JPH0728806A - Character processor and method therefor - Google Patents

Abstract

PURPOSE:To save the handling time of the selection of a homonym candidate and to improve operability by estimating the field information of a document at the time of calling the document and changing the sequence of the conversion candidate of a KANA(Japanese syllabary)/KANJI(Chinese character) according to the estimation. CONSTITUTION:A fixed memory ROM has an inter-field degree of association table WTBL and a keyword dictionary KDI, and a random access memory RAM is provided with an input buffer IBUF storing the reading column for which a key input is performed or the conversion result of a KANA/KANJI conversion word, a homonym buffer DOBUF storing the homonym candidate of the conversion result and a document buffer TBUF storing a field information table BTBL and document data. By the input from a keyboard KB, a CPU reads the control signal of the ROM and various kinds of controls are performed. Namely, the input is converted into a notation by referring to a dictionary KDIC, etc., and the document edited by the notation is stored. It is called, the field information corresponding to a routine document is called, the field information to which the document belongs is estimated, and the information is set to a storage means.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character processing apparatus and method for inputting kana characters and editing kana-kana mixed sentences by kana-kanji conversion.

[0002]

2. Description of the Related Art At present, in a character processing device such as a Japanese word processor, it is common to input kanji / kana mixed sentences using kana / kanji conversion. The biggest problem of the Kana-Kanji conversion method is the homophone. For example, read "Kiko"
There are many conversion candidates such as "climate", "mechanism", "eccentricity", "your school", etc., but it is impossible for the device to know which one the operator wants. Therefore, a general configuration is a system in which all possible candidates are displayed in a general Japanese language in a likely order, and the operator selects one from the displayed candidates.

[0003]

However, since the display order of the conversion candidates is determined based on a general viewpoint, the candidates desired by the operator are not always displayed at the top. For example, with regard to "Kikou" mentioned above, "climate" is the most natural conversion for people who enter documents that discuss geography and geology, but people who create engineering documents such as patents. For me, "mechanism" is the most natural transformation. In this way, the order of conversion candidates is usually different depending on the field in which the created document is a topic.
Of course, the word "climate" can never appear in an engineering document, but it is unlikely, and "mechanism" is the most natural first candidate in engineering documents.

However, in the conventional apparatus, the order of conversion candidates is fixed, and the conversion candidates are displayed in the same order both when inputting a document in the geographical field and when inputting a document in the engineering field. If the field information of the created document is known, more likely conversion candidates can be converted to higher ranks.

For this reason, there has been proposed an attempt to improve the conversion rate by extracting field information from the process of Kana-Kanji conversion during input and using it for the next Kana-Kanji conversion. According to this method, field information is gradually extracted in the input process, and when converting to Kana / Kanji, the field information is referred to and words in the field are preferentially converted. However, in that method, since the field information is stored by input, the field information is not stored when, for example, a standard document is read, and thus there is a drawback that the conversion rate cannot be improved.

[0006]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned drawbacks, and an object of the present invention is to estimate the field information of a document at the time of calling the document, and to determine the field information according to the estimated field information. It is an object of the present invention to provide a character processing device and a method for changing the order of conversion candidates for kana-kanji conversion to realize kana-kanji conversion input with high operability and with less trouble of selecting homophone words.

[0007]

[Means for Solving the Problems]
In order to achieve the above object, according to the present invention, in a character processing device for inputting a reading by a kana character string and editing a document including a kanji kana mixed notation, an input means for inputting the reading and a word reading , Dictionary information in which word information including notation and field information is stored in association with each other, field information storage means for storing field information of a document being edited, and reading input by the input means is stored in the field information storage. Unit and the dictionary unit, a conversion unit for converting into a notation, a document storage unit for storing a document edited by the notation obtained by the conversion unit, and a document saved in an auxiliary storage unit for the document A calling means for calling the storage means; an estimating means for estimating the field information to which the document belongs from the document called by the calling means; and a field information estimated by the estimating means. Characterized in that it comprises a setting means for setting the field information storage unit.

