JPH10289233A - Sentence correcting device, sentence correction method and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sentence correcting device that reduces labor and time which are required for correction work. SOLUTION: An object statement acquiring part 211 acquires an object statement that becomes an object to be corrected from a sentence before correction which is dealt with in an editing part 23. An analyzing part 82 analyzes an object statement and acquires morphemes and syntax information. A retrieving part 214 acquires the contents of correction work from a correction dictionary 31 based on these morphemes and syntax information, and a converting part 215 converts the object statement based on the correction work contents. The converted object statement is outputted to the position where the object statement in a sentence before correction exists. The correction of entire sentences is automatically performed by converting each statement of the sentence before correction. With this, sentence correction is performed without the intervention a proofreader who has correction ability, and labor and time which are required for correction work are reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sentence proofreading device, a sentence proofing method, and a recording medium for causing a computer to proofread a sentence in a printed matter.

[0002]

2. Description of the Related Art In a text editing operation performed prior to the production of a printed matter, a proofreading operation is performed according to a field of articles, coverage, commentary, and the like, and a reader group such as men, women, and occupations. Conventionally, this work has been performed manually by human tactics, and is performed according to text editing instructions distributed to the proofreaders or by using personal skills.

[0003] As a specific example, a description will be given of an operation of proofreading a sentence related to a certain field based on a sentence editing instruction. First, greetings and colloquial expressions are removed or changed from a sentence before proofreading. For example, "Oh my good morning"
Is deleted, and "various" is converted to "various", and "... was" is converted to "... was".

When a spoken expression is removed from a sentence, unnecessary phrases are deleted or changed. For example, delete "I mentioned before" and "So", and then "Fig.
However, "is converted to" A is as shown in FIG. 1 ".

[0005] When the unnecessary wording is eliminated, the sentence structure is converted next, for example, to avoid unnecessary compound sentences or to arrange the sentence structure. For example, a conversion such as "A is B and C is D." is converted to "A is B. C is D."

[0006] After the above-described deletion / conversion work, a question of the contents of the sentence is examined, a sentence whose necessity is unknown is determined, and a sentence quantity is examined.

[0007] Then, as a final operation, term unification and typographical errors are checked.

[0008]

As shown in the above examples, the proofreading work is performed manually, and requires a lot of time and labor costs. However, once the text field and readership are determined, it is true that there are many unified tasks. In particular, the current situation is that the amount of work is greater for unified formal calibration work. Also, the work of newly constructing and correcting the contents of the sentence editing instruction requires a great deal of labor and time.

In view of the foregoing, the present invention has been made in view of the above problems, and has an object to automate a text proofreading operation as much as possible, and to improve the efficiency of this automation through a text proofreading operation. It is an object of the present invention to provide a sentence proofreading method and a recording medium for causing a computer to proofread a sentence.

[0010]

The invention according to claim 1 is a sentence proofreading apparatus for proofreading a sentence before proofreading, and (a) a proofreading content storing means for storing a plurality of judgment conditions and work contents related to proofreading. And (b) a target sentence obtaining unit that obtains a target sentence to be corrected from the sentence before the proofreading, and (c) a target sentence analyzing unit that obtains a target sentence element by analyzing the target sentence, d) a search unit that searches for a judgment condition stored in the proofreading content storage unit that matches the target sentence element, and obtains a work content corresponding to the searched judgment condition; and (e) the search. Conversion means for converting the target sentence according to the work content acquired by the means.

According to a second aspect of the present invention, there is provided the sentence proofreading apparatus according to the first aspect, further comprising: (f) dictionary registration means for storing a combination of a judgment condition and a work content relating to proofreading in the proofreading content storing means, The dictionary registration means, (f-1) a sentence input means for inputting a sentence before proofreading and a sentence after proofreading, (f-
2) a proofreading input means for inputting proofreading contents corresponding to the proofreading sentence, (f-3) an input sentence analyzing means for analyzing the sentence before the proofreading and the sentence after the proofreading, and (f- 4) By storing, in the calibration content storage means, the analysis results of the sentence before the proofreading and the sentence after the proofreading, which are predetermined by the input proofreading content, the combination of the judgment condition and the work content regarding the proofreading is stored. Analysis result registration means for storing in the calibration content storage means.

According to a third aspect of the present invention, there is provided the sentence proofreading apparatus according to the second aspect, wherein the judgment condition stored in the proofreading content storage means is a fixed judgment condition stored in advance and the analysis result registration. Means for storing the variable work conditions stored in the calibration content storage means, the fixed work contents stored in advance and the variable work conditions stored in the analysis result registration means. With content.

According to a fourth aspect of the present invention, there is provided the sentence proofreading apparatus according to any one of the first to third aspects, wherein (g) a display corresponding to each of at least one determination condition searched by the search means. Means for displaying in list display, (h)
Means for determining one work content for the target sentence by accepting a selection input for the list display.

According to a fifth aspect of the present invention, there is provided the sentence proofreading apparatus according to any one of the first to third aspects, wherein the search means (d-1) sets a priority order among at least one determination condition to be searched. Means for deciding the work content corresponding to the highest one as the work content for the target sentence.

According to a sixth aspect of the present invention, there is provided the sentence proofreading apparatus according to any one of the first to fifth aspects, wherein (i) means for inputting a field of the contents of the sentence before proofreading; Means for selecting and determining the calibration content storage means according to the field.

The invention according to claim 7 is a sentence proofreading method for proofreading a sentence before proofreading, wherein: (a) a target sentence obtaining step of obtaining a target sentence to be proofreading from the sentence before proofreading;
(b) a target sentence element obtaining step of analyzing the target sentence by the target sentence analyzing means to obtain a target sentence element, and (c) searching from a proofreading content storage means for storing a plurality of combinations of judgment conditions and work contents concerning proofreading Means for searching for a judgment condition matching the target sentence element, and obtaining a work content corresponding to the judgment condition; and (d) converting the target sentence by the conversion means in accordance with the work content obtained in the search step. And a conversion step of converting

The invention of claim 8 is the sentence proofreading method according to claim 7, wherein (e) a repetition of repeatedly executing from the target sentence obtaining step to the conversion proofreading for each sentence of the sentence before the proofreading. It further has a step.

