JPH07319909A - Device and method for retrieving document, character recognizing device and preparation of dictionary - Google Patents

Abstract

PURPOSE:To quickly retrieve an objective document even when a retrieving word whose notation is unknown is inputted. CONSTITUTION:A control part 4 reads out a homonym corresponding to a phonetic character string inputted from an input part 1 from a homonym information storing part 3. Then, the control part 4 retrieves a document including the pertinent homonym out of respective homonyms read out from the storing part 3 from a document storing part 2 and then outputs the retrieval information from an output part 5 in accordance with a prescribed format. Thereby, even when the notation of a retrieving word is unknown, an objective document can be retrieved by inputting a phonetic character string corresponding to the retrieving word.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document retrieval apparatus for retrieving documents stored in a database, and more particularly to a document retrieval method for retrieving documents without using keywords.
The present invention relates to a dictionary creating method for creating a dictionary by registering a pattern for recognizing a character input by handwriting in a character recognizing device installed in a document creating device for creating the document or the like.

[0002]

2. Description of the Related Art Conventionally, in this type of document retrieval apparatus, a method has been used in which a keyword is added to a document in advance when a database is created, and the relevant document is retrieved based on this keyword at the time of retrieval. With this method, the search speed is relatively high, but there is a problem in that the work of adding keywords to the document stored in the database is burdensome. Moreover, it is not always easy to add an appropriate keyword to a plurality of documents, and if the appropriate keyword is not added, the document targeted by the searcher may not be obtained during the search.

On the other hand, in recent years, with the increase in processing speed and capacity of computers, a search method by full-text search is being put to practical use when searching for a desired document. In this method, since all the character strings in the document can be searched, it is not necessary to add a keyword when the document is stored in the database, and the database including the phrase specified by the searcher at the time of searching the document. You can get all the documents in it. Furthermore, a method of searching a plurality of words for one search word designated by the searcher has also been used by using a synonym dictionary or a synonym dictionary.

However, in the document search method by such a full-text search, only a word of which the notation (character) is known can be retrieved, and therefore the notation (character) is known even if the pronunciation (reading) is known.
I couldn't search for a word I didn't understand. In particular, there is a drawback in that it is not always possible to retrieve the target document when the kanji to be applied is not known and only the reading is known, such as proper nouns in Japanese.

By the way, a document preparation apparatus equipped with the above-mentioned document retrieval apparatus has an input display device in which a display device and a coordinate input device are laminated and integrated, and the input display device has characters and figures on paper. Since information can be entered as if writing it, it has recently come to be used in various fields. Along with this, devices that used to use the keyboard as an input device to input characters and create sentences are also input in the form of coordinate point sequence information using a stylus pen on the LCD tablet without using the keyboard. The information has been recognized as characters and input into the device. Along with this, a document recognition device of this type is equipped with a character recognition device including the input display device.

Here, in the method of recognizing a character in the character recognition device, the coordinate information input by the input person is matched with a plurality of patterns in a recognition dictionary registered in advance, and the most input coordinate information is obtained. A large number of recognition methods have been proposed, including a method in which registered characters are used as recognition candidates. However, no sufficient character recognition rate has been obtained in any of the recognition methods, and in order to improve the character recognition rate, a character recognition dictionary for each individual is prepared in addition to the character recognition dictionary given in advance. We are trying to improve the recognition rate by registering difficult-to-recognize characters in a personal character recognition dictionary. However, when the input person registers a new character pattern, if similar patterns exist in the basic recognition dictionary or the personal character recognition dictionary registered in the past, it is rather difficult to identify similar characters. Sometimes it became.

[0007]

As described above, in the conventional document search apparatus, a search method using a synonym dictionary or a synonym dictionary is added to the full-text search so that one search specified by the searcher is performed. Documents that include synonyms or synonyms of the search words can be searched from the database in response to the entered words, but for the search words whose notation (character) is not known, the target document is not always prompt. There was a drawback that it was not always possible to search.

Further, in a conventional character recognition device having an input display device, a character recognition dictionary for each individual is prepared in addition to a character recognition dictionary given in advance, and characters difficult to recognize are registered in this dictionary. It is considered to improve the recognition rate, but when the user registers a new character pattern, a pattern similar to the basic recognition dictionary or a personal character recognition dictionary registered in the past is used. However, the existence of the character makes it difficult to identify similar characters.

