JP5501090B2 - Phrase prediction candidate output device - Google Patents

Description

  The present invention relates to a phrase prediction candidate output device having a kana-kanji conversion function, and more particularly, to a phrase prediction candidate output device that performs a conversion process on a plurality of phrase character strings to reduce the input load of long sentences.

  The kana-kanji conversion function generally includes a predictive conversion function that narrows down conversion candidates by performing a forward match search from the predictive conversion dictionary using the input yomi character string as a search key, and the input yomi character string for each phrase. A phrase conversion function for creating a conversion candidate by performing a complete match search from the phrase conversion dictionary using the Yomi character string for each phrase as a search key is provided.

  According to the predictive conversion function, a word that matches forward with the input character string is searched from the predictive conversion dictionary to narrow down conversion candidates. For example, if the input string is “Hana”, the conversion results “Flower”, “Bouquet”, “Flower”, “Talk”, etc. that match the Yomi string “Hana” will be narrowed down. These conversion results are displayed and output. In the following description, the converted character string obtained by the predictive conversion is referred to as a predictive conversion candidate.

  According to the phrase conversion function, the input Yomi character string is divided into phrases, words that completely match the Yomi character string for each phrase are searched from the phrase conversion dictionary, and conversion candidates are created by combining these words. . For example, if the input string is “Oreina”, the phrase “Oreini” is matched with “Thank you”, and the phrase “Hana” is matched with “Hana”. “Thank you flowers” is created as a conversion candidate, and this conversion candidate is displayed and output. In the following description, the converted character string obtained by the phrase conversion is referred to as a phrase conversion candidate.

  By the way, for example, suppose that the user wants to input the text “Thank you for the bouquet” and inputs the Yomi character string “Oreinana” to a character string conversion device having a kana-kanji conversion function. Here, even if the conversion key is operated to instruct the character string conversion device to display and output phrase conversion candidates, the character string “thank you bouquet” consisting of two phrases “thank you” and “bouquet” Since it is not recorded (registered) in the predictive conversion dictionary, and when the phrase conversion function is executed, the predictive conversion function is not executed at the same time, so only “thank you flowers” are displayed and output as a phrase conversion candidate. The conversion candidate “Thank you for bouquet” to be used is not displayed and output.

  Accordingly, the phrase conversion candidate “thank you for the bouquet” is not displayed and output unless the complete text string “Ore ni nabata” is input.

As explained above, you can not get conversion candidates such as “Thank you bouquet” just by entering an incomplete text string such as “Oreina”. A character string predicting device is disclosed in Patent Document 1.
The character string prediction device creates a plurality of character strings that are shortened in order from the beginning of the input character string. For example, if the input character string is “Kakinokaigi”, the character string “Kakinokaigi”, “Kinokaigi”, “Nokikai”, “Kaigi”, “I” "".

Then, the conversion result obtained by performing the predictive conversion on the words (headings) that are forward-matched or completely matched with each of the created five character strings, and the phrase for the reading character string before the matching words The conversion result obtained by executing the conversion is combined. In this case, the predictive conversion process is executed on the matching character string “Kaigi”, for example, “Conference will be held as a conversion result (predictive conversion candidate)”, “ Get a “Conference”, “Conduct a conference”, “Conduct a conference”.
In addition, the phrase conversion process is executed for “Kakino” which is the reading character string before the matching word to obtain “below”. Finally, combining “below” with the prediction conversion candidate “please gather because a meeting will be held” etc., “please gather because the following meeting will be held”, “ “Conference will be held” is displayed and output.

JP 2000-276466 A

  However, in the technique disclosed in Patent Document 1, a plurality of character strings are created for the input character string, and predictive conversion is performed on words that match the plurality of character strings in front or completely. As a result, many conversion candidates that are not actually used by the user may be output. For example, in addition to “Kaigi”, when predictive conversion processing is executed for “Igi”, a large number of conversion candidates that are significantly different from the user ’s intention, such as “No objection” and “There is objection”, are output. Sometimes. As a result, the user must select a desired candidate from among a large number of conversion candidates, which impairs the operability of selecting conversion candidates.

  The present invention has been made in view of such circumstances, and an apparatus that outputs an optimal conversion candidate by executing a predictive conversion process in accordance with a user's intention even when an incomplete character string is input. The purpose is to provide.