According to a ninth aspect of the present invention, there is provided the sentence proofreading method according to the seventh or eighth aspect, further comprising (f) a dictionary registration step which is performed in advance before the target sentence obtaining step is started,
The dictionary registration step, (f-1) a sentence input step of inputting a sentence before proofreading and a sentence after proofreading, and (f-2) a proofreading content for inputting an input proofreading content corresponding to the sentence after proofreading Input process and (f-
3) an input sentence analyzing step of analyzing the sentence before the proofreading and the sentence after the proofreading by input sentence analyzing means, and (f-4) the input of the sentence before the proofreading and the analysis result of the sentence after the proofreading. An analysis result registration step of storing a predetermined combination of calibration conditions and work content in the calibration content storage means by storing a predetermined content in the calibration content storage means.

According to a tenth aspect of the present invention, there is provided the sentence proofreading method according to the ninth aspect, wherein the criterion stored in the proofreading content storage means is a fixed criterion stored in advance and the analysis result registration. And a variable work condition stored in the calibration content storage means, wherein the work content stored in the calibration content storage means is a fixed work content stored in advance, and a variable work stored in the analysis result registration process. With content.

The invention of claim 11 is the invention of claims 7 to 10
In the sentence proofreading method according to any one of the above, the search step, (c-1) a step of displaying in a list display a display corresponding to each of at least one determination condition searched by the search means (C-2) determining one work content for the target sentence by selecting the list display.

The twelfth aspect of the present invention relates to the seventh to tenth aspects.
In the sentence proofreading method according to any one of the above, the search step, (c-3) the work content corresponding to the highest priority of at least one of the determination conditions searched by the search means, A step of determining by the retrieval means as work content for the target sentence.

The invention of claim 13 is the invention of claims 7 to 12
In the sentence proofreading method according to any one of the above, before the target sentence obtaining step, (g) a step of inputting a field of the content of the sentence before the proofreading, and (h) the proofreading according to the field Selecting and determining the content storage means.

According to a fourteenth aspect of the present invention, there is provided a recording medium recording a sentence proofreading program for proofreading a sentence before proofreading by a computer, comprising: (a) an object sentence to be proofreaded from the sentence before proofreading; And (b) a target sentence element obtaining step of analyzing the target sentence to obtain a target sentence element, and (c) calibration contents for storing a plurality of combinations of judgment conditions and work contents related to calibration. A retrieval step of retrieving a judgment condition matching the target sentence element from the storage means, and obtaining a work content corresponding to the judgment condition; and (d) converting the target sentence according to the work content obtained in the search step. And a program for causing a computer to execute the conversion step.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION

<1. Outline of Apparatus Configuration> FIG. 1 is a diagram showing an outline of the configuration of a sentence proofreading apparatus 1 according to one embodiment of the present invention.

As shown in FIG. 1, a sentence proofreading apparatus 1 includes a program section 2 having a plurality of programs for proofreading a sentence, and a dictionary section 3 storing a set of information used for proofreading a sentence. A display 4 for outputting a display to a worker who handles the apparatus, a keyboard 51 and a mouse 52 for taking input from the worker into the apparatus, and a control unit 6 for controlling the entire apparatus are connected to the bus line 7. So that it has The exchange of information between the components via the bus line 7 is performed by the operation of the control unit 6.

The program section 2 includes a management section 21 for performing processing inside the text proofreading apparatus 1, a text proofreading / dictionary registration instructing section 22 operable to input and output a worker, and a pre-proofreading section. Has an editing unit 23 for reading the text and editing the text by an operator.

In the present embodiment, each component of the program section 2 operates in a state stored in a readable / writable semiconductor memory device as in a normal computer. It may be configured by hardware.

The dictionary unit 3 stores a set of information which is referred to when the sentence proofreading device 1 proofreads a sentence. The dictionary unit 3 stores a proofreading dictionary 31 for storing contents of proofreading work, and a morpheme of a sentence to be proofread. It has a morphological dictionary 32 and a syntax dictionary 33 used for analysis and syntax analysis.

In the present embodiment, these are provided in the sentence proofreading apparatus 1 in a state stored in a fixed disk.
Any storage device that can be written and read may be stored and provided.

FIG. 1 is a diagram showing an outline of the configuration of the sentence proofreading apparatus 1, in which various interfaces and required detailed configurations are omitted, and a means for performing other processing is provided if necessary. Is also good.

Although the configuration of the sentence proofreading apparatus 1 has been described above, such a configuration can be realized by a general computer configuration. That is, as shown in FIG. 1, a disk reading unit 9 is further connected to the bus line 7, and these programs are stored in a semiconductor storage from a recording medium 91 such as a magnetic disk, an optical disk, or a magneto-optical disk for storing various programs included in the program unit 2. The computer may be used as the sentence proofreading device 1 by reading the data into the device.

<2. Outline of proofreading work> Before explaining the operation of the grammar correction apparatus 1 in the proofreading work, proofreading work using the grammar correction apparatus 1 and proofreading dictionary construction work used by the grammar correction apparatus 1 (dictionary registration work). Will be described with reference to FIG.

In the text proofreading operation using the text proofreading apparatus 1, the existence of the proofreading dictionary 31 for storing the regularity of proofreading as information is indispensable. Dictionary construction work must be performed (step Sa1).

The dictionary construction work is a work in which a proofreader having proper proofreading ability registers the proofreading rules and the reason for proofreading in the proofreading dictionary 31 when proofreading a text before proofreading to be proofreaded (step). Sa11, Sa12), the dictionary construction work is performed by a proofreading work on typical sentences before proofreading. Further, in order to perform more appropriate proofreading using the sentence proofreading apparatus 1, it is preferable that a proofreading dictionary is constructed by proofreading work on many sentences before proofreading by a proofreader.

