JP3154875B2 - Kanji conversion learning device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a learning device for a kanji conversion device, and more particularly to a kanji conversion learning device using both long-term learning and short-term learning.

[0002]

2. Description of the Related Art In a Chinese word processor or the like, an input method by reading kanji is, for example, a word intended by a user from a keyboard (a word is often composed of a plurality of kanji, but is composed of one kanji. Wrap things.
It also includes phrases, short sentences, etc., as described later. ) And phonetic character strings that represent kanji readings (in Japanese, the distinction between singular and plural is not clear, so in this specification we will not use them strictly. If there is a "column", it indicates a plurality of characters.) The conversion of the phonetic character string into kanji is performed by giving priority to a word having a large number of kanji (the value is equal to the number of syllables in Chinese) firstly constituted by the word processor. The phonetic symbol composed of the input phonetic character string is cut out by the longest matching method or the like that adopts the two principles of giving priority to the phonetic character string input earlier (hereinafter, the cut-off phonetic character string is used). Is sometimes referred to as a "morpheme."), By using the cut-out reading symbols as a search key to search a dictionary part to find a corresponding word, and converting it into a kanji string constituting the word. (Note that the longest match method is described in Japanese Patent Application No.
As disclosed in Japanese Patent Application No. 5911, Japanese Patent Application No. 5-75912, etc., since it is a well-known technique, description thereof will be omitted. If the converted word is not intended by the user who is the character inputting person, the correction is made manually. For this reason, if there are homonymous kanji and words, if the correct kanji and word are quickly and accurately selected from them, the efficiency of this input method will be improved. By the way, this homonymous kanji,
As a means for solving the word selection problem, what is generally called a learning function is used. This method improves the selection success rate by recording the usage and the number of times of use of homonymous kanji and words used by the user, and reflecting this in the selection between homonymous kanji and words. It is assumed that. At this time, the content of the record such as the frequency of use,
There are two types of learning, short-term learning and long-term learning, depending on the length of time to be learned.

[0003] Short-term learning is to prioritize the latest use when selecting homonymous kanji and words. For example, the homonymous kanji corresponding to the Chinese phonetic symbol "i4" can be written in the order of normal usage as "mean, million, easy,
In addition, the characters are arranged in this order and stored in the dictionary. If the latest character used by the user among these homonymous kanji is ″
If it is "wing," the order of storage is "wing, meaning, billion, easy,
When the same reading symbol "i4" is input next time, "wing" is automatically selected with the highest priority. If "benefit" is selected when inputting "i4", this "benefit" will be used in the next "i".
When inputting "4", it is placed at the top to be the subject of kanji conversion with the highest priority, so that "wing" has one lower priority, and consequently the arrangement of homonymous kanji corresponding to "i4" Is "Masu, Tsubasa, Min, Billion, Easy, ..." The advantage of this method is that the most recently used kanji and words are automatically placed at the beginning, so they are generally used. The disadvantage is to select the correct kanji and words for the user's input.The disadvantage is that even rarely used kanji and words are targeted for learning, so the nature of the created document and the user's When kanji or words that happen to be used frequently due to the habit of selecting words or kanji, etc., are frequently used, these kanji or words have a large adverse effect on the conversion rate.

[0004] Long-term learning is to store the number of times of use of homonymous kanji and words for each word.
Place and store homonymous kanji and words. On this,
The previously stored homonymous kanji and words are preferentially selected. The advantage of this method is that the conversion rate of the kanji is not lowered accurately in exceptional cases where the kanji and words are used infrequently.
The disadvantage is that the frequency of use is also stored for each kanji and word, so that the amount of required storage resources increases. Also, the kanji used for the currently input document,
The disadvantage is that the response to words tends to be delayed.

[0005] The order of arrangement of reading symbols in the dictionary section,
Change the arrangement of words, remember the number of uses for each specific word, correct kanji if it is erroneously converted to kanji, and prioritize the word corresponding to the phonetic character string cut out for this operation In response, the message is displayed at a predetermined position on the CRT, and the user selects a desired one from the display. If the display does not include a desired word, the user further instructs to that effect. Displaying words of lower rank is a well-known technique widely used in Japanese word processors. Therefore, their description is omitted. In addition, registration, storage, and storage of reading sounds and words in the dictionary unit are actually coded and then processed in an electromagnetic format. Omitted.

