JP3876451B2 - Kana-kanji conversion device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a kana / kanji conversion device and a kana / kanji conversion method, and more particularly to a kana / kanji conversion device and a kana / kanji conversion method for capturing a character string including at least a kanji and converting it.
[0002]
[Prior art]
A kana-kanji conversion device created for the purpose of handling Japanese on a computer inputs a kana character string and converts it into a kanji-kana mixed sentence. On the other hand, in recent years, for the purpose of reconverting a character string that has already been confirmed, or for the purpose of reconverting a character string read by OCR or a character string read by handwriting recognition, Something that tries to convert columns has been proposed.
[0003]
In other words, if a conversion error is found in a character string that has been input to a word processor or the like, instead of inputting the kana character string again, a kana character string is created from the input and confirmed character string, It is used for reconversion.
[0004]
[Problems to be solved by the invention]
However, in the conventional kana-kanji conversion, the kana character string is obtained from the converted character string by referring to the word dictionary for reverse conversion (a dictionary having kanji headings and readings), so that the word is incomplete. There was a problem that re-conversion was not possible. For example, when recognizing handwritten characters, it may be possible to input “Higo” or “Ito-” instead of the trouble of inputting kanji with a large number of strokes for idioms such as “airplane” and “Ito”. If you try to convert these strings to the correct Japanese strings, you must convert these strings to kana strings once, but even if the words that have the heading "Airplane" are registered in the dictionary, Because the word “higo” is not registered, it invites a problem that it cannot be reconverted well.
[0005]
Moreover, even if it can be converted into a kana character string once, if the corresponding word is not found, the input has to be performed again from the beginning. For example, since various kanji characters can be applied to names and the like, there is a case where a desired kanji character cannot be obtained even if kana-kanji conversion is performed after the kanji character is converted back to kana once.
[0006]
The present invention has been made to solve the above problems, and an object of the present invention is to smoothly convert a kana-kanji character string including at least kanji characters.
[0007]
[Means for solving the problems and their functions and effects]
The kana-kanji conversion device of the present invention made to achieve such an object,
A kana-kanji conversion device that once converts a character string including at least kanji into a kana character string and then performs kana-kanji conversion again,
A character string input means for capturing a character string including the kanji,
First inverse conversion means for inversely converting the captured character string into a kana character string with reference to a first dictionary in which kanji headings and readings are recorded for independent words constituting a phrase;
If a kana character string is obtained by the reverse conversion, the kana character kanji is converted as a single phrase from the front of the kana character string and subjected to a connection test with the ending / attachment, and then a kana that outputs a single phrase conversion candidate Kanji conversion means,
If no conversion candidate is obtained by the kana-kanji conversion means, a character string including the kanji is converted back to a kana character string by referring to a second dictionary in which kanji headings and readings are recorded for single kanji. A second reverse conversion means for providing the reversely converted kana character string to the kana-kanji conversion means;
If a conversion candidate by the kana / kanji conversion means cannot be obtained based on the kana character string obtained by the second reverse conversion means, a single kanji character for performing single kanji conversion on the kana character string Conversion means;
When single kanji conversion is performed by the single kanji conversion means, a delimiter in which the position of the kanji included in the character string captured by the second reverse conversion means is the delimiter position of the kana character to be converted by the single kanji conversion means Positioning means and
It is a summary to provide.
[0008]
The kana-kanji conversion method of the present invention corresponding to the kana-kanji conversion device is realized by the computer once converting a character string including at least a kanji into a kana character string and then executing a process for performing kana-kanji conversion again. Kana-kanji conversion method to be performed,
A character string including the kanji is taken in via a means for inputting the character string,
The captured character string is converted back to a kana character string with reference to a first dictionary in which kanji headings and readings are recorded for independent words constituting a phrase,
If a kana character string is obtained by the reverse conversion, the kana character string is subjected to kana-kanji conversion means as a single phrase from the front of the kana character string and then subjected to a connection test with the ending / attachment, and then a single phrase conversion candidate Output
If a conversion candidate is not obtained by the conversion, the character string including the kanji character is inversely converted into a kana character string with reference to a second dictionary in which kanji headings and readings are recorded for the single kanji character, Providing the converted kana character string to the means for performing the kana-kanji conversion;
When the means for performing kana-kanji conversion fails to create a conversion candidate, the kana character string obtained by performing the reverse conversion is set with the position of the kanji included in the character string as the delimiter position of the kana character. Perform single kanji conversion
This is the gist.
