JPH0916597A - Sentence elaboration device and method therefor - Google Patents

Abstract

PURPOSE: To attain the 'sentence elaboration' function which is suitable for a user by learning information as to what proofreading is performed when the proofreading of a sentence is performed and registering the information as the error detection rule for sentence elaboration. CONSTITUTION: When a 'proofreading' is requested by an input par 1 in accordance with a user operation and the proofreading of the sentence is performed by an editing part 3 under the control of a control part 2, the information before/after the proofreading is delivered to a proofreading information learning part 10 via a sentence elaboration control part 5. A proofreading information learning part 10 generates a corresponded error detection rule and registers the rule in a user dictionary 12 if the information before/after the proofreading is suitable as the error detection rule to which the relation before/after the proofreading (that is, learning contents of the information as to what proofreading is performed) is reflected and the information is unregistered. An error detection part 15 refers to a user dictionary 12 and an error detection rule dictionary 16 and performs the error detection, when the sentence is elaborated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for selecting a sentence using an error rule, and in particular, learning information about how the sentence is proofread when the sentence is proofread. The present invention relates to a sentence revision device and method that uses the information as an error detection rule.

[0002]

2. Description of the Related Art Generally, a document creating apparatus such as a Japanese word processor provides various document creating / editing functions so that a user can efficiently create a desired document.
However, even if a document (sentence) is created using this type of document creating apparatus, it is difficult to prevent the document from containing an error.

For this reason, the user needs to revise the created text while looking at it on the screen or the like to search for an error portion, and then correct the character string by using the editing (calibration) function of the device. there were. However, it is difficult for the user to search for an error in a sentence created by assuming that it is correct, and there is a possibility of overlooking it.

Therefore, in recent years, various error detection rules have been prepared in advance on the system side, and the function of automatically refining a sentence using the error detection rule or the error detection rule program (errors in a sentence or inappropriateness) A document creation device that provides a function of detecting various expressions and unifying the style and notation has appeared.

[0005]

However, even with a device having a text revision function, the number of errors that can be detected by the error detection rule prepared in advance is limited, and detection is performed using this error detection rule. There was a problem that errors in sentences that could not be made could not always be detected.

The present invention has been made in consideration of the above circumstances, and an object thereof is to learn information about how a proof was corrected when a sentence is proofread, and to use the information as an error for sentence revision. By enabling registration as a detection rule,
It is an object of the present invention to provide an apparatus and method for writing text that can realize the function of “text editing” suitable for the user.

[0007]

The present invention is directed to a character string designating means for designating an arbitrary character string in a sentence, and a character string proofreading for calibrating the character string designated by the character string designating means. Means and a calibration information learning means for learning information on how the text is calibrated by the character string proofing means, and the calibration information learned by the calibration information learning means are registered as an error detection rule for sentence revision. A first storage means for storing the text, a second storage means in which an error detection rule for text revision is registered in advance, and a text is registered in the first storage means and the second storage means at the time of text revision. And an error detecting means for detecting an error in a character string in a sentence by using the error detection rule described above.

According to the present invention, the learned calibration information is added with date and time information at the time of calibration (at the time of registration) and registered in the first storage means (user dictionary) as an error detection rule, and a certain character string is stored. First rule candidate to be applied to error detection
When there are a plurality of candidates in the storage means, the latest candidate is selected and used based on the date and time information attached to the candidate.

Further, according to the present invention, when the learned proofreading information is to be registered as an error detection rule, the proofreading sentence is analyzed word by word by an analyzing means, and based on the analysis result. It is also characterized in that the calibration information selection / registration means determines whether or not the calibration information is suitable as an error detection rule, thereby preventing registration of inappropriate information as an error detection rule. And

Further, according to the present invention, when the learned proofreading information is to be registered as an error detection rule, the proofread sentence is analyzed word by word by an analyzing means, and the analysis result is used as a basis. It is also characterized in that it is determined whether or not the calibration information needs to be corrected, and if necessary, the calibration information correction means corrects the error for the error detection rule based on the analysis result.

Further, according to the present invention, a list of registration information in the first storage means is displayed on the screen by the user presenting means, arbitrary registration information is designated from the registration information list, and the information is applied to error detection. If it is instructed to display the application form at the time, the rule application result display means shows how the text is calibrated if the specified information is applied to error detection. When the deletion or correction of the registration information that is the target of this rule application result is instructed by displaying on the screen using, the corresponding editing information is deleted or modified by the information editing means. Also features. Here, when the method of adding date information to the registration information is applied, the registration information of the first storage means may be sorted based on the date and time information, and a list may be displayed in the sorted order.