Generally, the kanji input device employs one of the above two learning methods, and further improves the success rate of the kanji conversion by making some improvements.
The following describes an example of a homonym learning device in a Japanese word processor. FIG.
It is a block diagram of this apparatus shown in Japanese Patent No. 157972. In the figure, reference numeral 100 denotes a kana input unit;
0 is a word dictionary, 300 is a character converter, 4
00 is a homonym order storage unit, 500 is a homonym selection unit, 600 is a display unit, 700 is a word dictionary update unit, 800 is a word dictionary update designation unit, 900
Is a homophone designation part. The kana input unit 100 uses phonetic characters such as Japanese hiragana and katakana as reading symbols (strictly speaking, hiragana and the like are "phoneme characters", but they will not be distinguished in the present specification as described later. ). The word dictionary unit 200 makes a table in which the Japanese word readings in the order of Ai and the corresponding notations of words centered on kanji are associated with each other, and stores the correspondence table in the order of the Japanese syllabary. The character conversion unit 300 sequentially searches the homophone sequence storage unit 400 and the word dictionary unit 200 for a kana character string input from the input unit, and obtains a corresponding kanji or a word notation composed of kanji and kana characters. (Note that the term "word notation" is used here because the Japanese word processor does not always convert all the input kana to kanji. For example, the kana sequence "call out" ). However, there is no substantial difference from "kanji" and "word" in Chinese word processors.) Also,
In the search in the word dictionary unit 200, when there are a plurality of word expressions having readings that match the input kana sequence which is the key of the search, these are selected in the order stored first, and the same kana sequence is input together with the input kana sequence. Output to selection section 500. The homophone order storage unit 400 stores the homonymous word expressions selected from the homophone selection unit 500 in association with their readings in the order of selection. Therefore, the newly selected one is arranged at the rear. For this reason, the homophone order storage unit 4
The order of the search in 00 is reversed from the order in which the word notations were stored, so that the newer the use, the higher the priority. The homonym selection unit 500 allows the character utterer to select a word notation intended by the character input unit by using a homonym designation unit 900 described later when the word notation output from the character conversion unit 300 includes a homonymous one. . The display unit 600
By displaying the word notation selected by the homonym selection unit 500 on the screen, it is possible for the user to check the result of the kanji conversion and to specify a specific word notation. The word dictionary updating unit 700 counts the number of times that each of the homonymous word notations stored in the homonym order storage unit 400 has been duplicated and used, and this number is a prescribed value “3”.
Are rearranged so that the word notation comes first in the arrangement of words having the same sound as the word notation in the word dictionary unit 200. The word dictionary update designation unit 800 inputs a dictionary unit update instruction from the user. The homonym designation section 900 is used to designate a specific one of a plurality of homonymous word expressions by a user or to input a conversion end instruction.

Actually, in a Japanese word processor, not all hiragana characters are converted into kanji characters because there is a word notation such as "call". The longest matching method in a Chinese word processor is not used as it is because there is no fixed numerical relationship such as one syllable and one kanji between the hiragana and the kanji that can be converted in the summer. However, these differences are irrelevant to the gist of the present invention. On the other hand, the principle of conversion, that is, searching in the dictionary part using Hiragana as a search key, is the same, and it is a well-known technique. Therefore, description of an algorithm for extracting a hiragana input for a dictionary part search in a Japanese word processor will be omitted.

Next, a description will be given of how the homonym learning device performs the kanji conversion with reference to an operation flowchart shown in FIG. In S91, a kana sequence is input from a kana input unit. In S92, the character conversion unit confirms that the input kana is in hiragana notation. If it is in hiragana, go to S93. If it is not the hiragana notation, the process directly goes to step S95, where the converted character string is displayed, and then an input wait state is set.

In S93, a kanji conversion process is started, and a word search is performed. In S94, the homonym selection unit determines whether or not the kana sequence to be processed includes a homonymous word expression based on the searched result. If included, S
Go to 96. If it is determined that it does not include,
Go to 95. In S95, the display unit 600 displays the converted character string. Then, an input waiting state is set.

In S96, guidance for the user to select a homonym is displayed on the display screen. Next, a description will be given, with reference to an operation flow chart shown in FIG. 10, of a process at the time of selecting a homonymous word notation. In S101, the user uses the homophone designation unit to select an intended one from the homonymous word notations received from the character conversion unit.

In S102, it is confirmed that the selection is correct, and (confirmation) is performed. In S103, the determined word expressions are arranged in the selected order and stored in the homophone order storage unit, and then the input standby state is set again. Next, a process of updating the word dictionary will be described with reference to an operation flowchart shown in FIG.

In step S111, when the user inputs a word dictionary update instruction to the word dictionary update designation unit, the word dictionary update unit starts. In step S112, the word notation stored in the homophone order storage unit is checked, and if there is a word that is stored in duplicate, the number of duplications is counted. In S113,
For word notation with a duplication frequency of 3 or more in the word dictionary,
The array is updated so that the order of storage in the homophone is selected first. Then, the system enters the next input waiting state.