The first technical means includes a character input unit for inputting a reading character string, a clause delimiter for dividing the input reading character string for each clause, and a phrase for recording character data that completely matches the reading character string of one phrase. A conversion dictionary and a predictive conversion dictionary that records character data that matches forward with a character string, and the conversion candidates that completely match the character string for each phrase in the character string divided for each phrase And the conversion candidate for each acquired phrase is combined to create a phrase conversion candidate for the input reading character string, and the reading character string of the final phrase of the input reading character string Is obtained from the predictive conversion dictionary, the final phrase of the created phrase conversion candidate is replaced with the obtained predictive conversion candidate, and a phrase prediction candidate is generated and displayed. Are those that, the obtained from the prediction conversion dictionary, wherein if the final clause read character string and the forward matching prediction conversion candidates have multiple from the predictive conversion candidate of the plurality of, phrase conversion candidates created in the above This is a phrase prediction candidate output device that selects a prediction conversion candidate that coincides with the last phrase in advance and creates a phrase prediction candidate based on the selected prediction conversion candidate .

The second technical means is characterized in that, in the phrase prediction candidate output device of the first technical means, when a plurality of phrase conversion candidates are created, a phrase prediction candidate is created for each phrase conversion candidate. To do.

In the phrase prediction candidate output device of the first or second technical means, the third technical means acquires the input conversion candidate for the input character string from the prediction conversion dictionary, and the acquired prediction conversion candidate The phrase prediction candidate is created only when the number of the phrase is less than or equal to the predetermined number.

  According to the present invention, the final phrase of the phrase conversion candidate is replaced with the prediction conversion candidate that is obtained from the prediction conversion dictionary and matches the front character string of the final phrase, and thus the phrase prediction candidate is created. Even if an input character string is input, conversion candidates according to the user's intention can be created and displayed. As a result, it is possible to reduce the input load of long sentences and further improve the operability for selecting conversion candidates.

It is a figure explaining the phrase prediction candidate output process performed in the phrase prediction candidate output device of the present invention. It is a functional block diagram of the phrase prediction candidate output device of the present invention. It is a flowchart explaining the flow of the phrase prediction candidate output process performed in the phrase prediction candidate output device of this invention.

FIG. 1 is a diagram for explaining phrase prediction candidate output processing executed in the phrase prediction candidate output device 1 of the present invention.
When the user inputs a "Do to thank" character input key is operated readings string, as indicated by reference numeral X _1, the read character string is displayed on the display panel 2. The phrase prediction candidate output device 1, when two or more phrase readings are input, divides the input reading string for each phrase and executes phrase conversion processing for each phrase.

Here, “Oreiini / Hana” is divided into two phrases, and the phrase conversion process is executed for “Oreiini” and “Hana”. Results of the phrase conversion process, as phrase conversion candidate indicated by symbol X ₂ displays the "blossom in return". “/” Indicates the segment break position.
In this way, if phrase conversion candidates are displayed, phrase batch determination can be performed quickly.

  In addition, a predictive conversion process is performed on the last character string of the last phrase of the phrase conversion candidate. As a result of the predictive conversion process, as a predictive conversion candidate for “Hana”, which is the Yomi character string of the last phrase, words that match the Yomi character string (“Hana”) in front, such as “Bouquet”, “Flower” Suppose you get

Then, a character string in which the last phrase of the phrase conversion candidate is replaced with the prediction conversion candidate is created and displayed. Here, as the final clause "flower" to "bouquet", a string is replaced with "floral" of phrase conversion candidate "thank / Hana", indicated by the reference numeral X ₃ "bouquet to thank", "floral thank you ”Is displayed.
In the following description, a character string obtained by replacing the final phrase of the phrase conversion candidate with the prediction conversion candidate is referred to as a phrase prediction candidate.
Thus, if the phrase prediction candidates are displayed, a long sentence can be created effectively with a small number of key inputs, and the user's key input operation can be reduced. Furthermore, the operability of selecting conversion candidates can be improved.

FIG. 2 is a functional block diagram of the phrase prediction candidate output device 1. The phrase prediction candidate output device 1 can be mounted on an electronic device having a character input function, for example, a mobile phone, a PHS, a smartphone, a personal computer, a handy terminal, and the like.
The phrase prediction candidate output device 1 includes an operation unit 11, a kana / kanji engine 12, a dictionary recording unit 13, a conversion candidate creation unit 14, and a display unit 15.

  The operation unit 11 includes character input keys and operation keys, accepts a user's instruction operation, and functions as a character input unit that inputs a yomi character string created by the user's character input key operation to the kana / kanji engine 12. .

  The kana / kanji engine 12 executes a kana / kanji conversion process on the input character string. Furthermore, a phrase delimiter 12a, a phrase converter 12b, and a predictive converter 12c are provided.

