JPH0444301B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a Japanese sentence creation device that converts Japanese notation input in kana to notation mixed with kanji and kana (hereinafter abbreviated as "kanji notation").

2. Description of the Related Art In recent years, Japanese sentence creation devices using a kana-kanji converter system have become popular as they have the advantage of requiring less burden on the operator as a Japanese input device. Among these, a Japanese sentence creation device that uses a kana-kanji conversion device method that inputs solid writing is attracting attention. In general, the kana-kanji conversion method converts an input Japanese kana sequence into kanji notation using this kana sequence as the reading, but the solid input kana-kanji conversion method converts the above-mentioned solid input kana sequence directly into kanji notation. Because the machine determines phrase breaks and then converts them to kanji, if you enter the wrong reading or the reading of a word that is not in the machine's internal word dictionary, all input characters will be converted to hiragana. In some cases, incorrect kanji notation for bunsetsu delimiters may be output.

Hereinafter, one method of converting solid kana to kanji in a conventional Japanese sentence creation device will be explained. This method is used, for example, in Information Processing Society of Japan Transactions (July 1979)
Vol.20No.4) “Separation of solid text and kana-kanji conversion using two-clause longest match method” (Hiroshi Makino,
Makoto Kizawa). The configuration of the Japanese part creation device will be explained using FIG. 6.
In FIG. 6, 1 is a kana string input section, and 2 is an input kana string temporary storage section, which stores the kana sequentially input from the kana string input section 1 as an input kana string. 3 is an independent word dictionary that stores at least Japanese pronunciations, kanji notation, and connection information. 4 is an attached word dictionary, which stores at least Japanese pronunciations and a connecting dictionary. Reference numeral 5 denotes a phrase determination section, which has an internal register JL for storing the length of an independent word and an internal register FL for storing the length of an attached word string. To briefly explain these internal registers JL and FL, both are set to "0" in the initial state, and one internal register JL stores the reading length of an independent word when it is detected. The other internal register FL is used to add the reading length of an adjunct word when it is detected. Next, the phrase search process will be explained. The independent word dictionary 3 is searched for independent words that have partially the same pronunciation from the beginning of the input kana string given from the bunsetsu string search section 7, which will be described later. Register if there are independent words with partially the same pronunciation.
Store the reading length of the independent word in JL. If there is no independent word that has partially the same pronunciation, a specific adjunct word string search determination unit 6, which will be described later, is activated to determine the string from the beginning of the input kana string to the specific adjunct word string as a clause. Next, the adjunct word dictionary 4 is searched for adjunct words that have partially the same pronunciation after the point indicated by "JL+1". Add the reading length of the attached word to register FL. It is determined whether or not the detected independent word and adjunct word, as well as the adjunct word and adjunct word, can form a clause using the connection information of each word and the connection table shown in FIG. For those that hold true, add the reading length to register FL,
The adjunct word search and determination process are repeated until the phrase is determined. If the phrase does not hold true, the phrase that holds true is set as the detected phrase. Then, the stored values of registers JL and FL are returned to the phrase string search unit 7, which will be described later. 6
is a specific adjunct word string search confirmation unit which has a specific adjunct word string dictionary, and selects adjunct words that have partially the same pronunciation from the beginning of the input kana string given by the clause confirmation unit 5 in the specific adjunct word string dictionary. Search within. If there is a specific adjunct word string, it is determined up to the specific adjunct word string, and the determined number of characters is returned to the phrase string search unit 7, which will be described later. If not, return information to that effect. 7 is the phrase string search section,
Internal register that stores the clause search start point
Have BP. To explain this register BP, the register BP is set to "1" in the initial state, and the length of the phrase determined by the phrase determination section 5 "JL + FL" and the length determined by the specific adjunct word string search determination section 6 are stored. Additional memorization. Next, phrase string search will be explained. Of the above input examples, register BP
The phrase determining unit 5 is instructed to search for a phrase from the point indicated by . If a phrase is detected as a result of the phrase search, the phrase determining unit 5 registers it.
Return JL and FL. Then, the phrase string search unit adds "JL+FL" to register BP. The above-described phrase string search is repeated for all input kana strings until the processing is completed. Reference numeral 8 denotes a kanji output section which displays the kanji notation of the phrase string searched by the phrase string search section 7.