With the above procedure, Japanese kanji conversion and associated learning are performed.

[0014]

However, the kanji conversion learning device having the above-mentioned configuration has the following problems. (1) It is necessary to manually select kanji, words, and word notations having the same homonym and update the dictionary part manually.

(2) The storage order of homonymous words and the like stored in the word dictionary unit and the homophone order storage unit changes depending on the learning function. Therefore, when the character of the text to be converted and the user's habit of selecting characters and the like change, it is necessary to update the storage order of the word dictionary unit. (3) The dictionary unit is a ROM (Read only memory)
ry, a read-only storage element), the word dictionary cannot be updated.

(4) The object of learning is limited to the word dictionary section, and words defined by the user for the convenience of document creation are excluded. For this reason, there is a certain limit to the efficiency of kanji conversion. The present invention has been made in view of the above problems, and has been made to solve the above problems (1) to (4), and to provide a kanji conversion learning device excellent in kanji conversion efficiency. Things.

[0017]

In order to achieve the above object, according to the first aspect of the present invention, the reading symbols and the corresponding words are arranged in accordance with the frequency of use if there are a plurality of corresponding words. In addition, the registered fixed large dictionary unit, the reading symbols and the corresponding words and the number of times of use thereof are stored in an array in accordance with the principle that the number of times of use is given first, and the newness of use is given priority. A learning dictionary section, the input dictionary is used as a search key, and the learning dictionary section and the fixed large dictionary section are searched in this order to detect all the corresponding words. If there are a plurality of corresponding words, a conversion unit that selects the one that is arranged first as a conversion target, and if there is a plurality of corresponding words, if the selection result of the conversion unit is incorrect , A selection unit that allows a user to select a word intended by the user from among the other words detected by the conversion unit according to an instruction to that effect by the user, and sets a kanji conversion result, and a selection of the conversion unit Is incorrect, and when it is corrected by the instruction of the character input person via the selection unit, if the corrected word is not stored in the learning dictionary unit, the number of times of use is set to 1. The word is newly stored in association with the reading symbol used as the search key above, and if the word has already been stored, one is added to the number of times the word is used, and then the word in the learning dictionary unit at the input reading symbol is entered. An array reviewing unit for reviewing the sequence firstly for the large number of times of use and secondly for the principle of giving priority to newness of use; and a number of times the word has been used after adding 1 to the sequence reviewing unit. If is greater than or equal to a predetermined value, this is detected. Then, after reducing the number of uses of all the words in the learning dictionary unit by a certain number, comparing with a separately determined threshold value, deleting words whose value is less than the threshold value from the learning dictionary unit And a kanji conversion learning device, characterized by having an updating unit for performing the conversion.

According to the second aspect of the present invention, a user-defined word is created according to a user-defined word creation instruction,
2. The kanji conversion learning device according to claim 1, further comprising a user-defined word creating unit that stores the user-defined word in the learning dictionary unit after associating with the reading symbol input as the search key. I have. In the invention according to claim 3, the kanji conversion learning device according to claim 2, wherein the user-defined word creating unit can use an off-line method.

In the fourth aspect of the present invention, the arrangement reviewing unit may store, in the learning dictionary unit, another word having the same sound as the search-reading symbol of the user-defined word created by the user-defined word creating unit. If the same user-defined word is used, the number of times of use of the new user-defined word is stored in the number of times of use of the same homonymous word or the larger of the number of uses and the predetermined value. In this case, the kanji conversion learning device according to claim 2 or 3, wherein the number of times of use is set to the maximum value or the larger of the maximum value and a predetermined value.

[0020]

According to the above construction, in the invention of claim 1,
Register the reading symbols (phonetic characters) and the corresponding words in the fixed large dictionary section in an array according to the frequency of use if there are multiple corresponding words (pre-stored)
Have been. The principle that the learning dictionary unit gives priority to the reading symbol, the corresponding word and the number of times of use thereof, firstly, the number of times of use (or the size of the value), and secondly, the newness of use. (New storage and maintenance of the storage in such a manner that the arrangement can be changed).

The conversion unit searches the learning dictionary unit and the fixed large dictionary unit in order using the input reading symbols as a search key,
All corresponding words are detected, and if there is only one corresponding word, the word is selected as the target of kanji conversion if the corresponding word is present in plurals, if there is more than one corresponding word. Since there are a plurality of corresponding words, if the selection result of the conversion unit is incorrect, the selection unit is used from among the other words detected by the conversion unit in response to an instruction from the user to that effect. After selecting the word intended by the user, the result is converted to kanji.