Therefore, the present invention has been made in view of the above circumstances, and a document search apparatus and a document search method capable of quickly searching for a target document even when a search word whose notation is unknown is input. An object of the present invention is to provide a character recognition device and a dictionary creation method capable of improving the recognition rate of handwritten characters by the input person newly registering a character pattern in a personal character recognition dictionary.

[0010]

According to a first aspect of the present invention, there is provided a document retrieval device for retrieving a document stored in a database on the basis of a retrieval word, wherein a phonetic character string and a homophonic word corresponding to the phonetic character string are provided. The stored storage means, the homophone variant for obtaining homophones corresponding to the input phonetic character string from the storage means, and each homophone variant acquired by the homophone variant is the search word. And a search means for searching a document from the database.

According to a fourth aspect of the present invention, in the character recognition device for character-recognizing and inputting handwritten characters by collating the pattern of the coordinate point sequence input from the coordinate input means with the pattern in the character recognition dictionary, the coordinates are Registration means for registering the pattern of the coordinate point sequence input from the input means and the character code corresponding to this pattern in the character recognition dictionary, and the existing pattern similar to the pattern registered by this registration means for the character recognition. Search method to search from the dictionary,
And a modifying unit that modifies the candidate priority information in the character recognition dictionary for the pattern searched by the searching unit.

[0012]

In the document retrieval apparatus according to the first aspect of the invention, the storage means stores the phonetic character string and the corresponding homophones and acronyms. The homophone variant has a homophone variant corresponding to the input phonetic character string from the storage means. The search means searches the document from the database by using the homophones that have been acquired by the homophone variant as the search words. This allows

In the character recognition device of the invention of claim 4,
The registration unit registers the pattern of the coordinate point sequence input from the coordinate input unit and the character code corresponding to this pattern in the character recognition dictionary. The modifying means modifies the searching means for searching the character recognition dictionary for an existing pattern similar to the pattern registered by the registering means, and the candidate priority information in the character recognition dictionary for the pattern searched by the searching means. To do.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a document search device of the present invention. Reference numeral 101 is an input unit for inputting a character string to be searched, 102 is a document storage unit that stores a document to be searched, and 103 is a homophone character that stores a phonetic character string as a set of homophones corresponding to this character string. Word information storage unit, 10
Reference numeral 4 denotes a homophone group of a character string input from the input unit 101 after obtaining a homophone acronym group of the input character string using the information of the homophone acronym information storage unit 103. Reference numeral 105 is an output unit such as an LCD or a CRT that searches for a document including the search result.

Next, the operation of this embodiment will be described. The search character string (phonetic character string) input from the input unit 101 is sent to the control unit 104, and the control unit 104 uses the information in the homophone acronym information storage unit 103 to generate a homophone acronym group of the input search character string. To get Here, the homophone acronym information storage unit 103 has a structure as shown in FIG. 2, and stores a phonetic character string and a homophone acronym group corresponding thereto. For example, "Sakai", "Sakai", "Sakai", and "Sakai" are stored as homonyms for the search character string "Sakai", and finally the search character string "Sakai" is stored. . Furthermore,
For each of the obtained homophone variants, the control unit 104 searches the documents stored in the document storage unit 3 for a document containing the same character string.

FIG. 3 is a flow chart showing the flow of the search processing of the control unit 104 described above. First, the control unit 10
In step 301, after obtaining the search phonetic character string from the input unit 101 in step 301, in step 302, the search phonetic character string and the homophone different word information in the homophone different word information storage unit 103 shown in FIG. Compare with the first character string in the phonetic items in the table. The control unit 104 checks in step 303 whether or not the character string in the phonetic item and the search phonetic character string match, and if they match, in step 304, the corresponding homophone from the homophone / synonym information table. An allophone word group is obtained, and if they do not match, it is confirmed at step 305 whether all phonetic items in the homonym word information table have been referenced. If there is no match with the search phonetic character string for all phonetic items in the homophone synonym information table, the control unit 104 outputs information that there is no homophone variant in step 305.
Output to 05 and end the processing.

