JPH0836571A - Document processor - Google Patents

Abstract

PURPOSE:To convert the character string mixing KANJI and KANA by pen input and the converted character string mixing KANJI and KANA without using a large-scale dictionary registering the all combinations of KANJI (Chinese character) and KANA (Japanese syllabic writing). CONSTITUTION:A character string 1 to be converted mixing KANJI and KANA is converted into a middle KANA character group 3 being all KANAs by a KANJI/KANA conversion 2. Then, it is converted by a KANA/KANJI conversion 4 to generate a primary candidate character string group 5. The primary candidate character string group and the character string 1 to be converted are collated by collation processing 6. The ones fulfilling the condition is made the last candidate character string group 7.

Description

Detailed Description of the Invention

[0001]

The present invention relates to a word processor having a pen input function, a computer, a personal digital assistant, etc.
With a handwritten character input and handwritten character recognition function, a Japanese sentence processing device that can directly input kanji, or a function that can convert the once input Japanese text afterwards with a general computer and word processor. The present invention relates to a document processing device.

[0002]

2. Description of the Related Art Currently, there are many word processors, computers and the like having a pen input function using a pen or the like. FIG. 9 shows an example of a portable information terminal having a pen input function. The display unit A is a display-integrated tablet. Input can be made on the display unit A by an input pen B which is an input means. FIG. 10 is an example of a screen during input. By inputting with the pen B in the entry frame C, the kanji, kana, alphanumeric characters, etc. input by the pen are recognized, and the recognized characters are displayed on the input line D. The input box may be written in any of the boxes (C1 to C4), and the recognized characters disappear from the entry box. Therefore, it is sufficient to sequentially enter in the empty boxes, and the characters are displayed in the input line in the order of writing.

In the conventional keyboard input method, conversion from hiragana to kanji is performed, but in the pen input method, kanji can also be directly input. However, it is not possible to input mixed Kanji characters or convert the text of Kana-Kanji clauses that have already been converted. If the conversion separately for each before and after the kana part kanji, for example, type "Oyakeen" If converted part of the "ene public" Hiragana portion,
The conversion accuracy is very poor because the conversion unit is short and candidates that have nothing to do with the immediately preceding character "Kou" are also selected.

Japanese Patent Publication No. 4-35785 discloses a method for converting an input kana-kanji character string by using a dedicated dictionary storing kanji corresponding to each entry word in which kanji and kana are mixed. There is. An example of a dictionary used in this method is shown in FIG. The kana-kanji conversion dictionary uses a dictionary that describes character strings that also include kanji as headwords. However, regarding all the headwords for "Japanese", considering all combinations of kanji and kana, "Japanese,""Japanese,""Nihongo,""Japanese,""Japanese,""Japanese," and "Nihongo" 7 words are needed. Assuming that the number of characters in the Kanji string that is the conversion result is R, the number of headwords is 2 ^R -1, which is a huge amount. For this reason, it is necessary to prepare a fairly large dictionary, which causes problems such as search efficiency and maintenance load.

[0005]

As an input method other than the pen input, there is generally a keyboard input method.
As a means of inputting Japanese sentences in kana and kanji, enter the input sentence in hiragana from the keyboard, and then click [Convert].
There was a method to convert the Hiragana character string into Kanji by operating the keys.

In the pen input method, since it is possible to directly input Chinese characters, for example, when a character string "park" is desired to be input, "park" can be directly input and recognized in the above-mentioned entry box C. Although it can be done, actually, there are many strokes for the Chinese character "en", and it is very convenient for the user to input "Kouen" and make a conversion request, which will be converted to "park". In addition, even in a device using a general input method, there may be a case in which a part or all of the Japanese text that has already been input is desired to be converted into kanji again. O
There may be a case where it is desired to reconvert a part of a sentence input from a CR (Optical Character Reader). For that purpose, it is necessary to make it possible to perform conversion from a character string that is mixed with kana and kanji, which also includes kanji, rather than the conventional conversion target is a character string that consists only of kana.

