JPH0522269B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a method for automatically assigning kana pronunciations to numbers and the number particles following them.

[Prior art]

In the conventional method of assigning a reading kana to a kanji character string (counterword) immediately after a number, the reading of a kanji is uniquely determined for each character. If the kanji has a different reading, an exception dictionary of compound words is prepared that combines the characters before and after the kanji, and if the kanji to be converted during the conversion process is registered in the exception dictionary, the reading is We have adopted a method of prioritizing the following. Regarding number words, those whose reading changes depending on the number immediately before the particle are registered in the exception dictionary.

However, with this method, it is not possible to give pronunciation kana names for numbers, and for most of the number particles, it is necessary to have pronunciation kana in combinations such as 1, 3, 6, 8, ten, hundred, thousand, etc. in the exception dictionary. It was hot.

On the other hand, one possible use of the reading kana assignment method for number words is to add reading kana to number words when outputting the number words aloud. It is necessary to create a reading kana name using one of the following combinations: a number and a character representing the number of digits, a single digit number and a particle, or a character representing the number of digits and a particle.

[Purpose of the invention]

An object of the present invention is to provide a method for automating the pronunciation and naming of numbers and character strings following them (particles), in which the pronunciation of the number is determined according to the particle, and the reading of the particle is determined according to the number. It is about providing.

Another object of the present invention is to provide a system that minimizes the required computer memory capacity when registering the pronunciations of a large number of classifiers.

[Summary of the invention]

Depending on the number of digits and the following particle, the number is
The way it is read changes. Similarly, the readings of digit characters such as tens, hundreds, and thousands, which express the number of digits, and number words change depending on the number immediately preceding them.

Therefore, in the present invention, digit characters are divided into a number reading kana table that stores the reading kana of numbers and digit characters according to patterns according to the classifier, and a number reading kana table that stores the reading kana of digit characters and classifiers according to the classifier or digit character. A conversion type that can register the pronunciation conversion table for digit characters and classifiers in which types are stored by pattern, and the pronunciation and code indicating which pattern in the two tables above corresponds to each digit character and classifier.・Prepare a table and add reading kana to number words using the following processing steps.

In the description of this method, numbers shall be unified into arithmetic numerals.

(a) If the character immediately after the arithmetic numeral in a number word is a digit character, the expression should be unified to use only the arithmetic numeral.

(b) If the numbers in number words are written in Chinese numerals, they are from 1 to 9.
Converts to arithmetic numbers up to and digit characters such as tens, hundreds, thousands, ten thousand, etc.

(c) If the number in the number word is an arithmetic number string,
Insert a digit character immediately after each number to indicate the number of digits in that number.

(d) The unit of reading kana shall be one set of a 1-digit arithmetic number and a digit character, a 1-digit arithmetic number and a particle, or a digit character and a particle.

(e) Refer to the conversion type table to obtain the reading kana of digit characters and classifiers, and the pattern code of the reading kana table of numbers and the reading kana conversion table of digit characters and classifiers.

(f) Refer to the number reading kana table and find the number reading kana corresponding to the pattern code found in (e).

(g) Refer to the pronunciation conversion table for digit characters and classifiers, find the pronunciation conversion type that corresponds to the number and the pattern code obtained in (e), and convert the pronunciation of the digit character or classifier according to that type. demand.

(h) Perform the processes in (e), (f), and (g) for all the pronunciation units obtained in (d).

[Embodiments of the invention]

Hereinafter, the present invention will be explained in detail with reference to the drawings. The hardware configuration of one embodiment of the present invention is described in the first example.
As shown in the figure. In FIG. 1, 1 is a processor, 2 is a number word data storage memory such as a magnetic tape or magnetic disk, 3 is an output data storage memory for storing number words and assigned pronunciations, 4 is a program storage memory, and 5 is a work memory. area, 6 is Kanji numeral substitution table memory, 7 is digit character table memory, 8
is the insertion digit character table memory, 9 is the digit character substitution table memory, 10 is the conversion type table memory, 11 is the reading kana table memory for numbers, and 12 is the reading kana conversion table memory for digit characters and classifiers. be.

The Chinese numeral substitution table memory 6 stores a set of Chinese numerals and their corresponding arithmetic numerals as shown in FIG.