[0012]

In the above construction, when the text is proofread, the information about how the text was proofread is learned by the proofreading information learning means, and the proofreading information, that is, the information before and after proofreading is also stored. Since the error detection rule for sentence revision is generated in and, the function of "sentence revision" suitable for the user can be realized.

Further, in the above configuration, "sentence revision"
Therefore, when a plurality of error detection rules in the first storage means (user dictionary) are met, only the most recently registered rule can be validated and used according to the date and time information attached to the rule. Therefore, the “sentence revision” that the user currently intends can be realized.

Further, in the above configuration, the registration information (user registration information) in the first storage means (user dictionary) is presented to the user in order from the newest one in accordance with the date and time information. Therefore, the user can visually recognize which registration information is actually used as the error detection rule.

Further, in the above configuration, when the proofreading information is learned and registered as an error detection rule, the proofread sentence is analyzed word by word (morpheme analysis),
From the analysis result, it is determined whether the calibration information is inappropriate as the error detection rule. Therefore, it is possible to prevent the occurrence of a defect in which inappropriate proofreading information, for example, proofreading information when the punctuation mark “.” Is calibrated to the reading point “,” is registered as the error detection rule.

Further, in the above configuration, it is checked whether or not the proofreading information to be registered as an error detection rule needs to be corrected based on the result of analysis of the proofreading sentence, and if necessary, the proofreading correction is performed. The information is corrected and registered as an error detection rule. As a result, when, for example, the word “gaku” in the character string “gaku doll” is corrected to “manipulation”, not only the calibrated part but also the information to be registered is corrected,
It is possible to register an error detection rule that detects the character string “Gandou” as an error and corrects it as “puppet”.

Further, in the above configuration, since the rule application form when the registration information designated by the user is used for error detection is displayed on the screen with a simple example sentence, the user can see how the sentence is changed by using the information. You can check whether it will be calibrated. Further, as a result of the confirmation, if the information is unnecessary, it can be deleted, and if necessary, it can be corrected.

[0018]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a text revision device (a document creation device having a text revision function) according to an embodiment of the present invention.

The text selection device shown in FIG. 1 includes an input unit 1, a control unit 2, an editing unit 3, a kana-Kanji conversion unit 4, a text selection control unit 5, a storage unit 6, a display control unit 7, a display unit 8 and a document storage. From the unit 9, the proofreading information learning unit 10, the proofreading information correction unit 11, the user dictionary 12, the analysis unit 13, the sentence analysis dictionary 14, the error detection unit 15, the error detection rule dictionary 16, the generation unit 17, the information editing unit 18, and the like. Composed.

The input unit 1 receives various inputs by keys or a mouse. The control unit 2 controls the entire device. The editing unit 3 edits a sentence under the control of the control unit 2.

The kana-kanji conversion unit 4 performs kana-kanji conversion using a kana-kanji conversion dictionary (not shown) under the control of the control unit 2. The text revision control unit 5 controls “text revision” under the control of the control unit 2. This text revision control unit 5
A user presenting unit 51 that displays the information learned by the calibration information learning unit 10 on the display unit 8 in order from the latest date and time.
And a rule application result display that shows to the user what kind of sentence is to be proofread when the information displayed by the user presenting section 51 is used for error detection in "sentence revision" And a portion 52.

The storage unit 6 is a memory for storing a sentence or the like being created or edited. The display control unit 7 controls the display on the display unit 8. The display unit 8 is used to display a sentence being created / edited, and is a CRT display, a liquid crystal display, or the like.

The document storage unit 9 is an external storage device such as a floppy disk device or a magnetic disk device for storing sentences and the like. The proofreading information learning unit 10 performs what kind of proofreading by the user from the information of the proofreading process in the editing unit 3 (here, the information before proofreading and the information after proofreading) under the control of the sentence revision control unit 5. Learn Taka. This calibration information learning unit 1
0 has a calibration information selection / registration unit 101 that determines whether or not the calibration information is suitable for being applied as an error detection rule, and registers it in the user dictionary 12 as an error detection rule only when it is suitable.

The proofreading information correction unit 11 corrects the proofreading information for registration in the user dictionary 12. The user dictionary 12 is used to register information required by the user. In the user dictionary 12, the information learned by the calibration information learning unit 10 and the information corrected by the calibration information correction unit 11 are registered as error detection rules.