According to another aspect of the present invention, in a character processing device for inputting a reading by a kana character string and editing a document mixed with kanji and kana, the document stored in the auxiliary storage unit is stored in the document memory. To estimate the field information to which the document belongs from the called document, set the estimated field information in the field information memory, and input reading.
The input reading is converted into the notation by referring to the dictionary and the field information memory in which the word reading and the word information including the notation and the field information are stored in association with each other, and the notation obtained by the conversion And a character processing method for editing the standard document stored in the document memory.

[0009]

With this structure, the conversion unit converts the reading input from the input unit into a notation by referring to the field information storage unit and the dictionary unit, and the document edited by the obtained notation is stored in the document storage unit. Remember. When the calling unit calls the document from the auxiliary storage unit to the document storage unit, the field information corresponding to the standard document is called to the field information storage unit, and the estimation unit acquires the field information to which the document belongs from the called document. The field information which is estimated and set by the setting means is set in the field information storage means.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of the present invention will be described in detail below with reference to the accompanying drawings.

<First Embodiment> FIG. 1 is a block diagram showing a first embodiment of a character processing apparatus according to the present invention. In the figure, a CPU is a microprocessor, which performs operations for character processing, logical judgments, etc., and each component connected to these buses via an address bus AB, a control bus CB, and a data bus DB. To control.

The address bus AB is a microprocessor C
An address signal for instructing a component to be controlled by the PU is transferred. 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.

DIC is a dictionary, which is referred to when performing Kana-Kanji conversion.

Next, the ROM is a read-only fixed memory that stores control procedures by the microprocessor CPU, which will be described later with reference to FIGS. 12 to 18, and fixed data. WTBL is an inter-field relevance table. K
DIC is a keyword dictionary. KANJID is a Kanji dictionary.

The RAM is a writable random access memory having a structure of 1 word 16 bits, and is used for temporary storage of various data from each constituent element. IBU
F is an input buffer for storing the conversion result of the phonetic string or the kana-kanji conversion word keyed in. DOBUF is a homophone word buffer that stores the homophone word candidates of the conversion result obtained by the kana-kanji conversion. BTBL is a field information table. TBUF is a document buffer that stores document data.

Reference numeral KB denotes a keyboard, which is a character symbol input key such as an alphabet key, a hiragana key, or a katakana key, and various types of character processing devices such as a conversion key, an execution key, a document call key, and a document save key. It is equipped with various function keys for instructing functions.

DISK is an external storage for storing document data, and stores a document created in the document buffer. The stored document is instructed by the keyboard.
Called when needed.

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

DBUF is a display buffer memory for storing data to be displayed.

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 pattern of the dot configuration and the display of the cursor on the display device CRT are controlled by the CRT controller.

Further, CG is a character generator, which stores a pattern of characters and symbols to be displayed on the display device CRT.

The character processing apparatus of the present invention comprising the above-mentioned 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 and its microprocessor CP
U reads out various control signals stored in the ROM,
Various controls are performed in accordance with those control signals.

FIG. 2 is a diagram showing an example of field processing according to the first embodiment.

In the figure, an underline indicates a reading line being input, and a black rectangle means a cursor. Since (a) is talking about the geographic field, the field information of the document is inclined toward the geographic field in the process of Kana-Kanji conversion, and when "Kikono" is converted, it is converted to "climatic". Since (b) is talking about the engineering field, the field information of the document is inclined toward the engineering field in the process of Kana-Kanji conversion, and when "Kikono" is converted, it is converted to "Mechanical".

FIG. 3 is a diagram showing the structure of the dictionary DIC.