The digit character table memory 7 stores a digit character and a numerical value indicating what power of 10 it means, as shown in FIG. 3, as a set.

Here, the digit character refers to the representation of the number of digits using Kanji characters such as tens, hundreds, and thousands.

In FIG. 3, the digit character "100" indicates 10 squared.

The insertion digit character table memory 8 stores each digit of a numerical value and its corresponding digit character as a set, as shown in FIG. For convenience of processing, "point" and "zero" are temporarily associated with the first digit and the digits below the decimal point.

The digit character substitution table memory 9 stores combinations of consecutive 0s and digit characters as shown in FIG.
A replaceable one-character digit character is stored as a set.

The conversion type table memory 10 includes the sixth
As shown in the figure, digit characters and number words (Here, we refer to character strings other than digit characters that follow numbers. Examples include 年, ashi, hou, tsubo, meter, tsu, and rai.) , its reading is standard form, voiced sound form, semi-voiced sound form, code indicating the reading kana pattern of the number immediately before the digit character or classifier, and the reading kana of the digit character or classifier changes depending on the prefix number. A set of codes indicating a pattern to be used is stored as a set. For example, the number word ``hira'' can be read in the standard form ``hiki'', in the voiced form ``biki'', in the semi-voiced form ``piki'', and as kana patterns for reading numbers, 1 (itsu), 2 (ni), 3 ( San), 4 (Yon), 5 (Go),
6
(lots), 7 (nana), 8 (hatsu), 9 (kiyu), 10 (jiyutsu), 100 (hiyatsu), 1000 (sen), 10,000 (man),
The pattern/code "c" is pronounced as 100 million (oku), etc., and the pronunciation kana conversion patterns for digit characters/counters are 1 (piki), 2 (hiki), 3 (biki), 4 ( Hiki), 5 (Hiki), 6 (Piki), 7 (Hiki), 8 (Piki), 9 (Hiki),
Conversion patterns "b" that are converted into kana characters are stored, such as 10 animals (piki), 100 animals (piki), 1000 animals (biki), 10,000 animals (biki), 100 million animals (hiki), etc.

In the number reading kana table memory 11, as shown in FIG. 7, the reading of each number is stored by pattern. Here, a blank column indicates that no reading is given to the number. For example, no pronunciation is given to the number 1 in the numeral data "100".

As shown in FIG. 8, the reading kana conversion table memory 12 for digit characters and classifiers contains conversion information that determines which reading kana type, standard form, voiced sound form, or semi-voiced sound form, should be given to each number. Stored by pattern. The symbol in the table indicates that the reading kana conversion type is the standard form. Similarly, indicates that it is a dakuon form, and indicates that it is a semi-dakuon form.

For example, the conversion pattern "b" indicates that the number 1 has a semi-voiced form, and below,
2 (standard form), 3 (voiced form), 4 (standard form), 5 (standard form), 6 (semi-voiced form), 7 (standard form), 8 (semi-voiced form)
,
They are expressed as 9 (standard form), 10 (semi-voiced form), 100 (semi-voiced form), 1,000 (voiced form), 10,000 (voiced form), 000 million (standard form), etc.

The program has a modular structure as shown in FIG.

An overview of the pronunciation kana assignment process for number words is shown below.

First, each table used when assigning pronunciation kana is loaded onto the processor main memory 5, and the numeral data to be assigned pronunciation is taken out from the numeral data storage memory 2. If the numeral data is expressed in Chinese numerals, a subroutine for replacing kanji numerals is activated to replace the kanji numerals with arithmetic numerals, and if an exceptional reading kana assignment is required, an exceptional reading kana assignment subroutine is started. Start up and read and give a tentative name. In addition, number word data includes, for example, 1 million yen, arithmetic numerals (hereinafter simply referred to as numbers) and kanji numerals (hereinafter kanji numerals are referred to as digit characters). If the numbers are consecutive, the number of digits correction subroutine is activated to unify the numbers into arithmetic numbers.

The pronunciation is <1 character number><digitcharacter>,<
1-character number><counterword>, <1-character number><digit character><counterword> (If only 1-character number>,
Consider that there is a blank particle. ), a digit character insertion subroutine is activated to insert a digit character between arithmetic digits. In the pronunciation kana assignment subroutine, the pronunciation assigned to the number word is output to the output data storage memory 3 together with the number word.