Next, an example of the operation of the conventional Japanese sentence creation device described above will be described. First, it is assumed that the operator inputs ``Carter Koho ni Select'' in an attempt to obtain the kanji notation for ``Carter Candidate.'' The clause string search unit 7 activates the clause determination unit 5. The phrase determination unit 5 selects independent words with partially matching pronunciations from the beginning of the input kana example in the independent word dictionary 3.
(shown in FIG. 3), it is determined that there is no independent word with a partially matching pronunciation from the beginning of the input kana example, and the specific adjunct word string search determining unit 6 is activated. The specific adjunct word string search confirmation unit 6 searches for a specific adjunct word string (shown in Figure 7) starting from the character string that cannot be analyzed (``ka''), and searches for a specific adjunct word string (shown in Figure 7) up to the longest adjunct word string detected, 'ni'. is confirmed as a clause. Then, the specific adjunct word string search determination unit 6 returns the determined number of characters “8” to the bunsetsu string search unit 7. The phrase string search section 7 activates the phrase determination section 5 after point "9" in the input kana string, and the phrase determination section 5 performs the following determination processing. An independent word having a partially matching pronunciation is searched in the independent word dictionary 3 (shown in FIG. 3), and "Kari" is detected. Then, the adjunct word dictionary 4 is searched for an adjunct word having the same pronunciation as the character string after "ru", and "ru" is detected. Checks whether the detected independent word and adjunct word are connected, and determines that they can be connected. Here, ``decide'' is confirmed as a clause.
As described above, the kanji output section 8 displays the phrase strings ``Kataa Koohoni'' and ``Kateru'' detected by the phrase string search section 7.

As mentioned above, in the conventional Japanese sentence creation device, even if the word "candidate" is in the independent word dictionary 3, the word "carter" is in the independent word dictionary 3.
If it is not registered, even the "candidate" part will be displayed in hiragana. In addition, if the operator makes a mistake in inputting the pronunciation sequence, for example, when he should have entered "Mondai wo kaiketsu suru" but instead enters "Nmon dai wo kaiketsu suru", the conventional technology described above This is the one that says "Solve the problem".
That is, even though the independent word dictionary 3 includes the word "problem", this part is displayed in hiragana.

Problems to be Solved by the Invention In the conventional Japanese sentence creation device as described above, when there is no word in the dictionary in the device or when a reading sequence is input by mistake, the input kana sequence is located at position A.
When the phrase search from position A fails, a specific adjunct string is searched from position A onward, and if the adjunct word string is found, the phrase is determined at position B. Therefore, from position A to position B, The reading sequence is displayed as hiragana. However, in most cases, this hiragana string includes words registered in an independent word dictionary. As described above, even if there is a word in the dictionary within the device, there is a problem in that that part is not converted.

The present invention has been made in view of this point, and even if a word is not found in the dictionary in the device or if the pronunciation is incorrectly input, the words registered in the independent word dictionary are displayed as kanji as much as possible. The purpose of this invention is to provide a Japanese sentence creation device that makes it possible to create Japanese sentences.

Means for Solving the Problems In order to solve the above problems, the Japanese sentence creation device of the present invention includes at least a kana input means for inputting kana, an input kana string temporary storage section for storing input kana strings, and at least An independent word dictionary that stores the pronunciations, kanji notation, and connection information of Japanese independent words, an adjunct word dictionary that stores at least the pronunciations and connection information of adjunct words, and input from the above-mentioned kana input means. A clause string search unit that activates a clause determination unit, which will be described later, to search for clauses whose pronunciations partially match the input kana sequence, and repeats the process of determining the clause until the entire input kana sequence is completed. , a phrase determination unit that searches for a phrase having a pronunciation that partially matches the input kana string specified by the phrase string search unit, using the independent word dictionary and the attached word dictionary;
Activated when information indicating that the above clause could not be retrieved is retrieved from the clause confirmation unit, and one character confirmation is used to confirm the first character of the input kana string specified by the clause string search unit as a clause. and a kanji output section that displays the kanji notation of the phrase string searched by the phrase string search section.