When it is determined in step 305 that there is a phonetic character string that has not been referred to yet, the control unit 104 compares the next phonetic item and the retrieved phonetic character string in step 307, and then proceeds to step 303. Returning, it is checked whether the phonetic item and the search phonetic character string match, and the above process is repeated until a homophone group is obtained. If the homonym group is obtained in step 304, the control unit 104 proceeds to step 3
In step 08, the first candidate of the homonym group is compared with the character strings in the first document of the plurality of documents stored in the document storage unit 102, and in step 309, the candidates are included in the character strings in the document. It is determined whether there is the same character string as the above, and if there is, the search result information is created in step 310 by combining the acronym with the structure shown in FIG. 4 and the document name including the same for the relevant document. If not, the process jumps to step 311.

Next, at step 311, the document storage unit 1
It is checked whether all the documents in 02 have been searched, and if there is a document that has not been searched yet, the candidate is compared with the character string of the next document in the document storage unit 102 in step 312, and step 309 Go back to and see if you have the same string. The above process is repeated for all the documents in the document storage unit 102. If it is confirmed in step 311 that all the documents have been searched, the control unit 104 proceeds to step 313 and checks whether there are homophone variants that have not been searched here. As a result, if present, the control unit 104
Compares the next homophone candidate with the character string in the first document of the plurality of documents in the document storage unit 102 in step 314, and returns to step 309 again to check whether the same character string exists. , The above process is repeated for all candidates. If the control unit 104 confirms that all candidates have been searched in step 313, the control unit 104 proceeds to step 315 and determines from the search result as shown in FIG. 4 the homophones and the homophone types as shown in FIG. A document name corresponding to is created in the output information in a form that can be clearly shown to the user, and this is sent to the output unit 105 in step 316, and the processing ends.

The output unit 105 described above displays the output information sent from the control unit 104, and its operation will be described below using a specific example. For example, when the character string “Sakai” is input from the input unit 101, the character string is sent to the control unit 104. When the homophone variation word information storage unit 103 stores a homophone variation word table as shown in FIG. 2, the control unit 104 searches for a character string such as “Sakai” from the phonetic item and sets it as a corresponding homography variation group. "Sakai"
Get "Sakai", "Sakai" and "Sakai". For each word of this homophone group, a document containing the word is stored in the document storage unit 1.
Search from 02. With this search, as shown in FIG.
The control unit 104 selects "Document 1" as a document including "Sakai".
"Document 2""Document3""Document4","Document5" as a document containing "Sakai", "Document 3""Document6""Document7" as a document containing "Sakai", "Sakai""Document7" and "Document 8" are obtained as documents including ". In addition, the information "document 5" and "document 9" are obtained from the input character string "Sakai". Next, the control unit 104 uses the information, which is a combination of the homophones and the document names including the same as shown in FIG. 4, as to what the different acronyms exist in the user and the document name including the different acronyms. In order to clearly show what is, the form is modified as shown in FIG. 5 and displayed on the output unit 105.

In the English homophone search, the homophone search can be performed by the same process as the Japanese homophone search described above. When the information as shown in FIG. 6 is stored in the homonym word information storage unit 103, and when a phonetic character string called “rait” is input from the input unit 101, the control unit 104 selects [rait] from phonetic items.
"R" as the corresponding homophone group
ite ”“ right ”“ write ”“ write ”
t ”is obtained. The control unit 104 searches the document storage unit 102 for a document including the corresponding word for each word of the homonym word group. The control unit 104 outputs, from the output unit 105, information in which the homophones thus obtained and the document name including the same are combined.

According to the present embodiment, the character string including the homophones of this character string is used as the search word for the input character string input from the input unit 101, and the document storage unit 102 is searched. It is possible to search all documents that include the search word in the document. For this reason, it is possible that a search word whose notation (character) is not known can be searched with a character string that includes homophones. Therefore, at the time of search, the searcher uses the character string specified by the searcher for the purpose. Documents can be searched quickly. In particular, even if the notation is not known, if only the reading is known, the target document can be searched by using the phonetic character string indicating the reading as a search word. Therefore, it is not necessary to use the keyword when storing the document in the document storage unit 102, and the document can be easily stored in the document storage unit 102.

