JPH113331A - Document preparing device, predictive inputting method and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce an operating burden by a user by securely predicting a character string the user desires to input. SOLUTION: This device is provided with a dictionary for predictive input 42 registering information for inputting prediction based on a reading character string inputted for preparing a document, a predictive processing part 50 acquiring plural predictive character strings based on the read character string by referring to information registered in this dictionary 42, a dictionary 44 for specifying a document field registering information for specifying the field of a document based on the content of a preparing document, a document field specifying part 52 specifying the document field by referring to information registered in this dictionary 44, a first narrowing part 54 narrowing plural predictive character strings discriminated by the part 50 corresponding to the document field specified by the part 52, and a character string candidate display part 56 displaying a narrow predictive character string and inputs a predictive character string selected from the displayed predictive character strings.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document creating apparatus for creating a document by performing a kana-kanji conversion on an input reading character string, a predictive input method, and a recording recording a program for realizing a predictive input function. Regarding the medium.

[0002]

2. Description of the Related Art In general, in a document preparation apparatus for preparing a document in Japanese, a character string is input by performing a kana-kanji conversion on an input reading character string to prepare a document.

The kana-kanji conversion function for performing kana-kanji conversion converts a character string read in hiragana into a character string mixed with kanji and kana while referring to information registered in a kana-kanji conversion dictionary.

[0004] The kana-kanji conversion function generates a character string mixed with kana-kana corresponding to a reading character string input by a user. Therefore, the user must accurately input a hiragana reading character string representing all readings of the document to be created using an input device such as a keyboard.

[0005] However, the operation of inputting all the read character strings accurately requires a great deal of effort for a user who is not accustomed to keyboard operation, for example.

Therefore, in recent years, a predictive input function has begun to be used as a function for reducing the effort for inputting a character string to be read even a little. The predictive input function predicts a character string that is likely to be input by the user without inputting all the readings of the character string to be input.

FIG. 25 shows an example of character input using the predictive input function. For example, as shown in FIG. 25 (a), when “soft” is input, the predictive input function predicts a character string that the user intends to input based on the input character string, and A character string (predicted character string) to be an input candidate is acquired, and displayed and presented to the user, for example, as shown in FIG. In the example shown in FIG. 25B, since a plurality of predicted character strings have been acquired for the input character string “Soft”, the list of predicted character strings is displayed.

[0008] Here, the user can select a character string to be input from the presented predicted character string, and input the character string as it is. Furthermore, as shown in FIG. 26, a kana-kanji conversion can be performed by inputting a reading character following a character string selected from the presented predicted character strings.

In the example shown in FIG. 26, a plurality of predicted character strings (FIG. 25B) predicted from the read character string "SOFTO" input as shown in FIG. Is an example in which the character string “software” is selected as shown in FIG. Following the “software” character string, FIG.
As shown in FIG. 26C, when the character string of "No Hinashitsu" is input and execution of Kana-Kanji conversion is instructed, FIG.
As shown in (1), Kana-Kanji conversion is performed on the input reading character string, and a kanji-kana sentence of "software quality" is generated.

In addition, the input prediction function calculates the meaning vector of the document from the average score of each word in the document being created when performing the prediction of the character string. A method of presenting a near word to a user as a predicted character string has been considered (for example, Japanese Patent Laid-Open No. 7-271).
774).

[0011]

As described above, in the conventional document creation apparatus, an input prediction function is used as a method for improving input efficiency.

However, in the above-described conventional technology, even when there are many predicted character strings predicted based on the input reading character string, the predicted character obtained by the prediction is given to the user. All columns are presented. Therefore,
The user has to search for and select a desired character string from a large number of presented predicted character strings. Therefore, there is a problem that the more the predicted character strings are presented, the more the labor is spent on the processing for selecting the character strings.

[0013] In addition, the method of making a prediction using the semantic vector of a document also uses this method because the average value of the scores of each word in the document is simply obtained and used as the semantic vector of the document. It is considered that the accuracy of specifying the meaning of the document is low, and conversely, there is a problem that the character string desired by the user is not predicted because the document is determined to have the wrong meaning.

The present invention has been made in consideration of the above-described circumstances, and it is possible to accurately predict a character string that a user wants to input, thereby reducing a user's operation load. An object is to provide an apparatus, a prediction input method, and a recording medium.

[0015]

SUMMARY OF THE INVENTION According to the present invention, there is provided a document creating apparatus for creating a document by performing a kana-kanji conversion on an input reading character string, based on the reading character string input for document creation. A predicted input dictionary in which information for performing input prediction is registered, and with reference to information registered in the predicted input dictionary, based on a read character string input for document creation, Prediction processing means for obtaining a plurality of predicted character strings which are character strings predicted to be input, and a document field in which information for specifying the field of the document is registered based on the content of the created document Specific dictionary,
With reference to the information registered in the document field identification dictionary,
Based on a character string in the document being created, a document field specifying unit that specifies the field of the document, and a plurality of predicted character strings determined by the prediction processing unit, the document specified by the document field specifying unit. First narrowing means for narrowing down according to the field, and display means for displaying a predicted character string narrowed down by the first narrowing means, and a predicted character selected from the predicted character string displayed by the display means It is characterized by inputting a column.

Further, information indicating the frequency of use of each word for each document field is registered in the document field specifying dictionary, and the document field specifying means corresponds to each character string in the document being created. It is characterized in that the sum of the values indicated by the word frequency information registered in the dictionary for document field identification is obtained for each field and the document field is specified.

Further, the document field specifying means stores the document field specifying dictionary corresponding to each character string in the document being created in accordance with a relative positional relationship with the character string to be input predicted. It is characterized in that the information indicating the use frequency of the registered words is weighted to obtain the sum.

Further, in the predictive input dictionary, field-specific availability data in which data indicating whether or not each word is used is set for each document field is registered.

Thus, by specifying the document field using the character string information and the position information of the character string in the document currently being created, the accuracy of specifying the document field is improved, and the document field specified by this method is used. By narrowing down the prediction string candidates
Only the most frequently used optimal character string can be presented to the user, and the operation of the user selecting from the displayed predicted character string can be simplified.

[0020] The present invention is also provided for registering a character string determined to be input, wherein the predictive input user dictionary is referred to together with the predictive input dictionary when the predictive means obtains the predictive character string. Character string storage means for registering a character string selected and input from the predicted character string in the predictive input user dictionary, and a predictive character string acquired by the predictive processing means registered in the predictive input user dictionary A second narrowing-down unit that narrows down only the obtained character string as a predicted character string when the search has been performed.

[0021] This allows the user to narrow down the candidates for the predicted character string by using the character string previously determined, thereby simplifying the operation for the user to select from the displayed predicted character string candidates. Can be.

In addition, a character string determined to be input and frequency information indicating a use frequency corresponding to the character string are registered. When the prediction input user dictionary referred to together with the prediction input dictionary and the character string determined to be input are not registered in the prediction input user dictionary, this character string is A character string storage unit to be registered in the user dictionary, and a character string registered as input by the character string storage unit when the character string determined to be input is not registered in the prediction input user dictionary. Frequency information storage means for updating and registering frequency information corresponding to the character string when the frequency information is not registered in the prediction input user dictionary; Means for storing a plurality of predicted character strings obtained by referring to the information registered in the predictive input dictionary and the predictive input user dictionary, and storing the character strings registered in the predictive input user dictionary and the frequency information storage Reordering means for rearranging the order based on the frequency information registered by the means.

[0023] With this configuration, the order of the predicted character string candidates is rearranged and displayed using the character string previously determined by the user and the frequency information of the character string, so that the character string frequently used by the user is displayed. Thus, the operation selected by the user can be simplified.

Further, a character string analysis dictionary in which information for analyzing a character string is registered and a character string determined to be input are referred to the information registered in the character string analysis dictionary. A definitive character string analyzing means for performing analysis while determining a word to be a unit registered in the predictive input dictionary, and a word obtained by analysis by the definite character string analyzing means are registered in the predictive input dictionary. If the word obtained by the analysis by the confirmed character string analyzing means is determined not to be registered in the predictive input dictionary by the dictionary determining means for determining whether or not Character string storage means for registering words in the predictive input dictionary.

