JPH06314273A - Document preparing device and candidate output order control method - Google Patents

Abstract

(57) [Summary] [Object] The present invention aims to improve the conversion rate of kanji and kana mixed kanji. According to the present invention, a kanji / kanji / kanji conversion unit 5 converts kanji / kanji / kanji into a character string input to a conversion buffer 7, and stores the obtained conversion candidates in a conversion candidate buffer 11 with priorities. The input kanji character examining unit 6 examines the kanji in the kanji notation input history using the character string in the conversion buffer 7. The kanji notation learning unit 12 uses the conversion buffer 7
The input character string in the above is compared with the conversion candidate character string selected by the operator, and the kanji input in kanji notation is stored in the kanji bitmap 14. The candidate kanji character checking unit 10 finds a conversion candidate including the Kanji stored in the Kanji bit map 14 in the candidate buffer 11 and informs the conversion candidate control unit 9. When the conversion candidate including the Chinese character is written in Hiragana, the conversion candidate control unit 9 controls the output priority of the conversion candidate to be lowered.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention inputs a kana-mixed kana character string with a stylus pen, performs character recognition processing, and performs kana-kanji kana-kanji conversion using this as input character information to create a kanji-mixed document. The present invention relates to a document creation device, and more particularly to adjustment processing of output rank of conversion candidates.

[0002]

2. Description of the Related Art Conventionally, in this type of document creating apparatus, the output priority of a plurality of conversion candidates obtained by converting an input kana character string into kana / kana mixed kana / kanji character is the number of connection points between phrases and word usage frequency at the time of conversion. It was decided by learning information. By the way, in a document creating apparatus such as a handwriting word processor having an online handwriting character recognition means and capable of inputting kana mixed with kanji, in general, simple kanji is often input in kanji notation rather than kana notation. Therefore, in the opposite case, that is, when the conversion result input in hiragana includes a kanji that is normally input in kanji as a candidate, the candidate is often different from the operator's intended candidate, and the correct conversion candidate is selected. In order to do so, it was necessary to perform candidate selection processing. For example, if you enter hiragana as “Ki About” and there is a conversion candidate “About Tree” as the conversion candidate, and if “Tree” is entered in Kanji, the operator intends “Tree” as a candidate. It tends to be absent. In such a case, the operator has to perform an operation of selecting, for example, "ki" as the next candidate.

[0003]

SUMMARY OF THE INVENTION In a conventional document creating apparatus for recognizing a kana character string input by a stylus pen or the like on-line handwritten characters and converting the kana character into kana mixed kana kanji, the conversion result input in hiragana by the operator. When the kanji input in the normal kanji notation is included as a candidate in, the candidate is often different from the candidate intended by the operator, and it is necessary to perform candidate selection processing to select the correct conversion candidate. There was a drawback that the conversion rate became worse.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to eliminate the above-mentioned drawbacks and to provide a document creating apparatus capable of performing kanji and kana / kanji mixed kanji conversion as intended by an operator and improving the conversion rate. There is.

[0005]

According to the present invention, there is provided an on-line handwritten character recognizing means, wherein the recognized kanji-mixed kana character information is used as input character information, and kanji-mixed kana-kanji conversion is performed on the input character information. A learning means for comparing character information and character information of conversion candidates selected after conversion of kanji mixed with kanji and storing the character information input in kanji notation at the time of input, and conversion obtained by performing kana-kanji conversion with kanji mixed First determining means for determining whether or not there is a kanji stored in the learning means in the candidate character information, and the learning means in the conversion candidate character information by the first determining means. Second judgment means for judging whether or not the input character information to be converted includes the Chinese character when it is determined that the Chinese character stored in the second character exists; Having the configuration and a priority adjustment means for reducing the output priority of conversion candidates the Chinese characters comprising the kanji if not included in the converted character information by the decision means.

[0006]

In the document creating apparatus of the present invention, the learning means compares the input character string with the selected character string of the conversion candidate and stores the input character in Kanji notation. The first determining means determines whether or not the kanji stored in the learning means exists in the conversion candidate character string obtained by performing the kanji mixed kanji conversion. The second determination means is the first
When it is determined that the kanji stored in the learning means is present in the conversion candidate character string,
At this time, it is determined whether the input character string to be converted contains the Chinese character. The priority adjusting unit lowers the output priority of the conversion candidate including the Chinese character when it is determined by the second determining unit that the Chinese character is not included in the conversion target character string.