Word information for kana-kanji conversion is stored. The reading of the word is stored in the "reading". The “notation” stores the notation of the word. The "grammar information" stores word grammatical information such as part-of-speech information. The "frequency" stores the frequency of the word. The frequency is described within the range of 1 to 5, meaning that 5 is the most likely word. The "field information" stores the field id to which the word belongs and the certainty factor of the field. Multiple field ids can be described in one word. The field id is any of field 1 to field n.
“Geography”, “Geography”, and “Engineering” for clarity
It is described as "office work". The certainty factor takes a value in the range of 0 to 1. 1 means 100% surely a word in the field. 0 means that the word has no relation to the field. When the field is not described, it is interpreted as the confidence level 0, and the field with the confidence level 0 is not described in the dictionary. Words that do not belong to any category at all (for example, "this")
Etc.) means that there is no description of this “field information”.

When a word is selected, the field information of the document is updated according to the field to which the word belongs and its certainty factor.

FIG. 4 is a diagram showing the structure of the keyword dictionary KDIC.

The keyword dictionary is a dictionary in which reading strings specific to a field are stored, and the field of the document is updated only by inputting the reading written therein.

In “reading”, a reading string of keywords is stored. In the “field information”, information equivalent to that shown in the dictionary DIC (field ID and certainty factor) is stored. Multiple fields can be described in the field information.

FIG. 5 is a diagram showing the structure of the input buffer IBUF.

The input buffer IBUF stores a keystroked reading string, a conversion result obtained by converting the reading string to kana-kanji characters, and a notation of the next candidate when the next candidate display is instructed. The character string stored in the input buffer IBUF is displayed with an underline on the screen.

(A) is a diagram showing a state in which the reading string "Kikono" is input. When the conversion key is pressed here, the kana-kanji conversion is performed, and the conversion result "Mechanism" is stored as shown in (b). When the conversion key is further pressed, the next candidate “climate” is stored as shown in (c).

FIG. 6 is a diagram showing the structure of the homophone word buffer DOBUF.

When the reading string is converted into kana-kanji, the same-sound word information is stored in the same-sound word buffer DOBUF. When the next candidate display is instructed, the homophone word buffer DOBUF
In the input buffer I, the notation of the candidate indicated by the candidate No.
It is stored in BUF.

"Candidate No." is a variable indicating the currently displayed candidate. The “i-th candidate” stores information about the candidate displayed in the i-th rank of the kana-kanji conversion result.
The “notation” stores the notation of the conversion candidate. “Frequency” is a copy of the frequency of the independent word of the candidate (the value described in the dictionary DIC). "Field information j" is a copy of the field id and the certainty factor of the independent word from the dictionary DIC.

FIG. 7 is a diagram showing the structure of the field information table BTBL.

The field information table BTBL stores field information accumulated in response to kana-kanji conversion input of the document. For each of the fields 1 to n, the likelihood (likelihood) of the field is stored. In the figure, the field 1 is shown as "engineering field", the field 2 as "geographical field", and the field 3 as "medical field". The high likelihood of the field i means that the document is likely to belong to the field i.

FIG. 8 is a diagram showing the structure of the inter-field relation degree table WTBL.

The inter-field relevance table is a table showing relevance between fields. The value W _{ij at} the intersection of row i column j
Indicates the magnitude of the influence of the field j on the field i, and the larger the effect, the larger the value. In other words, the degree of influence of the field j on the field i means that when the word of the field j is selected, the likelihood of the field i of the document being input should be increased (the degree of association is large). It indicates whether the likelihood should be lowered (relevance is negative), or it is better not to change the likelihood of field i (relevance is 0). Intersection W _kk of row k column k
Is an effect of the field k on the field k itself and naturally has a positive value.

FIG. 9 is a diagram showing the structure of a Kanji dictionary KANJID which describes to which field each Kanji code belongs.

The field information stores the field id to which the word belongs and the certainty factor of the field. Multiple fields i in one kanji
d can be described. The field id is any of field 1 to field n. In the figure, "geography", "geography", "engineering", "office work", etc. are described for easy understanding. Confidence is 0
Take a value in the range of ~ 1. 1 means 100% surely a word in the field. 0 means that the word has no relation to the field. When the field is not described, it is interpreted as the confidence level 0, and the field with the confidence level 0 is not described in the dictionary. Kanji that do not belong to any field (for example, "Kono") does not have field information.