When the selection of the conversion unit is incorrect and the correction is made according to a user's instruction through the selection unit, the sequence review unit does not store the corrected word in the learning dictionary unit. Then, the number of times of use is set to 1, and newly stored in association with the reading symbol used as the search key. If it is already stored, 1 is added to the number of times of use, and then the input reading symbol is added. The word sequence of the first
Second, the number of times of use is reviewed based on the principle that the newness of use is prioritized. (Therefore, the present array reviewing unit performs three kinds of operations of reviewing the sequence, adding new storage, and adding the value of the number of times of use.) If the updating unit adds 1 by the sequence reviewing unit, If the number of times of use of the word is equal to or more than a predetermined value, after detecting this, the number of times of use of all words in the learning dictionary unit is reduced by a certain number, and compared with a separately determined threshold value Then, words whose values are less than the threshold value are deleted from the learning dictionary unit.

According to the second aspect of the present invention, the user-defined word creating section creates a user-defined word by combining words from kanji in response to a word definition creating instruction from the user, and inputs the word as a search key. After making them correspond to the reading symbols, both are stored in the learning dictionary unit. According to the third aspect of the present invention, the user-defined word creating unit performs an operation for character input by a user who is input to a storage medium such as a floppy disk using another character input device. Another device equipped with a disk or the like can know the content of the input operation by reading from the floppy disk or the like.) And use the created user-defined word together with its reading symbol. It has a function of storing in the learning dictionary section.

According to the fourth aspect of the present invention, if the sequence reviewing unit has already stored in the learning dictionary unit a word having the same sound as the search symbol for the word created by the user-defined word creating unit, If the number of uses of this new user-defined word is stored in the number of times of use of the word or the value having a predetermined value greater than the number of times of use of the word, if a plurality of homonymous words are stored, The numerical value is set to the larger of the maximum value or the number of times of use and a predetermined value.

[0025]

FIG. 1 is a block diagram of an embodiment of a kanji conversion learning apparatus according to the present invention. In this figure, 10 is an input unit, 20 is a conversion unit, 25 is a display unit, and 30 is a display unit.
Is a selection unit, 35 is a homophone / user-defined word designation unit, 40 is a learning dictionary unit, 45 is a fixed large dictionary unit, 50 is an update unit, and 60 is a first buffer. Yes, 65 is a second buffer, 70 is an output unit, and 80 is a user-defined word creation unit. Hereinafter, the operation, configuration, and the like of each of these units will be described.

The input unit 10 uses a keyboard,
The phonetic character string to be converted, which is input by the user while referring to the RT, is received. The conversion unit 20 converts the input phonogram string into a corresponding kanji string while referring to the fixed large dictionary unit 45 and the like. The display unit 25 is composed of a CRT, and is a candidate for selecting a converted kanji string or a kanji string having the same sound as this, a word, and a word notation (hereinafter, these three will be referred to as “words” in the clear case). Display as The selection unit 30
Interacts with the RT, keyboard, push buttons, etc. (operates while adjusting) to select words with the same sound, and to create user-defined words based on the kanji output from the conversion unit. In addition, it is possible to specify that. The homophone / user-defined word designation section 35 is used by the user to instruct “select a desired one from kanji, words, and word expressions of the same sound” or “create a user-defined word”. , CRT, keyboard, etc. The learning dictionary unit 40 is composed of a volatile high-speed semiconductor memory and stores reading symbols, all the words corresponding to the reading symbols, and the number of times of use, and for each word of the same sound, the word with the highest number of times of use is preferentially converted. And place it first. If the number of uses is the same, the latest use is prioritized for conversion. FIG. 6 conceptually shows the configuration. The fixed large dictionary unit 45 is made of a nonvolatile magnetic disk,
A phonetic symbol consisting of a coded phonetic character string and all corresponding words, that is, words that are homonymous, are stored. It depends on the frequency of use. FIG. 7 conceptually shows the configuration. The updating unit 50 has a built-in sequence reviewing unit 51, which updates the learning dictionary unit 40 in a predetermined procedure based on the reading symbols, the corresponding words and the number of times of use, and Review the sequence. When selecting the same sound word, the first buffer 60 temporarily stores all corresponding same sound words for each morpheme composed of the code of the cut-out phonetic character string. When updating the learning dictionary unit, the second buffer 65 temporarily stores the value of the number of times of use of the word serving as a reference for updating. The output unit 70 outputs the kanji-converted word. The user-defined word creating unit 80 has a central role in creating words that are frequently used by users and are not in the fixed large dictionary unit, and often consist of many kanji as user-defined words. In the creation, a keyboard, a CRT, an MCPU and the like are interlocked.