The present invention is not limited to the above embodiment. Although Japanese and English are used in the embodiment, the present invention can be applied to other original languages having homophones. or,
Although the input character string is also searched at the same time when searching, the same effect can be obtained by separately searching the input character string and the homophones of this input character string. In this case, the input character string is omitted from the homophone group of FIGS. Furthermore, the search term can be input by voice input. Here, there are various methods for finding a search word from a character string in a document, such as a method of using the title of the character string and a method of creating a search table in advance when inputting a document. Does not particularly matter.

FIG. 7 is a block diagram showing an embodiment of the character recognition device of the present invention. The character recognition device of this example uses a transparent tablet 1 and a coordinate input device including a stylus pen 2 for indicating coordinates on the transparent tablet 1, and two-dimensional coordinate point sequence information obtained from the coordinate input device. A control device 3 mainly composed of a microprocessor for performing dictionary registration based on the display device, a display device 4 such as a liquid crystal display for displaying handwriting data of a registration pattern input from the stylus pen 2, and the control device 3 perform dictionary registration. It is composed of an external storage device 5 that is accessed when performing.

However, as the display device 4, a plasma display or the like can be used in addition to the liquid crystal display. The liquid crystal display as the display device 4 is laminated and integrated with the transparent tablet 1. That is, the transparent tablet 1 laminated and integrated with this liquid crystal display forms the same coordinate plane with the same size, and the information displayed on the liquid crystal display can be visually confirmed through the transparent tablet 1. . By the transparent tablet 1 and the display device 4 which are laminated and integrated as described above, the coordinate pointing position on the transparent tablet 1 is displayed as information at the same position on the display device 4, and, for example, a character / graphic is displayed on paper. You can input information as if you were drawing.

FIG. 8 is a diagram showing an example of the internal configuration of the document creating apparatus shown in FIG. The control device 3 includes an initial setting unit 31,
The input unit 32, the personal dictionary creation unit 33, the character recognition unit 34, the recognition dictionary operation unit 35, the display control unit 36, and the storage unit 37. The storage unit 37 is an input data buffer 371 for temporarily storing information on the coordinate point sequence input from the transparent tablet 1, a character code buffer 372 for temporarily storing character codes of character patterns to be registered in the dictionary, and personal characters. A dictionary creation buffer 373, which is a temporary temporary area for creating a recognition dictionary, a display data buffer 374 for temporarily storing data to be displayed on the display device (liquid crystal display) 4, and at the time of system initialization. It is composed of screen image data 375 which stores an image of an initial screen to be displayed, a screen area table 376 which stores information on an input frame of the input screen displayed via the display device 4 and coordinate positions of various buttons. There is.

Next, the operation of this embodiment will be described. The initial setting unit 31 performs initialization processing such as displaying an initial screen on the display device 4 and clearing various buffers. Also, the input unit 3
2 carries out a process relating to the coordinate information inputted from the coordinate input device 1. Furthermore, the personal dictionary creation unit 33 uses the personal character recognition dictionary 5 based on the coordinate information input by the input unit 32.
2 is performed. The character recognition unit 34 converts the coordinate information of the dictionary registration pattern stored in the input data buffer 371 into the personal character recognition dictionary 5 of the external storage device 5.
The data in the personal character recognition dictionary 52 are recognized as reference designated characters. The recognition dictionary operation unit 35 determines whether a dictionary pattern similar to the dictionary registration pattern in the personal character recognition dictionary 52 exists in the standard character recognition dictionary 51 based on the evaluation value of the character candidate output in the character recognition unit 34. If a similar dictionary pattern is found, the dictionary pattern is selected as a recognition candidate in the subsequent recognition processing, and therefore processing for adding information that lowers the rank of the candidate is performed. Further, the display control unit 36 outputs the data stored in the storage unit 37 to the display device 4,
It controls the erasing process.

The display control unit 36 of the control device 3 is the display device 4
The input screen 40 as shown in FIG. 9 is displayed. The user uses the stylus pen 2 to input a character pattern to be registered in the dictionary in the input frame 41 of the input screen 40, and the character code of the character to be registered is further input in the character code input area 42.
To enter. Then, when the registration icon 44 is touched with the stylus pen 2, dictionary registration work and recognition dictionary correction work are started. After inputting the registered character in the input frame 41 and when wanting to input the data again, touch the cancel icon 43 to initialize the data in the input frame 41 and the data in the character code input area 42.

