JPS6049932B2 - Japanese information processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a Japanese language information processing system, and more specifically, to a Japanese language information processing system that handles phonetic character strings in which long sounds are used confusingly with long sounds and A-line characters. .

With the progress of the information society, a wide variety of search systems have appeared.

Particularly recently, as computers have become more popular, search systems are emerging that allow ordinary people to register and search for search information. However, in this case, it does not mean that the general public will operate it directly. A typical example of this system is a computerized telephone directory assistance system.

That is, the information to be searched in this system is registered information from the telephone owner, and the search information is the offer information from the telephone number inquirer. The kana-yomi name is used as key information for these search targets and searches. In this case, the problem lies in the ambiguity of the Japanese phoneticization rules. That is, when ideograms are converted into phonetic characters, multiple different expressions are commonly used for the same name due to old and new kana usage, homophones, etc. One of these is the problem of converting long sounds into phonetic characters. The principles of how to write long sounds are established in 1 modern kanazukaiyo. 1.Modern Kanazukayo is a set of rules for expressing modern words in kana based on modern speech sounds, and is mainly applied to colloquial sentences.

The long sounds in this first modern kana are written in their own kana. However, 0 o. The basic rule is to write only the long sounds in a row in one wo. For example, it is as follows. Mother: Oria-san Brother: Nii-san Evening: Yugata Older sister: Nee-san Father: Oto-san (Quoted from Kadokawa Japanese Dictionary Revised 101st Edition) However, there are cases where the above rules are not followed or the pronunciation and written characters do not match. be.

For example: big; big distance; tooi AA; mei-rei (pronounced as ko-retsu long sound mero-P] koele-e) Oral response: koutou (however, pronounced as ko-retsu long sound ko-o-to-o)
Therefore, in order to convert long sounds into phonetic characters, it is customary to write them using the long sound symbol and to write them using the 1-ayo character.

For example, 1 air is written as 0 kuukyo or 0 kuukyo.

In order to convert the long sounds in the Eyo and 1-O rows into phonetic characters, in addition to using the long sound symbol, two types of 1-O row characters are also used. For example, 1 orderyo becomes 1 mere/meereeyo or 1 meireyo, and
One answer is kotoyo, 1 kootooyo, or 1 koutoyo. In this way, there are many exceptions to how long sounds are written, and some spellings differ from the pronunciation, so it is difficult to write them accurately, so various ways of writing are customarily allowed.

Therefore, in the telephone number guidance system, the kana reading name (registered information) from the phone owner registered as searched information and the kana reading name (offer: information) that is converted into phonetic characters from the offer from the phone number inquirer. ) often do not match.

At this time, if these pieces of information are used as the searched key information and the search key information, the two will not match and a search solution will not be obtained. In order to eliminate this kind of inconvenience, it would be possible to establish a rule that matches the searched information with the searched information, but since registered information such as people's names and business names are related to individual subscribers, it is not mandatory. It is difficult to unify them in a single sense.

Generally, as a countermeasure for such cases, (1) prepare multiple pieces of searched information for the same subscriber corresponding to each type of kana reading of the name;
2) When searching, by creating multiple search keys corresponding to each type of kana reading of the name, matching solutions (
We are working on a method to obtain a search solution.

For example, for subscriber name 1 Sato Cleanyo,
There are six kana readings: 1 Sato Cleaning 2 Sato Cleaning 3 Sato Cleaning 4 Sato Cleaning 5 Sato Cleaning 6 Sato Cleaning, but if you prepare multiple pieces of searched information in (1) above, use 1 to 1 above.
You will need to register searched information with six different kana pronunciations (6), and if you create multiple pieces of search information (2) above, you will need to create six types of search keys (1 to 6). become.

However, in the case of (1) above, the capacity of the file accommodating the searched information increases.

In the case of (2) above, the search operation is troublesome or takes a long time. In other words, the operator has to create and input a large number of search keys, which takes time and effort, or Even if you create a search key and perform a search, the search time will be twice as long as the number of created search keys. That is, the conventional method has drawbacks such as an increase in file capacity and a time-consuming search. The present invention eliminates the above drawbacks and has the above-mentioned long sound, which makes it possible to complete the search in a short time without increasing the capacity of the file containing the searched information or increasing the number of search keys during the search. The purpose of this invention is to provide a Japanese information processing method for executing a search for searched information.