FIG. 2 is a flowchart showing the operation of kanji conversion in this embodiment. Hereinafter, the contents will be described with reference to FIG. In S201, the reading symbol of the input phonetic character string is cut out by the longest match method. In step S202, using the phonetic symbols cut out for each morpheme as search keys, word candidates of the pronunciation are searched in the order of the learning dictionary unit and the fixed large dictionary unit, and a word group as a detection result (in many cases, there are a plurality of words; The word “group” is added to the “group”, and the “group” has no significance.) Is stored in the first buffer.

In S203, the first kanji and word in the kanji and word group stored in the first buffer are set as conversion results. In S204, the conversion result is displayed on the display device. The correct selection from the word candidates and the learning dictionary update process are finally performed by the user confirming the conversion result on the display device. If the conversion result is correct, the user inputs a "confirmation" instruction code to the designated device, and the updating unit updates the learning dictionary unit of the reading sound with the word obtained as the correct conversion result. If the user determines that the conversion result is not intended, the selection unit corrects the conversion result. The corrected conversion result is displayed on the display device. Then, it waits for an instruction code from the user. If the user inputs a "user-defined word" instruction code from the homophone / user-defined word designation section, a desired user-defined word is created by the user-defined word creation section. The arrangement reviewing unit in the updating unit stores in the learning dictionary unit together with the reading symbols that specify the kanji string that constitutes the word. The value of the number of times of use set at that time will be described later.

FIG. 3 is a flowchart showing the operation of selecting a word having the same sound and updating the learning dictionary unit in this embodiment. Hereinafter, these operations will be described with reference to FIG. In S301,
The converted kanji string and the cursor position are displayed on a display device. In S302, the user inputs an instruction code from the homophone / user-defined word designation section.

In S303, it is determined whether or not the instruction code is "determined". If confirmed, go to S304. If not "determined", the flow proceeds to S306. In S304, the flag (FLAG) indicating the update processing is set to the value of “general”.
In S305, for each word of the conversion result, the learning dictionary unit is subjected to an update and review process, which will be described later, using the pronunciation symbol and the word, and the process ends.

In S306, it is determined whether or not the instruction code is "cursor movement". If "cursor movement", go to S307. If not, go to S308. S
At 307, the cursor is moved to the position of the word to be corrected, and the process returns to S301. In S308, it is determined whether or not the instruction code is "display of the same sound word". If so, go to S309. Otherwise, S311
Go to

In step S309, the same-word candidate word is extracted from the first buffer and displayed on the display device 25. S3
At 10, the user selects a desired word from homophonic word candidates by the homophone / user-defined word designation unit. Then, go to S315. In S315, S31
The process returns to S301 after replacing the word selected at 0 with the incorrectly converted word.

In S311, it is determined whether or not the instruction code is a "user-defined word". If so, go to S312. If not, the process returns to S301. In S312, a flag (FLAG, electronic mark) indicating the updating process is set by the value of the "user-defined word". In step S313, the kanji conversion is performed using an on-line method in which kanji conversion is performed interactively while looking at the word processor or an off-line method in accordance with an instruction stored in a floppy disk input by a separate word processor. Make user-defined words. In the case of the off-line system, various instructions and operation contents are obtained by a user's separate instruction when mounting the floppy disk and a predetermined signal stored in the floppy disk. The predetermined signal in this case is predetermined, and similar symbols such as "[" and "]" in the specification of the patent application are used. However, since this is a well-known technique, its description is omitted. Then, go to S314.

In S314, the reading symbols of the user-defined word and the kanji string constituting the word are stored in the learning dictionary unit, and at this time, the updating and arrangement review processing described later is performed, and the process returns to S301. When storing words in the learning dictionary,
When the processing flag is “general” and the input reading code and word are already stored in the learning dictionary unit,
Increase the number of times the corresponding word is used by one. If the phonetic symbol is stored but no word is stored, the word is added to the phonetic symbol column, and the number of times the word is used is set to one. When neither a reading symbol nor a word is stored, the reading symbol and the word are newly stored in the learning dictionary unit as one learning data entry. Then, the number of times of use is set to 1. If the processing flag is “user-defined word”, if the reading symbol of the user-defined word is stored in the learning dictionary unit, the number of times the user-defined word is used is set to the maximum value in the learning data entry. Set with. If nothing is stored, the reading symbol and the word of the user-defined word are newly stored in the learning dictionary unit as one learning data entry. Then, the number of times of use is set to 1.

FIGS. 4 and 5 show the detailed operation flow of new storage of words in the learning dictionary section, review of the arrangement of stored words, and update of the words stored in the learning dictionary section of the present embodiment, ie, steps S305 and S314 in FIG. It shows. Note that both drawings should be originally one leaf, but they are divided into two leaves to limit the size of input to the electronic information processing system at the time of filing. Hereinafter, the new storage, arrangement review, and update of words in the learning dictionary unit will be described with reference to FIGS.

