JPH1011443A - Document code check system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To attain the accurate elaboration support to the writing errors and then to secure a function to automatically produce an index list from the extracted codes by receiving the information on the intermediate codes and the sorting results of an entire document, sorting the code names in an ascending order and displaying the sorting results in a single line for every individual code. SOLUTION: An object code selection part 80 receives the intermediate codes from an intermediate code extraction part 60 and performs the decision processing of four types of sorting among these intermediate codes by a five steps selection processing, i.e., a noise code, a noise suspected code, a pseudo code and other object codes. A code totalization display part 10 receives the extraction and sorting result information on the intermediate codes and term parts of an entire document and sorts these information in an ascending order of code names to display the normal information in a single line for every code. That is, the extracted term parts, the code names, the area-based generation frequency of relevant codes and the reference terms are displayed. If an attentional display object recognized in the display mode, the part of the display object is shown in a corresponding attentional display form.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a code inspection system for facilitating inspection and elaboration of a code mainly in a text file and for automatically generating an index of an extracted code.

[0002]

2. Description of the Related Art Conventionally, there are various methods for checking a code in a text file. For example, there is a method in which, for a code added to a drawing in a created text, a code described in the text corresponding to the code and a term corresponding to the code are extracted and realized by a KWIC list method. As a first example of the method, there is a “document creation device” of Japanese Patent Application Laid-Open No. 3-132866. According to this, KWIC (keyword in conte
By providing a display function of xt) format and inputting a target code as a keyword, all target codes are displayed on the KWIC screen, thereby enabling online editing and correction. As a second example of the technique, there is a "text inspection apparatus" disclosed in Japanese Patent Application Laid-Open No. 4-296970. According to this, a display function of a KWIC format is provided, a target numeral string is extracted, the immediately preceding character string is used as a drawing reference word candidate by a word dictionary, and grammatical consistency is checked to remove unnecessary candidates, and sorting is performed. It is generated and output in the KWIC format. Here, the KWIC is called a headword with a context, and is used to display a screen on the target keyword (headword) and the context before and after the target keyword and sort them together to facilitate online editing.

[0003] Here, the code is, for example, a symbol for specifying a specific part in a drawing, that is, a graphic code, a definition word used in a document, or a technical word used in a specific field.
rms), etc., in order to make it easier to refer to. These symbols vary depending on the field, industry, and document file type, but are mainly used for numerical values, alphabets, Greek letters, subscripts, and combinations thereof.

[0004]

The number of appearances of an object code (and the object word of the immediately preceding word in a pair) in a generally prepared sentence is, for example, several hundred, but among these, a mistake is made. (Input mistakes) The number of occurrences is far less.

[0005] Incidentally, there are many noise codes (useless codes other than the target code) other than the target code in the text, and they are obtained as a group of extracted codes including these. The noise code generally appears much more times than the number of erroneous entries, depending on the type of text. FIG. 12 shows, as an example of the noise code, a unit numerical value, a part of a term, a figure number, and a frequency numerical value. Because of these noise codes, a large number of useless extracted codes are recognized as target codes by the conventional relatively simple noise code removing means. By the way, the main purpose of the code inspection system is a support system that accurately finds errors.
For this purpose, it is one of the most important issues to remove the noise code as much as possible and to accurately extract the target code.

Conventionally, there are various technical means for removing these noises and improving the recognition rate. However, from the above point of view, in the related art, firstly, as a result of outputting an extracted code including a noise code, there is a problem that the output display is messy and the elaborator is extremely obstructive. Second, K
When outputting in WIC format, the extracted code is displayed together with the contextual sentence in the text.
It is a good term, and it is unpleasant and unpleasant to be displayed as those that clearly do not require elaboration as an elaborator. Thirdly, as shown in FIG. 13, there are many description forms in which a plurality of codes are continuously described as a sentence description form.
Conventionally, proper term extraction has not been possible for such a description form, which is a practical difficulty. Fourth, as shown in FIG. 14, there is a use in which the extracted target code and target term are to be added to the end of the book as an index for indexing. For this reason, it has been difficult to automatically generate a practical index and has not been used in practice.

Therefore, the problem to be solved by the present invention is as follows:
In the present invention, noise codes are removed as much as possible, suspicious codes are recognized, erroneous writing is recognized, the target code recognition rate is significantly improved, codes judged to be questionable are displayed, and unnecessary display is eliminated. In addition, the present invention aims at realizing an accurate revision assist for erroneous writing by displaying a list of all codes according to fields, and further realizing a revision assistance system for automatically generating an index list from extracted codes.

[0008]

First, in order to solve the above-mentioned problems, in the configuration of the present invention, a sentence file is received.
A text file input unit 20 for outputting document data in a predetermined order in response to a setting condition from the extraction condition setting unit 40 is provided. In response to the code conformance condition 62, a code object that conforms to the condition is extracted as an intermediate code 102mid, a term part 101 corresponding to the intermediate code is extracted, and the intermediate code extraction unit 60 that counts the number of code occurrences for each region by the intermediate code is provided. In response to the intermediate code 102mid from the intermediate code extracting unit 60, the same as the basic filter processing unit 81, the field common filter processing unit 82, the user dictionary processing unit 84, and the code explicit term processing unit 86 for the intermediate code. By the classification processing by the term comparison determination processing unit 88, the noise code 106nz and the suspect code 107 (the noise suspect code 11
6nz) and a target code selection unit 80 for classifying and outputting the target code 104obj other than the target code 104obj, and receiving the intermediate code 102mid and the classification result information over the entire document obtained as described above, and changing the code name. Sort in ascending order, 1 for each individual code
Displayed in a row, and at least the extracted term part 1 as the display content
01, the name of the intermediate code 102mid is displayed, the code generation count 400 per code region is displayed, and the reference term 302 of the index list ("code description" is also similar) is displayed.
If there is an attention display target in this display row, the display portion is displayed in an attention display mode, and a code total display unit 10 for displaying a list of codes of all display targets is provided. This makes it possible to accurately remove unnecessary noise codes, recognize suspicious codes, significantly improve the target code recognition rate, and improve the possibility of erroneous writing in the inspection of codes written in the document file and the corresponding terms. Attention is paid to the part and all codes are displayed in a list to realize an accurate elaboration support system for erroneous writing.

Second, in order to solve the above-mentioned problem, according to the configuration of the present invention, a text file is received and the extraction condition setting unit 4
A document file input unit 20 for outputting document data in a predetermined order in response to setting conditions from 0, receiving document data from the document file input unit 20, receiving a code matching condition 62 from an extraction condition setting unit 40, Then, the figure number 102zu target that matches this is taken out, and an intermediate code extraction unit 60 that counts the number of occurrences of each figure number for each area is provided, and receives the figure number 102zu and the number of occurrences for each area. Sort in ascending order, display one line for each drawing number, display the number of occurrences for each region (for each region where similar regions are merged and compressed as desired), and if there is a drawing number index column, A figure number list display means for displaying a reference figure number, and thereafter, in the judgment result display unit 606, displaying in a noticeable display form whether or not there is an erroneous description / missing description / excessive description due to the presence / absence number of occurrences in all sentences. Means. As a result, in the inspection of the figure numbers described in the document file, a revision assisting system is realized which enables the revision judgment of the generation of the figure numbers in each area accurately.

Third, in order to solve the above-mentioned problem, in the configuration of the present invention, a text file is received and the extraction condition setting unit 4
A document file input unit 20 for outputting document data in a predetermined order in response to setting conditions from 0, receiving document data from the document file input unit 20, receiving a code matching condition 62 from an extraction condition setting unit 40, In response to this, the code object conforming to this is extracted as the intermediate code 102mid, the term part 101 corresponding to the intermediate code is extracted, and the intermediate code extraction unit 60 that counts the number of occurrences of each code for each intermediate code for each area.
And the intermediate code 102 from the intermediate code extraction unit 60
In response to the mid code, the intermediate code is subjected to classification processing by the basic filter processing unit 81, the field common filter processing unit 82, the user dictionary processing unit 84, the code explicit term processing unit 86, and the same term comparison processing unit 88, A target code selection unit 80 is provided for classifying and outputting the noise code 106nz, the suspect code 107 (including the noise suspect code 116nz), and the other target codes 104obj, and outputs the extracted intermediate code 102mid
, The useless phrase at the beginning of the term part 101 is deleted, and the intermediate code 102mid
And an automatic index generation means for generating and outputting as a term for, and, if desired, collecting and outputting the same term. As a result, an index automatic generation function is realized from codes described in a document file that requires creation of an index list and terms corresponding to the codes.

