JPH05233604A - Word registering method - Google Patents

Abstract

PURPOSE:To execute the management in the same way as a word registered by a user's instruction by registering as inputted reading, a selected notation, and a word having an undefined part of speech, when a word which does not exist in a dictionary is selected. CONSTITUTION:In response to a user's input converting operation, in the case a word which does not exist in a dictionary for storing word information for converting reading of the word to a notation is selected, it is registered in the dictionary as inputted reading, a selected notation, and an undefined part of speech. Also, based on the stored information, a part of speech of the word is estimated, and this word is registered in the dictionary as a word having inputted reading, a selected notation, and a part of speech obtained by estimation. In such a way, an automatically registered word is also registered in the same dictionary in the same way as a user's registered word, and a user can execute unitary management of a registered word dictionary. Also, by the estimated part of speech, with regard to the automatically registered word, as well, its effective use adapted to an analysis of an input sentence can be executed in its subsequent conversion.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to Japanese language processing, and more particularly to a processing apparatus for creating a correct sentence in Kana-Kanji conversion.

[0002]

2. Description of the Related Art Conventionally, in inputting Japanese, kana-kanji conversion is executed using a word dictionary that stores word reading, notation, part of speech, and connection information. Then, it is proposed that a user prepares a registered word dictionary for newly registering a necessary word, and when a word that does not exist in the word dictionary is used, this word is automatically registered in the registered word dictionary. ing.

[0003]

The word registered in this way (hereinafter, referred to as an automatically registered word)
Is different from the user-registered word that the user has instructed and registered, and no part of speech is specifically determined. Therefore, it cannot be managed in the registered word dictionary in the same manner as user registered words.

Further, since the part-of-speech information is lacking, it is not suitable for use in the form of analysis of the input sentence at the time of conversion.

[0005]

SUMMARY OF THE INVENTION In order to solve the above problems, the present invention stores word information for converting word reading into notation in response to an input conversion operation by a user. When a word that does not exist in the dictionary is selected, the word is registered in the dictionary as a word having the input phonetic reading, the selected notation, and the undetermined part of speech.

According to another aspect of the present invention, in response to a user's input conversion operation, the operation does not exist in the dictionary storing the word information for converting the reading of the word into the notation. When a word is selected, the part of speech of the word is estimated based on the stored information, and the word is regarded as a word having the input pronunciation, the selected notation, and the part of speech obtained by the estimation. , Register in the dictionary.

[0007]

As a result, the automatically registered word is registered in the same dictionary as the user registered word, and the user can centrally manage the registered word dictionary.

Further, the estimated part-of-speech enables the automatically registered word to be effectively used in the subsequent conversion in accordance with the analysis of the input sentence, and the accurate conversion suitable for the user's purpose can be realized. To do.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the drawings.

FIG. 1 shows an example of the overall configuration of the present invention.

In the configuration shown in the figure, the CPU is a microprocessor, which performs arithmetic operations for character processing, logical judgments, etc., an address bus AB, a control bus CB,
The respective components connected to those buses are controlled via the data bus DB.

The address bus AB is a microprocessor C
It transfers an address signal indicating a component to be controlled by the PU. The control bus CB transfers and applies a control signal of each constituent element to be controlled by the microprocessor CPU. The data bus DB transfers data between the constituent devices.

Next, the ROM is a read-only fixed memory. PA is a program area in which control procedures by the microprocessor CPU, which will be described later with reference to FIGS. 6 to 12, are stored.

The RAM is a writable random access memory having a structure of 1 word and 16 bits, and is used for temporary storage of various data from each constituent element.

TBUF is a document buffer, which is a memory for storing the document information input from the keyboard KB.

YBUF is an input reading buffer memory for storing readings input from the keyboard KB.

DIC is a word dictionary for performing Kana-Kanji conversion.

TDIC is a registered word dictionary in which words can be registered to perform Kana-Kanji conversion.

GDIC is a learning data dictionary for storing learning data for kana-kanji conversion.

DBPOOL is a homophone candidate storage memory for storing information obtained by analyzing and converting YBUF readings into phrases.

YJPOOL works in conjunction with DBPOOL,
It is a memory for storing information on examples applied to the same phonetic word candidates.