In this way, if the determined character string is a character string that is not in the dictionary for predictive input, the character string is added to the dictionary for predictive input. Can present the character string to the user as a predicted character string candidate, and can flexibly change the predicted input environment of the document creation device according to the usage environment of each user.

In the predictive input dictionary, usable status data for each field in which data indicating the use or non-use of each word is set for each document field is registered. When it is determined that the word obtained by the analysis according to the is registered in the dictionary for predictive input, it is determined whether or not data indicating which of the use of this word is set, When the dictionary determination unit registers the word obtained by the analysis by the confirmed character string analysis unit in the predictive input dictionary and determines that data indicating no use is set, it indicates that there is use. It is characterized by comprising a content changing means for changing to data.

Thus, if the determined character string state data stored in the predictive input dictionary is data that is not used in the field of the document that is currently being created, the data is now created. By updating the data to be a character string used in the document field in the middle, the user presents the character string determined this time to the user as a predicted character string candidate in the next prediction processing in the same document field This makes it possible to flexibly change the prediction environment of the document creation device according to the usage environment of each user.

A kana-kanji conversion dictionary in which information for performing a kana-kanji conversion using a predicted character string as a reading character string is registered, and whether the predicted character string is included in the read character string is determined by the prediction. Means for determining by referring to information registered in the input dictionary; and information registered in the kana-kanji conversion dictionary when the means determines that the predicted character string is included in the read character string. And Kana-Kanji conversion means for performing Kana-Kanji conversion using the predicted character string as a reading character string.

This makes it possible to use the predicted character string as it is as the reading character string of the kana-kanji conversion, and the user can easily distinguish the predicted character string in the reading character string.
In addition, since kanji conversion is performed in which the predicted character string can only be regarded as fixed character string information without ambiguity, the conversion accuracy can be improved.

Means for determining whether or not the read character string includes the reading of the selected predicted character string by referring to information registered in the predictive input dictionary; When it is determined that the reading character string includes the reading of the predicted character string, the kana-kanji conversion is performed in accordance with the phrase separation with priority given to the same position as the phrase separation for the reading character string of the predicted character string. Kana-kanji conversion means for obtaining the same phonetic candidate, and whether there is a homophone candidate corresponding to the same reading character string as the reading character string corresponding to the predicted character string among a plurality of the same phonetic candidates obtained by the kana-kanji conversion. A homophone candidate checking means for determining whether or not there is a homophone candidate by referring to the prediction input dictionary. First candidate checking means for determining whether or not the word candidate is a first candidate among a plurality of homophone candidates obtained by the kana-kanji conversion means; and When it is determined that the determined homonym candidate is not the first candidate, there is provided a homonym candidate replacing means for replacing the candidate order of the homonym candidate with the first candidate.

Thus, when the reading character string including the predicted character string is converted to the kana-kanji character, the predicted character string is displayed as the first candidate of the homophone candidate which is the conversion result, so that the user can perform the predicted character conversion even after the kana-kanji conversion. Columns can always be obtained without the operation of candidate selection.

Further, a read character string specifying means for specifying a read character string to be used for predictive input from the input read character string by referring to information registered in the predictive input dictionary, A reading character string distinguishing display means for distinguishing and displaying a reading character string used for prediction input specified by the specifying means and other reading character strings is provided.

Thus, when the user gives a prediction input execution instruction while inputting a reading character string, it becomes possible to know which reading character string is to be predicted input.

[0034]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the system configuration of the document creation device according to the present embodiment.
The document creation device according to the present embodiment is realized by a computer that reads a program recorded on a recording medium such as a magnetic disk and the operation of which is controlled by the program.

As shown in FIG. 1, the document creation device according to the present embodiment includes a CPU 10, a memory 12, a display device 14,
It comprises an input device 16, an output device 18, and an external storage device 20.

The CPU 10 controls the entire document creation apparatus, is connected to various hardware devices via a bus, and performs processing such as control of each device and transfer of data between the devices. The CPU 10 operates according to a program stored in the memory 12 to perform a prediction input function,
Implement Japanese processing functions including kana-kanji conversion functions.

The memory 12 is provided with a program area 12a and a data storage area 12b, and stores a program for controlling the operation of the CPU 10 and various data as needed. When a Japanese processing system is used, a Japanese processing program 24 is stored in the program area 12a. The Japanese processing program 24 includes a prediction input function program 24a for realizing a prediction input function and a kana-kanji conversion program 24b for realizing a kana-kanji conversion function. The data storage area 12b stores document data and the like of a document created by the kana-kanji conversion by the Japanese processing system.

The display device 14 comprises a color liquid crystal display and its controller.
Under the control of Japanese language processing, the reading character string entered to create a document by Kana-Kanji conversion, the Kana-Kanji conversion result (Kanji-Kana-mixed document), and the prediction input function The displayed character string (predicted character string) is displayed.

The input device 16 is composed of a pointing device such as a keyboard and a mouse, and inputs and outputs various data and instructions including a character string to be read for document creation by kana-kanji conversion.

The output device 18 is a printer device for performing printing on a paper medium, a device for performing data communication with the outside, or the like.

The external storage device 20 is composed of a storage medium such as a hard disk and a controller, and stores documents to be displayed and data of various dictionaries referred to when performing Japanese processing. The dictionary includes a kana-kanji conversion dictionary 20a,
It includes a prediction input dictionary 20b, a document field identification dictionary 20c, a prediction input user dictionary 20d, and a character string analysis dictionary 20e. The dictionary data is read out and stored in the memory 12 as needed. Details of the data registered in each dictionary will be described later.

Next, a description will be given of a functional configuration of the document creating apparatus according to the present embodiment realized by the system configuration shown in FIG. FIG. 2 is a block diagram illustrating a functional configuration of the document creation device according to the present embodiment.

As shown in FIG. 2, the document preparation apparatus includes a control unit 30, an input unit 32, an output unit 34, a data storage unit 36, and a kana-kanji conversion unit 38. Also, a kana-kanji conversion dictionary 40, a predictive input dictionary 42, a document field specifying dictionary 44, which is used when creating a document of a Japanese sentence,
Each dictionary of a prediction input user dictionary 46 and a character string analysis dictionary 48 is prepared.

The control unit 30 is realized by the CPU 10 operating according to a program. The control unit 30 performs a central function of the Japanese language processing system, and stores data and the like input from the input unit 32 by operating the input device 16. Each function unit is controlled accordingly. The control unit 30 also performs overall processing of the prediction input function.

The input unit 32 inputs data and a function execution instruction in accordance with an operation on the input device 16 and transfers the instruction to the control unit 30.

Under the control of the control unit 30, the output unit 34 displays and outputs a read character string, a kana-kanji conversion result, a predicted character string predicted by the prediction input function, etc. The output of data from the unit 34 is controlled.

The data storage unit 36 stores data used for various processes by the control unit 30. For example, data necessary for kana-kanji conversion processing, document (character string) data of a document in the grass description, character string data The data necessary for the predictive input function, such as the data for field identification using, is stored.

Under the control of the control unit 30, the kana-kanji conversion unit 38 stores the information registered in the kana-kanji conversion dictionary 40 based on the read character string data input from the input unit 32 for document creation. (Word information, co-occurrence information, etc.)
Converts an input character string into a document with kana-kanji mixed sentences.
The kana-kanji conversion unit 38 can perform kana-kanji conversion using a predicted character string predicted using a prediction input function and selected for document creation as a reading character string to be subjected to kana-kanji conversion. .

The kana-kanji conversion dictionary 40 registers word information necessary for kana-kanji conversion processing by the kana-kanji conversion unit 38, co-occurrence information indicating a co-occurrence relationship between words, and the like. The co-occurrence information registered in the kana-kanji conversion dictionary 40 is also referred to when performing kana-kanji conversion using a predicted character string as a read character string.

The predictive input dictionary 42 has registered therein information for predicting a character string by a predictive input function based on a read character string input from the input unit 32. In the prediction input dictionary 42, a character string presented as a candidate for a predicted character string is registered in association with a reading, and a data structure in which the availability of each character string in each document field is represented by a score. Has become.

The document field specification dictionary 44 specifies the field of the document under preparation in order to predict the input of the optimum character string based on the read character string input by the user or the field of the document under preparation. For example, information indicating the frequency of use of each word for each document field is registered.

The prediction input user dictionary 46 registers a character string input by performing a kana-kanji conversion on the input reading character string, that is, a character string determined to be input by the user. It is referred to together with the prediction input dictionary 42 when predicting an input character string by the input function.