The proofreading work (step Sa2) using the proofreading device 1 which is performed on the premise that the dictionary construction work (step Sa1) has already been completed includes an automatic proofreading work (step Sa21) and a semi-automatic proofreading work. (Step Sa22). The automatic calibration work is a work that does not depend on humans except for the instruction at the start of the work, and the semi-automatic calibration work is the work in which the operator selects from the candidates what kind of calibration is to be performed to achieve more appropriate calibration It is accompanied by. The operator in the semi-automatic proofreading work may be a proofreader having a high level of proofreading ability, but in the description of the text proofreading apparatus 1, an operator having only a proofreading ability capable of selecting a candidate. And

The automatic proofreading operation (step Sa21) or the semi-automatic proofreading operation (step Sa22) completes the proofreading of the sentence by the sentence proofreading device 1, but if necessary (for parts not corrected by the sentence proofreading device 1).
More appropriate calibration is performed by a proofreader having the proofreading ability (steps Sa3 and Sa31). In this work, the dictionary construction work before the proofreading work described above (step Sa1)
Similarly to the above, if necessary, a proofreading dictionary construction operation is performed (step Sa32), and a more appropriate proofreading dictionary 31 is constructed. As a result, an automatic or semi-automatic proofreading operation using the proofreading device 1 for other uncorrected sentences (from step Sa3 to step Sa2)
(Shown by an arrow heading toward), a more appropriate calibration is realized.

The dictionary construction work before the proofreading operation by the sentence proofreading device 1 (step Sa1), that is, the dictionary construction work by the proofreader proofreading a sentence that has not been proofread at all is renewed after the proofreading work by the sentence proofreading device 1. Of course, it may be performed (indicated by an arrow from step Sa3 to step Sa1).

<3. Structure of Information in Proofing Dictionary> Before describing the proofreading work of the sentence using the proofreading device 1 and the construction work of the proofreading dictionary, the information structure of the proofreading dictionary 31 will be described.

FIG. 3 is a diagram showing the structure of information registered (stored) in the proofreading dictionary 31. As shown in the figure, the information in the proofreading dictionary 31 is classified according to the field of the text to be proofread, such as "article" or "commentary". This classification 311 is used to limit the search range when searching the proofreading dictionary 31 and the like, and may not be present when the field of the text to be proofread is a single field.

A calibration reason 312 is stored in the lower layer of each classification 311, and a calibration content 313 is stored in each calibration reason 312.
Are stored correspondingly. The reason for proofreading is the reason (intent) for the proofreader to perform proofreading, such as "delete unnecessary words" to delete unnecessary words and "unification of notations" to unify the notation of each sentence. I have. The proofreading content is an operation content for the text proofreading device 1 to actually perform proofreading, and the detailed content will be described later.

Each proofreading reason 312 has a proofreading determination element 314 as a clue for determining whether the sentence should be proofread.
And a post-calibration element 315 serving as a material for performing a calibration corresponding to the calibration judgment element 314. In addition, a priority 316 is associated with the calibration content 313,
The priority 317 is also associated with the post-calibration element 315. These priorities indicate the priorities of information used when actually performing calibration.

FIG. 4 shows the reason for calibration 312, the contents of calibration 313,
Specific examples of the calibration judgment element 314 and the post-calibration element 315 are shown. This example is a part used to delete an unnecessary word “wa,” for a sentence in an article “was, for example, to begin explanation”.

In the proofreading operation in this example, first, elements of a sentence whose part of speech is "exclamation point" and whose character string is "wa" (hereinafter, the concept of the element including the part of speech is referred to as "morpheme"). If the part of speech of the morpheme that is included in the sentence to be proofread and that follows the morpheme of “in” is “reading point” and the character string is “,”, “determination” is performed, and then these This is a procedure for performing the “work” of deleting the morpheme array “the”.

In normal proofreading work, such a judgment and work are performed by the proofreader himself, but the sentence proofreading apparatus 1 efficiently performs the judgment and work so that the proofreading dictionary 31 can be mechanically performed. The contents are registered in.

In FIG. 4, the proofreading reason 31 of unnecessary word deletion is shown.
In the example of the second example, the determination conditions in the proofreading work of the above example sentence and the fixed part of the work content are registered in the proofreading contents 313. In addition, a variable part of the determination condition is registered in the calibration determination element 314, and a variable part of the work content (in this example, the deletion operation is fixed in the post-calibration element 315). So nothing is stored).

That is, in the proofreading judgment element 314, the morpheme whose part of speech is “exclamation” and the character string is “wa” is the first morpheme.
The morpheme whose part of speech is "reading point" and the character string is "," is registered as the second proofreading judgment element, and is registered as the second proofreading judgment element. Is a variable part of the conditions for judging. Also, in the proofreading content 313, the fact that the morphemes whose parts of speech and character strings match each proofreading judgment element are arranged in order is a condition for fixedly determining whether or not the proofreading reason 312 of unnecessary word deletion is applicable. ing. The sentence proofreading apparatus 1 performs proofreading when the fixed judgment condition and the variable judgment condition match the sentence to be corrected.

Here, the fixed judgment condition is treated as a judgment condition specific to the proofreading reason, but is registered in the sentence proofreading device 1 in advance. The variable judgment conditions serve as parameters for performing the same calibration reasons but different calibration operations. Therefore, the calibration judgment element 31 different from the example shown in FIG.
When 4 is associated with the proofreading reason 312 of unnecessary word deletion, deletion of unnecessary words other than “,” is realized. For example, if it is desired to delete the greeting "I'm sorry.", The morpheme of "I'm sorry." Is registered in the proofreading judgment element in order.

FIG. 5 is an example of the proofreading reason 312 of unification of notation, and the sentence "This is a desk." Is changed to "This is a desk." FIG. 9 is a diagram showing contents registered in a proofreading dictionary 31 when proofreading is performed to a sentence of FIG.

In this case, a morpheme whose part of speech is "judgment" and whose character string is "is" is registered as a first calibration judgment element of the calibration judgment element 314 which is a variable judgment condition, A morpheme whose part of speech is “phrase point” and whose character string is “.” Is registered as a proofreading judgment element. In addition, it is registered that the first calibration determination element and the second calibration determination element exist in order in the calibration content 313 as a fixed determination condition.