Further, in addition to the first constituent means, upon receiving a detailed display instruction for the designated intermediate code 102mid,
For each designated intermediate code 102mid, the area information 26a
rea and line number information 24gyou are displayed, the sentence is displayed from the beginning of the corresponding line, the designated intermediate code 102mid and the preceding term part 101 or term target 201yogo are displayed with attention, and the context after the intermediate code is designated by the specified number of characters (for example, 20 characters). There is a configuration means for providing a detailed code display means for displaying in ()). Also,
In addition to the first constituent means, a banned word dictionary (a banned word dictionary 271dic, an honorific word, a command tone word, etc.) in which a specific prohibited word according to the type and area of the sentence file is registered is provided. When the search is performed on the area or the entire text and a phrase corresponding to the prohibited phrase dictionary is detected, the area information 26area and the line number information 24gyou are displayed, the text is displayed from the beginning of the document line, and By providing a display that highlights the words you want to use,
It is possible to further easily determine the elaboration of the use prohibition phrase specific to the document file type. In addition, in addition to the first configuration means, receiving the intermediate code 102mid, receiving the term subject 201yogo,
There is a configuration unit that includes a KWIC editing unit that associates contexts before and after in a KWIC format, displays a term object on a screen with attention displayed, and performs online editing.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below in detail with reference to examples.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the classification definition of each code in the code extraction process will be described with reference to the relation diagram of FIG. The extraction target code 100all is all description character strings that can be codes in the document. The intermediate code 102mid is an extraction target code 100a.
It is a first-stage extraction code extracted from within ll in accordance with predetermined conditions described later. The target code 104obj is a code character string intended by the author in the document file, and is a code to be extracted. The noise code 106nz is a code to be removed from the intermediate code 102mid by a filter dictionary described later. The suspicious code 107 is a suspicious code that is difficult to determine whether it is the noise code 106nz or the target code 104obj. In this set classification, there are erroneous code 108miss and erroneous term 109 due to the writing error of the author, which are mixed in the target code 104obj, the noise code 106nz, the suspicious code 107 and outside the intermediate code area. ing.
As a code inspection system, it is desired that an elaborator can easily and easily find a target erroneous portion from these, and can perform correction and correction work.

Here, there are three types of erroneous writing of the erroneous code 108miss. First, there are cases where all of the terms corresponding to the object code that occurs a plurality of times are erroneous, and in this case, there are erroneous forms of both erroneous writing on the term side and erroneous writing on the code side. The second is a case where the target code itself is erroneously described. The third is a case where the description of the term corresponding to the objective code is omitted. In addition, the number of appearances for the same code may be only once or may occur plural times. It is desirable to be able to indicate the erroneous description as accurately as possible with respect to these described codes. On the other hand, since the description of the code is allowed by various and various code descriptions depending on the field and the author,
It is practically difficult to directly detect and point out the erroneous code 108miss except for a part. For this reason, in the code elaboration assistance system, as much as possible classifying means is provided, the noise code 106nz is excluded and the suspicious code 107 is classified,
It is required to classify the erroneous code 108miss and the erroneous term 109 and display them in an appropriate display form so that the elaborator can easily make a comprehensive judgment.

A specific example of a patent specification file will be described below as an example of a text file. (Embodiment 1) The processing configuration of a code checking system according to Embodiment 1 of the present invention includes a text file input unit 20 as shown in FIG.
A setting switch from the condition setting unit 40, an intermediate code extraction unit 60, a target code selection unit 80, and a definition code extraction unit 90
And a code tally display unit 10.

First, a text file input unit 20 shown in FIG.
Will be described. The sentence file input unit 20 includes the condition setting unit 4
The data is read out from the file to be inspected line by line in the order corresponding to the read setting condition 22 within 0 and output. By the way, three documents (abstract, specification, application) may be combined in an arbitrary order to form one file. For this reason, the reading setting condition 22 specifies in which document order to read. Normally, the application is read and output in the order of application, specification, and abstract, and skips if there is no corresponding document.
In some cases, it is desired to use the code in the “explanation of code” area in the specification as a reference for inspection. In this case, the condition may be set so that the area is output first. This is performed by the reading order management unit 21 shown in FIG. The processing configuration of the text file input unit 20 is shown in FIG.
As shown in (a), the read order management unit 21, the read setting condition 22, the row number management unit 24, and the area management unit 2
6, a heading order management unit 28, and a heading reference table 29
And a comment deletion unit 32. The line number management unit 24 shown in FIG. 2A has two cases: a line unit is a line feed symbol unit; and a line unit is a logical line unit and a logical page unit. In the case of a patent specification, since the logical line unit (72 bytes) and the logical page unit line (29 lines) are determined, the document data read in line feed units are counted and managed under this line / page condition. The line number information 24gyou is output.

The area management unit 26 shown in FIG. 2A may regard a chapter mark as an area or a specific heading as that area. In the case of a patent specification, since the regions are clearly divided by the black brackets, the heading names of the black brackets (for example, “document name”, “claims”
The area management is performed in units of “Brief description of drawings” and the like, and the area information 26area is output.

The heading order management unit 28 shown in FIG.
Manages the validity of the order of headings corresponding to the file type. The heading here means, for example, “Chapter 1” “1.1
Structure of the manual, "1.2 Precautions in handling", "1.
In the case where chapter numbers are assigned in ascending order as in “3 performance specifications”, or in patent specifications, the order in which explicit headings of black parentheses are generated in the order determined by the application format is defined. Manage legitimacy based on criteria. For this reason, the heading order management unit 28 is provided with a heading reference table 29 (or a numerical ascending order generating means) which is a reference corresponding to the heading and the file type (patent specification, instruction manual, etc.) input in order. 34, it is determined whether or not there is an error in the order of occurrence of the heading (or ascending numerical value) based on the heading relation before and after the heading. If there is an error, heading information 28titl with a warning tag is output together with the heading name, and FIG. (Ac)
A warning is displayed as shown in FIG.

The comment deletion section 32 shown in FIG.
A special internal comment text form that is described as a proof memo in the text by the function of the word processing software and is not output as a final product (document), image data, line drawing data, figure data other than character information , And link information data with the external diagram file. Here, the file format includes a text data format and a binary data format specific to a word processor. The binary data format includes a format in which a figure or image data is attached to a sentence in a single file format, or a format in which the link information data is stored in a separate file. When an information data portion other than the original text is detected, it is deleted and not supplied to the subsequent inspection. However, some non-image data may be in a data format in which a drawing number or the like is stored in a character code form and can be extracted. In this case, the drawing number information 32zu is output and erroneous writing for the extracted code described later is performed. It may be used as a reference for comparison.

Next, the intermediate code extracting section 60 shown in FIG. 1 will be described. The intermediate code extraction unit 60 mainly extracts a code object according to the code matching condition 62 and a term corresponding thereto. As shown in FIG. 2B, this processing configuration includes an intermediate code simple extraction unit 60b, a continuous code recognition processing unit 60c, a plural term recognition extraction processing unit 60d, and an intermediate code extraction exclusion processing unit 60e.

The intermediate code simple extraction unit 60b is a simple mechanical extraction focusing on the code word form, and receives the code adaptation condition 62 shown in FIG. 3 and simply extracts the intermediate code 102mid conforming to the condition from the text. Extract and output mechanically. The corresponding term part 101 located before this code is also output. The sign matching condition 62 eliminates unnecessary extraction, and provides an advantage that the subsequent filtering process is facilitated.