For example, the Chinese character "min" is 0.6 in the geographic field,
It can be read from this table that it belongs to the geological field 0.6.

FIG. 10 shows an evaluation formula for determining the first candidate in kana-kanji conversion. The phrase likelihood is calculated as the sum of the word frequency of the converted word and the “field likelihood” of the converted word. Here, the “category likelihood” of the converted word is the product of the certainty factor of the converted word with respect to the field k ( _denoted by C _k ) and the likelihood of the field k on the field information table at that time ( _denoted by B _k ). It is determined by the maximum value of B _k × C _k .

The converted word having the largest phrase likelihood obtained as described above is converted as the first candidate, and the second and subsequent candidates are displayed in descending order of the phrase likelihood.

FIG. 11 shows a method of updating the field information table when a word is selected.

When a word is selected, the likelihood of each field in the field information table is updated as follows according to the field to which the word belongs.

First, the certainty factor ( _denoted as C _k ) of the field k of the selected word is obtained from the dictionary DIC, then the influence degree W _ki that the field k has on the field i is obtained from the inter-field relevance table, and the certainty factor C _{k is obtained.} Get on. If there is no description of the field k in the field information of the selected word, the certainty factor of the field k is 0. The value obtained by this multiplication is summed from the fields 1 to n and added to the likelihood (denoted as B _i ) of the field i in the field information table. The calculation of the likelihood B _i of the field i is performed on the likelihoods of the fields 1 to n.

Further, when the word converted by the first candidate is not selected (when a word other than the first candidate is selected),
The likelihood of the field in the field information table is updated as follows.

First, the degree of non-selection influence (indicated by X in the figure) on non-selection of words in the field k on the field i is defined. Here, the degree of non-selection influence of the field i on the field i is the field i.
Is a negative value of the influence W _{ii on} the field i, i ≠ j
In this case, the degree of non-selection influence given to the field i by the field j is 0.

Next, the certainty factor ( _denoted by d _k ) of the field k of the word which has been converted by the first candidate but not selected (non-selected word) is obtained, and the non-selection influence of the field k on the field i is obtained. Multiply by degrees. The sum of the fields 1 to n is added to the multiplied value and added to the likelihood B _i of the field i in the field information table. The likelihood calculation for the field i is performed for the likelihoods from the field 1 to the field n. As a result, the field information table is affected in the opposite direction from when the field to which the word converted by the first candidate belongs is selected.

FIG. 12 shows a method of updating the field information table when the standard document is called and the field is estimated.

1 Kanji included in the called standard document 1
For kanji, the field information table is updated according to this formula.

Firstly seeking confidence C _k in the field k kanji, we obtain the degree of influence W _k that field k has on the field i from the field relevancy table, multiplied further multiplied by a certain value alpha. If there is no description of the field k in the kanji field information, the certainty factor of the field k is 0. α is a correction value for matching the effect of word selection on the field likelihood and the effect of Chinese characters on the field likelihood. For example, when one word consists of two Kanji characters on average, the value is 0.5. The sum of the fields 1 to n is added to the multiplied value and added to the likelihood B _i of the field i in the field information table. The likelihood calculation for the field i is performed for the likelihoods from the field 1 to the field n.

Next, the above operation will be described according to the flow.

FIG. 13 is a flow chart for explaining the operation of fetching the key input and processing according to the first embodiment.

In step 13-1, the key data to be keyed is input from the keyboard.

At step 13-2, the type of the fetched key is determined and the determination result is output. Depending on the determination result, the process branches to each key processing routine.

If it is a conversion key, step 13-3
And the conversion process is performed in step 13-3 as detailed in FIG. After the conversion process is completed, in step 13-8, it is determined whether or not the conversion process is completed by hitting the execution key. If the conversion key is other than the execution key, the process loops to step 13-2 to process the key.
When it is not the execute key, the process loops to step 13-1 to skip the key.