First, the operation shown in FIG. 4 will be described. S4
In step 01, the input reading symbols, words, and processing flags are stored in a built-in storage unit (not shown) as variables YOMI, LEX, and FLAG, respectively. Further, the maximum value of the number of times of use and the constant values used for updating, TH and FUNM, respectively, are also stored in the storage unit, and the process proceeds to S402.

In step S402, a search is performed in the learning dictionary unit using YOMI as a search key. Search value (word, number of uses)
In the register group A of the second buffer. In S403, it is determined whether or not the register group A is an empty set (nothing is input). If it is an empty set, go to S404.
Note that an empty set indicates that there is no reading symbol in the learning dictionary unit. If it is not an empty set, go to S406.

In S404, the number of times of use is set to 1.
In S405, YOMI, LEX and the number of times of use (in this case, 1) are stored in the learning dictionary unit, and then the process ends. S
At 406, it is determined whether the FLAG is a "user-defined word". If it is not a "user-defined word", S40
7 and if it is a "user-defined word", go to S414.

In S407, the register group A is searched using LEX as a search key. The search value (the number of times LEX is used) is stored in the register b of the second buffer. After a while, go to S408. In S408, it is determined whether or not the register b is an empty set. If it is not an empty set, go to S409. If it is an empty set, go to S412.

In S409, the value of the register b is incremented by 1, and the process proceeds to S410. In S410, the value of the number of times of use of the corresponding word in the register group A is updated with the value of the register b using LEX as a search key. In S411,
The order of the arrangement of the words stored for the input YOMI of the register group A is corrected based on the value of the number of times of use. Then, it enters the processing node B of FIG. 5 described later.

In S412, the value of the register b is set to 1. In S413, the value of the register b is stored at the end of the value of the register group A using LEX as a search key. Thereafter, the process goes to the processing node B shown in FIG. In S414, the maximum value of the number of times of use is extracted from the register A, and
To be stored.

In S415, the value of the register b is added to the word represented by LEX as the number of times of use, and the register group A
Is stored at the forefront of the word group indicated by the reading symbol. Thereafter, the process goes to the processing node A shown in FIG. Next, the operation shown in FIG. 5 will be described. From processing node B, S50
Enter 1.

In step S501, the LEX in the register group A and the number of times of use thereof are rearranged in descending order of the number of times of use, using LEX as a search key. In step S502, the user-defined word and the word arrangement in the learning dictionary unit are changed with the value of the register group A using YOMI as a search key. In S503, it is determined whether the value of the register b corresponds to TH. If it is TH, go to S504. If the value of the register b does not correspond to TH, the process ends.

In S504, the values of all the times of use in the learning dictionary section are reduced by FUNM. S505
go to. In S505, words whose usage count in the learning dictionary unit is 0 or a negative number are deleted from the learning dictionary unit, and the processing is terminated. The processing node A enters S502.

FIG. 6 is a conceptual diagram of the configuration of the learning dictionary unit in this embodiment. In this figure, the numbers written after the kanji and words indicate the number of times the kanji and the like are used. In addition, the arrangement for storing kanji and words also depends on the order of use. For this reason, the kanji and word placed at the forefront of each reading symbol are selected with the highest priority when the reading symbol is input as a search key.

FIG. 7 is a conceptual diagram of the configuration of the fixed large dictionary section of this embodiment. The arrangement of homonymous kanji and words is determined according to the results of surveys on the frequency of use in general users and documents (corpora, corpus, language materials). As a result, the kanji is preferentially output as a kanji conversion target. About the present embodiment configured as described above, specifically, “di2ren2g
ong1sh4men3lie4 "will be described below.

Referring to the learning dictionary section and the fixed large dictionary section,
After the conversion unit, the result is “enemy“ official ”fierce” (here, the word surrounded by “to” indicates a word to be described as an embodiment of the present invention). The result processed by the selection unit is “suitable“ offensive ”fierce”.
Next, the contents of the learning dictionary unit before and after the arrangement is updated by the updating unit at this stage will be described.

(Before) gong1sh4 formula 3, public affairs 3, offense 3, construction 2 (after) gong1sh4 offensive 4, formula 3, public affairs 3, construction 2 Then, "uo3xiou1gai3fang2yu4gong1sh4" is input,
The result of the kanji conversion by the conversion unit is “Improvement defense“ offensive ”.” The result modified by the selection portion is “Improvement defense“ construction. ”At this point, after being updated by the updating unit, The contents of the learning dictionary section are shown below.