Effects The present invention has the above configuration, so that when there is no word in the dictionary in the device or when a reading sequence is input by mistake,
Through the processing of the single character determination unit, words registered in the independent word dictionary can be displayed as Kanji characters as much as possible.

Embodiment An example of an embodiment of the Japanese sentence creation device according to the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a Japanese sentence creation device according to an embodiment of the present invention. In the figure, numerals 1 to 4 and 8 are the same as the components with the same numbers in FIG. 6 described above. Reference numeral 15 denotes a phrase determination section, which has an internal register JL for storing the length of an independent word and an internal register FL for storing the length of an attached word string. To briefly explain these registers JL and FL, both are set to "0" in the initial state, and one internal register JL stores the reading length of an independent word when it is detected. the other register
FL is such that when an adjunct word is detected, its reading length is added. Next, the phrase search process will be explained. The independent word dictionary 3 is searched for independent words that have partially the same pronunciation from the beginning of an input kana string given from a bunsetsu string search unit 17, which will be described later. Register if there are independent words with partially the same pronunciation.
Store the reading length of the independent word in JL. If there is no independent word with partially the same reading, use 1 as described below.
Activates the character confirmation section. Next, the adjunct word dictionary 4 is searched for adjunct words that have partially the same pronunciation after the point indicated by "JL+1". register
Add the reading length of the attached word to FL. It is determined whether or not the detected independent word and adjunct word or adjunct word and adjunct word form a clause using the connection information of each word and the connection table shown in FIG. 5. For those that hold true, the reading length is added to the register FL, and the adjunct word search and determination process are repeated until it is determined as a clause. If the phrase does not hold true, the phrase that holds true is set as the detected phrase. Then, the stored values of registers JL and FL are returned to the phrase string search unit, which will be described later. Reference numeral 16 denotes a one-character determining unit, which determines a phrase using only the input characters given from the phrase determining unit 15, and returns the determined number of characters "1" to the phrase string search unit 17, which will be described later. Reference numeral 17 denotes a clause string search unit, which has an internal register BP that stores the clause search start point. To explain this register BP, the register BP is set to "1" in the initial state, and the length of the phrase determined by the phrase determination section 15 "JL + FL" and the length determined by the one character determination section 16 is "1". Additional memorization. Next, phrase string search will be explained. An example of the input kana after the point indicated by the register BP in the input kana string is set as an input kana string to be passed to the phrase determining section 15, and the phrase determining section 15 is instructed to search for a phrase. When a phrase is detected as a result of the phrase search, the phrase determination unit 15 returns registers JL and FL, and additionally stores the determined number of characters in register BP. The above-described phrase string search is repeated for all input kana strings until the processing is completed.

An example of the operation of the Japanese sentence creation device configured as described above will be described below with reference to the flowchart shown in FIG.

First, in a step, the operator inputs a sequence consisting of a kana input section 1, and the kana sequence input as a step is temporarily stored in an input kana sequence temporary storage section 2. Next, the phrase string search unit 17 searches the internal register BP.
The input kana example stored in the input kana string temporary storage unit 2 after the point in the input kana string indicated by the value is set as the input kana string to be passed to the bunsetsu determination unit 15, and has the same reading as the subsequence of the input kana string. The phrase determination unit 15 is instructed to search for a phrase. Here, as a step, the phrase determination section 15 searches for an independent word from that point. Then, as a result of the step search (step), if there is an independent word, its reading length is set to lng1, then lng1 is stored in register JL and the process proceeds to the step; otherwise, the process proceeds to the step. If register JL is greater than the length of the given input sequence in step, proceed to step; otherwise proceed to step. In the step, the phrase determination section 15 converts the one-character determination section 16
is instructed to confirm only one character indicated by the register BP in the input kana string as a clause, and when the one character confirmation unit 16 is confirmed, it returns a message to that effect to the clause string search unit 17. Then, the phrase string search section 17 sets "BP+1" in the register BP, and sends the determined kanji notation of the phrase to the kanji output section 8 (step).
Next, in the step, if the register BP is greater than the length of the input sequence, proceed to the step, otherwise proceed to the step.