The character string analysis dictionary 48 is a predictive input user dictionary for character strings input by performing Kana-Kanji conversion on the input read character strings, ie, character strings determined to be input by the user. Information for performing analysis to obtain a character string to be registered in 46 is registered. As the analysis for the character string, morphological analysis, syntax analysis, and the like are performed, and information for these analysis is registered.

The control unit 30 shown in FIG. 2 includes a prediction processing unit 50, a document field specifying unit 52, a first narrowing unit 54,
Character string candidate display unit 56, prediction input user dictionary processing unit 5
8, the second narrowing-down unit 60, the character string candidate sorting unit 62,
Fixed character string analysis unit 64, dictionary determination unit 66, prediction input dictionary processing unit 68, homophone candidate check unit 70, first candidate check unit 72, homophone candidate replacement unit 74, read character string identification unit 76, and read Each functional unit of the character string distinction display unit 78 is provided.

The prediction processing unit 50 refers to the information registered in the prediction input dictionary 42 and the prediction input user dictionary 46 and, based on the read character string input for document creation, A process of acquiring a plurality of predicted character strings that are character strings predicted to be input is executed.

The document field specifying unit 52 refers to the information registered in the document field specifying dictionary 44, and specifies the character field in the document being created and the document field using the position information of the character string. Perform

The first narrowing unit 54 narrows the plurality of predicted character strings determined by the prediction processing unit 50 in accordance with the field of the document specified by the document field specifying unit 52.

The character string candidate display section 56 displays a predicted character string predicted by the prediction input function.
A predicted character string narrowed down by the first narrowing unit 54,
The predicted character strings sorted by the character string candidate sorting unit 62 are displayed.

The predictive input user dictionary processing section 58 performs processing on the predictive input user dictionary 46 accompanying processing by the predictive input function, and includes a character string storage section 58a and a frequency information storage section 58b. Character string storage unit 58a
Indicates that a character string input by performing a kana-kanji conversion on the input reading character string is converted into a prediction input user dictionary 4.
If the character string is not registered in the user dictionary 46, the character string is registered in the user dictionary 46 for predictive input. When the character string input by performing the kana-kanji conversion on the input read character string is not registered in the user dictionary for predictive input 46, the frequency information storage unit 58b
8a, the frequency information is registered in association with the registered character string, and the input character string is stored in the predictive input user dictionary 46.
Is registered, the frequency information corresponding to this character string is updated and registered.

When the predicted character string obtained by the prediction processing unit 50 using the input read character string is registered in the prediction input user dictionary 46, the second narrowing unit 60 A process of narrowing down only columns as predicted character strings is performed.

The character string candidate rearranging section 62 sorts the order of a plurality of predicted character strings obtained by referring to the information registered in the prediction input dictionary 42 and the prediction input user dictionary 46 by the prediction processing section 50. A process of rearranging based on the character string registered in the prediction input user dictionary 46 and the frequency information registered by the frequency information storage unit 58b is performed.

The determined character string analysis unit 64 performs kana-kanji conversion on the input read character string, and analyzes the character string determined to be input with reference to the information registered in the character string analysis dictionary 48. That is, a process for determining a word (for example, an independent word) to be a unit registered in the prediction input dictionary is performed.

The dictionary discriminating section 66 discriminates whether or not a character string determined to be input by performing kana-kanji conversion on the input read character string is registered in the predictive input dictionary 42. Perform processing.

The predictive input dictionary processing section 68 performs processing on the predictive input dictionary 42 accompanying the processing by the predictive input function, and includes a character string storage section 68a and a content change section 68b.
Is provided. When the determined character string is not registered in the predictive input dictionary 42, the character string storage unit 68a performs a process of adding the character string to the predictive input dictionary 42. When the determined character string is in the predictive input dictionary 42, but the determined character string is a character string that is not used in the document field at the time of determination, the content changing unit 68b The dictionary 42 for predictive input converts the determined character string state into a state frequently used in the document field currently being created.
Is performed to change the contents of.

The homophone candidate checking unit 70 detects homophone candidates corresponding to the same reading character string as the reading character string corresponding to the predicted character string predicted by the predictive input function, among the homophone candidates obtained by the kana-kanji conversion. A process is performed to determine whether or not there is any, with reference to the predictive input dictionary 42.

The first candidate checking section 72 converts the predicted character string predicted by the prediction input function into a kana-kanji conversion section 38.
Is performed to determine whether or not it is the first homonym candidate obtained by the kana-kanji conversion process for the input character string input by.

When the first candidate checking unit 72 determines that the predicted character string predicted by the predictive input function is not the first candidate of the homophone candidate, the homophone candidate replacement unit 74 replaces the predicted character string with the predicted character string. A process of exchanging the candidate order as the first homonym candidate is performed.

The read character string specifying unit 76 performs a process of specifying a read character string to be used for predictive input from the input read character string by referring to information registered in the predictive input dictionary 42.

The reading character string distinction display section 78 performs a process of distinguishing and displaying the reading character string used for predictive input specified by the reading character string specifying section 76 and other reading character strings.

Next, the operation of the document creating apparatus according to the present embodiment will be described with reference to flowcharts.

First, a basic procedure for creating a document using the predictive input function will be described.

The user inputs a character string to be read from an input device 16 such as a keyboard in order to create a document by kana-kanji conversion. If there is an instruction from the user to execute an input using the predictive input function during the input of the reading character string, the Japanese processing system determines whether or not the user has input the reading character string up to now. A character string likely to be input is predicted, and a predicted character string obtained by this prediction is presented to the user.

The user can directly select a desired character string from the predicted character strings presented on the screen by the Japanese processing system by clicking with a mouse or the like of the input device 16 or use a keyboard or the like. Move the cursor to the character string to be selected with the cursor key or the like, and press the return key to select the predicted character string as the read character string.

In the present embodiment, when the predicted character string is presented to the user by the predictive input function, in addition to the information on the read character string input by the user, the document data that has been input by the user and the user By using the previously input character string history data and the like, the user can be presented with a minimum required number of predicted character string candidates, and can easily select from a group of predicted character string candidates. .

Next, the processing of the predictive input function for discriminating the field of the document currently being created and narrowing down candidates for predicted character strings in accordance with the field of the document will be described with reference to the flowchart shown in FIG.

FIG. 4 shows a specific example of discriminating the field of the document currently being created, narrowing down candidates for predicted character strings, and presenting only the optimum character strings to the user. FIG.
FIG. 4A shows an example of a computer-related document 1, FIG. 4B shows an example of a food-related document 2, and FIG. 4C shows an example of a sports-related document 3, each having the same read character string "Soft".
3 shows a predicted character string presented when input prediction is performed.

First, when a user presses a specific key for instructing execution of the predictive input function of the input device 16, the input unit 32 notifies the control unit 30, and the Japanese processing system follows the predictive input function program. Is activated (step A1).

Here, as shown in FIG.
When the reading character string "Soft" is input from
It is assumed that execution of the prediction input function has been instructed.

When the prediction input function is activated, the control unit 30
The document field specifying unit 52 of the
The document data stored in the data storage area 12b (data storage unit 36) is checked to determine the document field (step A2). The specification of the document field is performed by referring to the document field specifying dictionary 44 in which the relationship between each word and each field is associated with numerical data (points) from 0 to 2 as shown in FIG. This is done by using.

In the document field identification dictionary 44, information indicating words frequently used in a document for each document field is registered. For example, as shown in FIG. Each word to identify ("Recent", "Personal"
Data indicating the frequency of use in the fields (“food”, “clothing”, “calculator”,...) Arranged in the horizontal direction of “computer” are registered. In the present embodiment, as data indicating the frequency, a score of “2” when frequently used, “0” when not used, and “1” when used at an intermediate frequency are registered. ing.

For example, since the word “computer” is not used in the field of “food”, a score of “0” is registered, and in the field of “computer”, a score of “2” is registered. Have been.

For example, the following formulas (1) and (2) are used to specify the field of the document being created. In the calculation formula, as the score of the word, the score of the word registered for each word in the dictionary for document field identification 44 shown in FIG. 5 is used.