[0007]

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 creating apparatus of the present invention. Reference numeral 1 denotes an input device such as a handwritten transparent tablet that performs online handwritten character recognition processing when a user inputs a character string with a pen and outputs a corresponding character code, which is superimposed on a display device 18 described later. Is located so that the displayed information can be seen. Two
Is an input control unit that sorts input character strings and setting data into necessary parts, 3 is an input buffer that temporarily stores the input character strings, and 4 is a format control unit that sets the format of the input document. Kanji-mixed kana-kanji conversion unit that converts kanji-mixed kana-kana strings to kanji-mixed kana-kanji, 6 is an input research part that checks whether or not the kanji code of the conversion candidate has been input in kanji notation in the past, and 7 is the kanji to be converted A conversion buffer in which mixed kana character strings are once stored, 8 a kanji mixed kanji kanji conversion dictionary, 9 a conversion candidate control unit for performing various processes for selecting conversion candidates obtained by kanji mixed kana kanji conversion, 10 Candidate Kanji Research Department that investigates whether or not there is a Kanji code in the character strings of all conversion candidates, and if so, 11 is a Kanji conversion for mixed Kanji characters. The conversion candidate buffer that temporarily stores the obtained conversion candidates, 12 is a kanji notation learning unit that investigates whether the kanji in the selected conversion candidate is the kanji / kana notation in the input state, and 13 is each of the selection candidates. A notation type counter that counts whether the kanji was written as kanji or kana in the input state, 14 is a kanji bitmap that stores the input kanji in the kanji notation, and 15 is kanji mixed kana kanji conversion A document buffer for storing the confirmed character code, 16 a display control unit for displaying the display data expanded in the display buffer 16 on the display device 18, and 17 expanding the display data for display on the display device 18. A display buffer, 18 is a display device such as a CRT or LCD for displaying documents, figures and the like. This display device 18
Has a structure that is integrally overlapped with the input device 1.

Next, the operation of this embodiment will be described. Usually, an operator writes a kana-mixed kana character string on the input device 1 such as a handwriting tablet with a pen or the like, and an online handwritten character recognition unit (not shown) performs character recognition to generate a recognized character string and document it. As input to the device. Kanji mixed kana character input / conversion is an important input / conversion means in a document creation device having a pen input means. Kanji with a small number of strokes can be input as it is, and kanji with a large number of strokes can be input with its reading, It can convert kana-kanji into characters, and its input and conversion speeds are higher than those of a document creation device having a conventional pen input means (only kana character input / conversion means). At this time, the operator refers to various messages displayed on the display device 18 and interactively proceeds with the document creation work.
The character string input from the input device 1 is temporarily stored in the input buffer 3 through the input control unit 2 while waiting for the conversion process, and then is again input from the input control unit 2 into the kanji / kanji conversion unit 5.
Sent to. The kanji / kanji / kanji conversion unit 5 temporarily stores the input character string to be converted into the conversion buffer 7 until the input character string has an appropriate length for forming a word, a phrase, or the like. At the stage when the length of the character string stored in the conversion buffer 7 is separated by the detection of the punctuation character code, that is, the length of the conversion target, the kanji-mixed kana-kanji conversion unit 5 converts kanji-mixed kana-kanji With reference to the dictionary 9, the kana-kanji conversion processing for the conversion target character string is started.

At the same time as the conversion of the Kanji-mixed Kana-Kanji conversion unit 5 is started, the input Kanji-character inspection unit 6 investigates the character string in the conversion buffer 7 to be converted, and stores it in the Kanji bitmap 14 in this character string. Check whether or not there is a kanji code that is displayed, and if there is, check its position. Here, the kanji character bitmap 14 is for inputting at least on / off information corresponding to a kanji code, and can represent a stored kanji in the on state. That is, if there is a kanji code that has a history of being directly input from the input device 1 in kanji notation, or if it has a notation type counter, the kanji bit map 14 corresponding to the kanji code for which the proportion input in kanji notation is higher than a certain value. The upper bit is turned on. Further, such a Kanji bit map 14 may be capable of inputting not only two levels of on / off but three or more levels.