In step, the segment determination unit 15 selects in the adjunct word dictionary 4 the adjunct word that has the same pronunciation as the subsequence after the point in the input kana sequence indicated by “JL+1” out of the input kana sequence given from the bunsetsu string search unit 17. Search. Let the reading length of the adjunct word detected here be lng2. And in the step,
It is determined whether or not the detected independent word and the dependent word or the dependent word and the dependent word are connected using the connection information of each detected word and the connection table. As a result, in the step, if the connection is possible, the internal register FL
Set "FL+lng2" to "FL+lng2" and proceed to step. If connection is not possible, proceed to step. In step, if "JL+FL+1" is greater than the length of the given input kana string, the clause determining unit 15 proceeds to step; otherwise, proceeds to step. In step, the phrase determination section 15 determines the connectable adjuncts as phrases, and the phrase string search section 17 adds "JL+FL" to the register BP. Then, the phrase string search section 17 sends the determined kanji notation of the phrase to the kanji output section 8 (step). Next, in step register BP
If is greater than the length of the input sequence, proceed to step; otherwise proceed to step.
Finally, in step, the kanji output section 8 displays the kanji notation sent from the phrase string search section 17.

The Japanese sentence creation device of the embodiment has been explained above using the flowchart shown in FIG.
This will be explained using FIG.

For example, let's assume that the operator is trying to get the kanji notation for ``I choose Carter as a candidate,'' and inputs ``I choose Carter as a candidate.''
This will be explained in accordance with the steps (101) to (139) below.

(101)...The operator inputs "Kata Koho ni Decide" from the kana input section 1.

(102)...When the input is completed, "Kata-kōho ni kimeru" is temporarily stored in the input kana string temporary storage unit 2 as an input kana string (step).

(103)...The phrase string search unit 17 initializes the internal register BP and sets it to "1".

(104)...The clause string search unit 17 sets the input kana sequence to the clause determination unit 15 from the point indicated by the register P onwards from the input kana sequence to the clause determination unit 15, and
Instructs to search for phrases for this kana string. The phrase determining unit 15 searches the independent word dictionary 3 shown in FIG. 3 for independent words having the same pronunciation from the beginning of the input kana string (step).

(105)...The phrase determination unit 15 determines that there is no independent word with the same reading in the independent word dictionary 3 (step).

(106)...Phrase determination unit 15 is a register for input kana string
The one-character determining section 16 is instructed to confirm with only one character indicated by BP, and the one-character determining section 16 decides the phrase only with the character indicated by register BP in the input kana string. Then, phrase string search unit 1
7 receives the fixed number of characters “1” and registers
Set "BP+1" to BP (step).

(107)...The phrase string search unit 17 sends the determined kanji notation of the phrase to the kanji output unit 8 (step).

(108)...It is determined that the register BP (=2) is smaller than the length of the input kana string "11" (step).

(109)...Next, the bunsetsu string search unit 17 instructs to search for a bunsetsu from the point with the value "2" of the register BP of the input kana string, and the bunsetsu determination section 15 searches for an independent bunsetsu that has the same reading from the beginning of the input kana string. The word is searched in the independent word dictionary 3 shown in FIG. 3 (step).

(110)...Repeat steps (104) to (107).

(111)...Register BP (=3) is determined to be smaller than the length of the input kana string "11" (step).

(112)...Next, the bunsetsu string search unit 17 instructs to search for a bunsetsu from the point of the value "3" of the register BP in the input kana example, and the bunsetsu determination unit 15 selects an independent bunsetsu string that has the same pronunciation from the beginning of the input kana string. The word is searched in the independent word dictionary 3 shown in FIG. 3 (step).