The analysis unit 13 analyzes the sentence (morphological analysis) using the sentence analysis dictionary 14 under the control of the sentence revision control unit 5. The sentence analysis dictionary 14 is used to previously register knowledge information required for sentence analysis by the analysis unit 13.
This knowledge information includes a headline for each word and part-of-speech information.

The error detecting unit 15 controls the error detection rules in the user dictionary 12 and the error detection rule dictionary 16 under the control of the sentence correction control unit 5 when a sentence is corrected, and further, an error detection rule program (not shown). ) Is used to detect text errors.

The error detection rule dictionary 16 is used to previously register various error detection rules used for detecting an error in a sentence. The generation unit 17 generates a recommended candidate when there is a recommended candidate (correction candidate for correcting the detected error) in the sentence error detected by the error detection unit 15.

The information editing unit 18 deletes or corrects the registered information in the user dictionary 12 according to an instruction from the user. Next, the overall operation of the configuration of FIG. 1 will be described.

First, in creating a document, the user operates the input unit 1 to input a character string, the input character string is converted to Kana-Kanji by the Kana-Kanji conversion unit 4, and further edited by the editing unit 3 according to the user's designation. It is created by processing. The document created in this way is stored in the document storage unit 9.

Now, one of the documents stored in the document storage unit 9 is read from the document storage unit 9 into the storage unit 6 as a result of being designated by the user through the input unit 1 for proofreading, and the document is further stored. It is assumed that (a part of) is displayed on the display unit 8 by the display control unit 7.

In such a state, it is assumed that "correction" is requested from the input unit 1 according to the user's operation and the sentence is proofread by the editing unit 3 under the control of the control unit 2. In the proofreading process in the editing unit 3, the user operates the input unit 1 to specify a character string to be calibrated on the screen of the display unit 8 with a cursor or the like, and a correct character string is input from the input unit 1. It is done by inputting.

When the text is proofread by the editing unit 3 in this way, the text revision control unit 5 receives the information before and after the proofreading from the editing unit 3, and the proofreading information learning unit 10 together with the proofread text. Send to.

The proofreading information selection and registration unit 101 in the proofreading information learning unit 10 uses the information before and after the proofreading sent from the text revision control unit 5 before and after the proofreading (that is, how the proofreading was performed). It is determined whether or not it is appropriate to register (in the user dictionary 12) as an error detection rule that reflects (learning content of).

For this determination, the calibration information learning unit 1
0 causes the analysis unit 13 to analyze the sentence sent from the sentence revision control unit 5 using the information in the sentence analysis dictionary 14.
The proofreading information learning unit 10 generates an error detection rule when it can determine from the information before and after the proofreading and the sentence analysis result in the analysis unit 13 that it is suitable for registration as an error detection rule. , (If the rule is not registered in the user dictionary 12), the rule is registered in the user dictionary 12.

If the calibration information cannot be used as it is as the error detection rule, the calibration information learning unit 10 causes the calibration information correction unit 11 to perform the correction. The error detection rule generated in the calibration information learning unit 10 is provided with information on the corresponding calibration date and time (registration date and time).

In this way, the error detection rule generated based on the information before and after the proofreading (and the proofread sentence) is automatically registered in the user dictionary 12. Next, it is assumed that, according to a user operation, the input unit 1 requests "sentence review" and the range of sentences to be the target of "sentence review" is specified. In this case, the control unit 2 activates the text revision control unit 5. As a result, the text revision control unit 5
The text in the designated range is sent to the analysis unit 13.

The analysis unit 13 obtains information necessary for analyzing a sentence from the sentence analysis dictionary 14 and analyzes the sentence word by word.
By the sentence analysis (morphological analysis) in the analysis unit 13, the sentence in the designated range is divided into word units, and the part-of-speech information is obtained for each word.

The error detection unit 15 detects the error in the text in the specified range using the sentence analysis result of the analysis unit 13 and the error detection rules registered in the user dictionary 12 and the error detection rule dictionary 16.

If the error detected by the error detection unit 15 has a recommended candidate (correction candidate, conversion candidate), the generation unit 17 generates a recommended candidate based on the corresponding error detection rule. The result of error detection and the selection candidates are displayed on the display control unit 7
Is displayed on the display unit 8 via. If the displayed recommended candidate is the one that the user has intended, the user performs an input operation to select and confirm the recommended candidate.

The overall flow of the error detection rule automatic registration processing and the text revision processing in the configuration of FIG. 1 has been described above. Next, the above-mentioned error detection rule automatic registration processing and sentence revision processing will be described by giving an example.