When the kanji-mixed kana-kanji conversion process is completed, the kanji-mixed kana-kanji conversion unit 5 passes all conversion candidates to which the output priority is added based on the number of connection points between phrases and the frequency of word use through the conversion candidate control unit 9. 1
Store in 1. Next, the candidate kanji character checking unit 10 checks whether or not all the conversion candidate character strings stored in the conversion candidate buffer 11 include the kanji code stored in the kanji bitmap 14, and if there is, Investigate its location. When the candidate kanji survey unit 10 finds a Kanji code stored in the Kanji bitmap 14 in the conversion candidates, the conversion candidate control unit 9 inputs the found Kanji code from the survey result of the input Kanji survey unit 6 in Kanji notation. Check whether it has been done. As a result, when the conversion candidate control unit 9 finds that the character corresponding to the Kanji code is input in Kana notation, it outputs the output priority of the conversion candidate including the Kanji candidate stored in the conversion candidate buffer 11. Lower.
After the adjustment of the output priority is completed, the conversion candidate control unit 9 sends the conversion candidate having the highest output priority to the display control unit 16 and displays it on the display device 18, thereby prompting the operator to select the conversion candidate. As a result, the candidates selected by the operator are sent to the kanji writing learning unit 12. This allows
The kanji display learning unit 12 compares the selection candidate character string with the input character string stored in the conversion buffer 7, investigates whether each kanji of the selection candidate is a kanji or kana notation in the input state, and The survey result is set in the corresponding kanji character counter of the notation type counter 13. In addition, Kanji notation learning unit 1
2 indicates that the value set in the notation type counter 13 indicates that a certain kanji is input in kanji notation at a certain ratio or more relative to kana notation input, and the kanji bitmap corresponding to such kanji code. 14 is turned on, and conversely, when it is below a certain value, the kanji bitmap 14 corresponding to the kanji code is turned off. At the same time, the conversion candidate control unit 9 stores the character code of the selected conversion candidate in the document buffer 15 and displays it via the display control unit 16 in accordance with the format preset in the format control unit 4. Expand the character code in the buffer 17,
It is displayed on the display device 18. When the operator starts inputting the next character without performing the selection process, the same process as above is performed assuming that the displayed first candidate is selected.

FIG. 2 is a flow chart showing the flow of processing from the input of a character string to the selection of conversion candidates in the apparatus shown in FIG. The character string input from the input device 1 in step 201 is temporarily stored in the input buffer 3 by the input controller 2, and then transferred in step 202 to the conversion buffer 7 through the kanji / kanji conversion unit 5. The kana / kanji / kanji conversion unit 5 monitors in step 203 whether or not the character string input to the conversion buffer 7 has an appropriate length such as a word or a phrase, and when the appropriate length is reached, the step is performed. Proceed to 204. When the character string in the conversion buffer 7 is to be converted, the character input from the input control unit 2 is stopped until the conversion process is completed. When all the character strings to be converted are stored in the conversion buffer 7, the input Kanji character checking unit 6 refers to the Kanji bitmap 14 in step 204 to display the Kanji notation in the character string in the conversion buffer 7. It is checked whether or not there is a kanji with an input history, that is, a kanji for which the kanji bitmap 14 is on, and if there is, the character position is stored. on the other hand,
In step 205, the kanji-mixed kana-kanji conversion unit 5 converts the character string in the conversion buffer 7 into kanji-mixed kana-kanji by referring to the kanji-mixed kana-kanji conversion dictionary 8. In step 206, the kanji-mixed kana-kanji conversion unit 5 stores the conversion candidates obtained by the kanji-mixed kana-kanji conversion in the conversion candidate buffer 11 together with the output priority obtained from the number of connection points between phrases and the word usage frequency.