  The phrase delimiter 12a executes a process of dividing the input character string for each phrase. In addition, the said process can use the various methods conventionally proposed.

  The phrase conversion unit 12b has a phrase conversion function, acquires from the phrase conversion dictionary 13a of the dictionary recording unit 13 a word (conversion candidate) that completely matches the Yomi character string for each phrase delimited by the phrase delimiter 12a. By combining the obtained conversion candidates for each phrase, a phrase conversion candidate for the Yomi character string is created.

  The predictive conversion unit 12 c has a predictive conversion function, and acquires a predictive conversion candidate that matches forward with the character string from the predictive conversion dictionary 13 b of the dictionary recording unit 13.

  The dictionary recording unit 13 is a recording device such as an HDD or a flash memory, and records a phrase conversion dictionary 13a and a prediction conversion dictionary 13b, which are various dictionaries (data) for character string conversion.

  The phrase conversion dictionary 13a is a dictionary that records character data that completely matches a half-character string of one phrase. For example, “Hana”, “Flower”, “Hana”, “Nose”, etc., which are character data that completely matches “Hana”, are recorded in correspondence with the Yomi character string “Hana”.

  The predictive conversion dictionary 13b is a dictionary that records character data that coincides with the front character string. For example, corresponding to the Yomi character string “Hana”, the character data “Bouquet”, “Flower pattern”, “Flower language”, “Flower shape”, “Story”, “噺”, which is the character data that precedes “Hana”. "Etc. are recorded.

The conversion candidate creation unit 14 performs operation control of the kana / kanji engine 12. Also, a prediction conversion candidate that matches the front character string of the final phrase of the phrase conversion candidate created by the phrase conversion unit 12b is acquired, and the final phrase of the phrase conversion candidate is replaced with the acquired prediction conversion candidate.
The prediction conversion candidate may be acquired by executing the following process. First, the conversion candidate creation unit 14 acquires the last character of the last phrase of the phrase conversion candidate created by the clause conversion unit 12b, and outputs the last character of the last phrase to the prediction conversion unit 12c.

  The predictive conversion unit 12 c acquires from the predictive conversion dictionary 13 b a predictive conversion candidate that matches forward with the last-character Yomi character input from the conversion candidate creation unit 14, and outputs the prediction conversion candidate to the conversion candidate creation unit 14.

  When the conversion candidate creation unit 14 acquires the prediction conversion candidate in this manner, the conversion candidate creation unit 14 creates and displays a phrase prediction candidate by replacing the final phrase of the phrase conversion candidate created by the phrase conversion unit 12b with the prediction conversion candidate. Display output to the unit 15.

  The display unit 15 includes the display panel 2 and the like, and displays the input character string, phrase conversion candidate, phrase prediction candidate, and the like.

  FIG. 3 is a flowchart for explaining the flow of the phrase prediction candidate output process executed in the phrase prediction candidate output device 1. In the following description, a case where the user inputs the reading character string “Oreina” will be described as an example.

  When the user inputs a reading character string via the operation unit 11 (step S1), the operation unit 11 outputs the input reading character string to the kana / kanji engine 12. In this case, it is assumed that “I am not good” is input. The phrase delimiter 12a of the kana / kanji engine 12 delimits the input yomi character string for each phrase (step S2), and outputs it to the phrase converter 12b and the conversion candidate generator 14. Here, “Oreini / Hana” is divided into two phrases.

  Next, the phrase conversion unit 12b acquires from the phrase conversion dictionary 13a of the dictionary recording unit 13 a conversion candidate that completely matches the Yomi character string for each phrase delimited by the phrase delimiter 12a, and creates a phrase conversion candidate (step S3) and output to the conversion candidate creation unit 14. Here, the phrase conversion process is executed for “Oreini” and “Hana”, and “thank you / flower” is output to the conversion candidate creation unit 14 as a phrase conversion candidate. In this embodiment, a case where only one candidate is output as a phrase conversion candidate will be described as an example. However, when a plurality of phrase conversion candidates are output, steps S4 to S4 described below for each of the plurality of phrase conversion candidates. The process of S7 is executed.

  The conversion candidate creation unit 14 acquires the final character string of the last phrase of the phrase conversion candidate input from the phrase conversion unit 12b (step S4), and outputs it to the prediction conversion unit 12c. Note that the final character's Yomi character may be acquired from the Yomi character string delimited for each phrase input from the phrase delimiter 12a. Here, since the input reading character string is divided into two phrases “Oreini / Hana”, the conversion candidate creation unit 14 acquires “Hana” as the reading character string of the final phrase.