In Japanese Examined Patent Publication No. 4-35785, as a method for converting a character string containing kana-kanji characters, a method of using a dictionary in which a character string including kanji in the entry word of a conventional kana-kanji conversion dictionary is also devised. There is. However, this method results in a very large number of dictionary items. Also,
As can be seen from the above example, while all possible combinations result in a dictionary that contains many headwords that are rarely referenced (such as "nihongogo"), If the registration is limited to the headwords that are likely to occur, it is not easy to draw the boundary line. Creating such a dictionary would be very burdensome, and it would take a lot of effort to make it consistent with other dictionaries in updating the dictionary and registering words by the user. .

According to the present invention, it is not necessary to have such a huge dictionary, and a character string containing Chinese characters including Chinese characters is once converted into a character string containing only Hiragana characters. Convert a string.

[0009]

According to the first aspect of the present invention, kana-kana conversion means for converting a kana-kanji magic character string that is a conversion target character string into a character string not containing kanji to obtain an intermediate kana character string, The intermediate kana character string is converted to a kana-kanji mixed character string to obtain a primary candidate character string, and the primary candidate character string and the conversion target character string are collated.

In the second aspect of the present invention, the partial character string in which Chinese characters are continuous in the conversion target character string is stored as it is in the primary candidate character string, and the partial character string in which Kana is continuous in the conversion target character string. Is provided with a collating unit that determines a final candidate character string when it corresponds to any non-empty character string in the primary candidate character string.

[0011]

According to the first aspect of the present invention, the kanji-kana conversion means converts the input character string, which is a mixture of kanji and kana characters, into a character string containing only kana characters. Here, the intermediate kana character string is obtained. By kana-kanji conversion means,
Converts to a kana-kanji character string. Here, the primary candidate character string is obtained.

According to the second aspect of the present invention, the collating means collates the conversion target character string with each of the primary candidate character strings, and the Chinese character included in the conversion target character string in the primary candidate character string is in the correct position. In addition, except those appearing in the same order, the remaining ones are output and displayed to the user as correct answer candidates.

[0013]

1 is a block diagram showing the flow of processing according to the present invention. By converting the conversion target character string 1 into the kana-kanji conversion 2, the intermediate kana character string group 3 is obtained. A primary candidate character string group 5 is obtained by adding a process by the kana-kanji conversion 4 to the intermediate kana character string group 3. The final candidate character string group 7 is obtained when the primary candidate character string group 5 and the conversion target character string 1 are matched by the matching process 6.

A block diagram of one embodiment of the present invention is shown in FIG. 8 is a tablet, 9 is a character recognition circuit, 10 is a conversion target character string buffer, 11 is a kanji kana conversion circuit, 12 is a kanji kana conversion circuit, 13 is an intermediate kana character string buffer, 1
4 is a kana-kanji conversion circuit, 15 is a kana-kanji conversion dictionary, 1
6 is a primary candidate character string buffer, 17 is a character string collating means,
18 is a final candidate character string buffer, 19 is a display circuit, 20
Is a display device, and 21 is a conversion process control circuit.

Each of the above blocks will be described in detail based on the processing. As shown in FIG. 9A, the tablet 8 normally also serves as a display unit. A character is input using the pen B into the entry area C in FIG. The input character is recognized by the character recognition circuit 9. In the character recognition circuit 9,
The input character is recognized using information such as image information of the input character and stroke information (writing brush) of the input input. The recognized character string is first stored in the conversion target character string buffer 10. The contents of the conversion target character string buffer are converted into a character string containing only kana by the kanji kana conversion circuit 11 using the kanji kana conversion dictionary 12 and stored in the intermediate character string buffer 13.

The character string containing only the kana stored in the intermediate kana character string buffer 13 is converted into the kana-kanji conversion dictionary 15.
The kana-kanji conversion circuit 14 using the converted data is converted into a kanji character string and stored in the primary candidate character string buffer 16. The contents of the conversion target character string buffer 10 and the contents of the primary candidate character string buffer 16 are compared by the collation circuit 17, and only when it is determined that the contents of both buffers do not conflict, the contents of the primary candidate character string buffer 16 are stored. The contents are copied to the final candidate character string buffer 18.