Next, the procedure for assigning pronunciation kana to number words will be explained based on a flowchart, and the pronunciation kana assignment process for "26659.3 tsubo" will be exemplified as specific number word data.

First, the main routine for assigning pronunciations to kana numbers will be explained based on the flowchart of FIG.

When this main routine is activated, it loads a Chinese numeral substitution table from a magnetic tape, magnetic disk, etc. onto the processor (100). Similarly, the digit character table (101), the inserted digit character table (102), the digit character substitution table (103), the conversion type table (104), the reading kana table for numbers (105), the reading kana for digit characters and classifiers. Load the conversion table (106) onto the processor. Next, the number word data to be assigned a pronunciation kana is taken out from the number word data storage memory 2 and stored in buffer A as illustrated in FIG. 16 (107). if,
If the target data is not in the numeral data storage memory 2, the processing of this main routine is terminated.

It is determined whether the number of the imported numeral is expressed in Chinese numerals (108), and if it is satisfied, the character string in buffer A is stored in buffer B (117), the Kanji digit replacement subroutine is activated, and the character string is expressed in Chinese numerals. After replacing the number with an arithmetic numeral (118), execute the process from 115 onwards, and if not satisfied, execute the process from 109 onwards. In step 109, the arithmetic numeral portion of the number word data is stored in buffer B as illustrated in FIG. 17, and the remaining portion is stored in buffer C as illustrated in FIG. 18 (109). It is determined whether the numeral data of Batsuhua A is exceptional reading data, such as the 1st day (tsukitachi), 2nd day (futsuka), ... 10th day (touka) of the date (110), and if it is satisfied, Activate the exception pronunciation kana assignment subroutine to assign pronunciation kana, and store it in buffer F as illustrated in FIG. 21 (118).
Execute the process from 116 onwards, and if not satisfied, execute the process from 111 onwards. In 111, it is determined whether the first character of Batsuhua C is a digit character (111), and if it is satisfied,
Activates the digit number correction subroutine, changes digit characters to numerical values (112), executes the processes from 113 onwards, and if not satisfied, executes the processes from 113 onwards.

113 starts the digit character insertion subroutine and
A digit character is inserted between the arithmetic numbers in buffer B and stored in buffer D as illustrated in FIG. 19 (113). The character strings in buffers D and C are combined and stored in buffer E as illustrated in FIG. 20 (114). Activate the reading kana assignment subroutine, add the reading kana to the character string Batsuhua E,
Store in buffer F (115). Output data storage memory 3 includes buffer A and buffer F as one set.
Output to (116). After that, repeat the process from 107 onwards.

For example, when the number word data "26659.3 tsubo" is extracted in 107, it is stored in buffer A as shown in Figure 16, and since the number word data is not expressed in Chinese numerals, the arithmetic number part "26659.3" is As shown in Figure 17, the remaining part "tsubo" is
is stored in buffer C as shown in FIG. The numeral data is not exceptional reading data, and the first character of Batsuhua C is not a digit character, so it is 113.
starts the digit character insertion subroutine, and digit characters are inserted between the arithmetic digits, as shown in Figure 19.
It is stored in buffer D. The character strings in buffers D and C are combined and stored in buffer E as shown in FIG. It is stored in the buffer F as follows.

Buffer A and buffer F are output as one set to the output data storage memory 3, and the process ends.

Next, regarding the Chinese numeral replacement subroutine, the first
This will be explained based on the flowchart shown in FIG.

When this subroutine is started, the buffer B
Determine whether the first character is a digit character (201), and if satisfied, add the number 1 to the beginning of the character string in buffer B, store it in buffer B (202), and execute the processes from 203 onwards. However, if not satisfied, the process from 203 onwards is executed. In 203, it is determined whether digit characters exist consecutively in the character string in buffer B (203), and if it is satisfied, the number 1 is inserted between those digit characters, and the number 1 is added to buffer B. After storing (204),
Repeat the process from 203 onwards, and if not satisfied, execute the process from 205 onwards. At step 205, it is determined whether or not there is a kanji numeral registered in the left column of the kanji numeral replacement table shown in Figure 2 in the character string in buffer B (205), and if it is satisfied, the kanji numeral is After replacing with the arithmetic numerals registered in the right column of the Chinese numeral replacement table (206), repeat the process from 205 onwards, and if the results are not satisfied, execute the processes from 207 onwards. In 207, after storing the character string of buffer B in buffer E (207),
Terminates the processing of this subroutine.

