the numerical relationship extraction means,
Numeric expression error detecting means for detecting a numerical expression in which the unit of counting is incorrect by referring to the counting information in the storage means corresponding to the word in the numerical expression. Document creation support device. 5. A document creation support method for dividing an input sentence into words, extracting relevant words, and detecting an error in the sentence based on the relevant words. The counting unit including the unit of the counting method and the numerical value condition when using the unit is registered in the storage unit, and the input sentence is detected in order to detect an error in the numerical expression in the input sentence. Is divided into words, and by analyzing the dependency relation of each of the divided words, a numerical expression having a relation of counting the number is extracted, and for each of the extracted numerical expressions, By referring to the counting information in the storage means corresponding to a word, the counting unit is wrong, or the counting unit is not wrong, but the expressed numerical value in the counting information is The above numerical condition Not Tashi, document creation support method and detecting a numeric representation erroneous. 6. An input unit for inputting a sentence, a word dividing unit for dividing the sentence input by the input unit into words, and analyzing a dependency relationship between the words divided by the word dividing unit. Word relation analysis means; and numerical relation extraction means for extracting a numerical expression having a relation of counting the number from the dependency relation between words analyzed by the word relation analysis means; For each word, a counting method storage means for storing counting method information including a unit of the counting method and a numerical condition when the unit is used, and a numerical expression extracted by the numerical relation extracting means. ,
By referring to the counting information in the storage means corresponding to the word in the numerical expression, the unit of counting is wrong, or the unit of counting is not wrong, but the expressed numerical value is A document creation support device, comprising: a numerical expression error detection unit that detects an erroneous numerical expression that does not satisfy the numerical condition in the counting method information. 7. A document creation support method for dividing an input sentence into words, extracting relevant words, and detecting an error in a sentence based on the relation. Register the counting method information including a unit of the counting method, a numerical condition when using the unit, and a recommended expression rule for generating an expression recommended when the condition is not satisfied, in the storage unit, In order to detect an error in the numerical expression in the input sentence, the input sentence is divided into words, and the dependency relation of each of the divided words is analyzed to count the number. By extracting a numerical expression having the following, and referring to the counting information in the storage unit corresponding to the word in the numerical expression for each of the extracted numerical expressions, the unit of counting is incorrect, or Simple counting Is not wrong, the detected numerical value does not satisfy the numerical condition in the counting method information, detects an erroneous numerical expression, the counting unit is not wrong, but the expressed numerical value is A document creation support method, characterized in that, when a numerical expression in the counting information that does not satisfy the numerical condition is detected, a recommended numerical expression is generated according to the recommended expression rule corresponding to the numerical condition. 8. An input unit for inputting a sentence, a word dividing unit for dividing the sentence input by the input unit into words, and a dependency relationship between the words divided by the word dividing unit is analyzed. Word relation analysis means; and numerical relation extraction means for extracting a numerical expression having a relation of counting the number from the dependency relation between words analyzed by the word relation analysis means; For each word, counting method information including a unit of counting method, a numerical condition when using the unit, and a recommended expression rule for generating an expression recommended when the condition is not satisfied is stored. For each of the numerical expressions extracted by the numerical relationship extraction means,
By referring to the counting information in the storage means corresponding to the word in the numerical expression, the unit of counting is wrong, or the unit of counting is not wrong, but the expressed numerical value is The numerical condition in the counting information is not satisfied, the numerical expression error detecting means for detecting an erroneous numerical expression, and the numerical expression error detecting means does not satisfy the numerical condition in the counting information. A recommended expression generating means for generating, from the numerical expression in which the error is detected, a numerical expression satisfying the numerical condition in accordance with the recommended expression rule stored in the counting method storage means in correspondence with the numerical condition. Characteristic document creation support device.