Next, the candidate kanji character investigation unit 10 performs step 207.
In order to sequentially process all conversion candidates, the counter i is initialized. Candidate Kanji survey unit 10 is step 2
At 08, referring to the Kanji bitmap 14, it is checked whether or not there is a Kanji for which an input history in Kanji notation exists in the character string of the i-th conversion candidate in the conversion candidate buffer 11, and if it exists, Advances to step 209, where it is checked whether or not the corresponding Chinese character is a Hiragana character in the input character string by referring to the result of the investigation by the input Chinese character research unit 6. If it is Hiragana, the conversion candidate control unit 9 performs step 210.
At, the output priority of the i-th conversion candidate in the conversion candidate buffer 11 is lowered. After that, the candidate kanji character research unit 10 advances the counter i by 1 in step 211, and then proceeds to step 2
In step 12, it is determined whether or not the above steps 208 to 210 have been performed for all conversion candidates in the conversion candidate buffer 11, and if so, the process proceeds to step 213. When the review of the output priorities for all conversion candidates is completed by the above-described processing, the conversion candidate control unit 9 displays the conversion candidate with the highest output priority on the screen of the display device 18 via the display control unit 16. After displaying in step 213, the operator waits for candidate selection processing in step 214.

After that, the candidate character string selected by the operator is sent from the conversion candidate buffer 11 to the Kanji writing learning unit 12 by the conversion candidate control unit 9. In step 215, the kanji notation learning unit 12 checks whether all the characters which are the kanji notation of the sent candidate character string are kanji or hiragana at the time of input, and for each kanji notated character. The result is set in the notation type counter 13. Further, the kanji notation learning unit 12 uses the notation type counter 1
After setting 3, the notation type counter 13 is checked for the number of kanji notation and kana notation input counts, and if it is determined in step 216 that the proportion of kanji notation input is greater than or equal to a certain value, it corresponds in step 217. When the kanji bitmap 14 is turned on, that is, it is stored that the kanji is input in the normal kanji notation, and if it is determined in step 216 that the kanji is less than a certain value, the corresponding kanji bitmap 14 is set in step 218. It is stored in the OFF state, that is, it is a kanji that is normally input in hiragana notation. When the setting of the notation type counter 13 is completed in this way,
The kanji writing learning unit 12 transfers and stores the selection candidate character string in the document buffer 15 in step 219. The display control unit 16 causes the character string stored in the document buffer 15 to be displayed in the display buffer 17 according to the format preset in the format setting unit 4.
After unfolding up, this is displayed on the screen of the display device 18 in step 220. Further, the kanji / kanji / kanji conversion unit 5 initializes the conversion buffer 7 in step 221 in preparation for the next input / conversion.

FIG. 3 shows the Kanji bitmap 14 shown in FIG.
It is the figure which showed the structural example. In this example, only the simplest two-step on / off information can be input to the Chinese character bitmap, and one bit is assigned to one character in the character code order. That is, this Kanji bitmap 14
Then, the bit corresponding to the character input in Kanji notation is turned on (= 1) even once. In this case, "A"
It is shown that only and "evil" have a history of being directly written in Kanji at the input stage. Further, FIG. 4 shows an example in which the Kanji bitmap 14 has a two-stage configuration in which it is linked to the notation type counter 13, and shows an apparatus capable of counting the number of times Kanji / Kana notation has been performed. ing. The 0th bit in the first row of the Kanji bitmap 14 shown in FIG. 4A is information indicating the history of the direct Kanji notation shown in FIG. The 1st to fh bits are a numerical value indicating an offset pointer to a counter that stores the number of times of notation. As shown in FIG. 4 (B), the notation type counter 13 is located at the address portion indicated by the base address X + offset pointer, and the upper and lower 4 bits respectively indicate the number of Kanji notation and the number of Hiragana notation. As shown in FIG. 4, “A” is input eight times in the Kanji notation, and the number of times the Kanji notation is selected in the character string subjected to the conversion process after the input in the Hiragana notation is 0. The kanji notation learning unit 12 changes the value of the notation type counter 13 from the notation state of characters, and sets 1 in the kanji notation information when a preset ratio, for example, 90% or more kanji notation. Therefore, for example, even if there is one input in Kanji notation, if there are two inputs in Hiragana notation, the Kanji notation information becomes 0 because the value is below the set value.