LRNDAT is a learning data storage memory that stores learning states of individual words and examples.

FZTBL is an adjunct word string conversion table for associating an adjunct word with connection information stored in the DIC.

The KB is a keyboard, which is provided with various function keys such as an alphabet key, a hiragana key, a katakana key and other character symbol input keys, and a conversion key for instructing conversion.

In FIG. 1, YOMI is a key for inputting a reading, CON is a conversion instruction key for converting the input reading, NXT is a next candidate conversion instruction for changing a conversion candidate and converting it to a next candidate. Key, SEL is a selection key for confirming the current homophone display candidate and learning the candidate notation at the same time, HIRA is a hiragana conversion key for confirming the input reading in hiragana notation, and KATA is the input reading. A katakana conversion key that is determined by katakana notation, TAN is a single-kanji conversion instruction key for converting the input reading at the single-kanji level.

DISK stores a document created in a memory for storing a standard document, and the stored document is called when necessary by a keyboard instruction.

CR is a cursor register. The CPU can read and write the contents of the cursor register. The CRT controller CRTC, which will be described later, displays a cursor at a position on the display device CRT for the address stored here.

DBUF is a buffer memory for display, and TB
The pattern such as document information stored in the UF is stored.

The CRTC plays a role of displaying the contents stored in the cursor register CR and the buffer DBUF on the display CRT.

The CRT is a display device using a cathode ray tube or the like, and the display of the dot configuration pattern and the cursor on the display device CRT is controlled by the CRT controller.

Further, CG is a character generator for storing patterns of characters and symbols to be displayed on the display device CRT.

The character processing device of the present invention comprising the above-described components operates in response to various inputs from the keyboard KB, and when an input from the keyboard KB is supplied, an interrupt signal is first sent. Sent to the microprocessor CPU, which microprocessor CP
U reads out various control signals stored in the ROM,
Various controls are performed in accordance with those control signals.

An example in which automatic registration is executed in the apparatus of this embodiment having the above configuration will be described below with reference to FIG.

(A) is the state of the CRT before characters are input. In the figure, 21 indicates a cursor.

In this state, when the characters "ki", "ka", and "i" are input, the state of the CRT becomes as shown in (b). The underline indicated by 22 indicates that the kana-kanji conversion result has not been determined.

When the conversion key is input from the state of (b), kana-kanji conversion is executed on the input character string, and as a result, the state of the CRT becomes as shown in (c). After converting the first conversion result "Machine" to "Strange" by the next candidate conversion,
When the selection key is input, the conversion result is determined, word learning is executed, and as a result, the state of the CRT becomes as shown in (g). In this state, again, "ki", "ka",
When the character "i" is input and the conversion key is further input, the word is converted into "a mystery" by learning, and as a result, the state of the CRT becomes as shown in (i).

When the hiragana conversion key is input from the state of (b), the hiragana conversion is executed on the input character string, and as a result, the state of the CRT becomes as shown in (d). When the selection key is input in this state, the words of reading "kikai" and notation "kikai" are automatically registered. Further, when the characters "ki", "ka", and "i" are input in the same manner and the conversion key is further input, it is converted into "kikai" by the automatically registered word, and as a result, the state of the CRT is ( j).

The katakana conversion is executed in the same manner as the hiragana conversion, and as a result, the states of the CRT are as shown in (e) and (k).

When the single-kanji conversion key is entered from the state of (b), the single-kanji conversion is executed on the input character string,
As a result, the state of the CRT becomes as shown in (f). By repeating the single-kanji conversion, it will be converted into the notation "kai", and when the selection key is entered, the conversion result will be determined,
As a result, the state of the CRT becomes as shown in (h). here,
The words of reading "kikai" and notation "meeting" are automatically registered. From this state, "ki", "ka", "i" in the same way
When the character "" is input and the conversion key is further input, it is converted into the automatically registered word "encounter", and as a result, the state of the CRT becomes as shown in (l).

FIG. 3 shows the state of the CRT when the registered word dictionary list display is executed.

The symbol "*" on the display indicates that it is an automatically registered word, and is distinguished from the user registered word registered by the user.

FIG. 4 is a diagram showing the structure of the registered word dictionary TDIC.