[0009]
Such a kana-kanji conversion apparatus and method uses a first dictionary that records kanji headings and readings for independent words that constitute a phrase from a character string including kanji captured by OCR, pen input, keyboard input, or the like. It reverse-converts to a character string, performs kana-kanji conversion based on the obtained kana character string, and outputs a single phrase conversion candidate. If such a conversion candidate is not obtained, the second dictionary in which kanji headings and readings are recorded for a single kanji character is referred to, and the captured character string is inversely converted into a kana character string, which is converted into a kana kanji character. It serves as a means for performing conversion. Therefore, according to the kana-kanji conversion apparatus and method, if a word constituting the phrase is included in the target character string and then the reverse conversion and the kana-kanji conversion are performed first, then conversion candidates cannot be obtained (for example, kanji) For example, when a part of the idiom is input as a kana), a reverse conversion is performed with reference to the second dictionary for single kanji characters, so that a desired conversion candidate can be easily obtained. Moreover, in the present invention, when the kana-kanji conversion means cannot create a conversion candidate based on the kana character string obtained by the second inverse conversion means, a single kanji character is used for the kana character string. Single kanji conversion means for performing conversion is further provided, and further, delimiter position determining means for setting the position of the kanji included in the captured character string as the delimiter position of the kana character to be converted into single kanji characters. Therefore, kanji candidates can be output for character strings that are not registered in the dictionary for kana-kanji conversion, which is effective for conversion of human names. Since the break position for single kanji conversion can be determined, the conversion accuracy of single kanji conversion can be improved. For example, when the character string “Hi-go” is imported, it is only necessary to perform single-kanji conversion for “hi” and “ki” with “line” as the delimiter position, and “hiko” and the like need to be conversion candidates. Absent.
[0010]
A request to reconvert a character string that includes kanji is considered to occur for a relatively short range of character strings, so grammar analysis can be performed by starting reverse conversion and kana-kanji conversion assuming a single phrase. The conversion candidate can be obtained in a short time. In addition, when conversion candidates cannot be obtained, since kana-kanji conversion is performed after reverse conversion is performed for each single kanji, there is an advantage that a kana character string that is the basis of kana-kanji conversion is always obtained. If single kanji conversion is finally used, there is a high possibility that a desired conversion result can be obtained.
[0011]
The kana character string obtained by the second reverse conversion means may be subjected to kana-kanji conversion by kana-kanji conversion means for performing kana-kanji conversion of a single phrase, but other kana-kanji conversion, for example, single kanji It can be used for conversion, continuous phrase conversion or automatic conversion.
[0012]
In this kana-kanji conversion device, when kana-kanji conversion is performed for the kana-kanji character string obtained by referring to the second dictionary, the kana-kanji conversion means converts the kanji and kana-kanji conversion included in the captured character string. It is possible to provide conversion candidate restriction means for converting only those that match a kanji included in a single phrase obtained later. For example, when “Koen” is input, “park” and “performance” are candidates for conversion, but “lecture” and “support” are not candidates. As a result, the conversion accuracy in single phrase conversion is improved in each stage.
[0013]
Further, in this kana-kanji conversion device, when the kana-kanji conversion means cannot create a conversion candidate based on the kana character string obtained by the second reverse conversion means, Single kanji conversion means for performing single kanji conversion can be provided. In this case, kanji candidates can be output for character strings that are not registered in the kana-kanji conversion dictionary. This is effective when converting a person's name. Furthermore, it is possible to provide delimiter position determining means for setting the position of the kanji included in the captured character string as the delimiter position of the kana character to be converted into single kanji characters. In this case, since the break position for single-kanji conversion can be easily determined, the conversion accuracy of single-kanji conversion can be improved. For example, when the character string “Hi-go” is imported, it is only necessary to perform single-kanji conversion for “hi” and “ki” with “line” as the delimiter position, and “hiko” and the like need to be conversion candidates. Absent.
[0014]
When a conversion candidate by the kana / kanji conversion means cannot be obtained, and when finally converting the single kanji by the single kanji conversion means, it is immediately determined which conversion means the conversion candidate obtained has been converted by. It can be difficult to do. Therefore, it is also preferable to display the conversion candidate as a single phrase output by the kana / kanji conversion means and the conversion candidate as a single kanji output by the single kanji conversion means in a distinguishable manner.
[0015]
In order to obtain a kana character string from a character string including kanji, a reverse conversion dictionary that extracts kana from kanji is required. This dictionary may be prepared exclusively, but can be generated from a dictionary for kana-kanji conversion that converts a kana character string into a kanji-mixed sentence. When a new word is registered in kana-kanji conversion, the inverse conversion dictionary can be updated each time. In this case, one type of dictionary can be prepared basically.
[0016]
In the various kana-kanji conversion devices described above, the second dictionary can be generated from a dictionary for single-kanji conversion. In this case, it is not necessary to prepare a special dictionary, and the second dictionary can be easily constructed.
[0017]
In addition, when a kanji candidate is output by the kana-kanji conversion means, conversion candidates by single kanji conversion can be output in the order according to the frequency of use. In the case of single-kanji conversion, there may be a large number of conversion candidates, and the method used to output the conversion candidates tends to affect usability.
