JPS6068425A - Kana/kanji converter having learning function - Google Patents

Abstract

PURPOSE:To attain the KANA (Japanese syllabary)/KANJI (Chinese character) conversion having high probability for selection of a correct conversion candidate by selecting a learning table corresponding to a user code according to the input of said code and performing the conversion based on the learning table. CONSTITUTION:When the input is through with a KANA sentence, a conversion key is operated to supply the KANA sentence data to a dictionary memory device 5 of a KANA buffer 42 in order to search the corresponding headword. If no headword of the same reading KANA is available, the KANA sentence of the buffer 42 is stored to a sentence buffer 41 in the form of a converted sentence. When a headword of the same reading KANA exists, the homonym number is transferred and set to a prescribed buffer of the device 5. At the same time, all converted sentences are transferred to a buffer 44 of an RAM4. Then a learning table buffer 43 of the RAM4 is searched. If no homonym number exists, the converted sentences of the buffer 44 are delivered in their orders. When the homonym number exists, the sentences of the buffer 44 are rearranged in the order of the priority data of the buffer 43.

Description

[Detailed Description of the Invention] Technical Field> The present invention relates to a kana-kanji conversion device that converts kana characters or kana sentences (or Roman characters or Roman characters) into corresponding kanji or kanji-mixed sentences. The present invention relates to a device having a learning function that prioritizes and selects the most recently used homophone in selection.

<Prior Art> In general, in a kana-kanji conversion device, 1. Selection of homophones is important, and several proposals have been made and put into practical use.

As a typical example of practical use, homophones are fixed and arranged in advance in the order of frequency of word occurrence, and when first selected, the words are output according to their frequency of occurrence, while a table called a learning buffer is set up. The most recently used words are registered as the highest priority, and in subsequent selection of conversion candidates, this learning buffer ranking, that is, the most recently used homophone, is given priority (learning function). ).

However, the problem with the selection method of conversion candidates using this learning function is that this device is operated by multiple people, and the chopsticks each create sentences with different characteristics. This is because the ranking of the learning table changes.

In other words, when creating a text, the nature of the text is determined to some extent by the person who created it, and the text also has its own characteristics, so the ranking of the learning table is determined to some extent for each operator (creator). It is something that will come.

However, the above conventional method has a problem in that the probability of selecting the correct conversion candidate decreases because the ranking of the learning table changes each time when multiple operators create sentences. .

Purpose> The present invention provides a conversion candidate selection method using the learning function described above.
Therefore, it is an object of the present invention to provide a kana-kanji conversion device that can maintain a learning table for each operator or text creator and has a higher probability of selecting the correct conversion candidate.

<Embodiment> FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 1 will be described below.

Reference numeral 1 denotes a keyboard device, which consists of a group of kana keys arranged in alphabetical order, and a group of editing control keys 419 including a conversion key and a previous candidate key.

The conversion key instructs conversion from a kana sentence to a kanji or a sentence mixed with kanji, and when the converted kanji or a sentence mixed with kanji is output and displayed, if it is not the target kanji, that is, a conversion candidate, it is another candidate. , which is manipulated to call a homophone.

Further, the previous candidate key is used to instruct to call up and display the conversion candidates that have already been outputted and displayed again in the operation of calling up homophones using the conversion key 1.

Reference numeral 2 denotes an editing control unit (c p u ), which performs editing control according to an editing program stored in a program memory.

3 is n1] RAM that stores the editing program described
This editing program is transferred from a floppy disk device 6, which will be described later, when the device is initialized and is written in 1. α is done. This editing program includes a program for executing the flows shown in FIGS. 4 and 5.

Reference numeral 4 denotes a memory device composed of a RAM, which includes a text buffer 41 for storing edited text data, a kana buffer 42 for storing kana sentences inputted from the keyboard device 1, and a floppy disk when initializing the device. A learning table sofa 43 into which learning tables are transferred from the disk device 6, a conversion result buffer 44 which temporarily stores all homophones of conversion candidates corresponding to the input kana sentence, and others. The area 45 includes various types of sofas, flags, etc.

Reference numeral 5 denotes a dictionary memory device consisting of a mask ROM or the like, in which a kana sentence is used as a headword and a kanji character or a sentence containing kanji characters corresponding to the kana sentence is stored. Homophones of each word are stored in advance in a fixed order of frequency.