The continuous code recognition processing unit 60c operates as shown in FIG.
As shown in the example of the document, there is a case where the first code, the second code, and the third code are used to describe the continuous code. In this case, the same term portion 101sa immediately before the first code is described.
me is regarded as an individual term part 101 of each continuous code, and is output. The continuous code recognition dictionary 60dic of FIG.
Shows a dictionary example for recognizing the description form of various continuous codes. FIG. 8C shows an example of a description document of a continuous code recognizable by the dictionary. As a result, there is obtained an advantage that it is possible to make the erroneous writing inspection target even for the continuous code description.

The multiple term recognition extraction processing unit 60d extracts the defined terms in parentheses as multiple terms (second / third terms). In the description form of the document, the intermediate code 102mid
The term object 201yogo in the foreground (this is the first term 201
yogo) is a term corresponding to the code itself, but in addition to this, a synonym is defined by adding parentheses, and is described and used together with the code in other places. In response to this, a plurality of terms are recognized. As the plurality of terms, as shown in the example of the document in FIG. 11A, a first term 201yogo before the intermediate code 102mid and a second term 202yogo by attaching parentheses between them.
There is a sentence form in which the second term 202yogo is defined and declared by defining or declaring the second term 202yogo with a parenthesis immediately after the code. These writing methods are commonly used. This second term 2
02yogo and the third term 203yogo describe synonyms or definition words, abbreviations, or English spellings for the first term 201yogo. The descriptive sentence in this case is a description of a code and a term that is normal for the author, and it is desirable that the description be not misjudged. Therefore, the multiple term recognition extraction processing unit extracts the second term 202yogo or the third term 203yogo defined in parentheses together with the first term 201yogo, and extracts the term identification marks 212mrk and 213mrk that are associated as a plurality of terms. To avoid misjudgment during post-processing. This definition declaration statement is usually made up of the second term 202
In many cases, one of yogo is defined and used, but since it can be described in a plurality of places, a large number of N-th terms 20Nyogo may be registered, and each is individually identified. This plural term recognition means has the advantage of eliminating the error of misjudgment even when multiple terms appear for the same intermediate code 102mid, and has the effect of eliminating unnecessary error display output.

As described above for the second and third terms, only the terms are directly described in parentheses, and the terms are described in a definition declaration sentence form as shown in FIG. 11B. There is a description form. In the case of this definition declaration sentence form, it is necessary to extract terms by excluding unnecessary declaration phrases. For this reason, two types of declarative term filter dictionaries 127dic shown in FIG. 11C are provided, and the part corresponding to the demonstrative pronoun of the front prefix is deleted by the front filter dictionary, and the part corresponding to the declarative predicate of the rear suffix is removed. On the other hand, it is deleted by the post-filter dictionary, and the remaining part is regarded as the second term and the extraction processing is performed. Of course, if there is an intermediate code 102mid description part in this, this term is also deleted.

The intermediate code extraction exclusion processing section 60e
The term immediately before the intermediate code 102mid is referred to by using the explicit exclusion term dictionary 60dic shown in (1), and the one that matches the dictionary is regarded as an explicit exclusion term and is excluded from the intermediate code 102mid. This explicit exclusion term dictionary 60dic is
This is an exclusion dictionary applied to all fields, and includes the immediately preceding term basic filter dictionary 60dic1 and the immediately following term basic filter dictionary 60dic2.
And According to the immediately preceding term basic filter dictionary 60dic1 shown in FIG.
Exclude as nz. Also, the case where the term immediately after the intermediate code 102mid corresponds to this according to the immediately following term basic filter dictionary 60dic2 shown in FIG. 23B is excluded as a clearly useless noise code 106nz. As a result, by clearly classifying and eliminating useless noise codes, there is obtained an advantage that the subsequent classification processing is facilitated, and an effect of reducing useless elaboration work is obtained. Of course users (author / editor, etc.)
Is an explicit exclusion term dictionary 60di
c It can be used by appropriately editing (adding / changing / deleting) the contents as desired. Thus, together with the output of the intermediate code 102mid, the byte position information 61pos of the intermediate code 102mid in the text and the position information, the line number information 24gyou, the area information 26area, and the heading information 28titl obtained by the text file input unit 20 are also encoded. Is stored in the intermediate code storage unit 68 in association with the data and is used for the subsequent processing.

The condition setting section 40 shown in FIG. 1 is a setting table of various setting conditions, and can arbitrarily change initial conditions in response to setting from a user, and also includes a code matching condition 62 described below. Supplies the execution control condition to each processing unit.

Here, the code matching condition 62 in FIG. 3 will be described in detail. The code matching condition 62 is a part of a setting switch of the condition setting unit 40, and has a setting condition for a code word form to be extracted as a target code, and can be set individually and arbitrarily. In each item, the double-byte numerical value setting is a switch (ON / OFF) for extracting a double-byte character or not, and the single-byte numerical value setting is a switch for determining whether or not a single-byte character is extracted. "" Means that the number of alphabetic characters immediately before the numerical value is within a predetermined number of characters (0 to 7) to be extracted, and "after the numerical value" indicates that the number of alphabetic characters immediately after the numerical value is within a predetermined number of characters (0 to 10). The target, full-width alphabetic setting before the numerical value is a switch whether or not to extract in the case of the full-width alphabet immediately before the numerical value, and the full-width alphabetic setting after the numerical value is a switch whether to extract in the case of the full-width alphabet immediately after the numerical value The setting of the first half-width alphabetic character is a switch of whether or not to extract in the case of the half-width alphabet immediately before the numerical value, and the setting of the second half-width alphabet of the numerical value is a switch for determining whether to extract in the case of the half-width alphabet immediately after the numerical value. It is. Further, the range of the numerical value is a numerical character set of 1 to 999.
An arbitrary range in the range of 9 (can be divided into multiple areas) is valid for extraction, and the specification of odd-even is to specify whether the code value is valid for odd, even, or both are valid. The line range means that an arbitrary range (which can be divided into a plurality of areas) whose file line number (or page / line number) is 1 to 99999 (or a specified line of a specified page) is valid. A region (a plurality of regions can be specified) that matches the specified matching heading character string (a chapter number is also regarded as a heading) is extracted and the line range is defined as a line number (or page / line number) of a file of 1 to 99999 (or specified). Extraction of an arbitrary range (can be divided into multiple areas) in the specified line of the page is valid. Superscript value setting is a switch of whether or not to extract a superscript value. Subscript value setting is a subscript value. Switch to extract or not The superscript alphabet setting is a switch of whether or not to extract a superscript alphabet, the subscript alphabet setting is a switch of whether or not to extract a subscript alphabet, and the uppercase setting is to extract uppercase letters Is a switch for determining whether or not to extract lowercase letters. A description example corresponding to each of these item conditions is shown in the code example on the right side of FIG.

Next, the target code selecting section 80 shown in FIG. 1 will be described. The target code selection unit 80 includes the intermediate code extraction unit 60
In response to the intermediate code 102mid from the intermediate code 102mid, the following four types of determination processing are performed from the intermediate code 102mid by a five-stage selection process. That is, the noise code 106nz, the noise suspect code 116nz, the suspect code 107, and the other object code 10
Four classification processing of 4obj is performed. This processing configuration includes a basic filter processing unit 81, a field common filter processing unit 82, a user dictionary processing unit 84, a code explicit term processing unit 86, and an identical term comparison processing unit 88, as shown in FIG. Become.

The first-stage basic filter processing section 81 performs a classification process using the basic filter dictionary 64dic shown in FIG. FIG. 4A shows an example of the immediately preceding term basic filter dictionary 64dic1 which is to be classified for the character string immediately before the intermediate code 102mid. FIG.
(B). FIG. 4C is an example of a term term basic filter dictionary 64dic2 which is to be classified for the character string immediately after the intermediate code 102mid. FIG. 4D shows an example of a corresponding document. If one of the codes immediately before and after the code matches, the classification is made as the suspect code 107, and if the code matches before and after the code, the classification processing is performed as the noise code 106nz. As a result, the intermediate code 102mid is further excluded from the noise code 106nz which is considered to be useless. Of course, the user can edit and use the contents of the basic filter dictionary 64dic as needed. However, there is a special condition of adding "(" and ")" before and after the code as in the example of the claim area in the patent specification, and this area is regarded as an extraction target.