The registered word dictionary is composed of a header including a management number of the dictionary, a main body which is a set of registered words, and a management table for managing the automatically registered words. For each registered word, a reading, a notation, a part of speech, and a flag for discriminating between a registered word by the user and an automatically registered word are stored. The example in the figure is an automatically registered word whose reading is “kikai”, notation “kikai”, and part of speech “undetermined part of speech”.

FIG. 5 is a diagram showing the structure of the management table. Addresses in the registered word dictionary are stored, and old and new management is performed according to the position in the table.

The operation of the above embodiment will be described according to the flow.

FIG. 6 is a flow chart showing the operation of the character processing apparatus of the present invention.

In step S6-1, some key is pressed from the keyboard and an interrupt is awaited. When a key is input, this key is identified in step S6-2, and depending on the type of key, step S6-3,
Step S6-4, Step S6-5, Step S6-
6, step S6-7, step S6-8, step S
6-9, step S6-10, step S6-11, and step S6-12.

Step S6-3 is a reading input key YOMI.
Is a process when is pressed, and the pressed reading code is stored in the input reading buffer memory YBUF.

Step S6-4 is a process when the conversion key CON is pressed, and is a character string which is input in step S6-3 and is stored in the input reading buffer memory YBUF, which is the target of Kana-Kanji conversion. To Kanji,
Output to the output buffer. When converting to kanji, the learning data dictionary is referenced to determine the first candidate of the homophone.

Step S6-5 is a single Kanji conversion key TAN.
This is the process when is pressed, and the character string that is input in step S6-3 and is stored in the input reading buffer memory YBUF is converted into the kana-kanji character at the single kanji level and output to the output buffer. To do.

In step S6-6, the hiragana conversion key HI is used.
This is the process when RA is pressed, and step S6-3
The character string to be converted into kana-kanji stored in the input reading buffer memory YBUF is converted into hiragana notation and output to the output buffer.

Step S6-7 is the katakana conversion key KA.
This is the process when TA is pressed, and step S6-3
The character string that has been input by and is stored in YBUF, which is the target of kana-kanji conversion, is converted to katakana notation and output to the output buffer.

Step S6-8 is the next candidate conversion key NXT.
This is the process when is pressed, and another candidate of the same word in the output buffer output in step S6-4 is displayed.

Step S6-9 is the processing when the selection key SEL is pressed, and the homophone in the output buffer displayed on the screen is fixed and the fixed character string is output to the document. Further, a process of learning the selected word is performed.

Step S6-10 is a process when the registration key TOR is pressed, and the word designated by the user is registered in the registered word dictionary TDIC.

Step S6-11 is a process when the list key LST is pressed, and a list of registered words in the registered word dictionary TDIC is displayed.

In step S6-12, YOMI, CO
This is a process when a key other than N, NXT, and SEL (for example, a key used for document editing such as a cursor movement key) is pressed, and a process corresponding to each key is executed.

Step S6-13 is a display process for displaying the changed part as a result of the above-mentioned respective processes. This is a widely-used process of reading one character of data in a document, developing it into a pattern, and outputting it to a display buffer.

FIG. 7 is a detailed flowchart of the conversion process of step S6-4.

In step S7-1, the character string to be converted into kana-kanji characters that has been segmented into phrase units and input is analyzed, and candidates for the output of kana-kanji conversion are stored as homophone candidate storage memory DBPOOL (same word). This is the process of outputting to the (pool). This is a process for extracting character strings in sequence into punctuated units, searching the word dictionary for analysis, and outputting only candidates that are recognized as syllables to the homophone word pool. Since the processing is generally performed and is well known, it will not be described here.

A step S7-2 checks whether or not the example pattern stored in the word dictionary exists in the analysis result output to the homophone word pool in the step S7-1. If it exists, the homonym candidate that is the target of the example is picked up as a priority candidate.

In step S7-3, the first candidate for kana-kanji conversion is determined from the priority candidates picked up in step S7-2 and the candidates for which words have been learned.

Step S7-4 is a process for displaying the output of the kana-kanji conversion stored in the output buffer, which is a process generally performed in a character processing device of the same kind, and is well known, so it will not be described here.