Gong1sh4 Offensive 4, Construction 3, Formula 3, Public affairs 3 Subsequently, “di2ren2de0gong1sh4shou4zu3” is input and processed by the character conversion unit. The result is “Enemy's“ Offensive ”obstruction”. Directly without being "blocked"
At this point, the contents of the learning dictionary unit processed by the updating unit are shown below.

Gong1sh4 Offensive 5, construction 3, formula 3, public affairs 3 If the input code received from the homophone / user-defined word designating unit is “user-defined word”, use Enter the user-defined word section. If the user inputs the reading symbol “gong1sh4” and the word “Miyamuro” of the user-defined word, the configuration of the column corresponding to this reading symbol in the learning dictionary unit after being processed by the updating unit is as follows. Show.

Gong1sh4 Miyamuro 5, Offensive 5, Construction 3, Formal 3, Public affairs 3 In this case, since “Miyamuro” is the earliest among the words having the maximum number of uses of 5, the next reading symbol ″
When converting gong1sh4 "to kanji, this user-defined word" Miyamuro "is selected with the highest priority. If it goes to the forefront of "gong1sh4" and the number of times of use reaches 6, and thus reaches a predetermined upper limit value TH, the number of times of use of all words for all the reading symbols stored in the learning dictionary unit Constant value FUNM from
(= 3) is subtracted.

Then, words whose number of uses corresponding to the result of the subtraction is 0 or less are deleted from the learning dictionary section. Also,
The pronunciation symbol from which all the corresponding words have been deleted is also deleted. This prevents an increase in the total number of words stored in the learning dictionary unit, improves the processing speed, and keeps the required amount of storage resources below a certain value.

Although the present invention has been described based on the embodiments, the present invention is not limited to the above-described embodiments, and may be modified as appropriate without departing from the scope of the invention. is there. For example, (1) when performing kanji conversion in a Japanese word processor, the present invention is used to update the learning dictionary section.

(2) The threshold value and the like used when updating the learning dictionary section can be appropriately set to optimal values according to words. For example, in Chinese, "ro" and "ze", which play a role of Japanese particle,
The predetermined value TH for the operation of the updating unit is set to be large, such as 50 or 60. (3) Phonetic characters and reading symbols mean characters and symbols that are input to identify ideographic characters, and ideographic characters such as alphabetic phonetic characters, kana-like phonemic characters, and other Arabic numerals. Characters that have many common elements, the note code of the Republic of China, the second note code, and the like. The “input” may be a handwriting input that has recently been put into practical use, instead of a keyboard input. Also,
Kanji is not limited to ideographic characters, but also includes phonograms and phonemic characters used integrally with ideographs, such as in kana in word notation in Japanese. Furthermore, the ideographic characters are not limited to pure Chinese characters in Chinese, but also Japanese characters such as "toge", which means foreign characters (as viewed from China), and the characters "nama", which means repeated Chinese characters. It also includes various symbols used in ordinary sentences, regardless of language and language.

(4) In the present embodiment, as in the case where the updating section and the array reviewing section are integrally manufactured, one component (requirement) of the present invention is changed for the convenience of program creation and manufacturing. A plurality of objects are physically formed, a plurality of components are physically integrated, or these are appropriately combined.
Also, the memory of the fixed large dictionary unit can be replaced according to a document to be created, such as Japanese or Chinese, irrespective of its type such as ROM and volatility.

(5) When a user-defined word is newly stored in the learning dictionary unit, the number of times that the word is used is at least subtracted by the updating unit even if the same sound word is not already stored. (FUNM) + threshold + margin (1). This prevents the user-defined word from being deleted from the learning dictionary unit by only one update operation.

(6) The pronunciation symbol associated with the user-defined word is not necessarily the exact pronunciation symbol of the word, but is abbreviated to improve user convenience.

[0058]

As described above, since the kanji conversion learning device according to the present invention has the advantages of both long-term learning and short-term learning, the following effects can be obtained and the conventional problems can be completely solved. Can be resolved. (1) It is possible to automatically learn kanji, words, and word notations having the same and different meanings independently of human hands, thereby improving the accuracy and input speed of kanji conversion.

(2) Since the user-defined words are stored in the learning dictionary section, the conventional user-defined word dictionary section becomes unnecessary. This saves storage resources and makes it difficult to determine which of the user-defined word dictionary unit and the learning dictionary unit should be searched first in order to select a candidate when there are multiple words with the same sound. Required. (3) Since only the kanji and words in the fixed large dictionary that have been used are stored in the learning dictionary, the storage resources can be further saved. Further, since there is no need to update the basic fixed large dictionary unit, the present invention can be applied to a dictionary unit using a ROM.