Document field score = Σ (word score × word weight) (1) word weight = (3- (total number of lines−line with word) / (total number of lines / 3)) (2) However, when the denominator of the fraction is 1 or less, it is set to 1. Also, 1
A line is a range up to a period or a line feed mark.

The document field specifying unit 52 calculates the sum of the scores of all the words in the document in each field by the calculation formulas (1) and (2).
And determine that the field with the highest score is the field of the document currently being created.

In the formulas (1) and (2), the reason that the word located near the character string to be predicted is often related to the field of the document currently being created and is likely to be used. Therefore, weighting is performed according to the relative positional relationship with the character string to be input predicted. That is, the score of a word located near the character string to be predicted is weighted so as to be higher than the score of a word located farther away. In the present embodiment, the score of a word is multiplied by three levels of 1 to 3 according to the position of the word.

FIG. 6 shows the scores for each field obtained from each word included in the document 1 in FIG. 4A by using the document field specifying dictionary 44 and the formulas (1) and (2). ing. In this case, since the input prediction for the character string “Soft” input on the third line of the document 1 is performed, 1
The word on the line is weighted "1", the word on the second line is weighted "2", and the word on the third line is weighted "3".

As shown in FIG. 6, for document 1,
Since the score of the field of “computer” is the highest at 18 points, it can be specified that the field of the document is the computer.

Here, the discrimination of the field using the above-mentioned formulas (1) and (2) is performed by simply adding the score of the word and specifying the field having the highest score as the field of the document under preparation. 7 and 8 show an example which is superior to the method and the method of specifying the document field from the average score of words. Here, it is assumed that the document field (document topic) changes in the middle of the document.

FIG. 7 shows an example in which the score of a word is simply added without weighting.

When specifying the document field of a document in which the document field changes in the middle, simply adding the score of each word in the document without weighting the score of the word gives the position of the character string targeted for prediction input. , And words used in a different document field that have been input before the document field changes have the same effect.

Accordingly, if the number of words in the document field input before the document field is changed is large, the influence of words in different document fields becomes stronger, and an incorrect document Fields can be easily identified.

In the example shown in FIG. 7, although there are many words in the field of "sports" among words located near the character string to be predicted and input (the lower part shown in FIG. 7),
Since there are many words in the field of "food", the score in the field of "sports" is 23 points,
Since the score of the "food" field is the highest at 25 points, the document field is determined to be "food". In this case, "food"
A character string such as “soft cream” or “soft drink” used in a related document is presented as a predicted character string for the input reading character string “soft”.

In the present embodiment, the weight applied to the score of the word is changed according to the position of the word, and the influence of the word in the vicinity of the read character string to be predicted is changed so that the erroneous document field is not discriminated as described above. , So that even if the field of the document changes on the way, the specification of the field is not mistaken.

FIG. 8 shows an example of calculating the score by weighting the score of the word.

In the example shown in FIG. 8, the words located near the character string to be predicted and input (the lower part shown in FIG. 7) are weighted heavily. Therefore, even if many words in the field of "food" are included, the score of the word in the field of "sports" is large, so that the score of the field of "sports" is the highest at 57 points, The score in the field of "food" is 48 points, so the field of documents is "sports"
Is determined. In this case, a character string such as "softball" used in a document related to "sports" is presented as a predicted character string for the input reading character string "soft".

Thus, the document field specifying unit 52
When the field of the document is determined by the document field specifying method that weights the score of the word according to the position of the word, the prediction processing unit 50 determines the reading character string “Soft” currently input by the user. Then, the prediction input dictionary 42 is searched to obtain a group of character string candidates that can be predicted to be likely to be input.

Further, the first narrowing-down unit 54 narrows down from the character string candidate group predicted by the prediction processing unit 50 to only the predicted character strings in the field specified by the document field specifying unit 52 (step A3). .

FIG. 9 shows a field-by-field availability status data structure of a character string candidate group predicted based on the read character string "soft" registered in the prediction input dictionary 42. . Note that the character strings registered in the predictive input dictionary 42 are associated with reading information so that a corresponding predicted character string can be detected based on the reading character string.

As shown in FIG. 9, the availability data for each field includes data indicating whether or not each character string (word) is used for each document field, that is, “1” for frequently used words, and “1” for not used words. “0” is set for the word. The first narrowing-down unit 54 acquires only a character string (word) in which “1” is set for the field specified by the document field specifying unit 52 as a predicted character string.

The character string candidate display section 56 displays a group of predicted character string candidates narrowed down by the first narrowing down section 54 on the screen of the display device 14 via the output section 34 and presents it to the user. This allows the user to select a desired character string from the group of predicted character string candidates.

For example, in the computer-related document 1 shown in FIG. 4A, the document 1 is converted to a computer-related (“computer”) from a character string such as “computer” or “hardware”.
Document field), a character string frequently used in computer-related documents in which “1” is set in the “computer” document field from the group of predicted character string candidates (see FIG. 9). “Software” is presented to the user as a prediction candidate.

In document 2 related to food shown in FIG. 4B, document 2 is determined to be a document related to food (document field of “food”) from character strings such as “bread” and “milk”, From the character string candidate group (see Fig. 9), "food"
The character strings “soft cream” and “soft drink” commonly used in food-related documents, in which “1” is set in the document field of “1”, are presented to the user as prediction candidates.

Similarly, in the case of the sports-related document 3 shown in FIG. 4C, it is determined from the character string such as "sports" that the document 3 is a sports-related document ("sports" document field). From the group of predicted character string candidates (see FIG. 9),
"1" was set in the "Sports" document field,
“Softball”, which is often used in sports-related documents, is presented to the user.

Further, even in the case of a document including the contents of different document fields as shown in FIG. 10, the above-described calculation formulas (1) and (2) and each field registered in the document field specifying dictionary 44 are used. By using the availability status data and calculating by weighting the score of the word near the character string "Softon" to be predicted, the document field can be accurately determined and the predicted character string Can be correctly narrowed down.

As described above, when a character string is input using the predictive input function, only the minimum necessary predictive character string that is optimal for the document (document field) currently being created is presented. An effect is obtained that the operation for selecting a desired character string from a group of character string candidates is made easier than before.

In the above description, the conventional method of simply adding the score of a word to specify a field with a high score as a field being created, or specifying the document field from the average score of the word As a method with higher accuracy in specifying the document field than the method, using the formulas (1) and (2), the points of words near the character string to be input predicted are weighted to improve the specific accuracy in the document field. However, as a method for improving the specific accuracy of the document field, a field judgment method based on the sum of the scores of words in a document and a field judgment method using only neighboring words are simply combined. A method of using the specification and a method of using the title of the document as one condition for specifying the document field can be considered. In the method using the title of the document, for example, a character string is registered in the document field specifying dictionary 44 in association with the document field, and the document field specifying dictionary 4 is registered.
Referring to FIG. 4, it is determined in which field the character string included in the title is included.

In the above description, the candidates for the predicted character string are narrowed down by specifying the document field by the document field specifying unit 52. However, the history information of the character string previously input by the user is further input to the user for predictive input. By registering the information in the dictionary 46 and using this information, predicted character string candidates can be further narrowed down. Hereinafter, the prediction input user dictionary 4
6 will be described.

FIG. 11 is a flowchart showing the flow of processing for registering the determined character string in the predictive input user dictionary 44.

The Japanese processing system registers the character string input by the user's operation in the document field identification dictionary 44.

When a character string is input using the function of the Japanese processing system, the fixed character string analysis unit 64 performs an analysis process while referring to information registered in the character string analysis dictionary 48, and inputs the character string. The input character string is decomposed into independent words which are registered in the predictive input dictionary 42 (step B1).

For example, a reading character string is input from the input unit 32, and the kana-kanji conversion unit 38 outputs the kana-kanji conversion dictionary 4.
It is assumed that the kana-kanji conversion is performed while referring to the information registered in 0, and a plurality of conversion candidates are obtained. Here, when an instruction to select a character string to be input is input from the input unit 32 and determined, the determined character string is to be analyzed by the determined character string analysis unit 64.

After the analysis by the confirmed character string analyzing section 64, the predictive input user dictionary processing section 58 checks whether each character string decomposed into an independent word is already registered in the predictive input user dictionary 46. Is determined (step B
2).

If the determined character string is not registered in the predictive input user dictionary 46, the predictive input user dictionary processing unit 58 predicts the character string by using the character string storage unit 58a. Is registered in the user dictionary 46 (step B3).