Reference numeral 6 denotes the floppy disk device mentioned above, and one of the floppy disks FD" has a program area 61 for storing an editing program and a table area 62 for storing a learning table. 2. The table area 62 is provided with a location ID for holding a user's (or creator's) code (ID code).

The other floppy disk FD2 serves as an external memory for semi-permanently storing the created text data.

7 is a display control section of the CRT display device 8, which has a built-in display memory.'J1. The input kana sentences, converted and edited data, etc. are stored and displayed on the device 8.

Reference numeral 9 denotes a printer control unit of the dot printer IO, which causes the printer 10 to create a large tree when text data from the text buffer 41 described above is supplied.

Now, to explain the configuration of the above-mentioned learning standard fluctuator 43 and conversion result buffer 44, all homophones that are conversion candidates are introduced into the conversion result buffer 44 from the dictionary memory device 5. are stored in the order determined by (order based on the limit frequency).

For example, the kana sentence "Kosei" is stored in the order shown in FIG. 2 (al).

Then, the kanji data serving as conversion candidates in the conversion result buffer 44 are displayed on the display device in the order of the table with reference to the learning table in the learning table buffer 43.

On the other hand, the learning table in the learning table buffer 43 is
As shown in the figure, the converted words or sentences that have homophones that have been converted during writing are stored sequentially as homophone numbers, and the homophones of these converted words or sentences are prioritized in the order of most recently used. A priority order key with an attached text is stored in correspondence with the homophone number.

In this figure 3, the homophone number of ``Kosei'' is 'J21'.
Corresponding to ``1.5'', the rank number of the homophone string shown in Figure 2(a) (this rank is fixed in advance as described above) is ``1.5'' according to the order of use.
．． 7.2.3.4.6'' is stored as the priority data.

The homophones of "al" are determined and displayed on the display device in the order shown in FIG.

The above-mentioned homophone number is a count value when searching for a headword in a dictionary, that is, the number of headwords from the beginning of a dictionary name, or the address value of its older sibling in a dictionary. Furthermore, when converted words or sentences are stored in the entire capacity of this learning table/<・sofa 43, the oldest registered one is deleted and the contents of the table are updated.

Next, the operation of the program diagram shown in FIG. 1 will be explained. This operation is performed according to the flowcharts in FIG. 554 and FIG.

First, when creating a text, the operator sets the special Nobby Disc board FD 74 and turns on the power, and initialization processing is executed according to FIG. 4.

This initializes each component and reads the editing program from the area 61 of the floppy disk FD1 and loads it into the program memory (RAM1) 3.

Here, if the floppy disk board FD1 is to be used for the first time, set the ID code registration mode from the keyboard device 4, and set and write the numerical code or character code by key operation. . Also, when the user wants to change the ID code, the user operates the screen. In this way, the floppy disk board FDI used by the user is identified as being exclusively used by the operator.

By the way, I just set up the floppy disk FD.

If an ID code has already been registered, then following the power-on, the operator inputs his own ID code from the keyboard device 1 and presses the ENTER key. Then, in response to this, the ID code just inputted is compared with the ID code stored in the learning table area 62 of the FD. Is this comparison +1+? ”-+l--two, tj+ntf・vn-+Vr<s-If it is not stored, the data of the learning chifur or the buffer 43 of the memory device 4 is stored.
loaded into.

On the other hand, if there is a mismatch in the comparison, an operator guidance message indicating the mismatch is displayed on the CRT device 8 and the floppy disk board FD of the floppy disk device 6 is displayed.
, indicates that it is not exclusive to the operator. Subsequently, when the ID code is input again, the above operation is repeated as the ID code is input again. This re-operation is based on the assumption that an input error has occurred in the I+) code.

Further, if the ENTER key is operated without inputting the ID code χ, the process shown in FIG. 4 is completed. In other words, the process is completed without loading the learning table. In this case, the learning function is created as a new one on the bath sofa 43, but since the ID code does not match on the disk board, as described in 1. I'm never alpha.

Next, the learning function will be explained (Fig. 5).