(4) It has the advantages of both a long-term learning experience and a short-term learning experience, and eliminates each disadvantage. For this reason, the kanji conversion accuracy rate is improved. (5) The frequency of use of words used in the present invention is general. As a result, the accuracy of character conversion is improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an embodiment of a kanji conversion learning device according to the present invention.

FIG. 2 is an operation flow chart at the time of a kanji conversion process in the embodiment.

FIG. 3 is an operation flowchart for selecting a homophone, updating a learning dictionary unit, and the like in the embodiment.

FIG. 4 is a more detailed operation flowchart at the time of updating a learning dictionary unit in the above embodiment.

FIG. 5 is a more detailed operation flowchart at the time of updating a learning dictionary unit in the above embodiment.

FIG. 6 is a diagram conceptually showing a configuration of storing a reading symbol in a learning dictionary unit, a kanji corresponding to the reading symbol, and the number of times of use thereof in the embodiment.

FIG. 7 is a diagram conceptually showing a configuration of registration of a reading symbol in a fixed large dictionary unit and a kanji corresponding to the reading symbol in the embodiment.

FIG. 8 is a configuration diagram of an example of a conventional kanji conversion learning device.

FIG. 9 is an operation flowchart when performing kanji conversion in the homonymous word learning device in the conventional example.

FIG. 10 is an operation flow chart when selecting a correct word from a plurality of homonymous words in the homonymous word learning device in the conventional example.

FIG. 11 is an operation flowchart when the dictionary unit is updated in the homonymous word learning device in the conventional example.

[Explanation of symbols]

 Reference Signs List 10 input unit 20 conversion unit 25 display unit 30 selection unit 35 homophone / user-defined word designation unit 40 learning dictionary unit 45 fixed large dictionary unit 50 update unit 51 array review unit 60 first buffer 65 second buffer 70 output unit 80 User-defined word generator

Continuation of the front page (56) References JP-A-4-199456 (JP, A) JP-A-4-77954 (JP, A) JP-A-63-136252 (JP, A) JP-A-63-106070 (JP, A) JP-A-3-282875 (JP, A) JP-A-5-2579 (JP, A) JP-A-6-124301 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB G06F 17/21-17/26

Claims (4)

(57) [Claims] 1. If there are a plurality of corresponding words, the reading symbols and the corresponding words are arranged in accordance with the frequency of use, and the registered fixed large dictionary section, the reading symbols and the corresponding words are registered. A learning dictionary unit for storing words and the number of times of use firstly in an array in accordance with the principle that priority is given to the number of times of use and secondly to the priority of newness of use; The dictionary part is searched in the order of the fixed large dictionary part, and all the corresponding words are detected.If the corresponding word is unique, the word is arranged first, and if there is a plurality of corresponding words, the word is arranged first. A conversion unit for selecting a target word as a target of kanji conversion, and when there are a plurality of corresponding words, if the selection result of the conversion unit is incorrect, the conversion unit is instructed by the user to that effect. Detected A selection unit that allows the user to select a word intended by the user from among other words, and selects the conversion unit as a result of the kanji conversion. If the corrected word is not stored in the learning dictionary section, the number of times of use is set to 1 and the word is newly stored as a search key in association with the reading symbol, and is already stored. If the word has been used, 1 is added to the number of times the word is used, and then the word arrangement in the learning dictionary section at the input reading symbol is firstly used by a large number of times, and secondly A sequence review unit that revises the rule based on the principle that priority is given to the word. If the number of times the word is used after adding 1 by the sequence review unit becomes a predetermined value or more, this is detected, and then the learning is performed. A fixed number of times all words are used in the dictionary A kanji conversion learning method comprising: an updating unit that compares a reduced value with a separately determined threshold value and deletes a word whose value is less than the threshold value from the learning dictionary unit. apparatus. 2. A method in which a user-defined word is created in accordance with a definition word creation instruction from a user, and the user-defined word is made to correspond to a phonetic symbol input as a search key, and both are stored in the learning dictionary unit. 2. The kanji conversion learning device according to claim 1, further comprising a user-defined word creation unit. 3. The kanji conversion learning device according to claim 2, wherein the user-defined word creating unit can use an off-line method. 4. The method according to claim 1, wherein the sequence reviewing unit is configured to perform a search if the search-reading symbol of the user-defined word created by the user-defined word creation unit is already stored in the learning dictionary unit when another word having the same sound is already stored. If the number of uses of the new user-defined word is the same as the number of uses of the homonymous word or the larger of the number of uses and the predetermined value, if a plurality of homonymous words are stored, The kanji conversion learning device according to claim 2 or 3, wherein the kanji conversion learning device is set to a maximum value or a larger value of the maximum value and a predetermined value among the number of times of use.