In the present embodiment, the prediction input user dictionary 4
In FIG. 6, as shown in FIG. 12, the read character string of the confirmed character string, the inputted confirmed character string, and frequency information indicating the frequency of use (the number of times of use) are registered in association with each other. I do.

In the example shown in FIG. 12, a fixed character string "software" is set for the read character string "software", a fixed character string "soft cream" is set for the read character string "software cream", and the read character string is set. This is an example in which a confirmed character string “softball” is registered for “softball”.

On the other hand, when the determined character string is already registered in the predictive input user dictionary 46, the predictive input user dictionary processing unit 58 uses the frequency information storage unit 58b to correspond to the character string. The frequency information is updated by adding 1 to the frequency information (step B4).

Next, input of a character string using the predictive input function using the predictive input user dictionary 46 in which the character string is registered as described above will be described.

FIG. 13 is a flowchart showing the flow of a process of narrowing down candidates for predicted character strings by using the user dictionary 46 for predictive input.

First, when the user inputs an instruction to execute the prediction input function from the input unit 32 (step C1),
The document field specifying unit 52 of the control unit 30 is the data storage area 12b (data storage unit 36) created up to now.
Check the document data stored by
The document field is determined (step C2). Note that the process of determining the document field by the document field specifying unit 52 is performed in the same manner as the process described in step A2 of FIG.

Further, the prediction processing unit 50 searches the prediction input dictionary 42 in which the use status data for each field is registered based on the currently read character string, and is likely to be input. Obtain a group of character string candidates that can be predicted to be high. Further, the first narrowing-down unit 54 narrows down only the predicted character strings in the field specified by the document field specifying unit 52 from the character string candidate group predicted by the prediction processing unit 50 (step C3). The processing in step C3 is as follows.
The detailed description is omitted as it is performed in the same manner as the process described in step A3 of FIG.

Thereafter, it is determined whether or not the predicted character string is a character string registered in the predictive input user dictionary 46 (step C4). The narrowing-down unit 60 narrows down only that character string as a candidate for a predicted character string. When there are a plurality of predicted character strings narrowed down by the second narrowing-down unit 60, the character string candidate reordering unit 62 determines the order of the plurality of predicted character strings with the character strings registered in the prediction input user dictionary 46. Sort based on the frequency information. That is, the order of the predicted character strings having a higher frequency value is changed to a higher order so that selection from a plurality of predicted character strings becomes easier. The character string candidate display unit 56 displays the predicted character strings whose order has been rearranged by the character string candidate rearrangement unit 62 and is presented to the user (step C5).

On the other hand, if it is not registered in the prediction input user dictionary, the character string candidate display section 56 proceeds to step C
The predicted character string acquired in step 3 is displayed as it is and presented to the user (step C6).

In the example described above, the user dictionary for predictive input 4
6 was presented to the user, but the number of predicted character strings presented to the user was the same,
Using the frequency information registered in the user dictionary for prediction input 46 corresponding to each character string, a required number of character strings can be obtained in order from the one with the highest frequency.

In this case, the character string candidate sorting unit 62
In the same manner as described above, the predicted character strings are sorted in descending order of frequency using the frequency information, and the character string candidate display unit 5
6 is displayed.

FIG. 14 shows an example in which candidates for predicted character strings have been narrowed down by the second narrowing-down unit 60 based on the character strings registered in the prediction input user dictionary 46 shown in FIG. ing.

If the instruction for predictive input by the predictive input function is given during the input of the reading character string “Soft”, the document field specifying unit 5
2. Based on the words in the document being created, the document field is determined to be “food” related, and “soft cream” and “soft drink” are often used for food-related documents.
Is obtained as a candidate for a predicted character string.

Here, referring to the predictive input user dictionary 46 in which information as shown in FIG. 12 is registered, it is determined that the character string “soft cream” is registered.

The second narrowing-down unit 60 acquires only the predicted character string “soft cream” that is the same as the character string registered in the prediction input user dictionary 46 from the plurality of predicted character strings. The character string candidate display unit 56 displays the predicted character string acquired by the second narrowing unit 60 as shown in FIG. 14 and presents it to the user.

As described above, by using the predictive input user dictionary 46 in which the character string previously input by the user is stored, the predictive character string can be further narrowed down according to the use environment of the user. Become. Therefore, the operation for selecting a desired character string from the group of predicted character string candidates presented by the prediction input function can be made easier.

In the above description, the character string registered in the predictive input user dictionary 46 is a character string definitely entered by the Kana-Kanji conversion, but is input from the predictive character string presented by the predictive input function. It is also possible to register a character string selected as a power character string.

Next, a description will be given of a function for flexibly responding to a user's request even when the user wants to input a character string which is rarely used in a document field which is being created.

FIG. 15 shows a case where a character string other than the predicted character string presented by the predictive input function is determined as a character string to be input, and the available status data for each field registered in the predictive input dictionary 42 is updated. 6 is a flowchart showing the flow of the processing performed. This will be described with reference to a specific example shown in FIG. 16 when a character string other than the predicted character string presented by the prediction input function is definitely input.

Here, as shown in FIG. 16, the user inputs “soft cream” while inputting a computer-related document.
It is assumed that at the point of inputting the reading character string "Soft" to input the product name of the software, execution of input prediction by the prediction input function is instructed.

In this case, since the document being created is a computer-related document by the predictive input function described with reference to FIGS. 11 and 13, the predictive character acquired based on the read character string “Soft” is used. Of the character strings, as shown in FIG. 16A, it is assumed that only the character string “software” frequently used in the computer-related document field is presented to the user as a prediction candidate.

In this case, since the user wants to input “soft ice cream”, the presented predicted character string “software” is not used for input, and is read continuously as shown in FIG. It is assumed that a character string has been input and kana-kanji conversion has been performed. That is, as shown in FIG. 16 (c), it is assumed that "software called software" is confirmed and input as a converted character string for the read character string including "soft cream".

When the character string is fixedly input, the fixed character string analysis unit 64 performs an analysis process while referring to the information registered in the character string analysis dictionary 48, and decomposes the input character string into independent words. (Step D1).

The dictionary discriminating section 66 includes a fixed character string analyzing section 64
Independence words (character strings that are confirmed and input) obtained by analysis
Is included in the candidate group of the predicted character string (step D2). In other words, the dictionary determination unit 66 determines that the finally input character string is a group of predicted character string candidates before narrowing the candidates of the predicted character string using the field information of the document (to the read character string “soft” shown in FIG. 9). It is determined whether or not the character string is included in the corresponding predicted character string.

[0138] If the character string that has been confirmed and input is included in the predicted character string, the dictionary discriminating unit 66 determines that the character string is not used much in the field of the document currently being created. Is checked (step D3). That is, the dictionary determination unit 66 refers to the field-specific availability data registered in the predictive input dictionary 42, and sets data (points) for the target word in the document field of the document currently being created. Is “1” indicating that it is used or “0” that is not used.

Here, if “0” indicating that the word is not used is set for the target word, the content changing unit 68 b of the predictive input dictionary processing unit 68 sends The data corresponding to the current field of the column is changed from "0" to "1" so that the target word becomes a word frequently used in the field of the document currently being created (step D4).

For example, as shown in FIG. 16, a character string “soft cream” that is rarely used in the computer field is used.
FIG. 17 shows an example of the change of the availability data for each field when the is input during the creation of the computer-related document. In the above-described example, as shown in FIG. 17, the score for the word “soft cream” in the document field of “computer” shown in FIG.
As shown in FIG. 7B, it is updated to “1”.

When the data is updated by the content change section 68b and the user instructs the prediction with the read character string “soft” during the creation of the document in the computer field next time, in addition to the predicted character string “software”, the user confirms the previous input. The “soft cream” that has been added is also presented as a candidate for the predicted character string.

On the other hand, if it is determined in step D2 that the character string that has been fixedly input is not included in the group of predicted character string candidates, the character string storage unit 6 of the predictive input dictionary processing unit 68
8a adds a newly input character string to the field-specific availability data registered in the predictive input dictionary 42 (step D5). Thus, when the next time the user inputs the same reading character string and performs an input using the predictive input function, the character string confirmed and input by the user last time is also added to the candidate of the predicted character string and presented. Become.