The character string obtained in the final candidate character string buffer 18 is displayed on the display device 21 through the display circuit 20 as a candidate of the conversion result, is output to the user, is displayed, and is confirmed when the user approves. If not, the next primary candidate character string or intermediate kana character string is obtained, and the same processing is repeated.

The activation of conversion processing and the presentation of candidates are input from the tablet 8 and passed to the conversion process control circuit 21 as a conversion command recognized by the character recognition circuit 9,
The conversion process control circuit 21 advances the processing based on the flowchart shown in FIG. One embodiment will be described in detail with reference to FIGS. 3 and 4. Here as a real example,
It is assumed that the character string “Kouen” is input from the tablet 8, recognized by the character recognition circuit 9, and converted.

First, in step 22, the character string to be converted is S. S is a conversion target character string buffer. That is, “public end” is stored in S here. Step 2
At 3, the empty set is substituted for the final candidate character string buffer RR to clear it. In step 24, S is converted into kanji and kana. In other words, kanji and kana conversion is performed on "Kouen". The resulting set of character strings is stored as II. Here, “Kouen” and “Kimien” are stored as II. This is the intermediate kana character string buffer. Step 2
In step 5, one character string of the element is extracted from II and set as I. In other words, "koen" is taken out as I. If there is no element to be taken out, the processing of step 30 is performed. In step 26, Kana-Kanji conversion is applied to I to obtain the resulting character string and CC. In the example, “park”, “performance”, “support”, “lecture”, and “takato” are obtained as CC. This is the primary candidate character string buffer. In step 27, one of the elements of CC is taken out and set as C. if,
If there is no element, the process returns to step 25. When all the elements of II are taken out in step 25, the process proceeds to step 30. In step 28, C and S are collated. If C and S are inconsistent, the process returns to step 27. If there is no contradiction, in step 29, C is added as an element of the set RR (final candidate character string buffer), and then step 2
Return to 7. That is, in steps 27 and 28, the CC elements "park", "performance", "support", "lecture", and "high distance" are compared, and "park" and "performance" that do not contradict S are stored in the final candidate character string buffer RR. Stored, return to step 25, take out "Kimien" as I, step 26
Apply Kana-Kanji conversion in. However, if the kana-kanji conversion is applied to "Kimien", no candidate can be obtained. Therefore, the process returns to step 25. II in step 25
Since there is no I to be taken out, the processing of step 30 is performed. In step 30, one character string of the element is taken out from RR and set as R. If there is no element, the process ends in step 31 with no candidate. Step 3
In step 2, R is displayed to the user. In step 33, if the user adopts R, the process ends, and if not, the process returns to step 30. In other words, "park" and "performance" were displayed as candidates.

The kanji / kana conversion in step 24 is a function that takes a character string S as an argument and returns a set of candidate character strings converted into kana only. The kana-kanji conversion in step 26 is a function that takes a character string I containing only kana as an argument and returns a set of character strings converted into a kanji-mixed character string.

In step 28, C and S are collated to determine whether there is a contradiction. The determination method will be described in detail with reference to FIG. In the above example, if the conversion target character string S is “Kouen”, the primary candidate character string group CC is ““ park ”“ performance ”.
It is assumed that "support", "lecture", and "high distance" are obtained. In "park" and "park", the kanji "Kou" appears at the same position as the conversion target character string S, and the kana "en" part is replaced with some kanji, while "support""lecture"" For "Takato", the kanji "Kou" does not appear, so it is clear that it is different from the character string that the user tried to input. As a result, “park” and “performance” remain as the final candidate character strings, and “support”, “lecture” and “takato” are rejected.