[0018]
Other aspects of the invention
As other aspects of the present invention, various aspects can be considered such that the input means is a handwriting recognition means, an optical character reader such as a scanner, and a speech recognition apparatus. Furthermore, this Kana-Kanji conversion device can be used alone as a word processor, or can be used by being incorporated in various devices that require Kana-Kanji conversion. Also, it can be realized in a form incorporated in an operating system of a computer. When it is realized by being incorporated as software in a device using a computer, it can be handled in the form of a flexible disk or CD-ROM on which this software is recorded, or can be distributed through personal computer communication or the like.
[0019]
For example, a portable storage medium that stores a software program that, once executed by a microprocessor of a computer system, converts a character string including at least kanji into a kana character string and then performs kana-kanji conversion again,
Import a character string containing the kanji,
The captured character string is converted back to a kana character string with reference to a dictionary,
Kana-kanji conversion of the kana character string obtained by the inverse conversion as a single phrase, and output a single phrase conversion candidate;
If the single phrase conversion candidate is not obtained, single kanji conversion is performed on the inversely converted kana character string, and a single kanji conversion candidate is output.
Anything that implements the function can be implemented.
[0020]
A software program (application program) for realizing the functions of these units is stored in a portable storage medium (portable storage medium) such as a floppy disk or a CD-ROM, and from the portable storage medium to the main memory of the computer system or externally. It is transferred to the storage device. A device for providing these software programs is provided via a communication line, and the software program for realizing the kana-kanji conversion method is transferred from the supply device to the main memory or external storage device of the computer system via the communication line. It may be transferred. For example, it is possible to download from a host computer for personal computer communication via a telephone line or to receive distribution using satellite broadcasting.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
Next, a preferred embodiment of a kana-kanji conversion apparatus according to the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram of a kana-kanji conversion apparatus according to the first embodiment, and FIG. 2 is a block diagram showing a schematic configuration of a computer on which the kana-kanji conversion apparatus is realized.
[0022]
For convenience of explanation, first, the hardware configuration of the computer 10 will be described with reference to FIG. As shown, the computer 10 includes an arithmetic processing unit 20 connected to a local bus 22, a PCI bridge 30 that connects the local bus 22 to a PCI bus 32 that is one of external buses, and an arithmetic via the PCI bus 32. The controller unit 40 that is accessed by the CPU 21 of the processing unit 20, the devices that control various I / O devices and the like are the I / O unit 60 connected to the ISA bus 42 that is a low-speed external bus, and peripheral devices. It consists of a keyboard 72, a speaker 74, a CRT 76, and the like.
[0023]
The arithmetic processing unit 20 is composed of a CPU 21 (using Pentium manufactured by Intel Corporation in the present embodiment), a cache memory 23, its cache controller 24, and a main memory 25 as a central processing unit. The PCI bridge 30 is a controller having a function of controlling a high-speed PCI bus 32. The memory space handled by the CPU 21 is expanded to a logical address wider than the actual physical address by various registers prepared in the CPU 21.
[0024]
The controller unit 40 includes a graphics controller (hereinafter referred to as VGA) 44 that controls display of an image on a monitor (CRT) 76, a SCSI controller 46 that controls data transfer with a connected SCSI device, a PCI bus 32, and a lower level It comprises a PCI-ISA bridge 48 that controls the interface with the ISA bus. The VGA 44 can display 640 × 480 dots and 16 colors with respect to the CRT 76. A character generator that stores display fonts, a graphic controller that receives a predetermined command and draws a predetermined graphic, and a video memory that stores a drawing image are mounted on the VGA 44. Since the configuration is well known, it will be described later if necessary, and is omitted in FIG.
[0025]
The ISA bus 42 connected via the PCI-ISA bridge 48 is a bus for input / output control to which various I / O devices are connected, and includes a DMA controller (hereinafter simply referred to as DMA) 50, a real-time clock (RTC). ) 52, two composite I / O ports 54 and 55, a sound I / O 56, a keyboard interface (hereinafter referred to as KEY) 64 for controlling the interface with the keyboard 72 and the mouse 73, and an interrupt controller for controlling interrupts with priority ( 66), a timer 68 for generating various time counts and beep sounds, and the like. The ISA bus 42 is connected to an ISA slot 62 in which an expansion board can be mounted.
[0026]
The composite I / O port 54 is provided with a port for inputting and outputting signals for controlling the floppy disk device 82 and the hard disk 84 in addition to parallel output and serial output. A printer 88 is connected to the parallel input / output via a parallel port 86, and a modem 92 is connected to the serial input / output via a serial port 90. The other composite I / O port 55 is connected to a scanner 93 and a tablet 94 capable of handwriting input. In addition to the speaker 74 described above, a microphone 96 can be connected to the sound I / O 56. In addition to these configurations, standardized I / O channels are often prepared in DOS / V machines, but illustration and description are omitted in this embodiment.
[0027]
Next, functions executed by the hardware thus configured will be described with reference to FIG. The configuration and operation of each unit shown in FIG. 1 will be outlined. The processing performed here is executed by the central processing unit (CPU 21) based on data input from the keyboard 72, scanner 93, or tablet 94. . All processing is performed by the CPU 21. With regard to kana-kanji conversion, when a character string is input from the keyboard 72 or the like and execution of the kana-kanji conversion is instructed by a predetermined operation, a predetermined interrupt process is started and the captured character string (kana character) A device driver that converts a string or a character string including kanji into a character string mixed with kana and kanji. Of course, in a computer capable of parallel processing, kana-kanji conversion may be performed by one application (input method), and the conversion result may be transferred to a required application. In this case, the input method collectively accepts input from the keyboard 72 or the like.
[0028]
As shown in FIG. 1, the kana-kanji conversion apparatus realized on the computer 10 inputs a character string through an input unit IPU. The input unit IPU basically captures a character string that has already been input as a document by specifying a range. However, the input unit IPU directly inputs a character string by a code, reads characters written on paper, and the like by character recognition. A scanner 93 and character recognition software for recognizing a character string, and a tablet 94 for inputting a handwritten character and recognizing it as a character string and handwritten character recognition software are included.
[0029]
The input character string may contain kanji, and the kanji-mixed part is converted into a kana character string by the first inverse conversion unit RC1 using the first dictionary stored in the hard disk 84. Converted. The first dictionary gives at least readings to kanji headings (words that can constitute a phrase). In the embodiment, as shown in FIG. 3, the first dictionary has a structure having part-of-speech information.
[0030]
Next, based on the kana character string obtained from the first reverse conversion unit RC1, the kana / kanji conversion unit KKC performs single phrase conversion. The single phrase conversion is a well-known one, and the conversion is performed on the assumption that a given kana character string is configured as one independent word + attached word. By limiting to single phrase conversion, morphological analysis for kana character strings can be simplified, and conversion candidates can be obtained by simple matching. Details of the processing will be described later. For the single phrase conversion, although not shown, an independent word dictionary and an attached word dictionary for kana-kanji conversion are used.
[0031]
When a conversion candidate is obtained by single phrase conversion, this is output, and the output conversion candidate is displayed in a predetermined area of the CRT 76. On the other hand, if the conversion candidate is not obtained, the second inverse conversion unit RC2 is activated next, and the second character string containing the reading and reading of the single kanji is read from the character string obtained by the input unit. Kana characters are written using the dictionary. As a result, even when a part of a word constituting the phrase and a part originally written in Kanji is kana, kana-kanji conversion is performed, and conversion candidates can be easily obtained. If a conversion candidate still cannot be obtained, a module for performing single-kanji conversion can be started, although not shown in FIG. In this case, a single kanji conversion dictionary (not shown) is used. With single-kanji conversion, conversion candidates can be shown for most character strings.
[0032]
A software program (application program) for realizing the functions of these units can be stored in a ROM so that it can be directly executed by the CPU 21, but a portable storage such as a flexible disk, MO, CD-ROM or the like. The program may be stored in a medium (portable storage medium), transferred from the portable storage medium to the main memory 25 or the hard disk 84 of the computer system, and executed. A device for providing these software programs is provided via a communication line, this device is connected to the communication line via a modem 92, and the software program is supplied from the supply device to the main memory 25 or the hard disk 84 of the computer system. It is good also as what is transferred to.
[0033]
Next, each processing routine will be described sequentially. 4 and 5 are flowcharts showing a re-conversion processing routine that takes in a character string including kanji and converts it again to kana-kanji. When an application program such as a word processor is being executed and a range of a character string that has already been input and confirmed is designated and a re-conversion is instructed, a program for kana-kanji conversion is started, and FIG. 4 and FIG. The process shown is started.
[0034]
When this routine is started, first, a process for acquiring a target character string is performed (step S100). The target character string is a character string that is passed to the kana-kanji conversion program of the present embodiment by specifying a range, and is a character string that is a target of re-conversion below. For example, as shown in FIG. 6 (A), it is assumed that the confirmed character string “Saga's is in the north of today” is acquired as the target character string. Next, a process for initializing the search character pointer and the number of searches is performed (step S110). First, a process for adding one search character is performed (step S120). As shown in FIG. 6B, since the initial value of the search character pointer is 0, the search character pointer is first set to the first character by performing the process of step S120 once.
[0035]
Next, a process of searching the reverse lookup independent word dictionary JD1 is performed (step S130). That is, the reverse lookup independent word dictionary JD1 corresponding to the first dictionary is searched for the character (“S” in FIG. 6) indicated by the search character pointer. This dictionary JD1 can be searched by kanji headings, and in the embodiment, its reading and part-of-speech information can be obtained. After the search, the search result is determined. The reverse lookup independent word dictionary JD1 is arranged in the order of kanji headings, and one word of the same kanji heading is read out in one search. Therefore, it is first determined whether or not the first word matches with the kanji heading (step S140). If no matching word is found as in the example of FIG. 6, it is determined whether there is a further forward match (step S150). If “Saga” or “Sato” is present for “S”, there will be a word that matches forward. In this case, the process returns to Step S120, and one search character is added and the reverse. The search of the free-standing word dictionary JD1 is executed again. That is, the reverse lookup independent word dictionary JD1 is searched for “Saga”.
[0036]
In this case, since the matching word “Saga” is found, the process proceeds to step S160 and subsequent steps shown in FIG. If there is no match at the number of searches and there is no forward match, it is determined that an appropriate kana character string cannot be obtained by this reverse lookup independent word dictionary JD1, and the routine proceeds to the additional conversion processing routine shown in FIG. However, this process will be described later.
[0037]
If the reverse lookup independent word dictionary JD1 is subtracted to find the exact number of searches (step S140), it can be determined that the corresponding word is registered in the independent word dictionary JD1, and the search character pointer is set to one. The process advances (step S160), and subsequently, the number of ending / attached word searches is initialized, that is, a process of setting the value to 1 is performed (step S170). Next, one search character is added, the pointer is placed on the first character following the independent word (“NO” in FIG. 6C), and the reverse lookup dictionary JD2 is searched for the ending / attached word (step S190). It is determined whether there is a match (step S200). If there is no match and a forward match is found (step S210), a search character is added (step S180) as in the case of searching the reverse lookup independent word dictionary JD1, and the reverse lookup ending / attachment search is performed again. This is performed (step S190). In FIG. 6, only the headword is shown for the reverse lookup ending / adjunct dictionary JD2, but the adjectives made up of kanji are used for the appendix (“K”, “Section”, “Machi”, “Please”, etc.) Therefore, it is the same as the reverse lookup independent word dictionary JD1 that is composed of kanji or kana headings and their readings and grammatical information.
[0038]
In the example shown in FIG. 6, since “no” is found in the reverse lookup ending / attached word dictionary, the determination in step S200 is “YES”. In this case, connection verification is performed (step S220). ). That is, a grammatical connection with an independent word already found by searching the reverse lookup independent word dictionary is examined to determine whether to connect (step S230). The case particle “NO” is connected to the place name “Saga”, but if it is not connected, the number of searches is increased (step S240), and the reverse ending / adjunct search (step S190) and the following processes are performed again. As a result, when the ending or attached word to be connected is found, this ending or attached word is added to the previously found independent word (step S250), and the target character string is added to all the obtained character strings. It is determined whether reconfiguration has been performed (step S260). In the example of FIG. 6, when the search pointer is placed at “no”, the connection verification is “YES”. Therefore, it is determined whether the reconstruction of the target character string is completed, and “ha” still remains. Therefore, the process returns to step S160, and the process is repeated from the position where the search character pointer is advanced by one. As a result, the noun “Nanoha” was found this time, and all the target character strings were reconstructed, so the kana character string “Saganoha” was obtained, and then the independent words were read. Kana-kanji conversion centering on the independent word search using (step S270) is executed (step S280). That is, as shown in FIG. 7, a dictionary is searched for independent words, and the independent word that has the longest match and also has an adjunct word connected is searched for, and conversion candidates “Saga no Ha”, “Saga no Ha”, “Sano no Ha” Are displayed, and selection of a candidate by the user is accepted (step S300). The details of the candidate selection process with display will be described later.
[0039]
In the search for ending / attached words, if there is no match at the number of searches and no forward match is found, it seems that there was a problem in cutting out independent words before entering the search for ending / attached words. The number of searches is increased (step S290), and the process returns to step S130 shown in FIG. 4 to resume the processing from the independent word search, that is, the reverse non-independent word search. In this case, if there is a match with the next word of the same heading, the above-described search for the ending / attachment word and the connection test are performed. For example, if the character string “actual size” is first searched up to “actual” by reverse lookup independent word search, and “actual, collapsible, noun” is found, the ending / adjunct search is once performed here. . However, for “Large”, the reverse ending / attached word dictionary JD2 has no corresponding heading, and no forward match is found. In this case, the number of searches is increased (step S290), and the search is performed again (step S130 and subsequent steps). Since there is no match for the number of searches (step S140) and a forward match is found (step S150), the next is independent. After the word “actual size” is cut out, the ending / attachment search is started. In this case, the attached word “no” is found, and the reconstruction of the target character string is completed.