FIG. 8 is a detailed flowchart of the selection process of step S6-9.

In step S8-1, if the selected notation exists in the dictionary, the learning process of step S8-3 is executed as it is. If it does not exist in the dictionary, step S8-
After performing the automatic registration process of 2, the learning process is performed. The learning process of step S8-2 is to learn the selected word and increase the priority of the word after conversion. After the learning process is completed, a confirmation process for confirming a homonym is performed. The learning process is to update the learning data corresponding to the word existing in the dictionary or the registered word, is a process generally performed in the character processing device of the same kind, and is known, and therefore will not be particularly described.

FIG. 9 is a detailed flowchart of the automatic registration process in step S8-2.

In step S9-1, it is checked whether or not there is a free space in the registered word dictionary, and if there is no free space, the process branches to the processing of deleting the oldest automatically registered word in step S9-3. If there is an open area in the dictionary, step S9-2
Then, it is checked whether or not there is a free space in the management table that manages the automatically registered words. If there is no free space, the process similarly branches to step S9-4 and the oldest automatically registered word is deleted. If it is confirmed that there is a free area, step S9
In -5, word registration is executed. The management table is updated in step S9-6 when the registration is completed.

The word registration process of step S9-5 is a process of newly registering a word in the registered word dictionary, and is a process generally performed in a character processing device of the same type.

FIG. 10 is a detailed flowchart of the oldest word deletion process in steps S9-3 and 9-4.

Step S10-1 is a process for retrieving the oldest automatically registered word from the management table. Succeedingly, step S10-2 is a process of deleting the retrieved automatically registered word from the registered word dictionary. The next step S10-3 is a process of updating the management table when the deletion is completed.

FIG. 11 is a detailed flowchart of the management table updating process in step S9-6 or step S10-3.

At step S11-1, 1 is set to the counter variable i. In step S11-2, it is checked whether the address of the i-th automatically registered word exists in the management table. If it exists, the process branches to step S11-3, and the address in the registered word dictionary of the automatically registered word is obtained from the management table. In step S11-4, it is updated according to the current registered word dictionary. In step S11-5, the data is put into the management table. In step S11-6, i is updated and the process is moved to the next automatically registered word in the management table.

FIG. 12 is a detailed flowchart of the user registration process of step S6-10.

In step S12-1, it is checked whether or not there is a free area in the registered word dictionary. If not present, step S
It branches to 12-2 to check whether the automatically registered word can be deleted. If it can be deleted, the empty area is created in the registered word dictionary by deleting it in step S12-3, and the process returns to step S12-1 again. If the word cannot be deleted, the process branches to step S12-5 to display that word registration is impossible. If it is confirmed in step S12-1 that there is a free area, word registration is executed in step S12-4.
Then, in step S12-6, the management table is updated.

As described above, according to this embodiment, the part of speech of the automatically registered word is set to "undetermined part of speech" and registered.
As a result, the automatically registered word can be managed in the same manner as the user registered word.

(Second Embodiment) In the above embodiment, the part of speech of the automatically registered word is uniformly "undetermined", but in the present embodiment, this part of speech is estimated from a known homophone.

This embodiment will be described below with reference to the drawings.

The overall structure of this embodiment is the same as that shown in FIG.
Only the parts different from the first embodiment will be described below.

FIG. 13 is a diagram showing the structure of the registered word dictionary TDIC in this embodiment. The structure of the dictionary itself is the same as in Fig. 3, but in the example in the figure, the words have the reading "kikai" and the notations are "kikai", "kikai", and "meeting", respectively. In each case, the part of speech is presumed to be a "noun".

FIG. 14 is a diagram showing the structure of the management table of this embodiment. The structure of the table is the same as that in FIG. 4, but the part of speech of the above-mentioned automatically registered word “Kikai” is a noun.

FIG. 15 is a diagram showing a homophone candidate storage buffer. This buffer is created in the homophone candidate storage memory DBPOOL during conversion. The figure shows a state in which the number of candidates is N. Each candidate has a dictionary that stores the words that make up the candidate, its notation, and part-of-speech information for that word.

Next, the automatic word registration processing in this embodiment will be described with reference to FIGS.