If the key is a reading key such as hiragana, the process branches to step 13-4 and the reading accumulation process described in detail in FIG. 14 is performed.

If the key is a document save key, step 13-5
And the document saving process described in detail in FIG. 18 is performed.

If the key is a document call key, step 13-6
And the document calling process described in detail in FIG. 19 is performed.

For other keys, the process branches to step 13-7 to perform other processes such as insertion, deletion, symbol input, etc. which are carried out in a normal character processing device.

When each process is completed, the process branches to step 13-1.

FIG. 14 shows step 13 according to the first embodiment.
4 is a flowchart illustrating in detail a "reading and accumulating process" of No. 4;

In step 14-1, the input readings are accumulated in the input buffer IBUF. In addition, the state of the reading column is displayed.

In step 14-2, it is searched whether or not the reading string stored in IBUF is registered in the keyword dictionary KDIC.

At step 14-3, it is judged whether or not it is registered, and if it is not registered, the process directly returns.

If it has been registered, the process proceeds to step 14-4 to initialize the loop variable i.

In step 14-5, it is determined whether or not i exceeds the number of fields n, and if not, step 14
Proceed to -6 to initialize the loop variable j.

In step 14-7, it is judged whether or not j exceeds the number of fields n, and if not, step 1
In 4-8, the field likelihood updating process shown in FIG. 11 is performed. The confidence factor C _j of the field j of the keyword obtained from the keyword dictionary is multiplied by the degree of influence W _ij of the field j obtained from the inter-field relevance table on the field i, and the i-th item of the field information table is added. Update the field likelihood B _i of Then, the process proceeds to step 14-9, increments the value of j, and loops to step 14-7.

When j exceeds n in step 14-7, the value of i is incremented in step 14-10.
Loop to step 14-5.

When i exceeds the number n of fields in step 14-5, the process returns.

FIG. 15 shows step 13 according to the first embodiment.
3 is a flowchart illustrating in detail a "conversion process" of No. 3;

At step 15-1, as detailed in FIG. 16, the reading string is converted into kana-kanji characters to create a homophone word buffer.

The notation of the first candidate of the homophone word buffer created in step 15-2 is set in the input buffer IBUF and displayed.

At step 15-3, the key data to be entered from the keyboard is fetched.

It is determined whether the key data acquired in step 15-4 is a conversion key. If it is not the conversion key (for example, the execution key is pressed or the next reading input is made), the candidate is confirmed without looking at the next candidate, so the process branches to step 15-7. Step 1 for conversion key
In step 5-5, the next candidate process is performed to update the value of the candidate No. in the homophone word buffer.

In step 15-6, the notation of the new conversion candidate in the homophone word buffer is transferred to the input buffer,
indicate. Then, it loops to step 15-3 and takes in the key input again.

In step 15-7, field updating processing is performed as described in detail in FIG.

At step 15-8, the candidate being displayed is confirmed, the process of embedding it in the document data or the like is performed, and the process returns.

FIG. 16 shows step 15 according to the first embodiment.
It is a flowchart explaining in detail the "kana-kanji conversion process" of -1.

In step 16-1, possible kana-kanji conversion candidates are created.

In step 16-2, the phrase likelihood corresponding to each conversion candidate is calculated as shown in FIG.

In step 16-3, the conversion candidates are sorted in the order of the phrase likelihood, and the first candidate is determined.

In step 16-4, the representation of the conversion candidates and the necessary information are registered in the homophone buffer according to the phrase likelihood order, and the process returns.

FIG. 17 shows step 15 according to the first embodiment.
It is a flow chart explaining "field update processing" of -7 in detail.

The field updating process is a part for performing the field likelihood updating process shown in FIG.

In step 17-1, the loop variable i
Is initialized.

In step 17-2, it is judged whether or not i exceeds the number of fields n, and if not, step 17
-3, the loop variable j is initialized.

