JPH0646423B2 - Word dictionary matching device - Google Patents

Description

TECHNICAL FIELD The present invention relates to a word dictionary matching device that recognizes a read character string by matching the input read character string with registered contents in a word dictionary.

[Conventional technology]

FIG. 3 is a block diagram showing a conventional example of this type of word matching device.

The input register 31 inputs and stores a character string read from a character reader (not shown), and further stores a word length table pointer.
Transfer to 32. The word length table pointer 32 classifies the read character string according to the word length (the number of characters) of the word string contained therein, and further converts this into a code suitable for the word dictionary. The word dictionary reading unit 33 inputs the code, issues a read command to the word dictionary 34 using this code, reads the character string corresponding to the code from the word dictionary 34, and stores it in the input register 31. The read character string is output in alignment with the read character string. The word collating unit 35 collates whether both character strings input from the word dictionary reading unit 33 match.

In this way, when the two character strings match, the word matching unit 35 outputs the code as the code of the read character string.

[Problems to be solved by the invention]

In the above-mentioned conventional word collating device, the number of bits per character increases as the number of characters forming a word increases, which has a drawback of increasing the capacity of the word dictionary and the collation time.

[Means for solving problems]

The word dictionary matching device of the present invention outputs the read character or a code representative thereof when the read character does not include a character having a predetermined appearance frequency that is low in frequency, when the read character string is confirmed to match. , If the read character includes a character with a predetermined occurrence frequency that is low in frequency, the dictionary is searched in the same way as a character string obtained by converting the character into a predetermined single code, and the matching of the character string is confirmed. When the first collating means for outputting pointer information stored in the character string in the dictionary, the pointer information input from the first collating means, and the read character string input from the character reader When it is confirmed that the read character string matches by searching and collating a second dictionary that stores a character string including a character with a low appearance frequency in advance from the character with a low appearance frequency indicated by the pointer information This read character or this The second to output the code that representative
It has a collating means.

In this way, since the read character string composed only of the frequently appearing characters is searched and collated with the small capacity dictionary composed only of the frequently appearing character, the dictionary having a large capacity like the conventional example is searched. The processing time is shorter than that in the case of collation, and the processing time can be totally saved even if the collation is performed twice when the read character string includes a character having a low appearance frequency.

〔Example〕

 Embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of a word dictionary matching device of the present invention, and FIG. 2 is a diagram showing an example of an address matched in this embodiment.

In this embodiment, the address of the mail is checked by checking it against an address dictionary created in advance, and the character reader 3 reads the target address portion of the input character string and performs the first reading. It is output to the register 4 as a character string. The register 4 outputs the input first read character string to the first collating means 1 and the second collating means 2, respectively. The first collating means 1 is composed of an index table pointer 11, a dictionary reading unit 12, a dictionary unit 13 and a collating unit 14, collates and outputs a character string (address) consisting only of frequently appearing characters, and if the character string When a character with a low appearance frequency is included, the matching is performed by the second matching means. The index table pointer 11 replaces the character of the first read character string input from the register 4 with a predetermined single code if the predetermined occurrence frequency is low. If not, it is output as the second read character string in any case.
The dictionary reading unit 12 inputs the second read character string, links the words contained therein to form a few word strings, and classifies them according to their respective word lengths (total number of characters). The dictionary unit 13 stores a large number of address name character strings in advance, and the dictionary reading unit
A character string corresponding to the word string input from 12 is output as a third read character string, and the dictionary reading unit 12 outputs the third read character string to the collating unit 14 together with the second read character string. To do. The collating unit 14 collates the second read character string and the third read character string, and when both match and the read character string does not include a single code, this is kept as it is, or A code representative of this is output. If the matching read character string includes a single code, the pointer information stored in the dictionary unit 13 corresponding to the character string including the single code is output from the collating unit 14 to the second collating unit. It is output to the means 2.
The second collating means 2 includes an index table pointer 21 and a dictionary reading unit.
A character string (address) including a character having a low appearance frequency is configured to be a matching target, which is composed of a 22, a dictionary unit 23, and a matching unit 24. The index table pointer 21 receives the first read character string from the register 4 and the pointer information from the first collating means 1, respectively, and corresponds to a single code in the word designated by the pointer value. Is converted into a predetermined code to form a fourth read character string, and the first and fourth read character strings are output to the dictionary reading unit 22. The dictionary unit 23 stores in advance a character string of an address name corresponding to a character having a low appearance frequency, and when the fourth read character string is input from the dictionary reading unit 22, the corresponding character string is changed to a fifth character string.
To the dictionary reading unit 22 as a read character string. Dictionary reading unit 22
Collates the fifth read character string with the first read character string.
Output to 24. The collating unit 24 collates both input read character strings, and outputs the collated read character string or a code representative thereof when the both match. The sorter 5 receives the output from the first collating means or the second collating means, and sorts the mail by the destination address.