In the second stage common filter processing section 82,
One or a plurality of desired dictionaries are selected for use from a plurality of common filter dictionaries 83dic for each application field. The common filter dictionary 83dic includes industrial fields (machinery,
Code names specific to the types of text files (chemical, electric, electronic) and text file types (manuals, manuals, dictionaries, specifications, official documents, etc.) and specific codes are registered in the filter dictionary. FIG.
(A) shows an example of a common filter dictionary 83dic using an immediately preceding filter term and an immediately following filter term specific to the electric field. In the same manner as in the case of the basic filter dictionary 64dic described above, the code dictionary immediately before and after the code is compared with the corresponding filter dictionary before and after the code.
, And when they match both before and after the code, the classification processing is performed as the noise code 106nz. Here, matching before and after the sign means matching before and after including the first-stage basic filter dictionary 64dic. For example, basic filter dictionary 64di
If the previous term matches in c, and the immediately following matches in the user dictionary 85dic, it is regarded as a front-back match. FIG. 5B shows an example of a document classified into the noise code 106nz and the suspicious code 107 corresponding to this. In this immediately preceding filter dictionary, “all hiragana” and the filter description mean that any of the hiragana is detected. FIG. 5C shows an example of a filter dictionary specific to the machine field as another example of the field common filter dictionary, and FIG. 5D shows an example of a corresponding document. FIG. 5E shows an example of a filter dictionary specific to the patent field and an example of a document corresponding to the filter dictionary. These are field common filter dictionaries 83 for a single field or a plurality of fields depending on the file type.
This filter processing is executed by citing one or more dic. Needless to say, the user can use the common filter dictionary 83di in this field.
c The contents can be edited as desired and used.

The user dictionary processing section 84 in the third stage is a filtering process using a user filter dictionary 85dic that is customized and used for each elaborator as needed.
The user filter dictionary 85dic1 immediately before the code and the user filter dictionary 85dic2 immediately after the code in FIG. 6C constitute a dictionary. This is because, in many cases, the definition code characters are determined differently depending on the personality, experience, or technical background of the manuscript author, or a unique code description that is characteristic is adopted. Is a personal user filter dictionary 85dic for individually registering and using filter words as needed according to the file type. The processing here is the same as the above-described filter processing except that the user filter dictionary 85dic is used.
If both immediately before and after mid match, noise code 10
The identification processing is performed as 6nz, and if only one of them matches, the identification processing is performed as the noise suspect code 116nz. Here, the match before and after the code means the common filter dictionary 83 in the second stage.
dic and match before and after including the first stage basic filter dictionary 64dic. For example, if the term immediately before matches in the common filter dictionary 83dic and the word immediately after matches in the user filter dictionary 85dic, it is regarded as a match before and after. This makes the first
The noise code 106nz and the noise suspect code 116nz are further selected from the suspect code 107 and the intermediate code 102mid after the step processing.
Classified as FIG. 6A shows an example of a document corresponding to an example of a user filter dictionary immediately before a code, in FIG. 6B. FIG. 6C shows an example of a document corresponding to the example of the user filter dictionary immediately after the code, in FIG. 6D. By the way, depending on the document description, there is a case where it is desired to explicitly exclude it as a non-sign.
As an example of this case, a half-width space character or a full-width space character is intentionally described immediately before or immediately after a character string to be filtered in the document, and the corresponding half-width / full-width space character is written in the user filter dictionary. There is also a usage method in which a filter is registered and a filter exclusion process is surely performed.

The fourth step, the code-specific term processing unit 86, performs a process of directly specifying the target code 104obj using the code-specific term dictionary 86dic shown in FIG. 27 (a, b). In general, there are many cases where terms specific to a field of use or an author are used. Focusing on this point, FIG.
FIG. 27A shows an explicit term dictionary 86dic1 immediately before a code shown in FIG.
An explicit term dictionary 86dic2 immediately after the code shown in (b) is provided.
If the code corresponds to the explicit dictionary, the target code 104
Classify directly as obj. FIG. 27C shows a document example corresponding to this. In this way, when the object code corresponds to the explicit term dictionary, the classification process is immediately performed as the object code 104obj, thereby obtaining an advantage that the recognition rate can be further improved. Conversely, the document often uses a number / standard name (eg, transistor standard name: 2SC173) of a specific part, IC product, device, recommendation, etc., and also wants to explicitly delete this. There is. For this purpose, a standard name filter dictionary 61dic as shown in FIG. 26 is provided.
Exclusion processing is performed assuming 06nz. In this dictionary, the “?” Symbol is an arbitrary single or plural numerical character strings (0 to 99999).

The same term comparison / determination processing unit 88 in the fifth stage
Is an erroneous detection detecting means that uses the extracted term, searches the entire sentence in reverse from this term, and classifies the detected thing as an erroneous note 200 err and takes it up. Secondly, there is an unsigned error detection means. Here, the intermediate code 102mid
May not be exhaustively detected. Figure 1 shows the normal case
9 (a) as "R12", and the code matching condition 6 shown in FIG.
A document example that conforms to 2. In this example, it is assumed that the immediately preceding term object 201yogo “reference resistance” is also described in the erroneous example in FIG. 19B. Here the term object 20
1yogo (eg "reference resistance")
Using the deleted word dictionary 150dic1 at the head of the term and the non-deleted Hiragana dictionary 150dic2 at the head shown in FIG. 9A, unnecessary words at the front are deleted to extract an appropriate term object 201yogo. On the other hand, in the document example of FIG. 19B, “R12” is an intermediate code 102
Assume a document example not extracted as d. As another document example, a document example in which the description of the code itself is missing is shown in FIG.
As shown in (c), also in this case, it is not extracted as the intermediate code 102mid.

In order to extract the erroneous document example shown in FIG. 19B, the first code erroneous detection means uses an intermediate code 1
The erroneous code is extracted from the extraction target code 100all excluded from 02mid. For this purpose, codes are extracted by temporarily maximizing the restriction of the code matching condition 62 shown in FIG. 3 and extracting the codes. As a result, the codes shown in FIG. 19B are detected as intermediate codes. For this new intermediate code, it is checked whether the term immediately before this matches the term object 201yogo in FIG. 19 (a), and if it matches, it is determined to be a code error 88miss and classified and output. Enable. Providing this detection means has the advantage that erroneous writing of the code itself can be detected.

Next, in order to cope with the erroneous document example shown in FIG. 19C, the second unsigned erroneous error detection means assumes that the term subject 201yogo is correctly described elsewhere. Take out. For this purpose, a full-text search is performed using all the term objects 201yogo that have already been obtained, and as shown in FIG. 19 (c), an unsigned one immediately after the matching term is found, and this is determined to be an unsigned mistake 89miss. Then, the output is classified and output as described later is enabled. By providing this detection means, it is possible to obtain an advantage that it is possible to detect even an erroneous description of a missing code description, thereby obtaining an advantage that the elaboration work is further facilitated.

If the number of occurrences of the noise code 106nz and the noise suspect code 116nz classified in addition to the classification means is plural, the probability of the target code 104obj is high.
By utilizing this, when there is a matching part of the term part 101 immediately before, a determination process that regards the target code 104obj as possible can be performed. That is, the same code count M as the execution selection is provided in the setting switch of the condition setting unit 40, and when it is to be executed,
The term part 101 immediately before the noise code 106nz and the noise suspect code 116nz for each individual code name is AND-processed,
If there are two or more remaining words in the result and the number of same codes is M or more, it is assumed that this is the target code 104obj, and processing is performed. As a result, the identification of the target code 104obj can be determined, and an advantage that unnecessary warning display can be reduced can be obtained.

In addition to the above classification means, in the noise code 106nz and the noise suspicion code 116nz classified in the above, in the character string from the end side of the extracted term display portion 101dsp described later, the number of continuous kanji / Katakana characters If there are many, the probability of the target code 104obj is high,
In some cases, the determination processing may be considered as 4obj. Therefore, the control of the execution selection and the conditions of the number P of the trailing side continuous Chinese characters, the number Q of the trailing side continuous katakana characters, or the number R of the trailing side continuous Kanji / Katakana mixed characters R are provided in the setting switch of the condition setting unit 40 and executed. If it should be, it is checked whether or not the number of the character strings at the end corresponding to each of the continuous character number conditions P, Q, and R. If so, it is assumed that this is the object code 104obj. As a result, the object code 104obj
, And the advantage that unnecessary warning display can be further reduced can be obtained.