First, a specific example of the error detection rule automatic registration process will be described with reference to the flowchart of FIG. now,
In response to [Sentence # 1] as shown in FIG. 3 (a), that is, "They felt a sense of strangeness and gazed at the movement of the doll."
The three calibrations shown in FIGS. 3B to 3D, that is, “.”.
Is corrected to "," [Calibration # 1], "Discomfort" is corrected to "Discomfort" [Calibration # 2], and "Reel" is "manipulated"
It is assumed that [Calibration # 3] is corrected to

Further, the condition for judging whether or not the calibration information learning unit 10 (in the calibration information selection registration unit 101) is suitable for the error detection rule is "unknown" as shown in FIG. "Condition # 1" that is a word, or Fig. 4
As shown in (b), "it is an independent word" [condition #
2] is satisfied.

In addition, the method of registration when [condition # 1] is satisfied is [action # 1] of "registering unknown word" as shown in FIG. 4C, and [condition # 2] is set. When the condition is satisfied, the registration method is assumed to be [action # 2] of "registering (pre) post information of the independent word and independent word" as shown in FIG. 4 (d).

Furthermore, the analysis result of the [sentence # 1] in the analysis unit 13 is [analysis # 1] as shown in FIG. 5, that is, they (pronouns) / is (particles) /. (Punctuation point) / Unpleasantness (unknown word) / To (particle) / feeling (The first line verb) / Te (particle) / Grip (Ra line 5th verb) / Doll (noun) / No particle / Movement (noun) / O (particle) / Gaze (Magoshita 1st verb) /. (Punctuation marks).

In such a case, [Calibration # 1] becomes [Analysis # 1].
1] to [Condition # 1] or [Condition # 2], the calibration information selection / registration unit 101 in the calibration information learning unit 10 determines that the information of [Calibration # 1] (before calibration) It is determined that the following information) is not suitable for the error detection rule (steps 21 and 22).

Since it can be seen that the following [Calibration # 2] applies to [Condition # 1] from [Analysis # 1], the above Step 2 is performed.
In 2, it is determined that the information of [Calibration # 2] is suitable for the error detection rule.

If this rule is not previously registered in the user dictionary 12, it is determined that it is not registered in the user dictionary 12 (step 23), and [Action #
The error detection rule registration in the user dictionary 12 according to 1] is performed by the proofreading information learning unit 10 (step 26). At this time, the system is inquired about the date and time, and the date and time information is added to the error detection rule.

Therefore, if the date and time of the inquiry is, for example, 18:50 on May 23, 1995, the figure is calculated based on the date and time information (calibration date and time information) and the information of [calibration # 1]. [Error detection rule # 1] shown in 6 (a), that is, discomfort (unknown word) → discomfort [time: 1995-05-23; 18: 50] is generated and registered in the user dictionary 12.

Here, the word "unnaturalness" on the left side of the arrow is information before proofreading and is used for error detection. The word "uncomfortable feeling" is attached with the part of speech (unknown word here) of the word analyzed by the analysis unit 13.

Further, the word "uncomfortable feeling" next to the arrow is the information after proofreading, and is corrected when the word "uncomfortable feeling" is detected in the target sentence at the time of sentence revision (that is, when an error is detected). It is used as information (recommendation candidate, conversion candidate).

Since it can be seen that the following [Calibration # 3] applies from [Analysis # 1] to [Condition # 2], the above Step 2 is performed.
In 2, the information of [Calibration # 3] is determined to be suitable for the error detection rule.

In this case, if the rule is not registered in the user dictionary 12 (step 23), [action # 2] is performed based on the information of [proofreading # 3]. In [Action # 2], it is necessary to register not only the proofreading information (independent word) but also the (previous) and subsequent information of the proofreading information in combination with the proofreading information (step 24). Therefore, the calibration information correction unit 11 corrects the calibration information (step 25).

In this example, based on [Analysis # 1] shown in FIG. 5 (by the analysis unit 13) for [Sentence # 1] in FIG. The previous word (independence word) “gaku” (ra line five-stage verb) and the proofread word (independence word) “manipulation” are corrected by the proofreading information correction unit 11 [sentence # 1].
Combined with the next word in "doll" (noun).

As described above, in [Action # 2], not only the calibrated portion but also the information to be registered is corrected by the calibration information correction unit 11. As a result, in the case of [Proofreading # 3], [Error detection rule # 2] shown in FIG. 6 (b), that is, Gari (La-line five-stage verb) + Doll (noun) → Manipulation + Doll (noun) [Time: 1995-05-23; 18: 51] is generated and registered in the user dictionary 12 (step 26).