For example, in the case of numeral data "1,112 yen,""1,112yen" is stored in buffer B because the beginning of buffer B is a digit character. Then 203
Since the digit characters 1,000 and 100 are consecutive, 1 is inserted between them, and "1,1120 yen" is stored in buffer B. At 205, since 2 in the number word data is registered in the Chinese numeral substitution table, it is replaced with the numeral 2, and at 207, the number 2 is replaced with the numeral 2.
"1,1120 yen" is stored in , and the process ends.

Next, regarding the digit number correction subroutine, the 12th
This will be explained based on the flowchart shown in the figure.

When this subroutine is started, the buffer B
In the numerical value, there are as many 0s as there are digits in Batsuhua C.
In order to add , the number of 0s to be added is stored in the parameter NUM by referring to the correspondence table of digit characters and the number of 0s to be added as shown in FIG. 3 (301).

Determine whether there is a decimal point in the value of buffer B (302), and if satisfied, execute the processes from 303 onwards,
If not satisfied, execute processing from 308 onwards. 303
Now, the number of digits below the decimal point of the numerical value in buffer B is determined and stored in parameter NN (303). Determine whether the number of 0s to be added is smaller than the number of digits after the decimal point (304), and if satisfied, move the decimal point of the number in buffer B to the right by the number of digits obtained by subtracting NN from NUM. (306) Execute steps 309 onwards, and if not satisfied, execute steps 305 onwards. 305 removes the decimal point of the number in buffer B (305) and converts it to NUM
The value obtained by subtracting NN from is stored in NUM (307).

Next, the number of zeros equal to NUM is added after the numerical value in buffer B (308), and the digit characters in buffer C are deleted (309). Thereafter, the processing of this subroutine ends.

For example, in the case of the numerical data "105,000 yen", refer to the digit character table and store the number 4 in the parameter NUM as the number of 0s to be added to the numerical value "10.5", corresponding to the digit character "10,000". do. Batsuhua B
Since there is a decimal point within the number "10.5", calculate the number of digits after the decimal point in 303 and store the number 1 in the parameter NN. Since the number of 0s to be added is not smaller than the number of digits to the right of the decimal point, the decimal point is deleted from the numerical value in buffer B. Then from NUM to NN
After subtracting the value 3 and storing it in NUM, add three digits 0 after the number in buffer B, delete the digit character "10,000" in buffer C, and end the process.

Next, regarding the digit character insertion subroutine, the first
This will be explained based on the flowchart shown in FIG.

When this subroutine is started, the buffer B
Find the number of digits in the integer part of the numerical value in (401), and (comma) and . (period, decimal point) is removed (402). Next, between the numerical values in buffer B, digit characters corresponding to each digit of the numerical value as shown in FIG. 4 are inserted and stored in buffer D (403). In the character string in Batsuhua D, 0 ten, 0 hundred,
If there is a character string 0,000, it is deleted (404). Furthermore, in the character string in buffer D, the fifth
If there is a character string as shown in the left column of the figure, it is replaced with the characters in the right column (405). Determine whether the last character of the character string in buffer D is a "point" or "zero" (406), and if satisfied, delete the last character in buffer D (407), The processing of this subroutine is ended, and if not satisfied, the processing of this subroutine is ended.

For example, in the case of the number word data "26659.3 tsubo", find that the number of digits in the integer part of the number "26659.3" in buffer B is 5, remove the decimal point in buffer B, and store "266593" in buffer B. do. next,
Add a digit character like the one shown in Figure 4 after each number and write "26,6159 points 3 zero" as a buffer D.
Store in. During Batsuhua D, 0 ten, 0 hundred, 0
Since there is no character string ``1000'' or a character string like the one shown in the left column of FIG. 5, it is determined whether the last character of the character string in buffer D is a "dot" or a "zero".
In this example, since the last character is "zero", that character is deleted, "26,6159 points 3" is stored in buffer D, and the processing of this subroutine ends.