According to the present invention, in a Japanese information processing system that handles phonetic character strings in which long sounds are used in combination with long sound symbols and 1-ayo characters, the long sound symbols are selected from among the phonetic character strings. a means for detecting the long sound symbol, a long sound conversion means for converting the long sound symbol into a one-ayo line character determined by the type of vowel of the character before the long sound symbol, and a means for storing data using the converted character string as key information. and the given search character 5
This is achieved by a Japanese information processing method that includes a device that uses the column as search key information by the conversion means and performs a search using this key information. Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an example in which the present invention is implemented in a telephone directory assistance system. In the figure, 1 is a character string input device for inputting offer information, and 2 is a character string input device for inputting offer information. A long sound correction device corrects long sounds in a character string to produce a certain pattern of key information; 3 is a file storage device that stores a large amount of information to be searched; 4 is a long sound correction device 2
Reference numeral 5 denotes a control device to be searched for searching information to be searched from the file storage device 3 using key information given from the key information given from the key information given from the key information given from the key information given from the key information given from the search target control device 5. In addition, 2 in the long sound correction device 2
1 is a sequence control device, 22 is an input information register, 2
3 is a character classification device, 24 is a front vowel extraction device, 25 is a long vowel conversion device, 26 is a vowel conversion device, 27 is an output information register, 28 is a long vowel conversion table, and 29 is a vowel conversion table.

FIG. 2 shows a part of an example of a file stored in the file storage device 3 of the embodiment shown in FIG. The information to be searched is stored.

The above key information is converted into a kana code based on the subscriber's name (person's name, trade name, etc.) with long sounds corrected according to the standards set by the present invention, and the file is created in the 5-dark and positive order of the above-mentioned kana code. There is.

The key information encoded in kana from the subscriber's name is created according to the following criteria.

(1) The long sound symbol 1-yo is set by converting the vowel type of the preceding character according to the vowel conversion table shown in FIG. 3b.

In Fig. 3b, I indicates the type of vowel before conversion, and ■ indicates the vowel after conversion. long symbol 1 for characters with
If -yo is added, that is, 1 car/keyo is converted to 1 car/keyo. For example, it is assumed that 1 Sankei Equipment has been converted to Sankei Setsubi, 2 Sanko Garden has been converted to Sankoga Aden, and 3 Sanko Shoko has been converted to Sankou Shiyoukou.

(2) If there is a 1 yo and 1 yo following the 1 yo string and 1 yo string, the 1 yo and 0 yo are converted and set as 1 yo and 1 yo, respectively.

For example, Sanko Denki becomes Sanko Denki. The key information to be searched in the file shown in FIG. 2 contains a kana code created in accordance with the above criteria.

Next, the operation of the embodiment shown in FIG. 1 will be explained.

Now, assume that the inquirer verbally (over the telephone) conveys the inquiry information 1, ``Sankosho'', to the directory assistance operator. The operator listens to the above-mentioned inquiry information, converts it into phonetic characters, and inputs it into the input device 1, not necessarily according to rules but according to customary practice.

Then, the input device 1 notifies the long sound correction device 2 that there has been an input, and the input information is transferred and stored in the input information register 22. Each time the register 22 is activated by the sequence control device 21, the register 22 sequentially registers 1 from the beginning of the input information.
Each character is extracted and sent to the character classification device 23.

In this case, first, the first character is extracted and passed to the character classification device 23. If the extracted character is the first character, the character classification device 23 passes the extracted character as is to the output information register 27 for storage.

In this case, 5 days are accumulated in the output information register 27. In the case of the second and subsequent characters, the characters extracted by the character classification device 23 are
- Classify as either 3 or 3.

The second and third characters are 1-cho 1-ko, which is the case in 3 above.

In this case, the same process as for the first character is repeated,
The output information register 27 stores the received characters as 1 in order from the beginning.
It accumulates letters one by one, and one sancho is accumulated here. Note that the fourth character is 1 o, which is the case of 1 above.