FIG. 16 is a detailed flowchart of the automatic registration process of step S8-2 of FIG. 8 in this embodiment.

Steps S16-1 to S4-4 are the same as steps S9-1 to S9-4 in FIG. When it is confirmed that there is a free area up to step S16-4, step S16-5
In, the part of speech of the word to be registered is estimated. continue,
Word registration is executed using the part-of-speech information estimated in step S16-6, and when the registration is completed, step S16-7
To update the management table. This step S16-6-
Step 7 is basically the same as steps S9-5 to S6.

FIG. 17 is a detailed flowchart of the part-of-speech estimation process of step S16-5.

In step S17-1, it is checked whether or not a homonym candidate exists, and if it does not exist, step S17-5.
The part-of-speech of the word that is automatically registered with is the undetermined part-of-speech. If there is a homophone candidate, the process branches to step S17-2.
Step S17-2 is a process of extracting part-of-speech of all homonym candidates. In step S17-3, it is checked whether all the parts of speech of the extracted homonym candidates are the same part of speech. If all the parts of speech of the same phonetic candidate are the same, the process branches to step S17-4, and the part of speech is used as the part of speech of the automatically registered word. decide. If they are not the same, the process branches to step S17-5, and the part-of-speech of the automatically registered word is set as an undetermined part-of-speech.

As described above, according to this embodiment, the part of speech of the automatically registered word is estimated and registered based on the part of speech information of the homonym candidate of the word, so that the word is input in the subsequent conversion. It is possible to convert in accordance with sentence analysis. As a result, it is possible to build a kana-kanji conversion device that can more accurately convert the input and suits the user's purpose.

(Third Embodiment) In this embodiment, the part of speech of the automatically registered word is estimated from the reading and the notation.

FIG. 18 shows an example of the overall construction of this embodiment.

In the illustrated configuration, JHTBL is a table for obtaining the certainty factor of a particle and the part of speech connected to the particle. The other structure is the same as that of FIG.

FIG. 19 is a diagram showing the structure of the particle determination table JHTBL. For each particle, a part of speech that may be connected to the particle and a certainty factor of connecting to the part of speech are stored as a pair.

The difference between this embodiment and the second embodiment is that
Since this is only the part-of-speech estimation processing in step S16-5 in FIG. 16, this will be described below.

FIG. 20 is a detailed flowchart of the part-of-speech estimation process (step S16-5) in this embodiment.

In step S20-1, a particle which seems to be connected to the end of the word to be automatically registered is extracted from the input phonetic reading and the selected notation. In step S20-2, the part of speech that may be connected to the extracted particle and its certainty factor are calculated as the particle determination table JHTBL.
Get from In step S20-3, the part of speech of the automatically registered word is determined based on the obtained part of speech and its certainty factor.
In this determination, when there is a single part of speech that may be connected, that part of speech is, of course, determined, but when there are multiple parts of speech, the part of speech with the highest certainty may be determined. However, the part of speech that satisfies a certain degree of certainty may be used as a candidate, and other conditions may be added to make the determination.

Further, in the above example, the judgment is made by using the particle connected at the end. However, for other parts of speech such as auxiliary verbs, a table for the certainty of connection is similarly provided and used for the judgment. Good. Also, the connection with the preceding part of speech may be used instead of the end.

As described above, according to this embodiment, the part of speech of the automatically registered word is read, the particle is extracted from the notation, and the part of speech of the automatically registered word is estimated and registered from the connectability of the particle. As a result, in the subsequent conversion, the word can be converted in accordance with the analysis of the input sentence. As a result, it is possible to build a kana-kanji conversion device that can more accurately convert the input and suits the user's purpose.

In each of the above embodiments, the user can change the part of speech set in the automatically registered word by allowing the user to call and update the dictionary information for the specific automatically registered word in the registered word dictionary. It can also be possible.

At this time, the word may be designated on the list display, or may be performed when the word is displayed as a candidate by a conversion operation.

In the above description of each embodiment, the management of the automatically registered words is performed by the management table storing the addresses in the registered word dictionary, but the old and new are managed according to the actual position stored in the registered word dictionary. You can also manage. It is also possible to manage the stored words themselves by having old and new management information. It is also possible to manage the usage status by updating the management information every time the automatically registered word is used for conversion.