As described above, the external storage device 5 has a standard character recognition dictionary 51 in which standard character patterns are registered.
And a personal character recognition dictionary 52 capable of registering the character pattern of the individual input person. Standard character recognition dictionary 5
1 and the personal character recognition dictionary 52 store character information referred to when the character recognition unit 34 recognizes a character. As shown in FIG. 10, the character information includes a dictionary number, variable rank data that stores information about how much the rank behind recognition is lowered in the character recognition process, a character, a character code, the number of strokes, and the start and end points of one image. Information and the like expressed in relative coordinates with the start point of one stroke as a reference are stored.

FIG. 11 is a flow chart showing the flow of processing for registering the character pattern of the individual input person in the personal character recognition dictionary 52 by the apparatus shown in FIG. 7 or 8. In step 501, the initial setting unit 31 of the control device 3 clears various buffers and the like in the storage unit 37 at the beginning of the process, and based on the image data of the screen image data 375, the initial screen 40 as shown in FIG. Is displayed on the display device 4 by the display control unit 36. Next, in step 502, the input person uses the stylus pen 2 to input a registration pattern in the input frame 41 of the input screen 40 shown in FIG. 9 or a character code to be registered in the character code input area 42. Then, in step 503, the input unit 32 refers to the values in the screen area table 376 and determines whether or not the values are input. here,
The screen area table 376 stores, for example, as shown in FIG. 12, an area number indicating the area of the input frame / various icons and the upper left and lower right coordinates of the area. If the registered character pattern is input in the input frame 41, step 50
In 4, the input unit 32 displays the handwriting on the display device 4 via the display control unit 36, stores the coordinate information in the input data buffer 371 in Step 505, and then returns to Step 502. Here, as shown in FIG. 13, the input data buffer 371 stores the x coordinate and the y coordinate of the data.
Further, the input data buffer 371 also stores a separator indicating a division of one stroke for use in the later character recognition processing.

Here, the judgment of the division of one stroke is made when the stylus pen 2 is separated from the transparent tablet 1.
If it is determined in step 503 that the input person has not input the characters of the registered pattern into the input frame 41, the input unit 32 causes the screen area table 376 to operate in step 506.
By checking whether or not the character code has been input based on, it is determined whether or not the character code of the registered character is input in the character code input area 42. If it is determined that the input unit 32 is inputting the character code for character registration, the corresponding character code touched in step 507 is displayed on the liquid crystal tablet 4 via the display control unit 36, and then in step 508. Character code buffer 37
2 and then returns to step 502.

Further, when it is determined in step 506 that the input person has not entered the character code input area 42, the input section 32 has entered the cancel button 43 based on the screen area table 376. No, step 509
Determines whether or not cancellation processing is being performed. If the cancel button 43 is touched, the input unit 32 causes the character code input area 4 via the display control unit 36 in step 510.
Initialization processing of the display of the character code in 2 and the display of the handwriting in the input frame 41 is performed. Then, in step 511, the contents of the input data buffer 371 and the character code buffer 372 are cleared, and the process returns to step 502. However, when it is determined in step 509 that the input person does not touch the cancel button 43 with the pen 2, the input unit 32 checks the data in the screen area table 376 to touch the registration button 44. In step 513, it is determined whether dictionary registration start processing is being performed. When the dictionary registration start process is not performed, it is determined that the pen 2 is placed at an invalid position, and the process returns to step 502 without doing anything.

If the registration button 44 is touched with the pen 2, the processing moves to the personal dictionary creating section 33, and the registration processing starts. First, the personal dictionary creation unit 33 uses the character code buffer 3
The value of 72 is checked to see if the character code to be registered is correct. When the character code is not correct, the personal dictionary creating unit 33 displays the warning sentence in step 515 via the display control unit 36 and the display data buffer 374, and returns to step 502. When the character code is determined to be correct in step 514, the input data buffer 371 is checked in step 516 to see if there is data to be registered. If the data is not entered, the personal dictionary creating unit 33 gives a warning message in step 517.
The display is performed via the display controller 36 and the display data buffer 374, and the process returns to step 502. When the registration data exists, the personal dictionary creating unit 33 registers in the personal dictionary as follows. That is, the personal dictionary creation unit 33 once transfers the contents of the personal character recognition dictionary 52 to the dictionary creation buffer 373, and works on the dictionary creation buffer 373. A registered pattern is added to the personal dictionary in the form of a data structure as shown in FIG. 10. When the work is completed in step 518, the added personal character recognition dictionary 52 is transferred to the external storage device 5.