For example, when “softhouse” which is not obtained as a predicted character string by input prediction based on the read character string “soft” is fixedly input, FIG.
As shown in FIG. 18, the content change unit 68b converts the finally input “softhouse” into the usable status data for each field in which the character string “softhouse” is not registered.
Newly added as shown in (b). In this case, the score for “softhouse” in the document field of “computer” is set to “1”.

In this way, the Japanese processing system
When the user inputs a character string using the predictive input function, the user has a function of updating the contents of the field-specific availability data registered in the predictive input dictionary 42 while checking the character string confirmed and input by the user. Can be changed so as to be optimal for each user and for each document field of the document to be created as the document is created.

Next, a case will be described in which the predicted character string obtained by the prediction input function is used for kana-kanji conversion to improve the conversion rate of the kana-kanji conversion result.

FIG. 19 is a flowchart showing the flow of the Kana-Kanji conversion process in which a character string predicted and input is used as it is as a reading character string for Kana-Kanji conversion.

First, in order to input a character string by Kana-Kanji conversion by the Kana-Kanji conversion unit 38, a reading character string is input from the input unit 32, and based on the reading character string, a character string is input using a predictive input function. Assume that it has been entered. In this case, even when a predicted character string including a kanji is selected and input, the character string including the kanji is treated as a read character string.

Here, when the conversion execution of the Kana-Kanji conversion is instructed, the Kana-Kanji conversion unit 38 determines whether or not the predicted character string is included in the read character string by the prediction input dictionary 42.
It is determined by referring to the information registered in (step E1).

When the predicted character string selected by the user is included in the read character string, the kana-kanji conversion section 38 uses the predicted character string included in the read character string and is registered in the kana-kanji conversion dictionary 40. By referring to the co-occurrence information, the kana-kanji conversion is performed with the homophone candidates narrowed down (step E2).

On the other hand, if the predicted character string is not included in the read character string, the kana-kanji conversion section 38 performs ordinary kana-kanji conversion (step E3).

FIG. 20 shows an example of kana-kanji conversion using a predicted character string as it is as a reading character string for kana-kanji conversion.

First, as shown in FIG. 20 (a), when the read character string "Yokohama" is input, execution of input by the predictive input function is instructed, and a predictive character string candidate is determined by the method described above. It is assumed that the character string “Yokohama Bay Bridge” is presented to the user and selected from these candidates as shown in FIG.

After the predicted character string “Yokohama Bay Bridge” is selected and input, as shown in FIG. 20C, the reading character string “Kakeru” is input as it is, and then the execution of Kana-Kanji conversion is instructed. Shall be.

In the kana-kanji conversion, the kana-kanji conversion section 38 performs segmentation on the read character string including the predicted character string "Yokohama Bay Bridge" as "Yokohama Bay Bridge / Kana". Cut).

The kana-kanji conversion unit 38 selects from among a plurality of homophone candidates (for example, bridge, lack, multiply, run, etc.) for “kake” obtained based on the word information registered in the kana-kanji conversion dictionary 40. With reference to the co-occurrence information, the candidates are narrowed down to "bridge" which has a co-occurrence relationship with "bridge" (the field of bridges, buildings, etc.), and "bridge" is output as a result of the kana-kanji conversion.

In this way, when performing the kana-kanji conversion using the predicted character string as a read character string as it is, it is possible to know which part of the read character string is the predicted character string selected by the user, and Can be treated as a determined character string (determined character string information with no ambiguity), and by using this information to narrow down homophone candidates, the conversion result can be improved.

In the above description, in order to improve the conversion rate of the Kana-Kanji conversion result, the Kana-Kanji conversion unit 38, which uses a predicted character string including Kanji as a reading character string as it is, has been described. The Kana-Kanji conversion process that gives priority to the predicted character string when only the character string read by the user is used will be described.

FIG. 21 is a flowchart showing the flow of the Kana-Kanji conversion process in which the prediction character string selected by the user has priority. Here, the kana-kanji conversion is performed by returning the selected predicted character string to a hiragana reading character string once, and the following processing is performed, whereby the predicted character string desired by the user can be obtained as a result of the kana-kanji conversion.

First, in order to input a character string by Kana-Kanji conversion by the Kana-Kanji conversion unit 38, a reading character string is input from the input unit 32. Based on the reading character string, a prediction character string is used by using a prediction input function. It is assumed that has been selected.
Here, the selected predicted character string is converted into Hiragana reading based on the reading information corresponding to the corresponding predicted character string registered in the predictive input dictionary 42, and the kana-kanji conversion unit is used. It is assumed that the character string to be read is a character string to be converted by Kana-Kanji conversion by the K. With respect to the reading character string obtained from the predicted character string, the reading character string can be continuously input or changed.

Here, when the conversion execution of the Kana-Kanji conversion is instructed, the Kana-Kanji conversion unit 38 determines whether or not the reading of the predicted character string selected by the user is included in the reading character string. The determination is made by referring to the information registered in the dictionary 42 (step F1).

If the predicted character string selected by the user is included in the read character string, the kana-kanji conversion unit 38
Performs Kana-Kanji conversion by performing a segmentation with priority given to the same position as the phrase delimiter for the reading character string of the predicted character string selected by the user, and referring to information registered in the Kana-Kanji conversion dictionary 40 ( Step F2).

Further, after the Kana-Kanji conversion unit 38 has performed the Kana-Kanji conversion, the homophone candidate check unit 70 checks the homophone candidates obtained by the Kana-Kanji conversion for homophones having the same reading character string as the predicted character string. It is determined whether or not there is, referring to the prediction input dictionary 42.

Here, if there is a homophone candidate that is the same as the reading of the predicted character string, the first candidate checker 72 determines whether the corresponding predicted character string is the first candidate in the homophone candidate. (Step F4).

As a result, if the predicted character string is not the first candidate in the homophone candidate, the homophone candidate replacement unit 74 replaces the predicted character string selected by the user with the kana-kanji conversion result in the homophone candidate. (Step F5).

On the other hand, if the predicted character string is not included in the read character string, the kana-kanji conversion unit 38 performs ordinary kana-kanji conversion (step F6).

FIG. 22 shows an example of kana-kanji conversion in which a predicted character string is prioritized.

First, as shown in FIG. 22 (a), when the reading character string "Jido" is input, execution of the input by the predictive input function is instructed. For example, it is assumed that "child pick-up bus" is presented to the user, and the predicted character string is selected.

In this case, in the predictive input dictionary 42, the reading “Jidosou Keibaisu” corresponding to the predicted character string “Child pick-up bus” is registered, and as shown in FIG.
This reading "Jidosoudaiba" is treated as a reading character string.

In order to perform the kana-kanji conversion for the read character string shown in FIG. 22 (b), the kana-kanji conversion unit 38 normally uses the information registered in the kana-kanji conversion dictionary 40 to perform “ji / dos / The reading character string is divided at various positions such as "Uge / Ibusu", "Jidoso / Geibasu", "Jido / Sogei / Basu".

In this embodiment, when a predicted character string is selected using the predictive input function, the kana-kanji conversion unit 38
Is a dictionary of kana-kanji conversion using the reading character strings "ji", "do", "so", "gei", "ba", and "su" corresponding to each character of the predicted character string "child pick-up bus" selected by the user. Find the longest word in, and divide the reading string in that unit.

More specifically, when the kana-kanji conversion dictionary 40 is first examined by reading “ji”, there are several words such as “next” and “time”. Next, if you look at "Jido", "Automatic"
There is a word "child", so "Jodo"
When there is no word corresponding to this reading, it is determined that the first break comes after "Jido".

By repeating this process, the break in the character string read for "Jidosou Geibisu"
Is determined. The kana-kanji conversion unit 38 performs kana-kanji conversion processing for each of the reading character strings separated by the breaks, using the information indicating the position of the break in the reading character string.

In this example, as the homophone candidates obtained by the kana-kanji conversion, for example, the first candidate “automatic” and the second candidate “child” are obtained for the read character string “jido”, and the read character For the row “Sogei”, a first candidate “pick-up” and a second candidate “creation” are obtained.

As a result, based on the co-occurrence information and the like, the first candidate "automatic transfer bus" and the second candidate "child pick-up bus" are obtained as homophone candidates as conversion results. As a normal conversion result, the first candidate “automatic transfer bus” is output.