The present invention is applicable not only to a single device but also to a system composed of a plurality of devices.
It is needless to say that the present invention can be applied not only to a device dedicated to character processing but also to a device or system having a function of outputting a character string mixed with kanji and kana by kana-kanji conversion. Further, it goes without saying that it can be realized by providing software to the device or system.

[0101]

According to the present invention, by setting the part of speech of an automatically registered word to "undetermined part of speech" and registering it, it becomes possible to manage the automatically registered word in the same manner as the user registered word.

Further, according to the present invention, by setting the part of speech of the automatically registered word to the estimated part of speech and registering the word, the word is used in the subsequent conversion in a form in accordance with the analysis of the input sentence. It becomes possible. As a result, it is possible to realize the kana-kanji conversion suitable for the user's purpose, which can convert the input reading more accurately.

[Brief description of drawings]

FIG. 1 is a block diagram showing the overall configuration of a character processing device of this embodiment.

FIG. 2 is a diagram showing an example in which automatic word registration is executed according to Kana-Kanji conversion.

FIG. 3 is a diagram showing an example of a state in which a registered word list is executed.

FIG. 4 is a diagram showing a configuration example of a registered word dictionary.

FIG. 5 is a diagram showing a configuration example of a management table of automatically registered words.

FIG. 6 is a flowchart showing an example of the processing procedure of the entire operation of this embodiment.

FIG. 7 is a flowchart showing an example of the processing procedure of the entire kana-kanji conversion.

FIG. 8 is a flowchart illustrating an example of a procedure of selection processing.

FIG. 9 is a flowchart illustrating an example of a processing procedure of automatic registration processing.

FIG. 10 is a flowchart showing an example of a processing procedure of the oldest word deletion processing.

FIG. 11 is a flowchart illustrating an example of a processing procedure of management table update processing.

FIG. 12 is a flowchart showing an example of a processing procedure of word registration processing by a user.

FIG. 13 is a diagram showing a configuration example of a registered word dictionary in another embodiment.

FIG. 14 is a diagram showing a configuration example of an automatic registration word management table in another embodiment.

FIG. 15 is a diagram showing an example of the configuration of a homophone table.

FIG. 16 is a flowchart showing an example of an automatic registration processing procedure according to another embodiment.

FIG. 17 is a flowchart illustrating an example of a processing procedure of part-of-speech estimation processing.

FIG. 18 is a block diagram showing an overall configuration of a character processing device of another embodiment.

FIG. 19 is a diagram showing an example of the configuration of a particle determination table.

FIG. 20 is a flowchart showing another example of the processing procedure of the part-of-speech estimation processing.

[Explanation of symbols]

 CPU Microprocessor ROM Read-only memory PA Program area RAM Random access memory YBUF Reading buffer TBUF Document buffer DIC Kana-Kanji conversion dictionary TDIC registered word dictionary GDIC learning data dictionary DBPOOL Homophone candidate storage memory YJPOL example information storage memory LRNDAT learning data storage memory FZTBL Adjunct word string conversion table JHTBL Particle determination table DISK External storage KB Keyboard CR Cursor register DBUF Display buffer memory CRTC CRT controller CRT Display CG Character generator AB Address bus DB Data bus CB Control bus

Front page continuation (72) Inventor Kazuyo Ikeda 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc.

Claims (4)

[Claims] 1. In response to a user's input conversion operation, when the operation selects a word that does not exist in a dictionary that stores word information for converting word reading into notation, the word is selected. Is registered in the dictionary as a word having an input pronunciation, a selected notation, and an undetermined part of speech. 2. When a word that does not exist in a dictionary that stores word information for converting a word reading into a notation is selected by the operation in response to a user input conversion operation, the word is stored. Estimate the part of speech of the word based on the information
A word registration method, characterized in that the word is registered in the dictionary as a word having an input reading, a selected notation, and a part of speech obtained by the estimation. 3. The word registration method according to claim 2, wherein the stored information is part-of-speech information of another word stored in the dictionary corresponding to the input reading. 4. The word registration method according to claim 2, wherein the stored information is information regarding a possibility of connection between a specific word and another part of speech.