As another example, it is assumed that "higo" is converted as the character string to be converted. "Go to S"
Is stored and Kanji and Kana conversion is performed. As a result of the kanji kana conversion, the I as an intermediate pseudonym string "shed-out this"
"-Out shed industry" "Ki has fire" (underlined the portion where the "line" has been converted) is obtained. By performing kana-kanji conversion for each of these intermediate kana character strings, "airplane", "delinquency period", "departure period", "pension", etc. can be obtained as the primary candidate character string C. Out of these, the "training period", "pension", etc., which do not include the kanji "line" included in the original conversion target character string, are excluded, and the remaining "airplane" and "delinquency period" are "higo".
Some kanji appear in the parts corresponding to "hi" and "ki", so these will be presented as candidates for correct answers.

In the above collation processing, the following method is used to determine whether or not to leave the primary candidate character string as the final candidate. The conversion target character string is divided into a partial character string in which only kana is continuous and a partial character string in which only kanji is continuous.

S = S1 & S2 & ... & Sn where NK (Si) [when i is an odd number] and K (Si) [when i is an even number] or K (Si) [when i is an odd number] NK (Si) It is represented by [when i is an even number]. Here, S is a character string to be converted, Si (i = 1,2, ...
,, n; n is a natural number), & is a concatenation operator of the character strings, K (X) is a predicate that the character string X contains only Kanji, and NK (X) is a character string Let X be a predicate indicating that it contains only kana. Such division is always uniquely determined.

That is, as an example, when "Higo" is a character string to be converted, S = [Hikou], S1 is [Hi] which is a partial character string in which kana is continuous, and S2 is Kanji. Is a continuous partial character string [row], and S3 is similarly a partial character string in which a kana is continuous [ki]. When the conversion target character string S is [Kouen], S1 is a partial character string in which Chinese characters are continuous [Kou], and S2 is a partial character string in which Kana is continuous [En].

At this time, if the primary candidate character string can be divided into partial character strings satisfying the following conditions, this character string does not contradict the conversion target character string and is left as the final candidate character string. To do. Such a primary candidate character string that cannot be divided is rejected and excluded from the final candidate character string.

T = T1 & T2 & ... & Tn (n is the number of partial character strings into which the character string S to be converted is divided) where Ti is an arbitrary character string [when i is an odd number] and Ti = Si and K (Ti) [when i is even] or Ti = Si and K (Ti) [when i is odd] and Ti is any non-empty character string [when i is even] where T is a primary candidate character Column, Ti (i = 1,2, ..., n; n
Is a natural number) represents the substring.

For example, the conversion target character string S is "Kouen".
If so, S1 = “public” and S2 = “en” can be uniquely divided (n = 2). If the primary candidate character string for this is “park”, it can be divided into T1 = “public” = S1 and T2 = “garden”, and the same is true for “performance”, but “lecture” and “support”
Etc. are excluded from the final candidate character string because there is no division in which T1 = S1 and T2 is not empty.

Further, if the conversion target character string S is “higo”, it can be divided into S1 = “hi”, S2 = “row”, and S3 = “ki”. If the primary candidate character string for this is “airplane”, it can be divided into T1 = “fly”, T2 = “line”, T3 = “machine”, and can also be divided in the “delinquency period”. However, for “training period” and “preferential”, T1 and T3 are not empty, and T2 = S
There is no division that results in 2.

That is, as shown in FIG. 6, a partial character string of continuous Kanji in the conversion target character string is stored as it is in the primary candidate character string, and a part of continuous kana in the conversion target character string. If the character string corresponds to any non-empty character string in the primary candidate character string, that primary candidate character string is left as the final candidate character string only then.

Further, in the above embodiment, when performing kanji / kana conversion and kana / kanji conversion, all candidates are stored in a buffer to generate a primary candidate character string, and the final candidate character string is presented by the collating means. Was there. The following embodiment is an embodiment which does not require a large buffer. Description will be made with reference to FIG. 7.