In step 17-4, it is judged whether or not j exceeds the number of fields n, and if not, step 1
In 7-5, the field likelihood updating process is performed. The certainty factor C _j of the field j of the selected word is multiplied by the influence degree W _ji of the field j obtained from the inter-field relevance table on the field i, and the i-th field likelihood B of the field information table is added. Update _i . Then, it progresses to step 17-6, the value of j is incremented, and it loops to step 17-4.

When j exceeds n in step 17-4, the value of i is incremented in step 17-7 and the process loops to step 17-2.

When i exceeds the number n of fields in step 17-2, the process proceeds to step 17-8.

At step 17-8, the non-selected word pointer is initialized to point to the first non-selected word.

In step 17-9, it is judged whether or not the processing has been completed for all the non-selected words, and if completed, the routine returns.

In step 17-10, the loop variable i is initialized.

In step 17-11, i is the number of fields n
It is judged whether or not it exceeds, and when it does not exceed, step 1
Proceeding to 7-12, the loop variable j is initialized.

In step 17-13, j is the number of fields n.
It is determined whether or not the field is exceeded, and when it is not exceeded, the field likelihood is updated in step 17-14. The certainty factor d of the field j of the non-selected word indicated by the non-selected word pointer
_j is multiplied by the degree of non-selection influence X _ij given to the field i by the field _j , and the i-th field likelihood B of the field information table is added.
Update _i . Then, it progresses to step 17-15 and j
Is incremented by one and loops to step 17-13.

When j exceeds n in step 17-13, the value of i is incremented in step 17-16 and the process loops to step 17-11.

In step 17-11, i is the number of fields n
When it exceeds, the process proceeds to step 17-17 to advance the non-selected word pointer to point to the next non-selected word. Then, it loops to step 17-9.

FIG. 18 shows step 13 according to the first embodiment.
It is a flowchart explaining in detail the "document storage process" of -5.

In step 18-1, the operator is made to input the document name of the document to be saved, and the process for accepting it is performed.

In step 18-2, the document data is transformed into a storage format for storing it on the disc.

In step 18-3, the document data saving format is set to ".doc" as the document name specified by the operator.
Save to disk with the file name added.

FIG. 19 shows step 13 according to the first embodiment.
It is a flow chart which explains the "document call processing" of -6 in detail.

At step 19-1, the operator is made to input the document name of the document to be called and the processing for accepting it is performed.

At step 19-2, a process of calling the document data having the designated document name from the disc is performed. Since the called document data is loaded on the document buffer, the document currently being created on the document buffer TBUF disappears and is replaced with new document data.

In step 19-3, the pointer "Kanji character pointer" indicating the Kanji of the retrieved document data is set at the beginning of the document.

At step 19-4, one Kanji character indicated by the Kanji pointer is fetched.

In step 19-5, it is determined whether the Kanji pointer indicates the end of the document. When the end of the document is indicated, the acquired Kanji code is invalid, so the process ends and returns. If the end of the document is not indicated, the acquired Kanji is valid, so the process proceeds to step 19-6.

In step 19-6, the loop variable i
Is initialized.

In step 19-7, it is judged whether or not i exceeds the number n of fields, and if not, step 19
Proceed to -8 to initialize the loop variable j.

In step 19-9, it is judged whether or not j exceeds the number n of fields, and if not, step 1
In 9-10, field likelihood estimation processing is performed based on the equation shown in FIG. The certainty factor C _j of the field k of the acquired kanji is multiplied by α times that obtained by multiplying the degree of influence W _ij of the field j obtained from the inter-field relevance table on the field i. The field likelihood B _i is updated. Then, the process proceeds to step 19-11, increments the value of j, and loops to step 19-9.

When j exceeds n in step 19-9, the value of i is incremented in step 19-12,
Loop to step 19-7.

When i exceeds the field number n in step 19-7, the process proceeds to step 19-13, the kanji pointer is incremented, and the process loops to step 19-5.