Next, a subroutine for assigning reading kana to numbers, digit characters, and number words will be explained based on the flowchart of FIG. 14.

When this subroutine is started, the buffer E
It is determined whether the character composition of the first character string is <number><digitcharacter><counterword> (501), and if it is satisfied, the process from 506 onward is executed; if not,
Execute processing after 502. At step 502, the reading kana for digit characters or classifiers, their conversion patterns, and the reading kana patterns for numbers or digit characters are checked from the conversion type table as shown in FIG. The pronunciation table for such numbers is referred to to find the pronunciation (503).
Next, refer to the reading kana conversion table for digit characters and classifiers as shown in Figure 8, and find the reading kana that corresponds to the conversion type of the reading kana for the digit character or classifier corresponding to the immediately preceding number or digit character. Select (504), and store the pronunciations of the numbers or digit characters in the buffer F (505).

The character string in buffer E without pronunciation is moved to the beginning of buffer E (512). It is determined whether the first character string of the character strings in buffer E is a number (513). If satisfied, the process from 501 onward is repeated; if not satisfied, the process of this subroutine is ended.

506 finds the reading kana pattern of the number immediately before the digit character from the conversion type table (506),
The pronunciation of the number is determined by referring to the kana pronunciation table (507).

Next, from the conversion type table, the pronunciation of the particle, its pronunciation conversion pattern, and the pronunciation kana pattern of the number are obtained (508). Refer to the number reading kana table and extract the reading kana of the digit characters (509).

Next, refer to the reading kana conversion table for digit characters and number characters, select the reading kana that corresponds to the reading kana conversion type for the number character according to the immediately preceding digit character (510), and select the reading kana for the number and digit character mentioned above. After storing the information together with the pseudonym in the buffer F (511), the processing from 512 onwards is executed.

For example, in the case of the numeral data "6 animals", the character string in buffer E is "6 animals", and the first character string structure is not <number><digitcharacter><counterpart>, so the conversion type table , the pronunciation kana of the particle ``person'', the standard form ``hiki'', the voiced form ``biki'', the semi-voiced form ``piki'', and the pronunciation conversion pattern ``b''.
and the number reading kana pattern "C".

Refer to the number reading kana table and select pattern C.
Find the pronunciation "rotsu" that corresponds to the number 6 among the kana pronunciations. Next, refer to the reading kana conversion table for digit characters and classifiers, and select the reading kana "Piki" that corresponds to the semi-dakuon form, which is the conversion type corresponding to the number 6, among the reading kana conversion types of pattern b. After storing the 6 animals (Lotupiki), including the number reading kana mentioned above, in Batshua F,
Finish the process.

Next, the exceptional pronunciation kana assignment subroutine will be explained based on the flowchart of FIG. 15.

When this subroutine is started, the buffer C
It is determined whether the first character of is "day" (601), and if it is satisfied, the process from 602 onward is executed; if not, other exception handling is executed (606), and then the process of this subroutine ends. In 602, the pronunciations of the auxiliary digit ``日'' are the standard form ``ka'' and the voiced digit ``tachi''.
and set it as a reading kana pattern for numbers,
"D", as the reading kana conversion type of the particle, "d"
Set. Referring to a number reading kana table as shown in FIG. 7, a reading kana is given to the numerical value of buffer B (603). Next, as shown in Figure 8, the reading kana for the number character is set based on the conversion type of the reading kana conversion table for digit characters and number words (604), and the reading kana for the number is combined with the reading kana for the numbers mentioned above to form a buffer F. Store (605).

For example, in the case of the number word data "3 days", the first character of Batshua C is "day", so based on the number reading kana pattern D, the reading kana "mitsu" is given,
Based on the reading kana conversion pattern d for digit characters and classifiers, the conversion type for the number 3 is the standard form,
After selecting the pronunciation "ka" and storing "Mitsuka" (combined with the numeric pronunciation) in the buffer F, the process ends.

〔Effect of the invention〕

According to the present invention, the following effects can be obtained.

(a) Pronunciations can be automatically assigned to numbers and particle words.