First, in step 35, the character string to be converted is S. Next, in step 23, Kanji-kana conversion is applied to S, and the result is set to I. If I is an empty character string in step 24, in step 25 there is no candidate and the process ends. If I is not an empty character string, then Kana-Kanji conversion is applied to I in step 26 and the result is set to C. If it is an empty character string in step 27, the process returns to step 23, the next candidate is set to I, and the process is continued. If it is not an empty character string, in step 28, S (conversion target character string buffer) is compared with C (primary candidate character string buffer). If there is a contradiction due to the collation, the process returns to step 23, the next candidate is set to I, and the process is continued. If there is no contradiction, the contents of C (primary candidate character string buffer) are substituted into R (final candidate character string buffer) in step 29, and presented to the user in step 30. If the user adopts it in step 31, the process ends in step 32. If the user requests the next candidate without adopting it, the process returns to step 23. It is also possible to repeat the backtrack as described above to generate candidates.

[0033]

EFFECTS OF THE INVENTION In a device that can input and recognize handwritten input characters, since Chinese characters can be directly input, a character string partially containing Chinese characters can be input. However, if you want to convert a part of the character string that was input with a kana character due to complexity or difficulty of input, or if you want to convert the already input Japanese sentence again, the input character string By effectively utilizing the information on the Chinese characters and their positions included in, it is possible to eliminate unnecessary candidates and improve the conversion accuracy.

A dictionary for converting a kanji kana-magic character string into a kana character string can be generally used for text-to-speech synthesis, generation of reading for sorting, etc., and it is not necessary to prepare a special dedicated dictionary. The conventional technique can be used as it is for the kana-kanji conversion. In this way, since it is a combination of general-purpose software technologies, it is easy to replace individual element technologies.

[Brief description of drawings]

FIG. 1 is a flowchart showing a flow of processing of the present invention.

FIG. 2 is a block diagram of a first embodiment according to the present invention.

FIG. 3 is a flowchart of Embodiment 1 of the present invention.

FIG. 4 is a diagram showing a conversion process in the embodiment of the present invention.

FIG. 5 is a diagram showing an example of an intermediate kana character string group, a primary candidate character string group, and a final candidate character string group for an input character string example in the present invention.

FIG. 6 is a diagram showing an example of conditions to be left as a final candidate character string by the matching process.

FIG. 7 is a flowchart of a second embodiment of the present invention.

FIG. 8 is a diagram showing an example of a dictionary in a conventional example.

FIG. 9 is a diagram showing a portable information terminal having a pen input function.

FIG. 10 is an example of a screen diagram in a handwriting input mode with a pen.

[Explanation of symbols]

 1 conversion target character string 2 kana kanji character conversion process 3 intermediate kana character string group 4 kana kanji character conversion process 5 primary candidate character string group 6 collation process 7 final candidate character string group 8 tablet 9 character recognition circuit 10 conversion target character string buffer 11 kanji Kana conversion circuit 12 Kanji kana conversion dictionary 13 Intermediate kana character string buffer 14 Kana kanji conversion circuit 15 Kana kanji conversion dictionary 16 Primary candidate character string buffer 17 Character string matching circuit 18 Final candidate character string buffer 19 Display circuit 20 Display device 21 Conversion process control circuit

Claims (2)

[Claims] 1. A kana-kana conversion means for converting a kana-kanji character string into a character string not containing kanji, and a kana character kanji character for converting a character string not containing kanji obtained by the means into a kana-kanji character string. A document processing apparatus, comprising: a conversion unit; and a matching unit that compares the Kana-Kanji character string that is the conversion target with the Kana-Kanji character string that is obtained by the conversion process. 2. A kana-kana conversion means for converting a kana-kanji kanji character string which is a conversion target character string into a character string not containing kanji and obtaining an intermediate kana character string, and the intermediate kana character string kana-kanji character kanji Kana-Kanji conversion means for converting to a string to obtain a primary candidate character string, and collating the primary candidate character string with the conversion target character string, and a partial character string of consecutive Kanji in the conversion target character string is the primary candidate character as it is. A collation means for storing a partial character string that is stored in a column and has consecutive kana characters in the conversion target character string as a final candidate character string when it corresponds to any non-empty character string in the primary candidate character string. A document processing apparatus comprising: