JPH0724054B2 - Data processing device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a text processing system for processing a Japanese sentence mixed with kana and kanji, and particularly to detecting or correcting a spelling error in such a text processing system. Related to the device.

[Background of the Invention]

With the development of OA, text processing by computer has been attracting attention. Especially in Europe and the United States, the computer processing of texts has become quite popular due to the small number of alphanumeric characters.In addition to simply inputting, storing and outputting texts, spelling errors in input texts can be detected and corrected. The system that does this is being commercialized. For example, UNIX utility program TYPO has a function to detect spelling errors in English words, and DEC10's SPELL program has a function to correct some spelling errors in English words. For example, JLPeterson “Computer Programs for Dete
cting and Correcting Spelling Errors ", CACM, Vo.123,
No.92 introduces such an English spelling verification system in detail.

However, in Japanese, (1) Kanji, Hiragana, Katakana and three character types are mixed.

(2) The number of kanji is extremely large.

(3) It is difficult to cut out a word because no space is inserted at the break of the word.

In Japan, it is difficult to develop a text input method and a spelling detection / correction system.

Furthermore, because the spelling error pattern itself in Japanese is different from that in Western languages, the processing methods developed in the West cannot be applied, and a unique Japanese method is required.

Japanese spelling patterns also depend heavily on the input method. At present, as Japanese input method, (1) touch type method (2) pen touch method (3) kana-kanji conversion method (4) handwritten character recognition method is known (JP-A-58-106663, JP-A-55-14368
1).

On the other hand, examples of the spelling error pattern are (a) user's memory error, (b) typo, and (c) device recognition error. Among them, (a) has many spelling errors in which homophones are used as target characters in Japanese, errors in sending kana, etc., which are often seen in the input methods (1), (2), and (4). In the method of 3) in which the phrase can be input, when the user inputs a phrase with a kana which is a phonetic character, the system converts it into a character string mixed with kana and kanji characters, so this kind of error can be caused by other methods. Less compared. (B) is an error peculiar to (1) and (2). (C) is (4)
This is a problematic error only in the method of.

As is clear from the above consideration, the kana-kanji conversion method (3), which is widely used in Japanese word processors at present, is considered to be a relatively error-free input method, but easy for beginners to use. , From the application of character recognition technology,
The present invention provides a convenient device for detecting and correcting errors of types (a) and (c), which is applied to the Japanese input method of the handwritten character recognition method of (4).

[Object of the Invention]

The present invention is based on the above consideration and its first
It is an object of the present invention to provide a means for detecting spelling errors included in a Japanese sentence in units of clauses in a text processing system for processing Japanese sentences mixed with kanji and kana. The second
It is an object of the present invention to provide a means for correcting an error in the detected spelling error in which a homophone is used as a target character.

[Outline of Invention]

The present invention, a Japanese sentence mixed with Kanji and Kana containing a spelling error, a means for dividing the word by referring to a word dictionary of mixed Kanji and Kana notation, and a means for presenting a clause in which the word division has failed to the user, With reference to the phonetic alphabet word dictionary, the conversion table that converts the phrase in which the word segmentation failed into the kana-kanji mixed notation to the phonetic alphabet (for example, Roman alphabet) notation and the obtained phonetic alphabet phrase It is characterized by having a means for dividing into words and a means for, when the division is successful, converting a phrase in the phonetic character notation into a kana-kanji mixed notation and presenting it as a corrected phrase to the user. .

Example of Invention

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the appearance of the document processing workstation taken as the present embodiment. Workstation 1 is a display 2, keyboard 3, tablet 4 for handwriting input,
And a floppy storage device 6. The user inputs a Japanese sentence from the keyboard 3 by hand or by handwriting on the tablet 4 with a stylus pen. The input Japanese sentence is displayed on the display 2 or stored in the floppy storage device 6. Further, the Japanese sentence stored in the storage device 6 can be displayed once on the display 2.

FIG. 2 is a block diagram showing details of the work station shown in FIG. The workstation includes a display 2, a display memory 7, a keyboard 3, a tablet 4, a floppy storage device 6, a processor 8, a handwritten character recognition device 9, and a printer 10.

The processor 8 is a device that performs various data processing such as processing of input text, and may be realized by a microprocessor or a dedicated logic circuit, or may be realized by a host computer and a terminal device. . The spelling error correction process of the present invention is executed by this processor.

The handwritten character recognition device 9 recognizes the kanji or kana characters input from the tablet, encodes them, and processes them.
The recognition rate of about 95-99% has been obtained for the device.

The printer 10 is a device that prints the Japanese sentence processed by the processor 8.

The floppy storage device 6 stores the input Japanese sentences as well as dictionaries and tables used for Japanese analysis. Other external storage devices such as a disk storage device may be used instead of the floppy storage device.

FIG. 3 is a diagram showing an example of a display screen display when the spelling error is detected and corrected in the present invention.
Japanese sentences input from the keyboard 2, the tablet 4, or the floppy storage device 6 are displayed in the upper half of the screen. When a spelling detection command is input from the keyboard 3, the system analyzes the Japanese sentence one phrase at a time starting from the beginning, and if an erroneous phrase is found, it is underlined. In this example, a false phrase "of a specialist" is detected. In contrast, the system displays three types of menus in the bottom half of the screen. This allows the user to select 1 if the error detected by the system is a special term that is not in the system's dictionary, and 2 if the spelling corrected by the system is correct,
If you want to enter a new correct spelling yourself, you can select 3. When the user's selection is complete, the system restarts parsing the sentence from the next clause.

FIG. 4 shows how the processor 8 and the floppy storage device 6 execute the spelling error detection / correction process as described above. The character string to be inspected is first stored in the character string buffer 20 and then the phrase segmentation means.
It is divided into clauses by 21. The bunsetsu consists of independent words (generally starting with Kanji or Katakana) followed by an arbitrary number of adjuncts (Hiragana), so if you detect a point where the type changes from Hiragana to Kanji or Katakana, Can be easily segmented. Next, an intra-segment morphological analysis 23 is performed for each segment extracted. Here, the morphological analysis is a process of dividing a bunsetsu into words, which is a concept handled in the kana-kanji conversion process of a Japanese word processor. Here, since the text is a mixed kana / kanji sentence, the analysis is advanced using the word dictionary 30 and the inter-word connection matrix 34 in kanji / kana notation. If the morphological analysis is successful, the bunsetsu is judged to be correct, and control is passed to the bunsetsu cutout means 21 to cut out the next bunsetsu. If the morpheme analysis fails, it is determined that the syllable contains a phonetic typographical error due to a homonym, and the bunsetsu is converted into the kanji-kana → romaji conversion means 24 and the kanji-kana → romaji correspondence table 33. It is used to convert to Roman alphabet and stored in Roman phrase buffer 25. Next, the Roman phrase is subjected to morphological analysis using the Roman morphological analysis means 26. This time, since it is a morphological analysis based on phonetic characters, a phrase containing spelling errors such that the sound is correct even if the sound is correct is correctly analyzed. The morpheme analysis in this case uses the word dictionary 32 and the inter-word connection matrix 34 in Roman alphabet. If the morpheme analysis fails here, it means that the original string contained errors other than the letters, and the system
Notify the user only that the phrase is incorrect.
If the morphological analysis is successful, the correctly analyzed phrase is converted into a mixed kanji / kana sentence by the Roman character → kanji kana conversion means 27, stored in the corrected character string buffer 28, and then presented to the user. The conversion is performed by referring to a pointer 31 from a word in the word dictionary 32 written in Roman alphabet to a word in the word dictionary 30 written in Kanji kana. When a plurality of Kanji / Kana mixed phrases correspond to the same romanized phrase, a plurality of candidates are presented to the user.

FIG. 5 is a diagram showing the flow of processing when the erroneous phrase “of a specialist” is cut out. When the morphological analysis of the phrase "of an expert" stored in the phrase buffer 20 is performed, the analysis fails because "the expert" is not in the dictionary. So, if you convert it into the Roman alphabet "SENMONKANO" using the Kanji-to-Romaji conversion table 33, the morpheme analysis succeeds because the noun "SENMONKA" is in the dictionary this time, and the noun "SENMONKA" and the case following the noun are succeeded. It is divided into particles "NO". Finally these words are pointers between the two dictionaries
Following 31, it is converted into a correction character string of "expert" and presented to the user.

FIG. 6 is a diagram showing the flow of processing when the erroneous phrase “I am not ashamed” is cut out. Also in this case, since the word "shame" is not in the dictionary, the morphological analysis using the mixed kanji kana notation fails. However, when converted to Roman alphabet, the morphological analysis succeeds and it is divided into the preformed form of the verb "HAJIRU" and the auxiliary verb "NAI". Finally, this is converted into a phrase “shameful” in which kanji and kana are mixed and presented to the user as a corrected phrase.

Although the example of Roman letters is shown as the phonetic alphabet, it is acceptable to use Kana letters.

〔The invention's effect〕

As described above, according to the present invention, firstly, there is an effect that a spelling error in a Japanese sentence mixed with kana and kanji can be detected. Errors include typographical errors, recognition errors in handwritten character recognition, user memory errors, and the like. Secondly, among the above errors, it has the effect of correcting spelling errors that are made by mistakenly using homophones and different letters.

[Brief description of drawings]

1 is an external view of a system, FIG. 2 is a block diagram, FIG. 3 is a diagram showing an example of a display, and FIG. 4 is a processor of FIG. Fig. 5 is a functional diagram showing the data flow in the floppy storage device. Fig. 5 is an explanatory diagram of the processing when a wrong character string "of a specialist" is input, and Fig. 6 is an error of "shame". It is explanatory drawing of the process at the time of the input character string. 1 ... Work station, 2 ... Display screen, 3 ... Keyboard, 4 ... Tablet, 5 ... Stylus pen, 6 ... Floppy storage device, 7 ... Display memory, 8 ... Processor, 9 ... Handwritten character recognition device, 10 …… printer, 20 …… character string buffer, 21
...... Phrase segmentation means, 22 ...... Kanji and Kana mixed phrase clause buffer, 23 ...... Korean and Kana mixed morphological analysis means, 24 ......
Kanji to Kana to Romaji conversion means, 25 …… Romaji buffer, 26 …… Romaji morphological analysis means, 27 …… Romaji to Kanji kana conversion means, 28 …… Corrected character string buffer,
29 …… Document memory, 30 …… Kanji kana notation word dictionary, 31 …… Pointer, 32 …… Romaji notation word dictionary, 33 …… Kanji kana → Romaji correspondence table, 34 …… Word connection matrix.

Claims (2)

[Claims] 1. A handwriting input means for inputting at least a kanji kana mixed sentence using handwritten characters, a storage means for storing a word dictionary in kanji kana notation and a word dictionary in phonetic alphabet notation, and the input kanji kana mixed A first conversion process for converting a part or all of a sentence into phonetic alphabet notation, a second conversion process for converting phonetic alphabet into a kanji kana mixed sentence, and the handwritten input kanji kana mixed sentence A processor for executing a first dividing process for dividing into words while referring to the word dictionary in Kanji and Kana and a second dividing process for dividing into words while referring to the word dictionary in phonetic character notation In the data processing device having the following, when the first division process is executed and the first division process fails, the first conversion unit causes one of the Kanji / Kana mixed sentences related to the failure to occur. A part or all is converted into a phonetic alphabet notation, the converted phonetic alphabet notation is subjected to the second division processing, and the second conversion means outputs the result of the second division processing to a kanji kana. A data processing device characterized by converting to a mixed sentence. 2. The data processing device according to claim 1, further comprising a printer or a display as a process for outputting the processing results of the first division processing and the second division processing, A data processing device, wherein when outputting the processing result of the division processing of 2, the output of the content instructing the selection of homonyms is displayed on the printer or the display.