[0040]
Next, the additional conversion process executed when no match is found in the reverse lookup independent word search will be described with reference to FIG. When the additional conversion shown in FIG. 8 is started, the character pointer is first initialized, that is, set to 1 (step S310), and one character is first extracted from the position of the pointer (step S320). It is determined whether or not this character is kanji (step S330). If the character is kanji, the reverse lookup single kanji dictionary KD corresponding to the second dictionary is searched (step S340). Kana) is acquired (step S350). The obtained reading (kana) is added to the end of the already acquired kana character string (step S360), and the character pointer is advanced by 1 (step S370). Thereafter, it is determined whether or not there are remaining characters (step S380), and if there are, the process returns to step S320 to repeat the processing from the extraction of one character.
[0041]
For example, in the case of the character string “HIGO” shown in FIG. 9, the re-conversion technique shown in FIGS. 4 and 5 fails at the time of reverse lookup independent word search, and the additional conversion shown in FIG. 8 is activated. Is done. For the character “hi” extracted when the character pointer is 1, the reading (kana) “hi” is acquired as it is, and the character “line” is read with reference to the reverse single kanji dictionary KD. Is obtained. If the processing is repeated for all the characters in this way, “hikoki” is obtained as a reading.
[0042]
When there are no remaining characters, single-sentence kana-kanji conversion is started with the acquired reading (step S390). This conversion is the same as step S280 shown in FIG. 5, and when a conversion candidate is obtained by a kana-kanji change, candidate selection processing (step S300) is executed next. Then, the detail of a candidate selection process (step S300) is demonstrated next. In the candidate selection process, as shown in the flowchart of FIG. 10, first, a process for receiving an instruction from the keyboard 72 or the like is performed (step S400). If this instruction instructs selection of a candidate, a process for replacing candidates is performed. This is performed (step S410). In the example of the result of the reconversion shown in FIG. 7, if the first candidate is “Sagano”, this is replaced with the next candidate “Saga” or “Sex”. Since “Sagano” is reconverted, the character string “Sagano” that is the target of reconversion should not be displayed as a candidate or may be the last candidate to be displayed. is there.
[0043]
If the instruction is a correction instruction (step S400), the displayed candidate is temporarily returned to the kana character string and a process for correcting it is performed (step S420). If the character string you are trying to obtain is “Sakanoha (Sanoha)”, and if you accidentally input “Saganoha” and it was misconverted, the fixed character string is the kana character. Even if it is converted back to a column and reconverted, the correct conversion result cannot be obtained. Therefore, in such a case, after returning to the kana character string, an operation of correcting the reading (kana character string) is performed. After correcting the reading, single-sentence kana-kanji conversion is executed (step S430). This conversion process is the same as the above-described single phrase kana-kanji conversion process (steps S280, 390). In the case of additional conversion, the character string correction and the subsequent single-sentence kana-kanji conversion process are performed in exactly the same manner.
[0044]
If the instruction from the keyboard 72 or the like is a confirmation instruction, first, a process of learning to preferentially output the selected candidate to the independent word dictionary KD is performed (step S440). This is the same as learning in normal kana-kanji conversion. Thereafter, a process of outputting a phrase composed of the currently selected candidates as a confirmed character string is performed (step S450), and this routine is terminated.
[0045]
With the above candidate selection process, the character string that has been subject to re-conversion or additional conversion can be easily converted into a desired Japanese character string.
[0046]
In the embodiment described above, the reverse lookup independent word dictionary JD1 and the reverse lookup ending / attachment dictionary JD2 are described as being generated from the independent word dictionary KD and the ending / attachment dictionary, respectively. A normal free-standing dictionary KD or the like has a reading as a headline and kanji and part-of-speech information as a converted character string, so that it is easy to generate the reverse lookup dictionary shown in FIG. Thus, whether it is generated from an independent word dictionary or when a reverse dictionary is prepared separately from the independent word dictionary, it is necessary to register it in the reverse dictionary when a new word is registered. FIG. 11 is a flowchart showing a word registration processing routine.
[0047]
As shown in the figure, when this routine is started, first, a process of inputting a word to be registered is performed (step S500). Subsequently, a process of inputting a reading is performed (step S510), and a process of inputting a part of speech is further performed (step S520). The part of speech may be input by selecting a part of speech prepared in advance. After sequentially inputting the above information, a process of registering in the independent word dictionary or the ending / attached word dictionary is performed (step S530), and further a process of registering in the reverse lookup dictionary is performed (step S540).
[0048]
By executing this process, the reverse lookup dictionary is updated simultaneously with the word registration, and can be a target of re-conversion or additional conversion.