The calibration operation includes a calibration judgment element 314.
Is registered as a content of the variable work corresponding to. In the calibration content 313, replacement of the first calibration determination element with the first post-calibration element is registered as a fixed work content.

As a result, the sentence "This is a desk." Has a morpheme of "is" and a morpheme of "." In order, and is determined to be a sentence to be corrected.
By replacing "is" with "is", proofreading to the sentence "This is a desk."

FIG. 6 is a diagram showing an example of the proofreading reason 312 that a sentence is divided. For example, a sentence "Penguin is a friend of a bird but wings cannot be degenerated and fly" is described as "Penguin is a bird. The feathers degenerate and cannot fly. "
And the registered contents of the proofreading dictionary 31 for dividing the sentence into two sentences.

In this example, the morpheme whose part of speech is “predicate connection particle” and whose character string is “ga” is registered as the first calibration judgment element in the calibration judgment element 314 which is a variable judgment condition. , A morpheme whose part of speech is “reading point” and whose character string is “,” is registered as a second proofreading judging element. In addition, as the fixed determination conditions, the first calibration determination element and the second calibration determination element are present in the calibration content 313 in order, and “subject” and “predicate” are respectively before and after the second calibration determination element. Is registered to exist.

In the post-calibration element 315, which is a variable work content, the character (string) “.” Is registered as the first post-calibration element. In the calibration content 313, which is a fixed work content, a work of deleting the first calibration judgment element and replacing the second calibration judgment element with the first post-calibration element is registered.

Thus, the morpheme “ga” and the morpheme “,”
Having a subject and a predicate before and after the morpheme ",", it is determined that the sentence "Penguin is a friend of birds, but feathers cannot degenerate and fly" is a sentence to be corrected.
"Ga" is deleted and "," is replaced with ".", So that "Penguin is a friend of birds. Feathers degenerate and cannot fly."

The structure of the information registered in the proofreading dictionary 31 has been described in accordance with the specific contents of the proofreading. The content of the work is registered as a pair, and the fixed part of the determination condition and the fixed part of the work content are registered in the calibration content 313. Then, the variable part of the judgment condition is registered in the calibration judgment element 314 in such a manner as to be associated with the calibration content 313, and the variable part of the work content is registered in the post-calibration element 315. ing. Therefore, the fixed part is registered without duplication in the proofreading dictionary 31, and the amount of registered contents is efficiently suppressed.

In the above description, it has been explained that the registered contents of the proofreading contents 313 are determined in advance in the sentence proofreading apparatus 1. However, any part of the judgment conditions and the contents of the work is assumed to be fixed and set in advance. Whether to register in the sentence proofreading device 1 is not limited to the above example, and may be arbitrarily determined according to the type of sentence to be actually proofread.

For example, in the unified example of the notation shown in FIG. 5, the work contents registered in the calibration contents 313 are replaced with the first calibration judgment element by the first calibration judgment element and the second calibration judgment element is replaced by the first calibration judgment element. (2) After the calibration, the element is replaced.
5, the same calibration operation can be realized by registering the first post-calibration element “is” and the second post-calibration element “.”.

Similarly, the contents registered in the proofreading judgment element 314 and the post-proofreading element 315 are not limited to morphemes. For example, syntax information such as the position of the subject / predicate may be registered in the proofreading judgment element 314. Alternatively, syntax information such as conversion of the position of the subject / predicate may be registered in the proofreading element 315.
This allows proofreading even for complex syntax.
That is, in the example shown in FIG. 6, syntax information that is a condition for a fixed determination that a subject and a predicate exist before and after the morpheme “,” is registered in the proofreading determination element 314 as a variable determination condition. Thereby, more various calibrations are realized.

In the above example, one calibration content 313 is registered for one calibration reason 312. However, as shown in FIG. 7, a plurality of calibration contents 313 are associated with one calibration reason 312. Each calibration content 313 further includes a plurality of calibration judgment elements 314 and post-calibration elements 315.
May be associated with each other. This is because even if the proofreader is considered to have the same proofreading reason, the number of conditions for determining whether or not to actually perform proofreading may be different.

<4. Description of sentence proofreading operation using sentence proofreading device><4.1 Configuration of Program Unit and Dictionary Unit> FIG. 8 is a diagram showing the configuration of the program unit 2 and the dictionary unit 3 of the sentence proofreading device 1 shown in FIG. Yes, solid arrows indicate the exchange of information in the automatic calibration work. Arrows indicated by broken lines indicate exchange of information in the dictionary construction work.

As shown in FIG. 8, a sentence proofreading / dictionary registration instructing unit 22 which handles input from an operator and display to the operator.
Is a field input section 220 for inputting a field of a sentence to be proofread, a candidate display selection section having a process of displaying candidates of a sentence corrected by a semi-automatic proofreading operation and accepting a selection of a candidate from an operator. 221, a sentence before and after the proofreading input unit 222 for receiving a sentence before proofreading and a sentence after proofreading from a worker in the dictionary construction work, and a proofreading reason 3
It has a calibration reason input unit 223 that accepts 12.

The management unit 21 retrieves one sentence (target sentence) to be handled from the text to be corrected which is being edited by the editing unit 23, a target sentence acquisition unit 211, and an analysis unit 8 that analyzes the target sentence.
2. From the proofreading dictionary 31 using the analysis results, proofreading contents 313
It has a search unit 214 for acquiring a sentence, a conversion unit 215 for generating a proofread sentence, and an output unit 216 for outputting the proofread sentence to the editing unit 23. A registration unit 217 for registering a dictionary during the dictionary construction work, and a proofreading dictionary 31
Also has a dictionary selection unit 210 for determining the use range of.

The analyzing unit 82 has a morphological analyzing unit 212 for obtaining a morpheme from a target sentence and a syntactic analyzing unit 213 for analyzing whether or not a subject / predicate exists, and the morphological dictionary 32 used for these analysis. And syntax dictionary 3
3 also belongs to the analysis unit 82.