Next, the operation of this embodiment will be described in the case of the destination address example shown in FIG.

Now, assuming that a number is specified as a character with a low frequency of occurrence and * is set as a single code, the address name "Ang Mo Kio Industrial Park 5" is read from the character reader 3 via the register 4 Index table pointer 11
Is input to the dictionary reading unit 12 as a second read character string. The dictionary reading unit 12 links the words of the second read character string to generate the following word string, for example.

 Park ＊ Mo Kio Industrial Park ＊ Kio Industrial Park etc.

These word strings are sequentially input to the dictionary unit 13 according to the word length (number of characters). Since the character string of the address name corresponding to the second read character string is stored in advance in the dictionary unit 13 according to the word length, the retrieved third character string is collated by the collation unit 14 via the dictionary reading unit 12. Then, it is checked whether or not it matches the second read character string. Usually, it is enough to search for a few words without one word. In the present case, "Ang Mo Kio Industrial Park
"*" Is searched and it is detected that both character strings match. If the single code * is not included in this character string, it is output as the collation result immediately, but in this case, it is included, so the pointer information stored in this part in the dictionary unit 13 is It is output to the second matching means.

On the other hand, in the second matching means 2, the index table pointer 21
Enter the pointer information and the first read character string,
A portion corresponding to the pointer information in the first read character string, that is, "5" in this case is converted into a predetermined code, and is read as a fourth read character string together with the first read character string in the dictionary reading unit. Output to 22. Dictionary part of the second matching means 2
Since the address name corresponding to this infrequently appearing character is stored in advance in 23, the first read character "5" is used as the first read character.
The fifth read character string is searched for by the same procedure as in the matching means 1 and the matching unit 24 matches whether or not the fifth read character string matches the first read character string. In this case, "Ang Mo Kio Indus
trial Park 5 ”is output as the result of the verification.

In this way, since an address name composed only of frequently appearing characters can be searched by a small capacity dictionary unit 13 composed only of frequently appearing characters, a dictionary unit having a large capacity is used as in the past. The processing time is shorter than that in the case, and the processing time can be saved as a whole even if the character including the character having a low appearance frequency is collated twice.

〔The invention's effect〕

As described above, according to the present invention, the characters of the input read character string are classified into a character having a high appearance frequency and a character having a low appearance frequency in advance, and a character string that does not include a character having a low appearance frequency is collated. A first collating unit for outputting and a second collating unit for collating a character string including the character are provided,
By matching the character strings separately according to the presence or absence of characters that occur infrequently, normally the matching of the read character string is completed only by matching the dictionary of the characters that appear frequently, and the dictionary capacity and matching time are reduced. There is an effect.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of a word dictionary matching device of the present invention, FIG. 2 is a diagram showing an example of an address input to this embodiment, and FIG. 3 is a conventional example of a word dictionary matching device. 3 is a block diagram showing the configuration of FIG. 1 ... First collating means, 2 ... Second collating means, 3 ... Character reader, 4 ... Register, 5 ... Sorting machine, 11, 21 ... Index table pointer, 12, 22 ... Dictionary reading section, 13, 23 ... Dictionary section, 14, 24 ... Collation section.

Claims (1)

[Claims] 1. A word string contained in a read character string read by a character reader is used to search and collate a word dictionary in which a character string corresponding to the character string is stored in advance.
In a word dictionary matching device that confirms whether the read character string matches a character string stored in a dictionary, when the read character does not include a character with a predetermined low appearance frequency, the read character When it is confirmed that the columns match, the read character or a code representative thereof is output, and when the read character includes a character with a predetermined low appearance frequency,
Similarly, the dictionary is searched as a character string obtained by converting the character into a predetermined single code, and when the matching of the character strings is confirmed, pointer information stored in the character string in the dictionary is output. The matching frequency, the pointer information input from the first matching means, and the character having a low frequency of appearance indicated by the pointer information in the read character string input from the character reader, A second dictionary storing a character string including a low character, and when the matching of the read character string is confirmed, the read character or a code representative of the read character is output. A word dictionary matching device characterized by the above.