[0049]
In the embodiment described above, reverse conversion is performed in units of registered words and then used again for kana-kanji conversion (re-conversion), and reverse conversion is performed in units of single kanji and then used again for kana-kanji conversion (additional conversion). ), But the kana-kanji conversion performed again is single phrase conversion. This is because requests for re-conversion and post-conversion are rarely made over multiple phrases, and rather the speed of conversion should be given priority. In simple phrase conversion, it may be difficult to process affixes (especially prefixes), but if "affix" + "word" are registered in advance, they can be easily converted by simple phrase conversion. it can. Of course, continuous phrase conversion can also be performed as morphological analysis.
[0050]
In the case of a person name or the like, there may be a case where a desired character cannot be obtained by single phrase conversion. Therefore, including such a case, it is also preferable to add a single kanji conversion option in the candidate selection process (step S300). An example of this is shown in FIG. If the input from the keyboard 72 or the like is an instruction to start single-kanji conversion, a process for dividing the reading is performed (step S460), and single-kanji conversion is started for each divided reading (step S470). According to such single kanji conversion, as shown in FIG. 13, when a character string such as “higo” is designated, the reading is separated before and after due to the presence of the kanji “line”. Each “Ki” will be converted to single kanji. In addition, regarding “line”, it is possible to deal with single-kanji conversion, and it is desirable to determine that it has already been input in kanji and remove it from the target of single-kanji conversion. In the example of FIG. 13, the part input in kanji is excluded from the target of single kanji conversion. If the input character string does not include the kanji part, the character string of the maximum length that can be a heading for single kanji conversion is cut out from the beginning and sequentially subjected to single kanji conversion.
[0051]
Such single-kanji conversion is useful not only when part of the characters are input in kanji like “Hi-Go”, but also when converting a fixed kanji character string as illustrated in FIG. It is. Especially in the case of a person's name, it is difficult to register everything, so it is useful to use single kanji conversion together. Even if the person name “Kitaoka” shown in FIG. 14 obtains the kana character string “Kitaoka” by reverse conversion, it is difficult to convert it to “Kitaoka” by ordinary phrase conversion. In such a case, single-kanji conversion is started, the kana character string “Kitaoka” obtained by reverse conversion is divided into the minimum combination of headings with single-kanji headings, and then single-kanji conversion is started. In this example, “Kita” + “Oka”, “Kita” + “O” + “Ka”, “Ki” + “Ta” + “Oka”, “Ki” + “Ta” + “Oka”, “Ki” ”+“ Ta ”+“ o ”+“ ka ”can be separated, and by specifying these appropriately, single kanji conversion for“ ki ”+“ ta ”+“ oka ”is started, and the desired It can be converted into the kanji “Taka” + “Tada” + “Oka”. In addition, it is desirable to display the conversion candidate obtained by the single kanji conversion and the candidate obtained by the single phrase conversion in different modes. In the example shown in FIG. 14, the conversion is divided into units for display so that it can be intuitively understood which conversion is performed. However, it is also preferable to clearly indicate by changing the display color or the like. It is.
[0052]
As described above, the kana-kanji conversion apparatus according to the present embodiment performs word units for a confirmed character string, a character string read by the scanner 93, a character string input by handwriting from the tablet 94, and the like. Then, the kana-kanji conversion is executed again, and kana-kanji conversion is executed again. If a significant candidate cannot be obtained by this process, the kana-kanji conversion is performed after reverse conversion of single kanji for each kanji. Therefore, kanji candidates can always be obtained, and corrections such as erroneous conversion can be easily performed.
[0053]
Although the embodiments of the present invention have been described above, the present invention is not limited to such embodiments, and can of course be implemented in various modes without departing from the gist of the present invention.
[0054]
[Brief description of the drawings]
FIG. 1 is an explanatory diagram showing a schematic configuration of kana-kanji conversion in the present embodiment.
FIG. 2 is a block diagram illustrating a hardware configuration of a computer 10 used in the embodiment.
FIG. 3 is an explanatory diagram showing a structure of a reverse lookup dictionary.
FIG. 4 is a flowchart showing the first half of reconversion processing.
FIG. 5 is a flowchart showing the second half of the reconversion process.
FIG. 6 is an explanatory diagram showing an example of a reconversion process.
FIG. 7 is an explanatory diagram showing a state of single phrase conversion after re-conversion processing.
FIG. 8 is a flowchart showing additional conversion processing;
FIG. 9 is an explanatory diagram showing the state of additional conversion.
FIG. 10 is a flowchart showing candidate selection processing.
FIG. 11 is a flowchart showing a word registration processing routine.
FIG. 12 is a flowchart showing processing when single kanji conversion is activated in candidate selection processing;
FIG. 13 is an explanatory diagram showing a state of single kanji conversion.
FIG. 14 is an explanatory view showing another example of single kanji conversion.