In case 1 above, the extracted character is 1 yo or 1 yo, and the character is passed to the front vowel extraction device 24. Then, the front vowel extraction device 24 extracts the front character of the received character from the input information register 22, and converts the extracted front character (in this case, the third character 1 J) into a known long vowel conversion table 28. Perform character conversion using this method. That is, the long sound conversion table 28 includes the conversion table shown in FIG. is extracted and returned to the front vowel extraction device 24. In this case, if the character - received by the front vowel extraction device 24 and the character converted by the long sound conversion table 28 match, that is, if they are both 1 eo or 1 yo, the character is transferred to the vowel conversion device. 26
If they do not match, the character 1 yo or 1 yo received by the front vowel extraction device 24 is passed to the output information register 27 . Therefore, in this case, 5 yo is passed to the vowel conversion device 26. For example, if it is assumed that 5 ko yo, which is not a long sound, is extracted, the 0 yo received by the front vowel extraction device 24 and the 1 oh of the output converted by the long sound conversion table 28 do not match, so the former character j character 1 Eo is passed as is to the output information register 27 and stored, and no long sound conversion is performed. The vowel conversion device 26 converts the received character 1e or 1
The vowel conversion table 29 is used to perform character conversion using a known method.

That is, the vowel conversion table 29 includes the conversion table shown in FIG. It is returned to the vowel conversion device 26. Vowel conversion device 26
passes this returned character 1yo or 1yo to the output information register 27. In this case, since the vowel conversion device 26 receives 1 oyo, it passes the 1 oyo to the output information register 27, and the output information register 27 accumulates 1 oyo. In the above process, if the character extracted from the input information register 22 is ``Yo'' and the vowel of the previous character is 1 ``Eo'', or the input information register 22
If the character extracted from is 1 oyo and the previous character has a strong vowel, the above process is repeated.

Also, if the character extracted from the input information register 22 is 1 yo and the vowel of the previous character is not 1 yo, or the character extracted from the input information/information register 22 is 1 yo and the previous character Even if the number of vowels is not 1yo, the process described above for such a case is repeated. Assume that the above-described processing progresses and that "0" is accumulated in the output information register 27. Next, the long sound symbol 1-yo is input to the character classification device 22 from the input information register 22.
passed to. This is case 2 above, which is classified by the character classification device 23 and the long vowel symbol 1-yo is passed to the long vowel conversion device 25. The long vowel conversion device 25 extracts the first character of the received long vowel symbol 1-yo from the input information register 22, sends the extracted character to the long vowel conversion table 28, and converts it to the conversion table [third] included in this long vowel conversion table 28. According to figure a],
Character conversion similar to step 2 performed by the front vowel extraction device 24 is performed.

If this converted character is ゝ 7 yo 1 yo,
This is passed to the output information register 27, and in the case of 1 yo 1 yo, this converted character is passed to the front vowel extractor 24. In this case, the long sound conversion device 25 converts the first character of the long sound symbol 1-jo extracted from the input information register 22 into the input side 1 from the conversion table set as shown in FIG.
(7) rco. The converted character of 0 on the conversion side (■) corresponding to is obtained and passed to the front vowel extraction device 24.

The front vowel extraction device 24 operates in exactly the same manner as in case 2 above, and sends the first character of the passed 1 o to the long vowel conversion table 28 again to obtain 1 o as a converted character, and extracts it from the long vowel converter 25. It is compared with the already sent character 1o, and since it matches, it is sent to the vowel conversion device 26, and in the same operation as already explained, it is converted into 1 by the vowel conversion table 29.
The output information is converted into a file and passed to the output information register 27.

In this case, this means that 1 count has been accumulated in the output information register 27.

When all the information stored in the input information register 22 is extracted in the above, the sequence control device 21 notifies the search control device 4 of the contents stored in the output information register 27.

The search control device 4 uses a public method to access the file storage device 3,
The notified information σ) is searched as search key information, and searched information that matches this is extracted.

In this case, search key information that matches this is obtained from the file storage device 3 set as shown in Fig. 2 using the search key information 1. Shinjuku ward. 364-24
68... and Sanko Shoko, Nakano Ward. 381-1357・
- Extract. The output device 5 outputs the searched information extracted by the search control device 4.

As a result, it was possible to search for the 1st Sankoshou that was requested by the inquirer. In the above search operation, the directory assistance operator gave an example of converting the information provided by the inquirer into phonetic characters, but after listening to the information from the inquirer, the pronunciation was the same. If you input one of the different phonetic character strings, such as 1 sanko shi yo ko cho J san ko shi yo ko yo... etc.,
As is easily clear from the above explanation, the long sound is uniquely corrected to the above-mentioned 1.
Similar to the above, the same searched information is searched.