The noise code 106nz, the noise suspicious code 116nz, and the suspicious code 107 classified by the respective classifying means are described.
For the target code 104obj, the number of occurrences is counted for each region and for each same code in order to display a list by region later.
However, the number of occurrences of an intermediate code having a term identification mark such as a second / third term is counted individually even if the code is the same.

Next, the definition code extracting section 90 shown in FIG. 1 will be described. The definition code extraction unit 90 is optional,
In the case where there is an “explanation of code” area column shown in the example of FIG. 7, the reference code 301 and the corresponding reference term 30 are used to obtain the code and term described here for comparison reference of the standard.
2 is extracted and supplied to the code tally display unit 10. Further, the intermediate code 102mid corresponding to the code name in the "description of code" area is directly classified as an obvious target code 104obj, and priority classification is performed even if this code is classified as a suspicious code or a noise code. In other document files, in general, an index list may be provided at the end of the book or the like, and reference numerals 301 and reference terms 302 are extracted by the same means from code objects in this index area. Also,
Figure number information 3 output by the comment deletion unit 32 shown in FIG.
When there is 2zu, if it is obtained from this, the same processing is performed, and it is also supplied to the code tally display unit 10. Also, depending on the setting of the code matching condition 62, the code may not be extracted as the intermediate code 102mid. As an optional operation for this, if the reference code 301 extracted here does not exist in the intermediate code 102mid, the search is performed over the entire text and the same code name found is searched for the intermediate code 102mi.
The classification processing may be immediately performed as d and the target code 104obj.

Next, the code tally display section 10 shown in FIG. 1 will be described. The code tally display unit 10 receives the intermediate code 102mid and the term portion 101 extracted and classified result information over the entire document obtained above and sorts them in ascending code name order as shown in the display output example of FIG. , Normal one for each code
Display on a line. That is, the extracted term part 101 is displayed, the code name is displayed, the number of occurrences of the code by region is displayed,
The reference term 302 in the “explanation of reference numerals” column is displayed. If there is an attention display target in this display, that portion is displayed in a corresponding attention display mode. As shown in FIG. 7, this processing configuration includes a code count display section 11, a multiple occurrence term section and processing section 12, a reference term comparison warning processing section 13, a useless reference code warning processing section 14, It comprises a code warning display processing unit 15, a full-width / half-width same-code identification warning processing unit 16, and a display line display control unit 17.

As shown in an example of a tabulated list display in FIG. 20, the code count display section 11 counts and displays the number of occurrences 400 over the entire area for the same code and the same term subject 201yogo, and generates an individual occurrence for each area. The number 401 is counted and displayed.
Further, the total number of code names 402 and the total number of code occurrences 403 are counted and displayed, and the non-code occurrence number 404 is counted and displayed. Since the number of non-code occurrences 404 may possibly be erroneous in some cases, this guideline is displayed. Here, the non-code counting target is the noise code 106nz and is not displayed and output.
On the other hand, the code objects are the noise suspect code 116nz and the suspect code 10
7 and the object code 104obj, and these code objects are
The list is displayed as shown.

In one display line, the same reference numeral 102mid and the same term subject 201yogo are displayed in one line. Naturally, the second term 202yogo and the Nth term 20Nyogo are individually displayed on separate lines. In addition, the display order in one line in the example of the summary list display shown in FIG.
The extracted term display portion 101dsp, the code name, the number of claims, the number of prior art occurrences, the number of occurrences of the solving means, the number of occurrences of the embodiment, and the reference term 302 of "explanation of the sign" are in that order. Here, in the display of the reference term 302, when there is a plurality of definition terms 305 (terms corresponding to the Nth term) of the reference term as shown in FIG.
Is displayed and output. In this display example, the area column where the count value is 0 is given a symbol “Δ” to make it easier to see. In addition, the area name 406 is compressed to four points (claims, conventional technology, solution, embodiment), and the other areas are the corresponding areas (for example, “embodiment” and “action”) among the four points.
This is an example in which the “commercial”, “effect”, and “abstract” areas correspond to the “embodiment” area) so as to fit in the 80-character display of a normal DOS screen.

The multiple occurrence term AND processing section 12 generates and outputs a character string to be displayed in the extracted term display portion 101dsp. However, the second term to the N-th term explicitly declared are individually AND-processed in order to be displayed on independent display lines.
First, the word part 101 corresponding to the same code 102mid is AND-processed, and the remaining matched character string part is displayed and output as an extracted terminology display part 101dsp. Further, in the AND processing, when the remaining character string of the AND processing is lost or becomes only one character, it is regarded as the erroneous term 109 or the erroneous sign 108miss, and the next line is divided into the following lines. Both are displayed and output. Here, the attention display is to give a display form that the elaborator can easily pay attention. For example, a plurality of types of color display, reverse display, blinking display, underline display, mesh display, etc. And the degree of suspicion. This attention display has an advantage that the erroneous code 108miss and the erroneous term 109 can be easily recognized and determined. In addition, by using these list display output modes, it is possible to easily confirm at a glance which codes are generated in which areas and how many times they are used, and it is possible to obtain a great advantage of facilitating the elaboration judgment for the erroneous writing of the codes from the distribution information of these codes. .

In the reference term comparison warning processing section 13, when there is a mismatch with the reference term 302, it is regarded as an erroneous description and attention display processing is performed. That is, as shown in FIG. 20, for an intermediate code 102mid display line that matches the reference code 301 in the “description of code” area obtained by the definition code extraction unit 90, the extracted term display part 101dsp and the reference term 302 Definition term 30
In the case where all the character strings do not match the reference term 302, a part that does not match the reference term 302 display part is displayed in the attention display form 521. As a result, the erroneous term 109 can be clearly pointed out, and an advantage of facilitating accurate elaboration determination can be obtained. Of course, the intermediate code 102mid which does not correspond to the reference code 301 is blanked out, and it is clear that it is not in the “description of code” column. can get.

When the reference code 301 obtained in the "explanation of code" area does not exist in the entire text, the warning processing unit 14 of the useless reference code displays a notice because it is considered to be erroneous or missing. That is, as shown in FIG. 24, the reference term comparison / warning processing section 13 checks whether there is an unused reference code 301, and if so, displays it as an unnecessary code declaration on an independent line in the attention display form 512. . This may be due to obvious missing code name description, erroneous reference code 108miss, or erroneous description in “description of code” or excessive description. In any case, the existence probability of erroneous description is high, and the advantage that accurate indication can be obtained can be obtained. .

The suspicious code warning display processing unit 15 displays the entire row of the suspicious code as a noise suspicious code 116nz or a suspicious code 107 as shown in the summary list display of FIG. 531 is displayed. This clearly indicates that the code is different from the target code 104obj, and indicates that the code is a detailed confirmation required code. In the case of a suspicious code, it is difficult to automatically determine whether or not it is an erroneous description due to filter conditions or dictionary contents, and it is appropriate for the elaborator to judge these. Usually, the user can easily edit various filter dictionaries as needed so as to minimize these suspicious codes, so that convenience can be easily improved. In displaying the suspicious code line, the elaborator specifies the AND processing display portion 101dsp and the number of occurrences of each area or the reference term 302.
It is relatively easy to judge whether or not there is an erroneous writing from the comprehensive information on the presence or absence of the error.

The full-width / half-width same-sign identification warning processing unit 16
Write the description for the same sign in both full-width and half-width,
There are both cases where both should be recognized as the same, and conversely, one of the descriptions should be determined to be erroneous code 108miss, and the processing changes depending on the condition setting switch. That is, the condition setting unit 40
Receiving a setting switch for full-width / half-width code identification from
First, when the same two-byte / one-byte codes are to be used, both codes are regarded as full-width codes, merged, counted and displayed on one line, and secondly, when only two-byte characters are to be used, as shown in FIG. Is displayed in the attention display form 541 in which the half-width description code is regarded as an erroneous description.
Third, when only a half-width code is to be used, the full-width description code is displayed and output in a noticeable display form in which the code is regarded as erroneous. This has the advantage of assisting erroneous writing decisions.