(113)...Repeat steps (104) to (107).

(114)...It is determined that the register BP (=4) is smaller than the length of the input kana string "11" (step).

(115)...Next, the bunsetsu string search unit 17 instructs to search for a bunsetsu from the point with the value "4" of the register BP in the input kana example, and the bunsetsu determination unit 15 selects an independent bunsetsu string that has the same reading from the beginning of the input kana string. The word is searched in the independent word dictionary 3 shown in FIG. 3 (step).

(116)...Repeat steps (104) to (107).

(117)...It is determined that the register BP (=5) is smaller than the length of the input kana string "11" (step).

(118)...Next, the clause string search unit 17 instructs to search for clauses starting from the point with the value "5" of the register BP in the input kana example, and the clause determination unit 15 searches for independent clauses that have the same reading from the beginning of the input kana series. The word is searched in the independent word dictionary 3 shown in FIG. 3 (step).

(119)...The clause determining unit 15 detects the "candidate" and sets "3" in the register JL.

(120)...Register JL (=3) is determined to be smaller than the length of the given input kana string "7" (step).

(121)...Next, the bunsetsu determination unit 15 searches the adjunct word dictionary 4 shown in FIG. 4 for an adjunct word having the same pronunciation from the point of value "4" of "JL+1" of the input kana string given, Send "ni" (step).

(122)...Furthermore, the phrase determining unit 15 uses the connection table shown in FIG. 5 to check whether or not the connection information of the independent word "candidate" detected above and the connection information of the adjunct word "ni" can be connected (step ).

(123)...Determines that connection is possible and sets "1" in register FJ. Then, it compares the length of the given input kana string, ``7'', with the value of ``JL+FL'', ``4'', and determines whether or not the input kana string ends (step).

(124)...The phrase determination unit 15 is the input kana column “JL+FP
The adjunct word dictionary 4 shown in FIG. 4 is searched for an adjunct word having the same pronunciation from the point with the value '5' of '+1', and 'ki' is detected (step).

(125)...The clause determination unit 15 uses the connection table shown in FIG. 5 to check whether or not the connection is possible using the connection information of the subordinate word "ni" detected above and the connection information of the subordinate word "ki" (step) .

(126)...The phrase determination unit 15 determines that connection is impossible (step).

(127)...The phrase determination unit 15 determines the connectable "candidate ni" as a phrase. Clause string search section 17
adds "JL+FL" to register BP (register BP is "5") (step).

(128)...The phrase string search unit 17 sends the determined kanji notation of the phrase to the kanji output unit 8 (step).

(129)...The length of the input Kana string "11" is compared with the value "9" of the register BP, and it is determined whether the input Kana string has ended or not (step).

(130)...The bunsetsu string search unit 17 instructs to search for a bunsetsu from the point of value “9” in the register BP of the input kana string, and the bunsetsu determination unit 15 searches for independent words with the same pronunciation from the beginning of the input kana string. A search is made in the independent word dictionary 3 shown in FIG. 3 (step).

(131)…Detects “decision” and puts “2” in register JL
Set (step).

(132)...Register JL (=2) is determined to be smaller than the input kana string length "3" (step).

(133)…Next, the phrase determination unit 15 inputs the input kana example “JL
The adjunct word dictionary 4 shown in FIG. 4 is searched for an adjunct word having the same pronunciation from the point with the value '5' of '+1', and 'ru' is detected (step).

(134)...Furthermore, the clause determining unit 15 uses the connection table shown in FIG. 5 to check whether the connection is possible using the connection information of the independent word "kari" detected above and the connection information of the adjunct word "ru" (step ).

(135)...Determines that connection is possible and sets "1" in register FL. The input kana string length "3" and the value "4" of "JL+FL+1" are compared to determine the input kana string circumferential force (step).

(136)...The phrase determination unit 15 determines "determine" as a phrase, and the phrase string search unit 17 adds "JL+FL" to the register BP (step).