In addition, in the above example, 0 engineering. 1 yo and 1 yo following characters in column and 0 o, column were converted to 1 yo and 1 yo, respectively, but conversely, 1 yo and 1 yo following characters in 1 yo column and 1 yo column, respectively, were converted to 1 yo and 1 yo, respectively. The same effect is obtained even if it is converted to 1 o and 1 o.

In addition, in this embodiment, 1 yo and 1 yo following the characters in the 1 yo column and J yo column, respectively, were converted to 7 yo and 1 yo, respectively, but when inputting information manually, 1 1 eyō following an eyō character is treated as 1 yo, and 0 ō following a 1 yo character is treated as 1 yo.For example, if 1 eyō is set as 1 eyō, and 1 ōyo is 0 ōyo, the long sound correction is corrected. In the device 2, sufficient effects can be obtained by only converting the long sound symbol without performing the above-mentioned conversion of 1 yo → 1 yo →, r yo → 1 yo.

In the above embodiment, the creation of search key information has been described, but it goes without saying that the present invention can also be applied to the creation of searched key information.

As explained above, according to the present invention, when listening information or ideographic information is converted into phonetic characters (kana characters),
In a Japanese information processing system that handles phonetic character strings in which long sounds are mixed with long sounds and A-line characters, it is easy to This has the effect of reliably correcting ambiguity in phonetic transcription of long sounds.

For example, when the present invention is applied to a telephone directory assistance system, when converting offer information from an inquirer into phonograms as input information, one key is required regardless of whether the long sound is expressed as a long sound symbol or as a 5-line character. By creating information and performing one search operation, the same search key information can be obtained, which has the effect of making it possible to obtain correct search results.

[Brief explanation of the drawing]

[Figure 1 is a block diagram of an embodiment of the present invention, Figure 2 is a diagram showing part of the contents of a file storage device used in the present invention, and Figure 3a is a long note conversion of the embodiment of Figure 1. Table 28
b is an explanatory diagram of the contents of the conversions included in the vowel conversion table 29. DESCRIPTION OF SYMBOLS 1... Input device, 2... Long sound correction device, 3... File storage device, 4... Search control device, 5... Output device, 21... Sequence control device, 22... Input information register, 23...Character classification device, 24...Front vowel extraction device) device, 25...Long vowel conversion device, 26...
・Vowel converter, 27... Output information register, 28・
...Long sound conversion table, 29...Vowel conversion table, K...Searched key information section, D...Searched information section.

Claims (1)

[Claims] 1. In a Japanese information processing system that handles phonetic character strings in which long sounds are used in combination with long sound symbols and "a" line characters, long sound symbols are detected from phonetic character strings. means, a long sound conversion means for converting the long sound symbol into an "A" line character predetermined according to the vowel type of the character before the long sound symbol, and means for storing data using the converted character string as key information. and a device for converting a given search character string into search key information using the conversion means and performing a data search using this key information. 2. A means for detecting a long sound symbol from a phonetic character string in a Japanese information processing system that handles a phonetic character string in which a long sound is used confusingly with a long sound symbol and an "A" line character, and a means for detecting the long sound symbol from a phonetic character string. a long sound conversion means for converting the character into a predetermined "a" line character according to the vowel type of the character before the long sound symbol, and further the character "E" and the character "E" following the character in the "E" row in the phonetic character string Means for detecting the characters "o" and "u" following the characters in the "i" and "o" strings, vowel conversion means for converting the detected characters into predetermined vowel characters, respectively; and a device for storing data using a given character string as key information, and a device for using a given search character string as search key information by the converting means and performing a data search using this key information. Word information processing method. 3 The vowel converting means converts the character "e" following the character of the "e" string detected from the phonetic character string into "i" and "e".
3. The Japanese language information processing method according to claim 2, wherein the character "o" following a character in the "o" string is converted to "u". 4. The vowel converting means converts the character "i" following the "e" string detected from the phonetic character string into "e" and the character "u" following the "o" string into "e". 3. The Japanese language information processing method according to claim 2, wherein the Japanese language information processing method is converted into "E".