The display line display control unit 17 is used when it is desired to suppress the display of a non-warning line that does not require elaboration. For example, after confirming the sign of the list display, there is a case where it is desired to display only the questionable code line displayed in the attention, and a display option corresponding to this, the display output control of whether to display only the attention display line is performed. Do. In the case where the setting switch is received from the condition setting section 40 and only the display row of interest is set to be displayed, as shown in FIG. 22, the display output is skipped for a non-warning row having no display section of interest. As a result, the list display screen in the case of a large number of codes does not flow and does not disappear, so that there is an advantage that only the target object to be revised is displayed and the revision is easy. In addition,
In the above description, the display control is performed for the attention display. However, the display control may be performed by providing a setting switch of the condition setting unit 40 that displays only the attention display mode designated as desired.

(Second Embodiment) In the description of the first embodiment, the noise code 106nz, the noise suspect code 116nz, the suspect code 107, and the target code 1
Although the case where the classification process is performed as 04obj is performed, the suspicion degree determination unit using a weighting method described below may be used. That is, a weight value counter is provided for each of the extracted intermediate codes 102mid, and if a corresponding weight value is obtained by each filtering process, the corresponding weight value (which can be arbitrarily changed by the user) is added to the weight value counter. For example, the basic filter dictionary 6 in FIG.
When 4dic is applied, 10 points are added. When the field common filter dictionary 83dic shown in FIG. 5 is applied, 5 points are added. When the user filter dictionary 85dic shown in FIG. 6 is applied, 2 points are added. A means for receiving the value of the weight value counter obtained in the final processing stage, dividing the display into a plurality of ranks according to a desired value, and arbitrarily selecting a display mode for each weight rank is provided. It is also possible to use the determination and attention display means to have a degree of freedom so as to display the corresponding attention.

(Embodiment 3) Intermediate code 1 in which the result of the above-mentioned list display form is pointed by a cursor (or a mouse) by an elaborator
Upon receiving 02mid or the intermediate code 102mid of the display object of interest, the KWIC editing means of the prior art may be used as the first means for editing as shown in FIG.
As a second means, using the multi-window function on WINDOWS, while viewing the list display window display, the position information of the code and the line number information 24gyou are given on another window screen to provide a document file to be inspected (source file). ) May be edited.

(Embodiment 4) As another embodiment, FIG. This figure number 1
02zu can also be considered as a kind of code, and an application example in which this is regarded as a code will be described below. It is generally used in the description with reference to a number of drawings. Of course, errors in the figure numbers themselves are not allowed and are subject to elaboration. Figure number 1
The numerical value of 02zu is generally in a continuous ascending order. Then, using the same technical means as the means for realizing the code list display of the above embodiment, the figure number 102zu is extracted, and the number of occurrences for each occurrence area is counted, and as shown in FIG. By displaying a list of numbers and noting the presence / absence of an erroneous description / missing description / excessive description for each figure number, accurate refining assistance can be provided to a refining person. By the way, the description form of FIG. 102zu varies and varies depending on the industrial field, file type, author, and the like. For this reason, the figure number 102zu in the diverse description form is compared using the figure number reference table shown in FIG. 31 (a), and converted into the unified drawing number shown in FIG. I do. In this dictionary, the "?" Symbol is any single or plural full-width / half-width numeric character strings (1 to 999), and the alphabetic characters "a, n" are any full-width / half-width alphabet. As an example of this unified conversion, see FIG.
(A) "," FIG. 4a "," FIG. 4A "," FIG. 4a ",
The description examples of "FIG. 4- (a)" and "fig.4a" are converted into a unified "FIG. 4 (a)" before use.

In the case of a patent specification, FIG.
There is an "explanation of figure" column area as an index of 2zu, and it is determined that the numerical values are consecutive in ascending order. Utilizing this, the figure number 102zuref of this area is extracted as the reference figure number 102zuref in the same manner as in the first embodiment, and is compared with the figure number 102zu extracted from the document in the other area. In this case, it is displayed in the attention display form 551, and further elaboration assistance display is enabled. An example of the diagram number list display shown in FIG. 30 will be described. The extracted figure numbers are sorted in ascending order and each figure number is displayed on one line. The display order in each line is the figure number
The order of occurrence of the prior art, the number of occurrences of the solving means, the number of occurrences of the example, the number of occurrences of others, the number of occurrences of "explanation of figure", and the determination result display unit 606 are in that order. Also in this display example, the area column in which the count value is 0 is provided with a symbol “Δ” to make it easier to see. The area name 406 is also merged and compressed in the same manner as in the first embodiment. For example, the “prior art” area is a combination of “conventional technology” and “problem to be solved”. The individual display can be optionally performed by changing the setting switch. In this display, the judgment result display unit 606 displays the figure number 102zure described in “Description of the figure”.
If f is not in any area, or conversely, if the figure number 102zu in the document is not described in the "Description of figure", it is displayed in the attention display form of "deletion / addition required", and a figure with an alphabetic subscript Numbers (for example, “FIG. 1 (a)”) are also extracted as being regarded as individual figure numbers, and when there is no corresponding figure number 102zu in the document, it is displayed in the attention display form of “addition request”. Further, the figure number specified in the document as "conventional" or "of the present invention" is detected in the "Description of figure" column, and the conventional / invention indication 604 is displayed in the "Description of figure" display field. May be provided. With these, it is possible to further facilitate the elaboration determination of which region should appear. FIG. 32 is a diagram number list display example showing the execution result of the present specification itself.

(Embodiment 5) An automatic index generation means in the form of "explanation of codes" which is an index list based on the codes and terms extracted by the means of the first embodiment will be described. For this purpose, there is a need for a means for appropriately deleting the non-phrase phrase at the head of the term part 101. In the example of the leading-side deleted phrase dictionary 150dic1 shown in FIG. 15A, a character string (mainly a case particle, an ending, a connecting particle, an auxiliary verb, and the like) at the head which is not normally regarded as a term is registered. I have. In addition, the example of the head side non-deletion hiragana dictionary 150dic2 shown in FIG. 15B is reverse to the head side deletion word dictionary 150dic shown in FIG.
A word with hiragana to prevent deletion is registered in dic1, and the corresponding hiragana is not deleted. For example, since "out" is detected in the term portion 101 "read memory by circuit", the "shi" at this position is not deleted and the final generated term is "read memory". Both dictionaries,
The user can add / modify the contents of the dictionary at any time, and the quality of automatic generation can be easily improved. Using the head side deleted phrase dictionary 150dic1, the non-term phrase part at the head side is deleted from the code example corresponding to the term part 101 shown in FIG. Then, the codes are sorted in ascending order and output in the form of “explanation of codes”. FIG. 15D shows an example of this automatically generated output. If there is an N-th term, it is output collectively in the form of a multi-defined term 305 with parentheses. In this case, "Description of the code" in the specification
This is the case when output is performed in the expression format corresponding to. Similarly, it is easy to cope with other output forms. For example, it may be automatically generated and output in the form of an index list attached to the end of a general document or the like. As shown in this embodiment, a highly practical index list can be mechanically and automatically generated and output from the extracted codes by means of an automatic generation means. This provides an extremely convenient advantage that the author can be freed from most of the work of creating the index list. can get.

(Embodiment 6) In addition to the means of the first embodiment, a detailed display function means for each designated intermediate code 102mid may be provided. The detailed display function of the designated code is a function of individually displaying the context before and after the instruction code 107 for each code generation in detail, thereby facilitating revision. Normally, the result is displayed first by moving the cursor (or mouse) on the screen to the position of the instruction code 107 to be examined. FIG. 16 shows the execution result. In this example of the display mode, for the line having the instruction code 107, the display head has the area information 26area display and the line number information 24gyou.
Attention is displayed on the display, the sentence is displayed from the beginning of the document line,
This is an example in which the instruction code 107 and the term part 101 in front are displayed with attention, and thereafter, the display is performed with the designated number of characters or less (the setting switch here is 20 characters) or less. As a result, there is obtained an advantage that it is possible to easily and accurately determine the elaboration of which area is to be used and what the description content is. Of course, an elaboration means for displaying a context before and after a designated code by a conventional KWIC method may be used.