When a predetermined kana sentence is input from the keyboard device 1, the kana sentence is written in the kana buffer 42. e and is displayed on the CRT device 8. When the input of this kana sentence is completed, a conversion instruction is given by operating a conversion key. In response to this instruction, the kana picture name sentence data in the kana buffer 42 is transferred to the dictionary memory device 5 to search for the corresponding headword. If there is no entry word with the same reading kana here, it is stored in the text buffer 41 as a kana sentence in the kana buffer 42 or as a henhoumata.

If a headword with the same pronunciation exists, the homophone number (the rank of the headword in the dictionary) is transferred to and set in a predetermined buffer of the memory device 5, and all converted sentences (homonyms) are stored as shown in Figure 2. Memory device (RAM2) as shown in
5 is transferred to the result buffer 44.

By searching the learning table 43 in the memory device (RAM) 5, it is detected whether or not the above-mentioned homophone number exists. If this homophone number is not in the learning table buffer 43, it will be output in the order of the converted sentence in the result buffer 44, that is, in the order of priority fixed in advance.

On the other hand, when a homophone number exists, the converted sentences in the result buffer 44 are sorted according to the order of priority data in the learning table buffer 43. That is, for example, if the homophone number ``1215'' is ``kosei'', the words are rearranged in relative priority order (in order of frequency of use) as shown in Figure 2 (bl).

Thereafter, the above conversion process is performed. That is, the results are output in the order of the result buffer 44. The operator performs the next candidate reading operation until the desired converted sentence is obtained. Alternatively, all the converted sentences may be displayed in the order of the resulting buffer 44 and the desired converted sentence may be selected.

In this way, when the desired converted sentence is output and selected, if there is no homophone number in the learning table buffer 43, the currently selected converted sentence is given the highest priority and the priority data is is written to the learning table buffer 43 along with the homophone number. That is, the data is written to the learning table buffer 43 using the rank number held in the result buffer 44 as rank data, with the number of the selected converted sentence at the beginning, and thereafter in accordance with the rank of the result buffer 44.

Further, when a homophone number exists in the learning table buffer 43, the priority order of the two selected converted sentences is updated to the highest priority order.

In other words, if "star" is selected in the ranking of Figure 2 (bl), the updated priority data is r
5.1.7.2.3.4.6J. Such operations are performed every time conversion is performed, and the learning table buffer 4
3 will sequentially hold the homophone numbers and their priority data when they are sequentially converted.

When storing data on the disk FD2, the data in the learning table buffer 43 is stored in the learning table area 62 of the floppy disk FD□. At this time, the ID codes are checked again to see if they match, and if they match or there is no ID code in the study table area 62, the ID code is stored, but if it does not match, it is not stored and the ID code is stored on the floppy disk.
The learning table of D is maintained as is.

Therefore, each operator has a dedicated floppy disk FD,
You can also have your own study table.

Other embodiments> In the above example, the disk board FD1 holding the editing program
A magnetic disk device (non-volatile memory) is provided inside the main body of the device, and this disk device stores and stores learning tables for each operator by ID code. It is also possible to have a configuration in which the corresponding learning table is loaded into the learning table buffer 43 by manual input of the ID code.

<Effects> As described above, in the present invention, in the selection method of conversion candidates using the learning function, a learning table is maintained for each operator (or text creator), and the device can be operated by multiple operators. It is possible to provide a kana-kanji conversion device which is capable of calling up conversion candidates using its own learning table, and has a higher probability of selecting a correct conversion candidate based on this even when used by a user.

[Brief explanation of the drawing]

Figure 1 is a book diagram showing the configuration of the device of the present invention, Figure 2
a) and (b) are diagrams showing a result buffer, FIG. 3 is a diagram showing a learning table buffer, and FIGS. 4 and 5 are flowcharts showing the operation of the apparatus. 1: keyboard device, 2: editing control unit, 4: memory device, 5: dictionary memory device, 6: floppy disk device,
4 is a text buffer, 42: a kana buffer, 431 is a learning table buffer, 44 is a Nil buffer, 61 is a nib program area, and 62 is a learning table table area.

Claims (1)

[Claims] 1. Has a learning function that outputs homophone conversion candidates based on priority order data (learning table) determined by learning so as to give priority to the most recently used candidate. In the kana-kanji conversion device. A non-volatile memory means is provided for storing a user's (or text creator's) code as well as a learning table learned by the user when writing the text, and the learning table corresponds to the code based on the input of the user code. A kana-kanji conversion device having a learning function that selects and performs conversion processing using the learning table.