Therefore, since the first candidate check unit 72 determines that the predicted character string “child pick-up bus” is not the first candidate of the two homophone candidates, the homophone candidate replacement unit 74 The selected predicted character string “child pick-up bus” is changed so as to be the first candidate among the homophone candidates. In this case, since "child" in the predicted character string has a lower priority than "auto", the priority of "child" is increased. As a result, as shown in FIG.
The “child pick-up bus” is output as a result of the kana-kanji conversion by the kana-kanji conversion unit 38.

In this way, the kana-kanji conversion unit 38
When performing a kana-kanji conversion process for only a character string read by a normal hiragana, in a kana-kanji conversion based on information registered in a kana-kanji conversion dictionary 40, a conversion result of “automatic transfer bus” is a first candidate. In such a case, the kana-kanji conversion is performed by returning the predicted character string to the hiragana, and the kana-kanji conversion is performed by giving priority to the predicted character string selected by the user, so that the “child pick-up bus” can be output as the conversion result.

Next, a case will be described in which a predictive input function is used when a plurality of phrase reading character strings are input.

FIG. 23 is a flowchart showing the flow of processing for displaying a character string range used for predictive input when a plurality of reading character strings are input, distinguished from other character strings.

When a user wants to give an instruction for predictive input while inputting a read string of a plurality of phrases, if no range of the read character string to be used for predictive input is indicated, the predictive input It is difficult for the user to understand which read character string is to be predicted when the instruction to execute is performed. Therefore, the range of the reading character string used for predictive input is displayed separately from other reading character strings according to the input of the reading character string, so that the user can easily understand the reading character string used for predictive input.

First, when a reading character string is input from the input unit 32 to create a document by Kana-Kanji conversion, it is sequentially displayed as a normal reading character string to be converted by the Kana-Kanji conversion unit 38. On the other hand, the reading character string specifying unit 76 refers to the prediction input dictionary 42,
The range of the reading character string used for predictive input is specified from the reading character string including the input character (step G1).

That is, from the read character string including the input character, the read character string specifying unit 76
It is determined whether or not at least one predicted character string registered in the input character string can be specified. If a corresponding predicted character string exists, the input read character string is input into the input prediction range (one phrase range). Applicable).

It is to be noted that, without using the information registered in the predictive input dictionary 42, the break position of the phrase is simply determined, and the character string corresponding to the phrase including the character input immediately before is used as the target of the predictive input. It may be specified as a range.

When the read character string used for predictive input can be specified by the read character string specifying unit 76 (step G
2) The read character string distinction display unit 78 displays the range of the read character string used for the prediction input with a display attribute that can be distinguished from other read character strings (step G3).

FIG. 24 shows an example in which the read character strings of a plurality of phrases are sequentially input.

As shown in FIG. 24 (a), when the reading character string "Taiiku" is input, it is determined that the input is within the range in which prediction input is possible, and the range of this character string is underlined. The range of the prediction input is specified. Here, when the execution of the predictive input is instructed, the predictive character string is presented by the processing by the predictive input function according to the above-described method based on the underlined read character string “Taiku”.

As shown in FIG. 24 (b), when the character string "Jikan" is input after the reading character string "Taiiku", the character string "Jikan" is input. At this point, it is determined that the input range is within the range in which prediction input is possible, and an underline is added instead of the underlined character string “Taiiku”. Here, when the execution of the prediction input is instructed, a prediction character string based on the underlined reading character string “Jikan” is presented.

Further, as shown in FIG. 24 (c), when a character string "Hashifuto" is input after the reading character string "Jikan", when the character string "Soft" is input, It is determined that the input range is within the range in which prediction input is possible, and an underline is added in place of the character string “Jikan” which is previously underlined. Here, when execution of prediction input is instructed, a prediction character string based on the underlined reading character string "Soft" is presented.

In this manner, the character string to be predicted is determined according to the input character string to be read, and display attributes different from the others are sequentially added only to the range of the corresponding character string. Is displayed, the user can easily grasp the read character string to be predicted input when instructing execution of the prediction input function.

In the example shown in FIG. 24, an underline is added to a character string as a display form for distinguishing a range of a read character string to be predicted and input. However, another display form is used. , Distinguish the reading character string used by the user for prediction from other reading character strings, such as changing the font of each character of the corresponding character string, changing the character size, changing the display color of the character, making italic, etc. You can use any method you can.

Each of the methods described in the above-described embodiments is, for example, a magnetic disk (floppy disk, hard disk, etc.), an optical disk (CD-ROM,
It is also possible to write the data on a recording medium such as a DVD or a semiconductor memory and apply it to various devices, or to transmit it via a communication medium and apply it to various devices. A computer that realizes the present apparatus reads the program recorded on the recording medium, and executes the above-described processing by controlling the operation of the program.

[0191]

As described above in detail, according to the present invention, the document field is specified using the character string information and the position information of the character string in the document currently being created, thereby improving the accuracy of specifying the document field. By using the document fields identified in this way to narrow down the predicted character string candidates, it is possible to present only the most frequently used optimal character strings to the user. The operation of selecting can be simplified.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a system configuration of a document creation device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a functional configuration of the document creation apparatus according to the embodiment.

FIG. 3 is a flowchart for explaining processing of a prediction input function for determining a field of a document currently being created and narrowing down candidates of a predicted character string according to the field of the document.

FIG. 4 is a diagram showing a specific example in which the field of a document currently being created is discriminated, candidates of predicted character strings are narrowed down, and only an optimal character string is presented to a user.

FIG. 5 is a view for explaining the contents of data registered in a document field identification dictionary 44;

FIG. 6 is a dictionary 44 for specifying a document field and calculation formulas (1) and (2).
FIG. 5 is a diagram showing scores for each field obtained from each word included in the document 1 of FIG.

FIG. 7 is a diagram showing an example in which a score is simply added without weighting a score of a word in a process for specifying a document field.

FIG. 8 is a diagram illustrating an example in which a score is calculated by weighting a score of a word in a process for specifying a document field.

FIG. 9 is a diagram showing a field-usable state data structure for a group of character string candidates predicted based on a read character string “soft” registered in a prediction input dictionary 42;

FIG. 10 is a diagram showing an example of a document including contents of different document fields.

FIG. 11 is a flowchart showing a flow of processing for registering a determined character string in the user dictionary for predictive input 44 in the embodiment.

FIG. 12 associates a read character string of a confirmed character string, an inputted confirmed character string, and frequency information indicating the frequency of use (the number of times of use) registered in the user dictionary for prediction input 46. The figure which shows an example.

FIG. 13 is a user dictionary 4 for predictive input in the present embodiment.
6 is a flowchart showing the flow of processing for narrowing down candidates for predicted character strings using No. 6;

FIG. 14 is a diagram showing an example in which a second narrowing-down unit has narrowed down candidates for predicted character strings based on a character string registered in the user dictionary for prediction input 46 shown in FIG. 12;

FIG. 15 is a flowchart of a process for updating field-specific availability data registered in the predictive input dictionary 42 when a character string other than the predicted character string presented by the predictive input function is determined as a character string to be input. 5 is a flowchart showing a flow.

FIG. 16 is a diagram showing a specific example in a case where a character string other than the predicted character string presented by the prediction input function is definitely input.

FIG. 17 is a diagram illustrating an example of a change in field-specific availability data when the character string “soft cream” is fixedly input during creation of a computer-related document.

FIG. 18 is a diagram illustrating an example of updating field-specific availability data in which a character string “softhouse” is not registered.

FIG. 19 is a flowchart illustrating a flow of a kana-kanji conversion process in which a character string predicted as input is used as a reading character string for kana-kanji conversion as it is in the embodiment.

FIG. 20 is a diagram showing an example of kana-kanji conversion using a predicted character string as it is as a reading character string for kana-kanji conversion.

FIG. 21 is a flowchart illustrating a flow of a kana-kanji conversion process in which a prediction character string selected by a user is prioritized.

FIG. 22 is a diagram showing an example of kana-kanji conversion in which a predicted character string is prioritized.

FIG. 23 is a flowchart showing the flow of processing for displaying a character string range used for predictive input when a plurality of reading character strings are input, distinguished from other character strings.

FIG. 24 is a diagram illustrating an example of a case where reading character strings of a plurality of phrases are sequentially input;

FIG. 25 is a diagram showing an example of character input using a conventional predictive input function.