In FIG. 8, a proofreading pre / post sentence input unit 222, a proofreading reason input unit 223, and a registration unit 217 indicated by reference numeral 81 are shown.
Is used only for dictionary construction work, and the analysis unit 82 operates in both proofreading work and dictionary construction work. Note that the components indicated by reference numeral 81 and the analysis unit 82 are parts that operate in the dictionary construction work, and constitute the dictionary construction unit 8.

<4.2 Automatic Calibration Operation> Solid arrows in FIG. 8 indicate the exchange of information between components in the automatic calibration operation. Hereinafter, the operation of the text proofreading apparatus 1 in the automatic proofreading operation will be described according to the arrows. In the automatic proofreading operation, the operator does not make any input other than the initial setting to the grammar correction device 1, and thus the grammar correction / dictionary registration instructing unit 22 basically does not perform any operation after the initial setting.

FIG. 9 is a diagram showing a flow of the operation of the sentence proofreading apparatus 1 in the automatic proofreading operation. In the automatic proofreading operation, first, an initial setting (step S11) for setting a field of a sentence to be proofread is performed. Thus, the field input unit 220 receives the input field and receives the input of the field.
Determines the area to be used among the registered contents in the proofreading dictionary 31. When the initial setting is completed, the target sentence acquiring unit 211 reads the sentence to be corrected read from the editing unit 23.
Acquires one sentence (target sentence) (step S12).

The obtained target sentence is sent to the analysis unit 82, where morphological analysis and syntax analysis are performed. As a result, the morpheme and syntax information (the subject position and the number of predicates) of the target sentence are acquired as target sentence elements (step S1).
3).

The target sentence element is sent to the search unit 214 and compared with a condition for determining whether the sentence is a sentence to be corrected (step S14). That is, the target sentence element is compared with the fixed judgment condition registered in the proofreading content 313 and the variable judgment condition registered in the proofreading judgment element 314, and if the target sentence element matches these judgment conditions, It is determined that the target sentence is a sentence to be proofread.

Here, if it is determined that the sentence is not to be proofread, the operation shifts to an operation of acquiring the next target sentence (steps S18 and S12). Content 313 and post-correction element 315, which are information on how to perform the correction, are acquired as the target sentence correction content and the target sentence post-correction element (step S1).
5).

The operation up to this point will be described with reference to the example shown in FIG. 6. The target sentence “Penguin is a friend of birds, but the wings degenerate and cannot fly” is obtained by the target sentence obtaining unit 211. Then, the morphological analysis and the syntax analysis are performed in the analyzing unit 82 (steps S12 and S13). As a result of the morphological analysis, it is clear that this sentence has a morpheme whose part of speech is "predicate connecting particle" and its character string is "ga" and a morpheme whose part of speech is "reading point" and whose character string is "," ,
It is determined that there is a morpheme whose part-of-speech and character string match those of the first and second calibration determination elements, which are the variable determination elements shown in FIG. Further, according to the result of the parsing, the subject "Penguin" and the predicate "is" exist before the second proofreading judgment element ",", and the subject "Hana" and the predicate "Fly" do exist behind. As a result, the fixed judgment condition in the proofreading contents 313 is “the first proofreading element and the second proofreading element are arranged side by side in the target sentence, and the subject is located before and after the second proofreading judgment element, respectively. And a predicate exists "(step S14).

As a result, the calibrated element 315 shown in FIG.
Is acquired as the target sentence proofreading element, and the contents of deleting the first proofreading judgment element and replacing the second proofreading judgment element with the first proofreading element in the proofreading contents 313 are obtained as the target sentence proofreading contents. (Step S15).

When the target sentence proofreading content and the target sentence proofreading element are obtained, the target sentence is converted according to these contents (step S16). That is, in the example shown in FIG. 6, “GA” is converted to “.”. As a result, the sentence is divided into two sentences, "Penguin is a friend of birds. Feathers can degenerate and cannot fly."

When the proofread sentence is acquired, the output unit 216 outputs the proofread sentence (including two or more sentences) at the position where the target sentence exists in the unread proofread text read into the editing unit 23. ) Is written (step S17). By performing this operation on all the sentences in the sentence before proofreading, the proofreading of the entire sentence before proofreading is automatically performed (step S18).

The above description is for the case where only one combination of the proofreading content 313 and the proofreading judgment element 314, which is a condition for judging whether or not the sentence should be proofread, matches the target sentence element. When a plurality of calibration determination conditions are satisfied, the priority 316 or 317 shown in FIG. 3 is used to determine a combination of one calibration content 313 and a calibration determination element 314.

That is, the higher the priority 316 associated with the proofreading content 313 is, the higher the priority is applied, and if a plurality of proofreading judging elements 314 associated with the same proofreading content 313 match the target sentence element, the proofreading is performed. Element 31
The higher the priority 317 associated with No. 5 is, the higher the priority is applied. As a result, the search unit 214 can determine only one combination of the proofreading content 313 and the proofreading determination element 314 that match the target sentence element.

The structure of the information in the proofreading dictionary 31 is shown in FIG.
Also in the case of the structure shown in FIG. 7, the priority 316a associated with the calibration reason 312 is used with the highest priority, the priority 316b associated with the calibration content 313, and the priority 317 associated with the post-calibration element 315 in this order. Calibration contents 3 used by the conversion unit 215 by using the priority order
13 (target sentence proofreading content) and one combination of the post-proofreading element 315 (target sentence post-proofreading element).

As described above, the sentence proofreading device 1 automatically performs the proofreading of a predetermined sentence without the intervention of an operator. Therefore, a formal proofreading operation of a sentence that has conventionally required an enormous labor is required. Can be greatly reduced. Further, since the necessity of proofreading is determined using the target sentence element generated by the morphological analysis / syntax analysis, errors such as oversight of proofreading work can be reduced.

<4.3 Semi-Automatic Calibration Work> FIG. 10 is a diagram showing how information is exchanged between the program unit 2 and the dictionary unit 3 in the semi-automatic calibration work by solid arrows.
11 to 13 are diagrams showing the operation of the sentence proofreading apparatus 1 and the flow of the work of the worker.