Next, a specific example of the text revision process will be described. Now, for example, it is assumed that "sentence refinement" is requested at the beginning of the sentence "[I got a doll]." ([Statement # 2]).

In this case, the error detection unit 15 follows the error detection rule # 2 registered in the user dictionary 12 and shown in FIG. Is detected, and as a conversion candidate (recommended candidate), a “puppet” paired with the “puppet” in [Error detection rule # 2] is acquired. The conversion candidate “puppet” is passed to the text revision control unit 5 and displayed on the display unit 8 via the display control unit 7.
Is displayed to the user.

If there are a plurality of error detection rules with different conversion candidates in the user dictionary 12, the new error detection rule of the registered time (calibration date / time) is used. For example, [Error detection rule # 2] Repetition (La-line five-stage verb) + Doll (noun) → Manipulation + Doll (noun) [time: 1995-05-23; 18: 51] [Error detection rule # 3] Revolution When there are two error detection rules that detect a "gray doll" as an error, such as (Rayuki 5 verbs) + doll (noun) → Marionette [time: 1995-05-20; 17: 35]. The latest [Error Detection Rule #
2] applies.

Next, a case will be described in which the contents of the user dictionary 12 are requested by the input unit 1 according to the user operation. In this case, the user presentation unit 51 in the text revision control unit 5 reads each registration information (error detection rule) of the user dictionary 12 through the information editing unit 18, and reads each registration information according to the date and time information in the registration information. Sort by newest first.

Next, the user presenting section 51 presents a list of the read registration information (user dictionary information) on the display section 8 in the sort order and presents it to the user. FIG. 7 shows an example of a list display of this user dictionary information.

Next, in the list display state as shown in FIG.
A case will be described in which arbitrary registration information is selected and designated by the input unit 1 according to a user operation, and it is requested to display what happens when the registration information is used as an error detection rule (rule application result). .

In this case, the text revision control unit 5 activates the error detection unit 15 to acquire the rule application result information regarding the user dictionary registration information (error detection rule) designated by the user. Here, the part of speech before correction in the registration information is used, if it is a noun, it is not used, and if it is a verb, it is put at the end of its usage.

For example, if the character is a definite verb (verbal verb) + doll (noun) → puppet + doll (noun) [time: 1995-05-23; 18: 51], it is a noun. Become.

Further, if "strengthen (sa change) → emphasize [time: 1995-05-19; 14: 20], then it is sa strange, so that the inflection ending is added to emphasize-> emphasize.

The above situation is shown in FIG. When the rule application result display unit 52 in the text revision control unit 5 acquires the rule application result information regarding the user-specified user dictionary registration information (error detection rule) from the error detection unit 15, the rule application result display unit 52 displays it on the display unit 8. And present it to the user.

If the user sees this and the rule is not necessary, the user can operate the input unit 1 to instruct the deletion. For example, in the case where “strengthen → emphasize” is presented, if the user gives an instruction to delete, the sentence revision control unit 5
Activates the information editing unit 18 to delete the corresponding registration information (error detection rule) from the user dictionary 12, that is, the enhancement (sa change) → emphasis [time: 1995-05-19; 14: 20]. In addition, according to the correction instruction from the user,
It is also possible to correct the registration information in the user dictionary 12 by the information editing section 18.

In the above embodiment, the user presenting section 51
Although the rule application result display unit 52 is provided in the text revision control unit 5, it may be independent of the text revision control unit 5. Further, the calibration information selection / registration unit 101 in the calibration information learning unit 10 may also be independent of the calibration information learning unit 10.

[0067]

As described above in detail, according to the present invention, when a sentence is proofread, information about how the sentence is proofread is learned and the information is used as an error detection rule for sentence revision. By doing so, it is possible to automatically create an error detection rule suitable for the user, and it becomes possible to detect an error that could not be detected by the error detection rule prepared in advance by the system side. "Reform" can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of a text selection device according to an embodiment of the present invention.

FIG. 2 is a flowchart for explaining an error detection rule automatic registration process in the embodiment.

FIG. 3 is a diagram showing a specific example of a sentence proofread by the editing unit 3 and proofreading contents (character strings before and after proofreading).

FIG. 4 is a diagram showing a specific example of a condition for judging whether or not the calibration information is suitable for an error detection rule and a method of registering an error detection rule when the condition is satisfied, which is applied in the embodiment.