  The predictive conversion unit 12c acquires from the predictive conversion dictionary 13b a predictive conversion candidate that matches the front character string of the last phrase input from the conversion candidate generating unit 14 (step S5), and outputs the prediction conversion candidate to the conversion candidate generating unit 14. Here, it is assumed that the prediction conversion candidate 1 “bouquet”, the prediction conversion candidate 2 “flower pattern”, and the prediction conversion candidate 3 “story” are acquired.

  The conversion candidate creation unit 14 replaces the final phrase of the phrase conversion candidate input from the phrase conversion unit 12b with each prediction conversion candidate input from the prediction conversion unit 12c, and generates a phrase prediction candidate (step S6). . Here, it is assumed that three phrase prediction candidates “Thank you bouquet”, “Thank you floral pattern”, and “Thank you / story” are created.

  The conversion candidate creation unit 14 outputs the phrase conversion candidate created in step S3 and the phrase prediction candidate created in step S6 to the display unit 15, and displays and outputs them (step S7). In addition to the phrase prediction candidates “Thank you / Bouquet”, “Thank you / flower pattern”, “Thank you / story”, you can display and output a total of four candidates for phrase conversion candidate “Thank you / flower”. These four candidates can be selected as conversion candidates for the reading character string “Oreinana”.

  Only when the number of clauses delimited in step S2 is two or more clauses, the processing of steps S4 to S6 may be executed.

(Appendix 1)
In the above embodiment, as described in step S6, the phrase prediction candidates are created using all the prediction conversion candidates obtained from the last character string of the last phrase, but only the prediction conversion candidates that satisfy the predetermined condition are used. By creating a phrase prediction candidate using, a phrase prediction candidate closer to the user's intention can be displayed.

  For example, when there are a plurality of predictive conversion candidates that coincide with the half-character string of the last phrase of the phrase conversion candidate, the conversion candidate creation unit 14 selects a predetermined number or less of the predictive conversion candidates from the plurality of predictive conversion candidates. Then, a phrase prediction candidate may be created based on the selected prediction conversion candidate.

For example, when the condition “3” is set in advance in the conversion candidate creation unit 14 as the predetermined number, it is assumed that there are seven prediction conversion candidates. At this time, the conversion candidate creation unit 14 selects three prediction conversion candidates from the seven prediction conversion candidates, and creates a phrase prediction candidate based on the selected prediction conversion candidates. At this time, three may be selected in ascending order of the number of characters, or three may be selected in the order of increasing number of characters.
The number condition can be arbitrarily changed (customized) based on an instruction operation from the user.

(Appendix 2)
In addition, when there are a plurality of predictive conversion candidates that coincide with the last character string of the final phrase of the phrase conversion candidate, the conversion candidate creation unit 14 creates the phrase conversion unit 12b from among the plurality of prediction conversion candidates. It is also possible to select a prediction conversion candidate that matches the final phrase of the phrase conversion candidate, and create a phrase prediction candidate based on the selected prediction conversion candidate.

  For example, as described in the flowchart of FIG. 3, the final character string of the phrase conversion candidate is “Hana”, and the prediction conversion candidate 1 “bouquet” is input as a plurality of input prediction conversion candidates. Assume that there is conversion candidate 2 “flower pattern” and predictive conversion candidate 3 “story”. At this time, the conversion candidate creation unit 14 selects the predicted conversion candidate 1 “bouquet” and the predicted conversion candidate 2 “floral pattern” that coincide with the final phrase “flower” of the phrase conversion candidate.

  As a result, when the kana-kanji conversion engine creates conversion candidates in consideration of the connection of the sentences, the “thank you flowers” and “thank you flowers” that are correlated with the phrase conversion candidates “thank you flowers”. Only the “pattern” is displayed to the user as a phrase prediction candidate, and the phrase prediction candidates to be output are excluded by eliminating conversion candidates that are not correlated with the phrase conversion candidates (“thank you flowers”), such as “Thank you for thanks”. It becomes possible to limit to what is useful for a user.

  In accordance with this, if there is no predictive conversion candidate that matches the final phrase of the phrase conversion candidate created by the phrase conversion unit 12b, the conversion result of the predictive conversion is not considered in consideration of the correlation with the phrase conversion candidate. The phrase prediction candidates may be created in the order of candidates. Examples of the candidate order include alphabetical order and alphabetical order.

  Note that, based on an instruction operation from the user, it is also possible to determine whether or not to select a prediction conversion candidate that matches forward with the final phrase after the phrase conversion, from among the plurality of prediction conversion candidates.