FIG. 26 is a diagram for explaining a situation in which a reading character is input following a character string selected from a conventional predicted character string and kana-kanji conversion is performed.

[Explanation of symbols]

 Reference Signs List 10 CPU 12 Memory 14 Display device 16 Input device 18 Output device 20 External storage device 30 Control unit 32 Input unit 34 Output unit 36 Data storage unit 38 Kana-Kanji conversion unit 40 Kana-Kanji conversion dictionary 42 predictive input dictionary 44 document field specifying dictionary 46 predictive input user dictionary 48 character string analysis dictionary 50 predictive processing unit 52 document field specifying unit 54 first narrowing unit 56 character string candidate display Unit 58: prediction input user dictionary processing unit 58a: character string storage unit 58b: frequency information storage unit 60: second narrowing unit 62: character string candidate rearrangement unit 64: fixed character string analysis unit 66: dictionary determination unit 68 ... Prediction input dictionary processing unit 68a ... character string storage unit 68b ... content change unit 70 ... homophone candidate check unit 72 ... first candidate check unit 74 ... homophone candidate replacement 76 ... to read the character string specifying unit 78 ... read character string distinguish between the display unit

Claims (13)

[Claims] 1. A document creation apparatus for creating a document by performing a kana-kanji conversion on an input reading character string, wherein information for performing an input prediction based on the reading character string input for document creation is provided. A predicted input dictionary in which is registered, and characters predicted to be input based on a reading character string input for document creation with reference to information registered in the predicted input dictionary. Prediction processing means for acquiring a plurality of predicted character strings as strings, a document field specifying dictionary in which information for specifying the field of the document is registered based on the content of the created document, and the document field With reference to the information registered in the specific dictionary,
A document field specifying unit that specifies a field of the document based on a character string in the document being created; and a plurality of predicted character strings determined by the prediction processing unit, the document specified by the document field specifying unit. A first narrowing means for narrowing down according to the field of; and a display means for displaying the predicted character string narrowed down by the first narrowing means, and a predicted character selected from the predicted character string displayed by the display means A document creation device characterized by inputting a column. 2. The document field identification dictionary registers information indicating the frequency of use of each word for each document field, and the document field identification means corresponds to each character string in a document being created. 2. The document creating apparatus according to claim 1, wherein a sum of values indicated by frequency information of words registered in the dictionary for document field identification is obtained for each field to specify the document field. 3. The document field identification dictionary corresponding to each character string in a document being created according to a relative positional relationship with a character string to be input predicted. 3. The document creation apparatus according to claim 2, wherein the information indicating the use frequency of the registered words is weighted to obtain a total sum. 4. A field-based use status data, in which data indicating whether or not each word is used is set for each document field, is registered in the prediction input dictionary.
Document creation device. 5. A predictive input user dictionary for registering a character string determined to be input and referred to together with said predictive input dictionary when obtaining a predictive character string by said predicting means. Character string storage means for registering a character string selected and input from a predicted character string in the user dictionary for predictive input, and a predictive character string acquired by the predictive processing means in the user dictionary for predictive input 2. The document creating apparatus according to claim 1, further comprising a second narrowing-down unit that narrows down only the acquired character string as a predicted character string when the character string is obtained. 6. A character string determined to be input,
For registering frequency information indicating a use frequency corresponding to the character string, and a prediction input user dictionary that is referred to together with the prediction input dictionary when the prediction processing unit acquires the predicted character string; A character string storage means for registering the character string in the predictive input user dictionary when the character string determined to be input is not registered in the predictive input user dictionary; If the registered character string is not registered in the predictive input user dictionary, frequency information is registered in association with the character string registered by the character string storage unit, and the frequency information is registered in the predictive input user dictionary. If not, frequency information storage means for updating and registering frequency information corresponding to the character string, and the prediction input dictionary and the prediction input user by the prediction processing means. A plurality of predicted character strings obtained by referring to the information registered in the dictionary are arranged in order based on the character strings registered in the predictive input user dictionary and the frequency information registered by the frequency information storage unit. 2. The document creating apparatus according to claim 1, further comprising a rearranging means for rearranging. 7. A character string analysis dictionary in which information for analyzing a character string is registered, and a character string determined to be input is referred to information registered in the character string analysis dictionary. A definitive character string analyzing means for performing analysis while determining a word to be a unit to be registered in the predictive input dictionary; and a word obtained by analysis by the definite character string analyzing means is registered in the predictive input dictionary. A dictionary determination unit that determines whether or not a word obtained by the analysis by the confirmed character string analysis unit is not registered in the predictive input dictionary. 2. The document creation apparatus according to claim 1, further comprising: a character string storage unit for registering a word in the prediction input dictionary. 8. The predictive input dictionary registers field-specific availability data in which data indicating whether or not each word is used is set for each document field, and wherein the dictionary discriminating means includes the determined character string analyzing means. When it is determined that the word obtained by the analysis according to is registered in the dictionary for predictive input, it is determined whether or not data indicating which of the use of this word is set, When the dictionary determination unit registers the word obtained by the analysis by the confirmed character string analysis unit in the predictive input dictionary and determines that data indicating no use is set, it indicates that there is use. 8. The document creating apparatus according to claim 7, further comprising a content changing unit that changes the data. 9. A kana-kanji conversion dictionary in which information for performing a kana-kanji conversion using a predicted character string as a reading character string is registered, and whether the predicted character string is included in the reading character string is determined by the prediction. Means for determining by referring to information registered in the input dictionary; and information registered in the kana-kanji conversion dictionary when the means determines that the predicted character string is included in the read character string. 2. The document creating apparatus according to claim 1, further comprising: a kana-kanji conversion unit that performs kana-kanji conversion using a predicted character string as a reading character string while referring to. 10. A means for determining whether or not a reading of a selected predicted character string is included in a read character string by referring to information registered in the predictive input dictionary. When it is determined that the reading character string includes the reading of the predicted character string, the kana-kanji conversion is performed in accordance with the phrase separation with priority given to the same position as the phrase separation for the reading character string of the predicted character string. Kana-Kanji conversion means for obtaining the same phonetic candidate, and a plurality of homophone candidates obtained by the Kana-Kanji conversion,
A homophone candidate checking unit that determines whether there is a homophone candidate corresponding to the same reading character string as the reading character string corresponding to the predicted character string by referring to the prediction input dictionary, When the checking means determines that there is a homophone candidate, a first candidate check for judging whether or not this homophone candidate is a first candidate among a plurality of homophone candidates obtained by the kana-kanji conversion means. A homophone that replaces the candidate order of the homophone candidate with the first candidate when the homophone candidate determined by the homophone candidate check unit is not the first candidate. 2. The document creation apparatus according to claim 1, further comprising candidate replacement means. 11. A reading character string specifying means for specifying a reading character string to be used for predictive input from an input reading character string by referring to information registered in the predictive input dictionary; 2. The document creation apparatus according to claim 1, further comprising: a reading character string discriminating display unit that displays the reading character string used for predictive input specified by the specifying unit and other reading character strings. . 12. A prediction input method for performing input prediction from an input reading character string when performing a kana-kanji conversion on the input reading character string, the method comprising: Refer to the registered information for input prediction based on the reading character string input for document creation, and input based on the reading character string input for document creation. Obtain a plurality of predicted character strings, which are character strings predicted to be created, and refer to the information for specifying the field of the document registered in the dictionary for document field specification, and refer to the characters in the document being created. Identify the field of the document based on the column, narrow down the obtained plural predicted character strings according to the field of the specified document, display the narrowed predicted character string, and display the displayed predicted character Input predicted string selected from column Predictive input method comprising Rukoto. 13. A program for predicting an input from an input reading character string when a document is created by performing a kana-kanji conversion on the input reading character string, the program being registered in a prediction input dictionary. In addition, referring to information for performing input prediction based on a reading character string input for document creation, input is performed based on a reading character string input for document creation. Obtains a plurality of predicted character strings, which are character strings that are predicted to be, and refers to information for specifying the field of the document registered in the dictionary for document field specification, and determines the character string in the document being created. Based on the specified document field, the obtained predicted character strings are narrowed down according to the specified document field, the narrowed predicted character string is displayed, and from the displayed predicted character string Enter the selected prediction string Computer readable recording medium storing a program for controlling a computer to.