(Embodiment 7) In addition to the means of the first embodiment, a use prohibition phrase detecting function means for limiting the description style depending on the type of the document file may be provided. This use prohibited phrase detection function may be a special use prohibited phrase depending on the industrial field, manuscript file type and area,
It is a kind of culmination as a so-called "do not collect". In the patent specification, there is a prohibited word in the claim area where the right range is ambiguous, as in the use prohibited word dictionary 271dic shown in FIG. For this reason, only the claim area document is inspected and, if applicable, as shown in the warning display example of FIG. 28B, the 26 area display and the line number information 24gyou display are noticeably displayed, and the text is displayed from the beginning of the document line. This is an example in which the corresponding phrase is noticeably displayed.
In this case, not the erroneous writing of the code but the erroneous writing of the document phrase, but the object of the elaboration is improved, and the convenience of the elaboration work is improved. In other file types, similar use-prohibited words may be described in the instruction manual in the form of "Damasuma-tone" without using honorific words as shown in FIG. 29 (a). Further, the command tone phrase shown in FIG. 29B may be prohibited, or the ambiguous tone shown in FIG. 29C may be prohibited. As described above, the use prohibited word dictionary corresponding to the file type in the unique field is provided, and the erroneous writing of the sentence word is detected. In this way, a prohibited word dictionary (such as a useless fixed-word dictionary, a redundant word dictionary, a sentence kana unified dictionary, a katakana notation dictionary, etc.) is provided to limit the description style according to the document file type. By executing the process, the erroneous description of the prohibited word is easily displayed in a noticeable manner, the elaboration work with few oversights can be performed, and the convenience is further improved.

In the example of the display output described in the first embodiment, the horizontal direction is adjusted according to the screen size on MSDOS.
In the specific example of the case where the area is merged and compressed so that it fits into 0 characters (40 double-byte characters) and the display is output as in the example of the total list display shown in FIG. 20, the number of merged compressions is reduced (or No merged), count information is output for each individual area in a list, and the display of the part protruding from the screen is displayed in a wrapped manner, or a vertical / horizontal scroll display function of the display screen is provided so that a desired part can be displayed on the screen. You may do it. In addition, in the case of printer output, multi-character printing is generally possible. In response to this, the number of merged compressions may be reduced to print more detailed information in a list than on the screen. Since the number of characters in the vertical and horizontal directions can be displayed on the WINDOWS upper screen by changing the character font size, a means for displaying a list of more detailed information as shown in FIG. 21 may be provided if desired.

In the description of the first embodiment, a specific example of a patent specification is described. An index list corresponding to “description of reference numerals” is provided in another document.
Obviously, a similar technique can be used for this. In the description of the first embodiment, the noise code 106
Although nz was excluded from the list display, if necessary, a setting switch was provided in the condition setting unit 40 to set the noise code 10
6nz may be displayed in the noise code attention display form in the same manner as the warning display of the suspicious code, and may be displayed and output as an aid in elaboration determination. In the description of the first embodiment, the case where the target code selection unit 80 divides the dictionary into a plurality of dictionaries (a basic filter dictionary 64dic, a common filter dictionary 83dic, and a user filter dictionary 85dic) has been described. A configuration may be adopted in which batch processing is performed using a set of dictionaries. In the description of the first embodiment,
Although the description has been given of the case where the constituent means of the code-specific term processing section 86 and the same term comparison / determination processing section 88 are provided in the target code selecting section 80, a constituent means in which both or one of these constituent means is deleted as desired may be adopted. good. Alternatively, a setting switch may be provided in the condition setting unit 40 to perform each execution selection control.

In the description of the first embodiment, the case where the definition code extracting unit 90 is configured has been described. However, if desired, this configuration may be deleted or the condition setting unit 40 may be provided with an execution selection switch. Execution selection control may be used. Also, in the description of the fifth embodiment, the automatic generation means in the form of “description of code”, which is an index list, is described in parallel with the code totaling display unit 10. If desired, FIG.
As shown in FIG. 5, the configuration of the code tally display unit 10 may be omitted.

Further, in the description of the first embodiment, a case has been described in which the code object always has a numerical value as shown in the example of the code conformity condition 62 shown in FIG. 3 as a code object. However, if desired, the number of numeric characters shown in FIG. 3 is set to 0 and only alphabetic characters (for example, “fosc, Ts
tart, Din ”) may also be executed in association with the subsequent processing.

[0060]

According to the present invention, the following effects can be obtained from the contents described above. Code matching condition 6 shown in FIG.
Reference numeral 2 is an extraction condition table for simply and mechanically excluding unnecessary noise codes, whereby many explicit useless noise codes are extracted and excluded, thereby obtaining an advantage of facilitating subsequent filtering. Intermediate code 102mid from the document
And, for the code itself, words before and after it, multiple dictionaries that are excluded from the target code, multiple dictionaries that specify the target code, dictionaries that specify erroneous codes, and dictionaries that recognize continuous codes From 102mid, the target code 104obj, the suspicious code 107, the noise code 106nz, the code error 88miss, the non-code error 89miss, and the error code 10
8miss, and the number of occurrences for each area is counted. If there is an index column (eg, "description of code" column), the reference code and the reference term are taken out, and each code is extracted from these for each region. By displaying in a list display form and in a noticeable display form, it is possible to display at a glance whether the code is erroneous or suspicious or a code to be refined. As a result, there is obtained an advantage that the elaboration work can be realized based on accurate judgment.

FIG. 25 shows the result of the implementation of the present invention. This is a part of the result of performing the code check of the present specification itself,
It shows the result in an output form in which the code collation area and the number of occurrences and the "code description" are displayed automatically at a glance, and it is clear that the elaboration judgment is easy and effective.
In addition, this display form can dramatically reduce the display amount compared to the conventional KWIC system, greatly reduce the confusing of the elaborator due to unnecessary redundant display, and furthermore, elaborate by paying attention to the part which seems to be erroneous. There are many advantages that can be made easier.

FIG. 32 shows a part of the execution result of the figure number inspection of the present specification itself, and shows an output result in which the figure number generation area and the number of occurrences are sorted. The effect that the validity of the figure number can be confirmed at a glance from the result display unit 606 is obtained, and it is understood that this is a practically effective figure number inspection function. FIG. 33 shows a part of the automatically generated and output result of the “description of code” column according to the execution of the present specification, and automatically generated and output in the “description of code” of the patent specification from all the extracted codes and terms. The results show that the system has a sufficiently practical automatic index generation function. It is sufficient to write only a part of the index, and the writing work for the index part can be largely deleted. The effect is great.

[Brief description of the drawings]

FIG. 1 is a processing configuration diagram of a code checking system according to a first embodiment of the present invention.

2A and 2B are a processing configuration of a text file input unit 20 and a processing configuration of an intermediate code extraction unit 60 according to the present invention.

FIG. 3 shows an example of a word limiter condition setting table and an example of a character string according to the present invention.

FIG. 4 shows (a) an immediately preceding term basic filter dictionary, (b) an immediately preceding term applied document example, (c) an immediately following term basic filter dictionary, and (d) an immediately following term applied document example of the present invention.

FIG. 5 shows (a) an example of a filter dictionary specific to the electric field, (b) an example of this document, (c) an example of a filter dictionary specific to the mechanical field, (d) an example of this document, and (e) of the present invention. And (f) an example of this document.

FIG. 6A shows an example of a user filter dictionary immediately before a code, (b) an example of a corresponding document of a user filter dictionary immediately before a code, (c) an example of a user filter dictionary immediately after a code, and (d).
It is an applicable document example of a user filter dictionary immediately before a code.

FIG. 7 is a processing configuration diagram of a code tally display unit according to the present invention.

8A is an example of a continuous code description form document, FIG. 8B is an example of a continuous code description word dictionary of the present invention, and FIG. 8C is an example of a document in which a continuous code is described using a continuous code description word.

FIG. 9 is a processing configuration diagram of a target code selection unit according to the present invention.

FIG. 10 is an explanatory diagram of a classification definition of each code in a code extraction process.