(Appendix 3)
In addition, the predictive conversion unit 12c of the kana / kanji engine 12 acquires in advance the prediction conversion candidate of the input yomi character string from the prediction conversion dictionary 13b, and the conversion candidate creation unit 14 acquires the acquired prediction conversion candidate. The phrase prediction candidate may be created only when the number of is less than a predetermined number.

  For example, the prediction conversion candidate, “meeting room”, “meeting”, which is character data that matches forward with “Kaigi” corresponding to the reading character string “Kaigi”, is recorded in the prediction conversion dictionary 13b. It is assumed that “conference hall” is not recorded as a predictive conversion candidate of “Kaigi”. Further, it is assumed that the prediction conversion candidate “international conference hall” of the Yomi character string “Kokusaikaigi” is recorded in the prediction conversion dictionary 13b.

  At this time, when the conversion candidate creation unit 14 executes the phrase prediction candidate creation process for the input character string “Kokusaikai”, the phrase for “Kokusai (international) / kaigi (meeting)”. “International / meeting room” and “international / meeting” are displayed and output as prediction candidates. However, since “Conference Hall” is not recorded in the prediction conversion dictionary 13b as a predictive conversion candidate for the reading character string “Kaigi”, “International Conference Hall” is not displayed and output. In such a case, it is inconvenient for the user if the user intends to acquire “international conference hall” as a predictive conversion candidate.

  Therefore, the predictive conversion unit 12c acquires in advance the predictive conversion candidate for the input character string from the predictive conversion dictionary 13b. Then, the conversion candidate creation unit 14 performs the phrase described in steps S2 to S7 in FIG. 3 only when the number of obtained prediction conversion candidates is a predetermined number, for example, 0 or 1 to 3 or less. A prediction candidate creation process is executed.

  By doing in this way, since the conversion process according to a user's intention can be performed, a user's convenience improves further.

(Appendix 4)
The functions of the phrase delimiter 12a, the phrase converter 12b, the predictive converter 12c, and the conversion candidate generator 14 can also be realized by a program. In this case, the phrase prediction candidate output device 1 includes a ROM that records programs for realizing the functions of the phrase delimiter 12a, the phrase converter 12b, the prediction converter 12c, and the conversion candidate generator 14, a CPU that executes the program, and data A decompression RAM is provided. When creating the phrase prediction candidate of the present invention, the CPU reads the program from the ROM and executes it.

DESCRIPTION OF SYMBOLS 1 ... Phrase prediction candidate output device, 11 ... Operation part, 12 ... Kana kanji engine, 12a ... Phrase delimiter part, 12b ... Phrase conversion part, 12c ... Prediction conversion part, 13 ... Dictionary recording part, 13a ... Phrase conversion dictionary, 13b ... Prediction conversion dictionary, 14 ... Conversion candidate creation part, 15 ... Display part, 2 ... Display panel.

Claims (3)

A character input part for inputting a Yomi character string;
A phrase delimiter that delimits the input text string for each phrase;
A phrase conversion dictionary that records character data that exactly matches the Yomi string of one phrase;
A predictive conversion dictionary that records character data that matches the Yomi character string
A conversion candidate that completely matches the Yomi character string for each phrase in the Yomi string divided for each phrase is acquired from the phrase conversion dictionary, and the input is obtained by combining the acquired conversion candidates for each phrase. Create a phrase conversion candidate for the Yomi string,
Obtaining a predictive conversion candidate that matches the read character string of the last sentence of the input read character string from the predictive conversion dictionary;
The final phrase of the created phrase conversion candidate is replaced with the acquired prediction conversion candidate, and a phrase prediction candidate is created and displayed .
When there are a plurality of predictive conversion candidates that are obtained from the predictive conversion dictionary and that are forward-matched with the Yomi character string of the final phrase, the final phrase and the forward of the created phrase conversion candidate are selected from the plurality of predictive conversion candidates. A phrase prediction candidate output device that selects a matching prediction conversion candidate and creates a phrase prediction candidate based on the selected prediction conversion candidate . In the phrase prediction candidate output device according to claim 1 ,
A phrase prediction candidate output device characterized in that, when a plurality of phrase conversion candidates are created, a phrase prediction candidate is created for each phrase conversion candidate. In the phrase prediction candidate output device according to claim 1 or 2 ,
The prediction conversion candidate of the input reading character string is acquired from the prediction conversion dictionary, and the phrase prediction candidate is created only when the number of the acquired prediction conversion candidates is equal to or less than a predetermined number. Phrase prediction candidate output device.

Images