In the semi-automatic proofreading work, first, the initial setting is performed in the same manner as the automatic proofreading work.
One sentence (target sentence) is obtained by the target sentence obtaining unit 211 (FIG. 11: steps S21 and S22). When the target sentence is obtained, the analysis unit 82 analyzes the target sentence and obtains a target sentence element (step S23).

Thereafter, the search unit 214 checks the proofreading dictionary 3
Among the proofreading contents 313 and proofreading determination elements 314, which are the conditions for determining whether or not to perform proofreading stored in No. 1, those which match the target sentence element are searched for and acquired. At this time, when there are a plurality of combinations of the proofreading contents 313 and the proofreading judgment elements 314 to which the target sentence element matches (step S2).
4, S31), the calibration content 313 and the post-calibration element 3 by a predetermined number (candidate number) preset according to the priority order.
15 is obtained (FIG. 12: step S311). That is, the target sentence proofreading contents and the target sentence proofreading elements are obtained by the number of candidates.

When the target sentence proofreading content and the target sentence proofreading element are obtained, the target sentence is converted for each of these combinations, and candidate sentences which are proofread by the number of candidates are converted by the conversion unit 215. It is created (step S312). These candidate sentences are sent to the candidate display selection unit 221 and displayed in a list on the display 4 (step S313).

Here, the operator selects and inputs the most appropriate sentence from the list of candidate sentences (step S3).
14). The input selected sentence is sent to the output unit 216, and the target sentence in the editing unit 23 is replaced with the selected candidate sentence (step S315).

When the proofreading of one target sentence is completed, the target sentence obtaining unit 211 obtains the next target sentence from the uncorrected sentence of the editing unit 23, and repeats this in order to obtain all the sentences of the uncorrected sentence. Calibration is performed (Step S
32, S26).

When acquiring the target sentence proofreading contents and the target sentence proofreading elements, the proofreading contents 3 matching the target sentence elements are obtained.
If there is only one combination of the number 13 and the proofreading determination element 314 (steps S25 and S33), there is no choice for the operator, so the corresponding proofreading content 313 is acquired as the target sentence proofreading content as in the automatic proofreading work. Then, the proofreading element 315 corresponding to the corresponding proofreading determination element 314 is acquired as the target sentence proofreading element (FIG. 13: step S331).
Thereafter, the target sentence is converted and output to the editing unit 23 (steps S332 and S333).

Also, the proofreading content 31 that conforms to the target sentence element
If the combination of No. 3 and the proofreading determination element 314 is not found, the next target sentence is obtained from the editing unit 23 (Steps S25, S26, S22).

As described above, in the semi-automatic proofreading operation, more appropriate proofreading can be realized only by the operator selecting a candidate sentence.

The flow of information indicated by the dashed arrow in FIG. 10 shows another operation mode in the semi-automatic calibration work. Note that the difference between the operation of this operation mode and the operation modes shown in FIGS. 11 to 13 is that the operation content shown in FIG. 12 becomes the operation content shown in FIG.

That is, when a plurality of combinations of the target sentence proofreading contents and the target sentence proofreading elements are acquired by the search unit 214 (steps S35 and S351), these are converted to the conversion unit 21.
5 is sent to the candidate display selection section 221 instead of being sent. Then, the candidate display selection unit 221 does not display the candidate sentences, but lists the target sentence proofreading contents and the target sentence proofreading elements as candidate elements (Step S).
352). Therefore, the operator determines and selects the most appropriate proofreading content from the list of candidate elements instead of the candidate sentences (step S353). The candidate element selected in this way, that is, the target sentence proofreading content and the target sentence proofreading element are set as the work contents for converting the target sentence (step S354). Thereafter, the target sentence is converted and output to the editing unit (steps S355, S356, S3
6).

In this operation mode, since the most appropriate conversion content is selected before converting the target sentence, the conversion of the target sentence is performed only once, unlike the operation content of FIG. Efficiency can be improved.

<5. Explanation of Construction Work of Proofing Dictionary> Next, the construction work (registration work) of the proofreading dictionary 31 which is a premise of the proofreading work or performed to realize more efficient proofreading work will be described.

As already described with reference to FIG. 8, in the proofreading dictionary construction work, the pre- and post-proofreading sentence input unit 222, the proofreading reason input unit 223, the registration unit 217, and the analysis unit 82 are used. The construction unit 8 is configured.
In FIG. 8, the exchange of information between these components is indicated by broken-line arrows, and the flow of the proofreading dictionary construction work is shown in FIG.

In the proofreading dictionary construction work, first, an initial setting for setting a field of registered contents and the like is performed (step S4).
1). Thereby, the registration area in the proofreading dictionary 31 is determined. Thereafter, a proofreader having proofreading ability inputs a sentence before proofreading and a sentence after proofreading (including two or more continuous sentences) (step S42). When the proofreader proofreads the text being edited in the editing unit 23, the proofreader writes the text before proofreading with the mouse 52.
The user may select and input using the keyboard 51 or the like, and input a sentence corrected using the keyboard 51 or the like as a sentence after the correction. May be input. The input work of the sentence before and after the proofreading is performed via the sentence input unit 222 before and after the proofreading of the sentence proofreading / dictionary registration instruction unit 22.

The proofreader also selects and inputs the reason for proofreading the sentence before and after proofreading (step S43), whereby the proofreading reason 312 (see FIG. 3) that matches the input reason for proofreading is selected. Note that in FIG.
And the calibration content 313 are in one-to-one correspondence, so the input of the calibration reason 312 is equivalent to the input of the corresponding calibration content 313. The calibration content 313 corresponding to the proofreading reason 312 is previously determined in the text proofreading apparatus 1 as described above, and is selected by the proofreading reason 312 and is associated with the proofreading determination element 314 and the post-correction element. The contents to be registered in 315 (such as the number of parts of speech and character strings to be registered, etc.) are determined.