FIG. 11 (a) a description explanatory diagram of a second term and a third term, (b) a description example of a second term in the form of a definition declaration sentence,
It is an example of the multiple term filter dictionary for (c) definition declaration sentence form of this invention.

FIG. 12 is an example of a noise code.

FIG. 13 is an example of a continuous description form of a plurality of codes.

FIG. 14 is an example of an index document for an index.

FIG. 15A shows an example of a deleted phrase dictionary at the beginning of a term, (b) an example of a non-deleted Hiragana dictionary at the beginning, (c) an example of a term part and a code in a document, and (d) of the present invention. It is an output example automatically generated in the “description of code” format.

FIG. 16 is a reference output example of a context before and after a code description in an instruction code according to the present invention.

FIG. 17 is an example of description contents in a “description of reference numerals” area of the patent specification;

FIG. 18 is an example of an editing screen using KWIC editing means.

FIG. 19A is an explanatory diagram of a term portion and a term object corresponding to an intermediate code in a document, FIG. 19B is a document example of a non-intermediate code, and FIG. .

FIG. 20 is a diagram for explaining total list display output according to the present invention.

FIG. 21 is an example of displaying many pieces of detailed information in one line according to the present invention.

FIG. 22 is an example of a display mode of the present invention in which only a code line included in a display portion of interest is provided.

FIG. 23 is an example of an explicit exclusion term dictionary of the present invention.

FIG. 24 is an example of a noticeable display form of a useless code declaration according to the present invention.

FIG. 25 shows a part of the execution result of the present specification itself of the present invention.

FIG. 26 is an example of a standard name filter dictionary of the present invention.

FIGS. 27A and 27B are an example of an explicit term dictionary immediately before a code, an example of an explicit term dictionary immediately after a code, and an example of a document corresponding to the explicit term.

FIG. 28 shows an example of (a) a prohibited phrase dictionary according to the present invention;
(B) It is a sentence example of a use prohibition phrase.

FIG. 29 shows an example of a (a) honorific tone phrase dictionary according to the present invention;
(B) An example of a command tone phrase dictionary, and (c) an example of an ambiguous tone dictionary.

FIG. 30 is a diagram illustrating a figure number list display mode according to the present invention.

FIG. 31 shows (a) an example of a diagram number reference table and (b) an example of a diagram number for uniform conversion according to the present invention.

FIG. 32 is a part of a diagram number list display example of an execution result of the present specification itself of the present invention.

FIG. 33 is a part of an example of an automatically generated output in the “explanation of reference numerals” column by executing the present specification itself of the present invention.

FIG. 34 is an example of a warning display due to an incorrect heading order according to the present invention.

FIG. 35 is a processing configuration diagram of a code checking system that realizes an automatic index generation unit according to the present invention.

[Explanation of symbols]

 10 Code total display section 11 Code count display section 12 Timed term section and processing section 13 Reference term comparison warning processing section 14 Warning processing section 15 Suspicious code warning display processing section 16 Full-width / half-width same sign identification warning processing section 17 Display row Display control unit 20 Text file input unit 21 Reading order management unit 28 Index order management unit 24 Line number management unit 24gyou Line number information 26 Area management unit 32 Comment deletion unit 40 Extraction condition setting unit 60 Intermediate code extraction unit 60dic Continuous code recognition dictionary 60b Intermediate code simple extraction unit 60c Continuous code recognition processing unit 60d Term recognition extraction processing unit 60e Intermediate code extraction exclusion processing unit 61dic Standard name filter dictionary 64dic Basic filter dictionary 68 Intermediate code storage unit 80 Object code selection unit 81 Basic filter processing unit 82 Field common filter processing unit 83dic Common filter dictionary 84 Chromatography The dictionary unit 85dic This user dictionary 86 code explicitly term processor 86dic code explicitly term dictionary 88 term comparison determination unit 90 defines the code extraction unit 127dic declared term filter dictionary 150dic delete the phrase dictionary 271dic banned word dictionary

Claims (6)

[Claims] In the inspection of a code described in a document file and a term corresponding to the code, a document file is received, and a setting condition from an extraction condition setting unit (40) is received. A text file input unit (20) for output is provided, and receives document data from the text file input unit (20), and receives a code matching condition (6) from an extraction condition setting unit (40).
In response to 2), the code object conforming to this is changed to an intermediate code (1
02mid), a term part (101) corresponding to the intermediate code is extracted, and an intermediate code extracting unit (60) is provided for counting the number of code occurrences for each area of the intermediate code, and the intermediate code extracting unit (60) Intermediate code from (102mi
In response to d), the intermediate code is compared with the basic filter processing unit (81), the common field filter processing unit (82), the user dictionary processing unit (84), and the code explicit term processing unit (86). An object code selecting unit (80) for classifying and outputting a noise code (106nz), a suspicious code (107), and another object code (104obj) by classification by the judgment processing unit (88); Intermediate code (102mid) over the entire document obtained in
And code information, sort the code names in ascending order,
One line is displayed for each individual code, and at least the extracted term part (101) is displayed as display contents, and the intermediate code (102mi) is displayed.
d) The name is displayed, and the code generation frequency (40
0) is displayed, and if there is a noticeable display object in this display line, the display portion is displayed in a noticeable display form, and a code totaling display unit (10) for displaying a list of codes of all display targets is provided. A document code inspection system, comprising: 2. A document file input for receiving a document file and receiving setting conditions from an extraction condition setting unit (40) and outputting document data in a predetermined order in checking a figure number described in the document file. Unit (20) for receiving the document data from the text file input unit (20) and receiving the code matching condition (6) from the extraction condition setting unit (40).
In response to the 2), an intermediate code extraction unit (60) for extracting a figure number (102zu) object corresponding to the figure number and counting the number of occurrences of each figure number for each area is provided. Receiving the number of times,
Sort the figure numbers in ascending order, display each figure number on one line,
A document code inspection system, comprising: a figure number list display means for displaying the number of occurrences for each area; 3. In the inspection of the code described in the document file and the term corresponding to the code, the document file is received, and the setting data from the extraction condition setting unit (40) is received. A text file input unit (20) for output is provided, and receives document data from the text file input unit (20), and receives a code matching condition (6) from an extraction condition setting unit (40).
In response to 2), the code object conforming to this is changed to an intermediate code (1
02mid), a term part (101) corresponding to the intermediate code is extracted, and an intermediate code extraction unit (60) is provided for counting the number of occurrences of each code for each area of the intermediate code, and the intermediate code extraction unit (60) Intermediate code from (102mi
In response to d), the same term comparison is performed on the intermediate code by the basic filter processing unit (81), the field common filter processing unit (82), the user dictionary processing unit (84), and the code explicit term processing unit (86). By a classification process performed by the determination processing unit (88), a target code selection unit (80) that classifies and outputs a noise code (106nz), a suspicious code (107), and another target code (104obj) is provided. Receiving the term part (101) corresponding to the intermediate code (102mid), deletes unnecessary phrases at the head of the term part (101), and automatically generates and outputs an index as a term for the intermediate code (102mid). A document code inspection system, comprising: means for providing a document code; 4. In addition to the configuration means according to claim 1, upon receiving a detailed display instruction for the designated intermediate code (102mid), each of the designated intermediate codes (102mid) is
The area information (26area) and the line number information (24gyou) are displayed, the text is displayed from the head of the corresponding line, and the designated intermediate code (10
(2mid) and a term part (101) or term object (201yogo) in front of the document code, and a detailed code display means for displaying a context after the intermediate code by a designated number of characters is provided. 5. In addition to the constitutional means according to claim 1, a use prohibited word dictionary in which a special use prohibited word is registered according to the type and area of the sentence file is provided. When a phrase corresponding to the prohibited phrase dictionary is detected, the area information (26area) and the line number information (24gyou) are displayed, a sentence is displayed from the beginning of the document line, and the phrase corresponding to the dictionary content is displayed. A document code inspection system provided with a display means, comprising: 6. In addition to the constituent means according to claim 1, receiving the intermediate code (102mid), receiving the term subject (201yogo), and combining the context before and after in the form of a keyword in context (KWIC), A document code inspection system characterized in that KWIC editing means for displaying an object on a screen with attention and displaying the object and editing it online is provided.