FIG. 5 is a diagram showing an example of storage in the conversion candidate buffer 11 when "seikai de," is converted as an example sentence. FIG. 5 shows an example of storing the Kanji bitmap 14, and shows that “star”, “correct”, “raw”, “voice”, and “west” have been directly input in Kanji in the past. Figure 6
(A) is a diagram showing an example of data when a conversion candidate obtained when converting the example sentence “Seikai Ide” is stored in the conversion candidate buffer 11. The conversion candidates are stored in the conversion candidate buffer 11 with the output priority added together with the candidate character strings. If there is no effect from the Kanji bitmap 14,
The candidate with the highest output priority becomes the first candidate. However, in this example, as shown in FIG. 6 (A), it is stored in the Kanji bitmap 14 that “correct” has been input in kanji in the past, and the input sentence may not be “correct”. Since the input is “SEI”, the conversion candidate control unit 9 lowers the output priority of “correct answer” as shown in FIG. 6B, and as a result, “politics” is the first candidate. The output priority is changed so that In this example, the output priority is halved. However, depending on the ratio of the notation type stored in the notation type counter 13 or the value if at least three levels of information can be stored in the Kanji bitmap 14, The fluctuation range of the output priority can be changed.

According to this embodiment, it is memorized whether a certain kanji was in kana or kanji notation at the time of input, and then, in the conversion candidates obtained when kana-mixed kana kanji conversion is performed on the input character string. If it is found that a kanji input in the normal kanji notation exists by referring to the stored data and the kanji is the kana notation in the input character string, the conversion candidate of the kanji notation is selected. By lowering the output priority, it is possible to suppress the output of extra conversion candidates that the operator does not intend, and to increase the kanji conversion rate where kanji is mixed.

[0017]

As described above, according to the document creating apparatus of the present invention, kanji and kana mixed kanji conversion as intended by the operator can be performed, and the conversion rate can be improved.

[Brief description of drawings]

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

FIG. 2 is a flowchart showing a flow of processing from input of a character string of the device shown in FIG. 1 to selection of a conversion candidate.

FIG. 3 is a diagram showing an example of the structure of the Kanji bitmap shown in FIG.

FIG. 4 is a diagram showing a structural example of the Kanji bitmap shown in FIG. 1 and a notation type counter associated therewith.

FIG. 5 is a diagram showing another example of the structure of the Kanji bitmap shown in FIG.

FIG. 6 is a diagram showing an example of contents of a conversion candidate buffer shown in FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input device 2 ... Input control unit 3 ... Input buffer 4 ... Format control unit 5 ... Kanji mixed kana / Kanji conversion unit 6 ... Input Kanji research unit 7 ... Conversion buffer 8 ... Kanji mixed Kana / Kanji conversion dictionary 9 ... Conversion candidate control unit 10 ... Candidate Kanji research unit 11 ... Conversion candidate buffer 12 ... Kanji notation learning unit 13 ... Notation type counter 14 ... Kanji bitmap 15 ... Document buffer 16 ... Display control unit 17 ... Display buffer 18 ... Display device

Claims (2)

[Claims] 1. An online handwritten character recognition means is provided,
In the document creation device that performs the kana-mixed kana-kanji conversion on the recognized kanji-mixed kana-character information, compares the input character information with the conversion candidate character information selected after kanji-mixed kana-kanji conversion. A learning unit that stores the character information input in Kanji notation at the time of input, and whether or not the Kanji stored in the learning unit exists in the character information of the conversion candidates obtained by performing the Kana-mixed Kanji conversion. A first determining means for determining whether or not there is a Chinese character stored in the learning means in the character information of the conversion candidate by the first determining means, and Second determining means for determining whether or not the Chinese character is included in, and the Chinese character if the Chinese character is not included in the conversion target character information by the second determining means. A document creation apparatus comprising: a priority adjustment unit that lowers the output priority of conversion candidates including characters. 2. An online handwritten character recognition means is provided,
A method for creating a document in which kana-kana-kana character information that is recognized is used as input character information and kanji-kana kana-kanji conversion is performed to obtain conversion candidates, and characters that are normally input in kanji display are included in the conversion candidates, and A candidate output rank control method, wherein the output rank of the conversion candidates is lowered when the input character information is in Hiragana notation.