As described above, according to the first embodiment, the field information is estimated based on the Kanji information included in the document data of the standard document, and the field information is estimated and used at the same time when the standard document is called. Therefore, in the kana-kanji conversion when calling and editing a fixed form document, the order of candidate display can be accurately performed, so that a comfortable character processing device with high operability can be realized.

<Second Embodiment> In the above first embodiment, the field information is estimated based on the Kanji information included in the document data of the standard document. However, the present invention is not limited to this, and it is also possible to perform morphological analysis on document data and estimate the field information assuming that one word included in each clause is selected.
Most of the drawings are the same as those of the first embodiment, but the flow at that time is shown in FIG. 20 instead of FIG.

Along with this, the control procedure stored in the ROM of FIG. 1 is as described in FIGS. 13 to 18 and 20.

Next, the above operation will be described according to the flow.

Since the flow different from the first embodiment is only in FIG. 20, only that part will be described.

FIG. 20, which replaces FIG. 19 of the first embodiment, is a flow chart for explaining in detail the "document calling process" of step 13-6.

In step 20-1, the operator is made to input the document name of the document to be called, and processing for accepting it is performed.

In step 20-2, a process of calling the document data having the designated document name from the disc is performed. Since the called document data is loaded on the document buffer, the document currently being created on the document buffer TBUF disappears and is replaced with new document data.

At step 20-3, the retrieved document data is morphologically analyzed.

In step 20-4, a pointer "bunsetsu pointer" indicating a phrase in the document data subjected to morphological analysis.
Is set to indicate the clause at the beginning of the document.

At step 20-5, one clause indicated by the clause pointer is fetched.

In step 20-6, it is determined whether the phrase pointer indicates the end of the document. When the end of the document is indicated, the clause cannot be acquired, so the processing ends and returns. If the end of the document is not indicated, the acquired clause is valid, so the process proceeds to step 20-7.

In step 20-7, the loop variable i
Is initialized.

In step 20-8, it is judged whether or not i exceeds the number n of fields, and if not, step 20
Proceed to -9 to initialize the loop variable j.

In step 20-10, j is the number of fields n.
It is judged whether or not it exceeds, and when it does not exceed, step 2
Go to 0-11. In Step 20-11, as shown in FIG.
Based on the formula shown in, the field likelihood is estimated assuming that the words constituting the bunsetsu are selected. The certainty factor C _j of the field j of the word constituting the phrase pointed to by the phrase pointer is multiplied by the influence degree W _ij of the field j obtained from the inter-field relevance table on the field i, and the result is added to the field information table. Update the i-th field likelihood B _i . Then, step 20-
Go to 12, increment the value of j and loop to step 20-10.

When j exceeds n in step 20-10, the value of i is incremented in step 20-13 and the process loops to step 20-8.

When i exceeds the field number n in step 20-8, the process proceeds to step 20-14, the phrase pointer is incremented, and the process loops to step 20-6.

As described above, according to the second embodiment, the morphological analysis is performed on the document data of the fixed form document, and the field information is estimated as one word included in each clause is selected, Since the field information is estimated and used at the same time when the standard document is called, the order of candidate display can be accurately performed in the Kana-Kanji conversion when the standard document is called and edited, resulting in high operability and comfort. A character processing device can be realized.

[0135]

As described above, according to the present invention,
When the document is called, the field information is estimated from the called document, so the order in which candidates are displayed can be accurately set during kana-kanji conversion when the document is called and edited. A processing device can be realized.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of a character processing device according to the present invention.

FIG. 2 is a diagram showing an example of field processing according to the first embodiment.

FIG. 3 is a diagram showing a configuration example of a dictionary DIC according to the first embodiment.

FIG. 4 is a diagram showing a configuration example of a keyword dictionary KDIC according to the first embodiment.

FIG. 5 is a diagram showing a configuration example of an input buffer IBUF according to the first embodiment.

FIG. 6 is a homophone word buffer DOBUF according to the first embodiment.
It is a figure showing an example of composition.