When the sentence before and after the proofreading and the reason for the proofreading are inputted, first, the sentence before the proofreading and the sentence after the proofreading are analyzed by the analyzing unit 82 to obtain morpheme and syntax information (step S44). Then, of the analysis results of the sentence before the proofreading, those to be registered as the proofreading judgment element 314 and those of the analysis result of the sentence after the proofreading which are to be registered as the post-correction element 315 are acquired by the registration unit 217 (step S45), and input. It is registered in the proofreading dictionary 31 in association with the proofreading content 313 corresponding to the proofreading reason (step S46). In addition, if it has already been registered, it will not be registered. At this time, if necessary, the priority order may be input by the proofreader, or the priority order may be set in the order of registration.

In the above dictionary construction work, the proofreader makes a one-to-one correspondence with the proofreading reason 312 instead of the proofreading content 313.
Is input in consideration of the easiness of the work of the proofreader, and the proofreading contents 313 may be directly selected. The proofreading dictionary 3 shown in FIG.
In the case of the structure of the information in 1, the proofreading content 313 is not fixed to one by inputting the proofreading reason 312, so the proofreading content 313 is input instead of the proofreading reason 312 in the dictionary construction work.

Although the dictionary construction work has been described as being performed through the proofreading work, the proofreader may make additional corrections independently of the proofreading work.

As described above, in the sentence proofreading apparatus 1, the provision of the dictionary constructing section 8 allows the proofreader to construct the proofreading dictionary 31 through the normal proofreading work performed by the proofreader. Is unnecessary. Further, since the proofreading dictionary 31 is constructed differently for each proofreader, the proofreading dictionary 31 can be constructed according to the intention of the proofreader.

<6. Modifications> The sentence proofreading apparatus 1 according to the embodiment of the present invention has been described above.
The present invention is not limited to the above embodiment.

For example, in the above-described embodiment, the editing unit 23 proofreads the text being edited.
A sentence proofreading device without an editing function may be used. Also, the proofreading may not be performed for the entire text being edited, but may be performed only for the text in a specified range.

In the above-described embodiment, the analysis is performed for each target sentence. However, the morphological analysis / syntax analysis of all sentences may be performed first.

The contents of the information in the proofreading dictionary 31 shown in FIG. 3 and FIG. 7 are merely examples, and other proofreading reasons 312 include “unification of kana”, “unification of kanagana”, and “unification of kanji”. "Can be adopted. For example, this case corresponds to a case where it is desired to perform calibration such as “to” for “tori”, “to represent” to “represent”, and “and” to “and”.

Further, as a complicated example, a "double negation sentence"
“Complex wording” and “duplicate expression” can also be adopted as the proofreading reason 312. For example, it is a calibration such as "It cannot be said that this device will never fail" or "This device may fail." In this case, by registering morphemes such as “absolute”, “failed”, “no”, “no”, “no”, etc. as calibration judgment factors, the same calibration can be performed even if the device is a TV or car. Become.

In the above embodiment, the inside of one proofreading dictionary 31 is divided into fields, but a proofreading dictionary may be constructed for each field, and the dictionary selecting unit 210 may switch a plurality of proofreading dictionaries.

In the above embodiment, Japanese sentences have been described. However, even in English or the like, similar processing can be performed by registering the conditions for judging proofreading and the contents of proofreading work in the proofreading dictionary 31. It is possible. For example, in English, it is possible to convert "wanna" to "want to" and "gonna" to "going to".

Further, various modifications can be made to the structure of the sentence proofreading apparatus 1 and the structure of information in the proofreading dictionary 31 according to the contents and use of proofreading.

[0107]

As described above, claims 1 to 1
In the invention described in Item 4, the target sentence analyzing means (or computer) analyzes the target sentence to obtain the target sentence element, and the search means (or computer) outputs the target sentence among the judgment conditions stored in the proofreading content storage means. The work content corresponding to the element is acquired, and the conversion means (or computer) converts the target sentence based on the acquired work content, so the labor and time required for the formal text proofreading work Can be greatly reduced. Further, it is also possible to accurately perform such a calibration work.

According to the second and ninth aspects of the present invention,
The judgment condition and the work content can be registered in the calibration content storage means through the normal calibration work by the proofreader, so that the appropriate judgment conditions and the work content can be registered in the calibration content storage means without any extra effort. it can.
In addition, since the stored contents of the proofreading content storage means can be different for each proofreader, proofreading according to the proofreader's intention is realized.

According to the third and tenth aspects of the present invention, judgment conditions and work contents are divided into fixed and variable contents and stored in the calibration contents storage means. Can be memorized.

According to the inventions described in claims 4 and 11, since an appropriate one is selected from a list of displays according to the judgment condition, more appropriate calibration is realized.

According to the fifth and twelfth aspects of the present invention, since one work content is determined in accordance with the priority order, the labor and time required for the calibration work can be further reduced.

According to the inventions described in claims 6 and 13, since the proofreading content storage means is selected and determined according to the field of the sentence before proofreading, an efficient proofreading operation is realized.

[Brief description of the drawings]

FIG. 1 is a diagram showing an outline of a configuration of a sentence proofreading device 1. FIG.

FIG. 2 is a diagram showing an order relationship between a dictionary construction work and a sentence proofreading work.

FIG. 3 is a diagram showing a structure of information in a proofreading dictionary.

FIG. 4 is a diagram showing an example of the contents of a proofreading dictionary used for deleting unnecessary words.

FIG. 5 is a diagram showing an example of the contents of a proofreading dictionary used for unifying notations.

FIG. 6 is a diagram showing an example of the contents of a proofreading dictionary used for dividing a sentence.

FIG. 7 is a diagram showing a structure of information in a proofreading dictionary.

FIG. 8 is a diagram showing a flow of information in an automatic proofreading operation and a dictionary construction operation.

FIG. 9 is a diagram showing a work flow in an automatic calibration work.

FIG. 10 is a diagram showing a state of information flow in a semi-automatic calibration work.

FIG. 11 is a diagram showing a work flow in a semi-automatic calibration work.

FIG. 12 is a diagram showing a work flow in a semi-automatic calibration work.

FIG. 13 is a diagram showing a work flow in a semi-automatic calibration work.

FIG. 14 is a diagram showing a work flow in a semi-automatic calibration work.

FIG. 15 is a diagram showing a work flow in a dictionary construction work.