(b) A table that stores reading kana for numbers and digit characters by pattern, and a table that stores reading kana conversion types for digit characters and particle words for each pattern,
By adopting a method of assigning pronunciation based on the table showing which pattern in both tables each classifier and digit character follows, the classifiers and digit characters that are subject to pronunciation can be assigned pronunciations and digits. By registering a code indicating the pattern of the pronunciation kana of , and a code indicating the pattern of the pronunciation conversion type of digit characters and classifiers, it is possible to automatically assign the pronunciation kana.

(c) In the process of assigning pronunciations of number words, by unifying numbers to arithmetic numerals or kanji numerals, it is possible to extract pronunciations from the pronunciation table for numbers, and to convert pronunciations of digit characters and classifiers according to numbers. It simplifies the type retrieval process.

(d) one-digit number and digit letter; one-digit number and particle;
Since the assigned pronunciation is output using either a digit character and a number word, or a one-digit number, a digit character, and a number word, it can be used as a pronunciation unit of a number word in audio output.

[Brief explanation of drawings]

Fig. 1 is a hardware configuration diagram of the numerical reading kana assignment method according to the embodiment of the present invention, Fig. 2 is a diagram showing the correspondence between Chinese numerals and arithmetic numerals, and Fig. 3 is a diagram showing the correspondence between digit characters and arithmetic numerals.
A diagram showing what power of 10 is expressed, Figure 4 is a diagram showing the digit character corresponding to each digit of a numerical value, and Figure 5 is a diagram showing a single digit character that can be replaced with consecutive 0s and digit characters. Figure 6 shows the pronunciations of digit characters and particle words,
Figure 7 shows how to read numbers by type; Figure 8 shows how to read digit characters and classifiers by type; FIG. 9 is a diagram illustrating the software module configuration of the embodiment of the present invention, FIGS. 10A and B are diagrams illustrating the processing procedure of the main module for adding pronunciations of number words, and FIG. Figure 12 is a diagram showing the processing procedure for replacing digits with arithmetic numerals, when the numerical data subject to pronunciation is composed of numerals and Chinese numerals. Figure 13 is a diagram showing the digit number correction processing procedure, and Figure 14 is a diagram showing the digit character insertion procedure for inserting a digit number character between arithmetic numbers.
The figure shows a processing procedure for assigning pronunciations to number words, Fig. 15 shows an example of processing when it is necessary to assign pronunciations to numerals, and Fig. 16 shows an example of the configuration of buffer A. FIG. 17 is a diagram illustrating the configuration of buffer B, and FIG. 18 is a diagram illustrating the configuration of buffer C.
19 is a diagram illustrating the configuration of buffer D, FIG. 20 is a diagram illustrating the configuration of buffer E, and FIG. 21 is a diagram illustrating the configuration of buffer F. .

Claims (1)

[Claims] 1. An input means for inputting numeral data consisting of a combination of a number or digit character followed by a digit character or a number word, and a first storage means that stores patterns of pronunciations of the digit or digit character. , a second storage means that stores the pronunciation kana of the digit character or particle word following the number or digit character for each pattern; and a third storage means that stores the type of pronunciation of the digit character or particle word for the number or digit character. , means for combining the pronunciations of the first and second storage means based on the third storage means and assigning the pronunciations to the input numeral data; and means for outputting the assigned pronunciations. A device for assigning kana pronunciations to number words. 2. In the number word pronunciation kana assigning device as set forth in claim 1, means for detecting that the input numbers and digit characters are Chinese numerals and that there is no number from 1 to 9 immediately before the digit characters. and means for inserting 1 in front of the digit character according to the detection result of the detection means. 3. In the number word pronunciation kana assigning device according to claim 1, when a number and a number word following it are input, it is detected that the input number is an arithmetic numeral and does not include a digit character. and means for inserting a digit character indicating the number of digits of the input number immediately after each character of the input number according to the detection result. 4. In the device for assigning pronunciation kana for number words as set forth in claim 1, the output means may output the assigned pronunciation as one digit number and digit character, one digit number and digit character, one digit number and digit character, and one digit number. A kana pronunciation device for number words, characterized in that it outputs one set of a digit character and a particle, or a digit character and a particle.