FIG. 7 is a field information table BTBL according to the first embodiment.
It is a figure showing an example of composition.

FIG. 8 is an inter-field relevance table WT according to the first embodiment.
It is a figure showing the example of composition of BL.

FIG. 9 is a diagram showing a configuration example of a Kanji dictionary KANJID according to the first embodiment.

FIG. 10 is a diagram showing an evaluation formula for determining a first candidate in kana-kanji conversion according to the first embodiment.

FIG. 11 is a diagram showing an update formula of a field information table at the time of word selection according to the first embodiment.

FIG. 12 is a diagram showing an update formula of a field information table at the time of selecting a word according to the first embodiment.

FIG. 13 is a flowchart illustrating an operation of capturing a key input and performing a process according to the first embodiment.

FIG. 14 is a flowchart illustrating in detail the “reading and accumulating process” in step 13-4 according to the first embodiment.

FIG. 15 is a flowchart illustrating in detail “conversion processing” in step 13-3 according to the first embodiment.

FIG. 16 is a flowchart illustrating in detail a “kana-kanji conversion process” of step 15-1 according to the first embodiment.

FIG. 17 is a flowchart illustrating in detail the "field updating process" of step 15-7 according to the first embodiment.

FIG. 18 is a flowchart illustrating in detail the “document saving process” of step 13-5 according to the first embodiment.

FIG. 19 is a flowchart illustrating in detail the “document calling process” of step 13-6 according to the first embodiment.

FIG. 20 is a flowchart illustrating in detail the “document calling process” of step 13-6 according to the second embodiment.

[Explanation of symbols]

 CPU Microprocessor DIC Kana-Kanji conversion dictionary ROM Read-only memory RAM Random access memory DISK External storage KB Keyboard CR Cursor register DBUF Display buffer memory CRTC CRT controller CRT display CG character generator WTBL Inter-field relation table KDIC keyword dictionary IBUF input Buffer DOBUF Homophone buffer BTBL Field information table TBUF Document buffer KANJID Kanji dictionary

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Atsushi Sakakibara 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Kazuyo Ikeda 3-30-2 Shimomaruko, Ota-ku, Tokyo Canono (72) Inventor Yuji Kobayashi 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Hitoshi Haruri, 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Co., Ltd. (72) Inventor Chika Kogusu 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Mayumi Morita 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Mariko Kishino 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Risuna Adachi 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon within Co., Ltd. (72) inventor Suzuki ShigeruKyo Ota-ku, Tokyo Shimomaruko 3-chome No. 30 No. 2 Canon within Co., Ltd.

Claims (4)

[Claims] 1. A character processing device for inputting a phonetic reading by a kana character string to edit a document containing kanji and kana mixed notation, inputting means for reading the reading, word reading, and word information including notation and field information. And a dictionary means for storing the field information of the document being edited, and a reading input by the input means by referring to the field information storage means and the dictionary means. A converting means for converting into a notation, a document storing means for storing the document edited by the notation obtained by the converting means, a calling means for calling the document stored in the auxiliary storage section to the document storing means, The estimating means for estimating the field information to which the document belongs from the document called by the calling means, and the field information estimated by the estimating means are stored in the field information storage means. Character processing apparatus characterized by having setting means for. 2. The character processing apparatus according to claim 1, wherein the estimating unit estimates the field information of the document based on the Chinese character data in the called document. 3. The estimating unit estimates the field information of the document based on the field information of the word obtained as a result of the morphological analysis in the called document. The character processing device according to 1. 4. In a character processing device for inputting a reading with a kana character string and editing a document mixed with kanji and kana, a document stored in an auxiliary storage unit is called into a document memory, and the document is read from the called document. Estimating the field information to which the document belongs, setting the estimated field information in the field information memory, inputting the reading, and inputting the reading as word reading and word information including notation and field information A character processing method, which refers to the dictionary and the field information memory that are stored in association with each other, converts into a notation, and edits the standard document stored in the document memory according to the notation obtained by the conversion. .