[Explanation of symbols]

 1 sentence proofreading device (computer) 4 display 8 dictionary construction unit 9 disk reading unit 31 proofreading dictionary 51 keyboard 52 mouse 82 analysis unit 91 recording medium 210 dictionary selection unit 211 target sentence acquisition unit 214 search unit 215 conversion unit 217 registration unit 220 Input section 221 Candidate display selection section 222 Pre- and post-correction sentence input section 223 Calibration reason input section 311 Classification 313 Calibration contents 314 Calibration judgment element 315 Calibrated element 316, 316a, 316b, 317 Priority

Claims (14)

[Claims] 1. A sentence proofreading device for proofreading a sentence before proofreading, comprising: (a) a proofreading content storing means for storing a plurality of combinations of judgment conditions and work contents concerning proofreading; A target sentence obtaining means for obtaining a target sentence to be proofread; (c) a target sentence analyzing means for analyzing the target sentence to obtain a target sentence element; and (d) stored in the proofreading content storage means. (E) searching means for retrieving a work condition corresponding to the target sentence element among the judgment conditions, and obtaining a work content corresponding to the searched judgment condition; and (e) the target sentence according to the work content obtained by the search means. And a conversion means for converting the text. 2. The sentence proofreading device according to claim 1, further comprising: (f) dictionary registration means for storing a combination of a judgment condition and a work content relating to proofreading in the proofreading content storage means, wherein the dictionary registration means However, (f-1) sentence input means for inputting the sentence before proofreading and the sentence after proofreading, and (f-2) proofreading content input means for inputting input proofreading content corresponding to the sentence after proofreading, (f-3) input sentence analyzing means for analyzing the sentence before the proofreading and the sentence after the proofreading, and (f-4) the input proofreading content among the analysis results of the sentence before the proofreading and the sentence after the proofreading. And an analysis result registering means for storing a combination of a judgment condition and a work content relating to calibration in the calibration content storage means by storing a predetermined one in the calibration content storage means. Calibration device. 3. The sentence proofreading device according to claim 2, wherein the judgment condition stored in the proofreading content storage means is stored by a fixed judgment condition stored in advance and the analysis result registration means. With variable judgment conditions,
A sentence proofreading apparatus, wherein the work contents stored in the proofreading contents storage means include fixed work contents stored in advance and variable work contents stored by the analysis result registration means. . 4. The sentence proofreading device according to claim 1, wherein: (g) displaying a list according to each of at least one determination condition searched by the search unit in a list display. And (h) means for determining one work content for the target sentence by accepting a selection input to the list display. 5. The sentence proofreading device according to claim 1, wherein the search means (d-1) sets the highest priority among at least one determination condition to be searched. Means for deciding a corresponding task content as a task content for the target sentence. 6. The sentence proofreading device according to claim 1, wherein: (i) means for inputting a field of the content of the text before proofreading; Means for selecting and determining a proofreading content storage means. 7. A sentence proofing method for proofreading a sentence before proofreading, comprising: (a) a target sentence obtaining step of obtaining a target sentence to be proofread from the sentence before proofreading; and (b) a target sentence analysis. Means for obtaining the target sentence element by analyzing the target sentence by means; and (c) the target sentence element by a search means from a proofreading content storage means for storing a plurality of combinations of judgment conditions and work contents concerning proofreading. A search step of searching for a judgment condition that satisfies the above, and obtaining a work content corresponding to the judgment condition; and (d) a conversion step of converting the target sentence by a conversion unit in accordance with the work content obtained in the search step. A sentence proofreading method comprising: 8. The sentence proofreading method according to claim 7, wherein (e) a repetition step of repeatedly executing from the target sentence acquisition step to the conversion proofreading for each sentence of the sentence before the proofreading,
A sentence proofreading method, further comprising: 9. The sentence proofreading method according to claim 7 or 8, further comprising: (f) a dictionary registration step performed before the target sentence acquisition step is started. (F-1) a sentence input step of inputting a sentence before proofreading and a sentence after proofreading; (f-2) a proofreading content inputting step of inputting input proofreading content corresponding to the sentence after proofreading; f-3) an input sentence analyzing step of analyzing the sentence before the proofreading and the sentence after the proofreading by input sentence analyzing means; and (f-4) an analysis result of the sentence before the proofreading and the sentence after the proofreading. An analysis result registration step of storing, in the calibration content storage means, a combination of the determination condition and the work content relating to the calibration by storing the predetermined content based on the input calibration content in the calibration content storage means. Characteristic proofreading method. 10. The sentence proofreading method according to claim 9, wherein the judgment condition stored in the proofreading content storage means is stored in a fixed judgment condition stored in advance and in the analysis result registration step. The work content stored in the calibration content storage means has a fixed work content stored in advance and a variable work content stored in the analysis result registration step. A sentence proofreading method characterized in that: 11. The sentence proofreading method according to claim 7, wherein the search step (c-1) is performed according to each of at least one determination condition searched by the search means. A step of displaying the display in a list display;
A sentence proofreading method comprising: 12. The sentence proofreading method according to claim 7, wherein the search step includes: (c-3) a priority order of at least one determination condition searched by the search means. A step of the search means deciding a task content corresponding to the highest task as a task content for the target sentence. 13. The sentence proofreading method according to claim 7, wherein before the target sentence obtaining step, (g) inputting a field of the content of the sentence before proofreading; (h) a step of selecting and determining the correction content storage means according to the field. 14. A recording medium storing a sentence proofreading program for proofreading a sentence before proofreading by a computer, comprising: (a) a target sentence for obtaining a target sentence to be proofread from the sentence before proofreading; An obtaining step, (b) a target sentence element obtaining step of obtaining the target sentence element by analyzing the target sentence, and (c) the correction target storage means for storing a plurality of judgment conditions and work contents related to the correction from the correction content storage means. A search step of searching for a judgment condition matching the sentence element, and obtaining a work content corresponding to the judgment condition; (d) a conversion step of converting the target sentence according to the work content obtained in the search step; A computer-readable recording medium characterized by recording a program for causing a computer to execute the program.