When the work of registering in the personal character recognition dictionary 52 is completed, the processing moves to the recognition dictionary operation unit 35. The recognition dictionary operation unit 35 changes the registration information for the dictionary pattern similar to the character pattern registered in step 518 as follows. First, the recognition dictionary operation unit 35 character-recognizes the registered character data stored in the input data buffer 371 via the character recognition unit 34 in step 519. This character recognition processing is performed by the standard character recognition dictionary 51 having the same number of strokes as the input data stored in the external storage device 5.
It is performed as follows using the registration pattern of. First, the two-dimensional coordinate data stored in the input data buffer 371 is converted into the form of relative coordinates based on the starting point of the first image stored in the standard character recognition dictionary 51. Next, the relative coordinate information and the distance at each of the start and end coordinate points of one stroke are calculated. Then, the total value of the distances of the coordinate points is obtained, and in step 519, the value obtained by dividing the total value by the number of strokes is calculated as the evaluation value. Based on the calculated evaluation value, the evaluation value is compared with a preset threshold value to check in step 520 whether the evaluation value is smaller than the threshold value. When the evaluation value is smaller than the threshold value, the recognition dictionary operation unit 35 determines that the dictionary pattern of the recognition candidate for which the character is recognized is similar to the registration pattern registered in step 518. In that case, the recognition dictionary operation unit 35 increments the value of the variable rank data of the similar dictionary pattern in the standard character recognition dictionary 51 by 1 in step 521, and when this step 521 ends, the processing ends. If the evaluation value is greater than the threshold value in step 520, the processing of this example is ended.

After that, in the character recognition processing of the character recognition unit 34, character candidates are calculated, and the candidate rank is changed using this variable rank data. Specifically, the character candidate is "i"
When the ranking data of "ri" and "" are "2", "1" and "0" respectively, the rank of the first candidate is lowered by two and the rank of the second candidate is lowered by one. Is changed to "", "ri", and "i".

According to this embodiment, when registering a new character pattern in the personal character recognition dictionary 52, a pattern similar to the registered character pattern and already registered in the standard character recognition dictionary 51 is Even if the input person newly adds a registered character to the dictionary 51 by lowering the candidate rank when a candidate is picked up in the subsequent recognition processing, it is possible to avoid causing the recognition error due to the addition. As a result, the input person can successively register character patterns that he or she can easily recognize, and the character recognition rate can be improved accordingly.

The present invention is not limited to the above embodiment. For example, in the present embodiment, the variable rank data is changed to the dictionary pattern only with the standard character recognition dictionary, but the information of the change of the candidate rank of the dictionary pattern is added to the plurality of recognition dictionaries including the personal character recognition dictionary. You may go. Further, in the present embodiment, the variation rank data of the recognition pattern similar to the registered pattern was directly changed, but assuming that the recognition dictionary is on the ROM, the correspondence between the variation rank data and the dictionary number is determined by the dictionary body. R as a separate table
The data structure of the recognition dictionary may be such that the dictionary is modified by holding the information on the OM and changing the information.

In the present embodiment, the character recognition method is recognized by matching the distance between the character registered in the dictionary and the input coordinate, but the character is abstracted into a certain basic form. Other character recognition methods may be used as long as the evaluation value can be calculated, such as matching with the shape.
The data structure of the recognition dictionary may be changed accordingly. In the present embodiment, the rank changing process of lowering the candidate rank of the corresponding dictionary pattern based on the variable rank data by the number of values of the variable rank data is performed. The same effect can be obtained by performing the replacement process of the candidate ranks other than the present embodiment, such as lowering the rank.

[0038]

As described above, according to the document search device, the document search method, the character recognition device, and the dictionary creation method of the present invention, a plurality of homophones are input for one search phonetic character string input designated by the searcher. A document search device and a document search method that enable a search for a document including a character string having an acronym and can quickly search for a target document even when a search word whose description is unknown is input, and an input method. By newly registering a character pattern in the personal character recognition dictionary,
By storing the homophone conversion information that can improve the recognition rate of handwritten characters, it is possible to search for one or more homophones only by inputting one phonetic character string.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a document search device according to the present invention.