(137)...The phrase string search unit 17 sends the determined kanji notation of the phrase to the kanji output unit 8 (step).

(138)...The bunsetsu string search unit 17 compares the length of the input kana string "11" with the value of register BP "12" and determines that the input kana string has ended (step).

(139)...The kanji output section 8 displays the kanji notation ``choose as a candidate for ka-ta'' sent from the phrase string search section 17 (step).

As shown in the above operation example, according to this embodiment, even if the operator inputs the pronunciation string ``Carter Koho ni Kimeru'' that includes the word ``Carter'' that is not in the device's internal dictionary, it will not be entered in the device's internal dictionary. If a word does not exist, each character is determined as a phrase, and the message ``Select as a candidate'' is displayed. Comparing this with the case of the prior art, the portion that is displayed as hiragana without conversion is ``Ka-ta-Kohoni''. That is, the portion to be displayed without being converted is greatly reduced in this operation example, and the operator's correction operations can be reduced.

Effects of the Invention As described above, the Japanese sentence creation device of the present invention has an input kana string temporary storage section that stores input kana strings, and an independent input kana string temporary storage section that stores at least the pronunciations and kanji notations of Japanese independent words. Activates the phrase determination section, which will be described later, to search for phrases whose pronunciations partially match the input kana sequence with the word dictionary and the attached word dictionary that stores at least the pronunciations and connection information of the attached words; A phrase string search unit that repeats the process of determining a phrase until all of the input kana strings are completed, and an independent word dictionary that searches for phrases with pronunciations that partially match the input kana string specified by the phrase string search unit. and a bunsetsu determination unit that searches using the clause determination unit and an attached word dictionary, and is activated when receiving information from the clause determination unit that the clause cannot be searched, and performs an input specified by the clause string search unit. By being equipped with a single-character confirmation section that confirms the first character of a kana string as a bunsetsu, if the word is not found in the device's internal dictionary or if the pronunciation string is entered by mistake, it will not be converted and will be displayed as hiragana. This greatly reduces the number of parts that need to be adjusted, and reduces the amount of correction operations required by the operator, which has an extremely large practical effect.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a Japanese sentence creation device according to an embodiment of the present invention, FIG. 2 is a flowchart showing the operation of the Japanese sentence creation device according to an embodiment of the present invention, and FIG. 3 is an example of an independent word dictionary. Figure 4 is a diagram showing an example of an adjunct word dictionary, Figure 5 is a diagram showing an example of a connection table, Figure 6 is a block diagram of a conventional Japanese sentence creation device, and Figure 7 is a diagram of a conventional Japanese sentence creation device. It is a figure which shows an example of the specific attached word string dictionary of a word sentence creation apparatus. 1...Kana input section, 2...Input kana example temporary storage section, 3...Independent word dictionary, 4...Adjunct word dictionary, 8...
...Kanji output section, 15...Phrase determination section, 16...1
Character confirmation section, 17... Clause string search section.

Claims (1)

[Claims] 1. A kana input means for inputting kana, an input kana string temporary storage section for storing input kana strings, and an independent word dictionary for storing at least the pronunciations, kanji notation, and connection information of Japanese independent words. , an adjunctive word dictionary that stores at least pronunciations and connection information of adjunct words, and a clause confirmation process described later to search for a clause whose pronunciation partially matches the input kana string input from the above-mentioned kana input means. A bunsetsu string search section that activates the bunsetsu string search section and repeats the process of determining bunsetsu until the entire input kana string is completed, and a bunsetsu string search section that searches for pronunciations that partially match the input kana string specified by the bunsetsu string search section. A phrase determination unit that searches for a phrase with the phrase using the independent word dictionary and an attached word dictionary, and a phrase determination unit that is activated when receiving information from the phrase determination unit that the phrase cannot be retrieved, Accordingly, the present invention includes a one-character determination unit that determines the first character of a specified input kana string as a clause, and a kanji output unit that displays the kanji notation of the clause string searched by the clause string search unit. Characteristic Japanese sentence creation device.