[Explanation of symbols]
10 ... Computer
20 ... arithmetic processing unit
21 ... CPU
22 ... Local bus
23 ... Cache memory
24 ... Cash controller
25 ... Main memory
30 ... PCI bridge
32 ... PCI bus
40: Controller section
42 ... ISA bus
44 ... VGA
46 ... SCSI controller
48 ... ISA bridge
54, 55 ... Composite I / O port
54 ... Composite I / O port
55 ... Composite I / O port
56 ... Sound I / O
60 ... I / O section
62 ... ISA slot
68 ... Timer
72 ... Keyboard
73 ... Mouse
74 ... Speaker
76 ... CRT
82 ... Floppy disk device
84: Hard disk
86 ... Parallel port
88 ... Printer
90 ... Serial port
92 Modem
93 ... Scanner
94 ... Tablet
96 ... Microphone
IPU ... Input section
KEY ... Below
PIC ... Below
Pentium ... made by Intel
RCU ... Inverse conversion unit
SBC: Single phrase conversion section
SKC ... Single kanji conversion part

Claims (8)

A kana-kanji conversion device that once converts a character string including at least kanji into a kana character string and then performs kana-kanji conversion again,
A character string input means for capturing a character string including the kanji,
First inverse conversion means for inversely converting the captured character string into a kana character string with reference to a first dictionary in which kanji headings and readings are recorded for independent words constituting a phrase;
If a kana character string is obtained by the reverse conversion, the kana character kanji is converted as a single phrase from the front of the kana character string and subjected to a connection test with the ending / attachment, and then a kana that outputs a single phrase conversion candidate Kanji conversion means,
If no conversion candidate is obtained by the kana-kanji conversion means, a character string including the kanji is converted back to a kana character string by referring to a second dictionary in which kanji headings and readings are recorded for single kanji. A second reverse conversion means for providing the reversely converted kana character string to the kana-kanji conversion means;
If a conversion candidate by the kana / kanji conversion means cannot be obtained based on the kana character string obtained by the second reverse conversion means, a single kanji character for performing single kanji conversion on the kana character string Conversion means;
When single kanji conversion is performed by the single kanji conversion means, a delimiter in which the position of the kanji included in the character string captured by the second reverse conversion means is the delimiter position of the kana character to be converted by the single kanji conversion means A kana-kanji conversion device comprising position determining means.   The second reverse conversion means sends the kana character string obtained by the reverse conversion prior to the single kanji conversion means by the single kanji conversion means to the kana / kanji conversion means for outputting the single phrase conversion candidates. The kana-kanji conversion device according to claim 1, which outputs the kana-kanji conversion device. In the kana-kanji conversion device according to claim 2,
When the kana-kanji conversion means performs kana-kanji conversion on the kana character string obtained by referring to the second dictionary, the kanji contained in the captured character string and the single phrase obtained after the kana-kanji conversion A kana-kanji conversion device comprising conversion candidate restriction means for converting only those that match the kanji included in the conversion candidates.   The display unit according to claim 1, further comprising: a display unit configured to display a conversion candidate as a single phrase output by the kana-kanji conversion unit and a conversion candidate as a single kanji output from the single kanji conversion unit in a distinguishable manner. Kana-kanji conversion device. A kana-kanji conversion device according to any one of claims 1 to 4,
A kana-kanji conversion device comprising first dictionary dictionary generating means for generating the first dictionary for reverse conversion from a dictionary for kana-kanji conversion that converts a kana character string into a kanji-mixed sentence. In the kana-kanji conversion device according to any one of claims 1 to 5,
A kana-kanji conversion device comprising second dictionary generation means for generating the second dictionary from a dictionary for single-kanji conversion.   7. The kana-kanji conversion device according to claim 1, wherein said kana-kanji conversion means includes means for outputting conversion candidates by single phrase conversion in order according to the frequency of use. A kana-kanji conversion method realized by a computer once converting a character string including at least kanji into a kana character string and then executing a process of performing kana-kanji conversion again,
A character string including the kanji is taken in via a means for inputting the character string,
The captured character string is converted back to a kana character string with reference to a first dictionary in which kanji headings and readings are recorded for independent words constituting a phrase,
If a kana character string is obtained by the reverse conversion, the kana character string is subjected to kana-kanji conversion means as a single phrase from the front of the kana character string and then subjected to a connection test with the ending / attachment, and then a single phrase conversion candidate Output
If a conversion candidate is not obtained by the conversion, the character string including the kanji character is inversely converted into a kana character string with reference to a second dictionary in which kanji headings and readings are recorded for the single kanji character, Providing the converted kana character string to the means for performing the kana-kanji conversion;
When the means for performing kana-kanji conversion fails to create a conversion candidate, the kana character string obtained by performing the reverse conversion is set with the position of the kanji included in the character string as the delimiter position of the kana character. Kana-kanji conversion method that performs single-kanji conversion.

Images