FIG. 2 is a diagram showing an example of the structure in the homonym word information storage unit shown in FIG.

FIG. 3 is a flowchart showing a search process of a control unit shown in FIG.

FIG. 4 is a diagram showing an example of a result of a search performed by the control unit shown in FIG. 1 for a document including each homophone.

5 is a diagram showing a format example displayed on the output unit shown in FIG. 1 as an example of the result shown in FIG.

FIG. 6 is a diagram showing homophones when the character string input by the input unit shown in FIG. 1 is English.

FIG. 7 is a schematic configuration diagram showing an embodiment of a character recognition device of the present invention.

8 is a block diagram showing a detailed example of the control device shown in FIG.

9 is a view showing an example of a personal character pattern input screen displayed on the display device shown in FIG.

10 is a diagram showing an example of the data structure of the standard character recognition dictionary shown in FIG.

11 is a flowchart showing a character pattern registration process of the control device shown in FIG.

12 is a diagram showing an example of the data structure of the screen area table shown in FIG.

13 is a diagram showing an example of the data structure of the input data buffer shown in FIG.

[Explanation of symbols]

1 ... Transparent tablet 2 ... Stylus pen 3 ... Control device 4 ... Display device 5 ... External storage device 31 ... Initial setting part 32 ... Input part 33 ... Personal dictionary creation part 34 ... Character recognition part 35 ... Recognition dictionary operation part 36 ... Display Control unit 37 ... Storage unit 371 ... Input data buffer 372 ... Character code buffer 373 ... Dictionary creation buffer 374 ... Display data buffer 375 ... Screen image data 376 ... Screen area table 51 ... Standard character recognition dictionary 52 ... Personal character recognition dictionary Reference numeral 101 ... Input unit 102 ... Document storage unit 103 ... Homophonetic word information storage unit 104 ... Control unit 105 ... Output unit

Claims (8)

[Claims] 1. A document retrieval device for retrieving a document stored in a database based on a retrieval word, comprising storage means for storing a phonetic character string and a corresponding homophone acronym, and an input table. A homophone variant for obtaining a homophone word corresponding to a phonetic character string from the storage means, and a search means for searching a document from the database using each homophone variant acquired by the homophone variant word as the search word. A document retrieval device comprising: 2. The document search device according to claim 1, wherein the search means also searches the database for a document including an input phonetic character string. 3. A document retrieval method for retrieving a document stored in a database based on a retrieval word, wherein homophones corresponding to an input phonetic character string are obtained, and then the obtained homophones A document search method, wherein a document is searched from the database using a word as the search word. 4. A character recognition device for character-recognizing and inputting a handwritten character by collating a pattern of a sequence of coordinate points input from the coordinate input means with a pattern in a character recognition dictionary, wherein the coordinate input means From the character recognition dictionary, a registration means for registering the pattern of the coordinate point sequence input from and the character code corresponding to this pattern in the character recognition dictionary, and an existing pattern similar to the pattern registered by this registration means A character recognition device comprising: a search means for searching; and a correction means for changing candidate priority order information in the character recognition dictionary of a pattern searched by the search means. 5. The character recognition device according to claim 1, further comprising a dictionary in which a new pattern and a character code corresponding thereto are registered by the registration means. 6. The similarity between the pattern registered by the registration means and an existing pattern in the character recognition dictionary is determined based on an evaluation value obtained when character-recognizing the registered pattern. The character recognition device according to claim 4 or 5. 7. The correcting means, when the pattern searched by the searching means is listed as a character recognition candidate in the character recognition processing, changes the candidate priority information in the character recognition dictionary related to the pattern. 7. The character recognition device according to claim 4, wherein the character recognition device is a character recognition device. 8. A character recognition device for character-recognizing and inputting handwritten characters by collating a pattern of a coordinate point sequence input from coordinate input means with a pattern in a character recognition dictionary, wherein the coordinate point sequence In the dictionary creating method for creating a dictionary by registering the pattern of the above in the character recognition dictionary, the pattern of the coordinate point sequence input from the coordinate input means and the character code corresponding to this pattern are registered in the character recognition dictionary. A method of creating a dictionary, characterized in that, when performing the search, an existing pattern similar to the registered pattern is searched from the character recognition dictionary, and candidate priority information in the character recognition dictionary of the searched pattern is changed.