5 is an analysis unit 13 for the sentence ([statement # 1]) in FIG.
The figure which shows the analysis result by.

6 is a diagram showing a specific example of an error detection rule generated based on the calibration contents ([calibration # 2], [calibration # 3]) shown in FIG.

7 is a diagram showing an example of a list display of user dictionary information according to date and time information by the user presenting unit 51. FIG.

FIG. 8 is a diagram showing a rule application result display example by the rule application result display unit 52 in association with an error detection rule.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input part (character string designating means, rule application result display instructing means, deletion / correction instructing means), 2 ... Control part, 3 ... Editing part (character string proofreading means), 4 ... Kana-kanji conversion part, 5 ... Sentence revision Control unit, 6 ... Storage unit, 7 ... Display control unit, 8 ... Display unit, 9 ... Document storage unit, 10 ... Calibration information learning unit, 11 ... Calibration information correction unit, 12 ... User dictionary (first storage unit) , 13 ... analysis unit, 14 ... sentence analysis dictionary, 15 ... error detection unit, 16 ... error detection rule dictionary (second storage means), 17 ... generation unit, 18 ... information editing unit, 51 ... user presentation unit, 52 ... Rule application result display section, 101 ... Calibration information selection registration section.

Claims (7)

[Claims] 1. A character string designating unit for designating an arbitrary character string in a sentence, a character string calibrating unit for calibrating a character string designated by the character string designating unit, and the character string calibrating unit. Calibration information learning means for learning information on how the proofreading has been performed, and first storage means for registering the calibration information learned by the calibration information learning means as an error detection rule for text revision. A second storage means in which an error detection rule for text revision is registered in advance, and an error detection rule registered in the first storage means and the second storage means is used at the time of text revision. And an error detecting unit for detecting an error in a character string in a sentence. 2. A character string designating unit for designating an arbitrary character string in a sentence, a character string calibrating unit for calibrating the character string designated by the character string designating unit, and the character string calibrating unit. A calibration information learning means for learning information on how the calibration is performed by the calibration information learning means for adding date and time information to the calibration information, and the date and time information learned by the calibration information learning means. A first storage means for registering the proofreading information as an error detection rule for text revision, a second storage means in which an error detection rule for text revision is registered in advance, and An error detecting unit for detecting an error in a character string in a sentence by using the error detecting rules registered in the first storing unit and the second storing unit, wherein the corresponding error detecting rule is the first storing unit. Multiple in the means In some cases, texts elaboration apparatus characterized by comprising the error detecting means for selecting the latest error detection rules of which are based on the date and time information. 3. An analyzing unit for analyzing a sentence proofread by the character string proofing unit on a word-by-word basis, and whether the information proofread by the character string proofing unit is suitable for the error detection rule. Judgment based on the analysis result of the analysis means, and if it is not suitable, calibration information selection registration for preventing the calibration information from being learned by the calibration information learning means and registered in the first storage means 3. The method according to claim 1, further comprising means.
The written text revision device. 4. An analysis unit for analyzing a sentence proofread by the character string proofing unit on a word basis, and when the proofreading information is learned by the proofreading information learning unit, the proofreading information is analyzed by the analysis unit. 3. The text revision device according to claim 1, further comprising a calibration information correction means for correcting for an error detection rule based on the analysis result of 1. 5. A user presenting unit that displays a list of registration information in the first storage unit on a screen, and one of the registration information presented by the user presenting unit is designated and applied to error detection. The rule application result display instructing means for instructing the display of the application form of, and in accordance with the instruction by the rule application result display instructing means, what text is proofread when corresponding information is applied to error detection. Rule application result display means for displaying on a screen using an example sentence, deletion / correction instruction means for instructing deletion or correction of the registered information which is the display target by the rule application result display means, and this deletion / correction It further comprises: information editing means for deleting the corresponding registration information from the first storage means or correcting the corresponding registration information in response to an instruction from the correction instruction means. Sentence Elaboration apparatus according to claim 1 or claim 2, wherein that. 6. The sentence revision device according to claim 2, further comprising a user presenting unit that sorts the registered information in the first storage unit based on the date and time information and displays the screen. 7. When an arbitrary character string in a sentence is proofread, information on how the character string is proofread is learned, and the text is refined based on the information before and after the proofreading. The error detection rule is generated and registered in the user dictionary, and when the text is refined, the error detection of the character string in the text is performed by referring to the user dictionary and the error detection rule dictionary in which the error detection rule is registered in advance. A text revision method characterized by that.