JPH0612406A - Kanji conversion processing system for kana address notation and kana corporation name notation - Google Patents

Abstract

PURPOSE:To execute a KANJI (Chinese character) conversion with respect to a c/o name and a corporation name entered in a full name entry area, with regard to the system for reading optically a KANA (Japanese syllabary) address notation and a KANA corporation name notation, and executing its KANJI conversion. CONSTITUTION:The system is provided with a character reading part 3 for reading and recognizing optically a KANA character array of a KANA address notation and KANA c/o name notation entered into an input business form 1, an address KANJI dictionary 10 for storing the corresponding KANJI notation with regard to a KANA notation of an element for showing address on an administrative district such as a place name and an address, etc., a c/o name KANJI dictionary 11 for storing the corresponding KANJI notation with regard to the KANA notation of a c/o name being an auxiliary address display of a building, etc., and an address KANJI converting part 5 for generating the KANA character array from the KANA notation read by the character reading part 3, and converting it to a KANJI address and a KANJI c/o name by referring to the address KANJI dictionary 10 and the c/o name KANJI dictionary 11. Also, this system is constituted to that the KANA address notation and the KANA c/o name notation entered into the input business form 1 are converted to the KANJIs, respectively and outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for optically reading an address (kana address notation) and a corporation name (kana corporation name notation) written in kana characters on an input form and converting kanji characters. Conventionally, Kanji conversion of Kana address notation in the OCR reading system can only convert notation on the administrative division such as chome and address in the address, and Kanji conversion of person name such as building name which is an auxiliary description of address. I couldn't.

Further, in the conversion of kana corporation name into kanji, the entry area of the corporation name in kana notation in the input form is
It was provided separately from the name entry area. In the corporate name conversion process, Kanji characters are converted based on the corporate name dictionary registered by the user.

For this reason, the kanji conversion process for the kana address notation must be corrected on the correction screen on which the conversion result is displayed after the kanji conversion of the address.
It was inefficient. In addition, the entry of the corporation name was troublesome because it was necessary to enter the corporation name in the input area separately from the name on the input form.

According to the present invention, in the OCR system, kana address notation is also converted into kanji for a name,
Regarding kana corporate name notation, the purpose is to convert the corporate name entered in the name entry area of the input form into kanji in common with the name.

[0005]

2. Description of the Related Art FIG. 11 shows a kanji conversion system for kana address notation in a conventional OCR system. A description will be given according to the numbers shown. S1 Read the data written on the input form.

S2 Of the read data, the data written in the address field is interpreted. S3 Using the recognized kana character data and candidate characters up to the first level (prefecture name), second level (city group name), third level (town name), and fourth level (chome, street address) in address notation Then, the collation process is performed by referring to the dictionary.

S7 Kana-Kanji conversion processing is performed. S5: A combination of kana data and kanji data is generated. S6 A frame indicating the Kanji display area of the address is displayed on the correction screen (frame WRITE).

S7 The converted Kanji data is displayed on the correction screen (Kanji data WRITE). S8 Write alphanumeric characters and kana (ANK) on the correction screen (WRITE of ANK data).

S9 The name data is corrected on the correction screen. S10 Output the address data converted into Chinese characters to the user. FIG. 12 shows a kanji name conversion process and a kana corporation name notation conversion process in the conventional OCR system.

In the figure, S1 to S9 and S12 to S1
The series of processing of 3 is processing of name, S1, S10-S
11, a series of processing of S5 to S13 is processing of a corporate name. It will be described according to the numbers shown. S1 Read the data entered in the input form.

S2 Interpret the data entered in the name field. S3 It is determined how many words (surname and given name) are written in the name field (separated into surname and given name).
If the number is one, it means that the family name is entered, and the process proceeds to S3 '. Two
If it is an individual, since the family name and first name are entered, the process proceeds to S4.

Since there is only one word in S3'S3, the surname dictionary is accessed. Since there are two words in S4 and S3, dictionary access processing is performed on the surname dictionary and the first name dictionary. S5: It is judged whether or not the character string exists in the dictionary. If there is a character string, proceed to S6. If not, the process proceeds to S12.

S6 Kanji conversion processing and correction of the kana character array are performed. For example, even if the dakuten "" in the kana character array "Yamada" cannot be read and is recognized as "Yamata?", It is converted to "Yamada" in the collation. Then, the array of kana characters is corrected to "Yamada".

S7: A combination of kana data representing the reading of the name and the converted kanji data is generated. S8 Display a frame indicating the display area of the name on the correction screen (frame WRITE) S9 Display kanji data on the correction screen (WRITE of kanji data).

S12 Kana characters that are not required to be converted or do not exist in the dictionary are displayed on the correction screen (WRITE of ANK data). S13 Display the alphanumeric characters and kana (ANK) that do not require conversion on the correction screen. If the dictionary does not exist in S5, Kana is displayed on the correction screen (WR of ANK data).
ITE).

Next, the processing of the corporate name will be described. The corporate name is entered in the corporate name entry area provided separately from the name on the input form. S1 Read the data entered in the input form.

S10 Interpret the data of the corporation name field. S11 Access processing of the corporate name dictionary is performed. The corporate name dictionary is created by the user registering the corporate name as one word. After that, the processes of S5 to S9 and S12 to S13 are the same as those of the name process, and therefore the description thereof will be omitted.

[0018]

[Problems to be Solved by the Invention] In the conventional KCR conversion process of Kana address notation in the OCR system, only the address and the address notation on the administrative division can be converted to Kanji. It was corrected on the screen and converted into kanji. Therefore, the conversion process of addresses into Kanji was inefficient. Further, since the kana notation corporation name and the kana notation full name kanji conversion processing were processed separately, it was necessary to separately provide the name entry area and the corporation name entry area of the input form. Therefore, it was troublesome to fill in the input form of the name or the corporation name.

An object of the present invention is to improve a kanji conversion system for kana address notation and kana corporation name notation in an OCR system.

[0020]

The present invention provides a kanji conversion dictionary for kana address notation, in addition to a conversion dictionary for address notation on administrative divisions, a kana notation for kana and a kanji notation for kanji. ∙ As for the name in the address notation, kanji conversion is performed in common with the kanji conversion in the address notation.

Regarding the kana corporation name notation, in addition to the surname conversion dictionary and the name conversion dictionary in which the kana notation of the surname and given name in the name are compared with the kanji notation, the kana of the common name such as "stock corporation" in the corporation name is also used. By providing a general name conversion dictionary that contrasts the notation with the kanji notation, and referring to the kana notation of the name or corporate name in the name entry area, the surname conversion dictionary, name conversion dictionary, and general name conversion dictionary The kana corporation name notation written in the area is also converted to kanji.

FIG. 1 shows a basic configuration (1) of the present invention, which is a system for optically reading a kana address notation and converting it into kanji. In the figure, reference numeral 1 denotes an input form, which includes an address field in which an address is written in kana. Reference numeral 2 is an address field, which is an entry area for writing a kana address. Three
Is a character reading unit, which optically reads the kana address notation written in the address field to recognize kana.

Reference numeral 5 denotes an address / kanji conversion unit, which refers to a dictionary of kana address notation read by the character reading unit 3,
It is converted into kanji. A kana character array portion 6 stores kana address expressions and recognition candidates read and recognized by the character reading portion 3. Reference numeral 7 is a portion for storing a kana character array of kana address notation of an address on the administrative division in the kana character array part. Reference numeral 8 is a part for storing a kana character array of kana notation of a name in the kana character array part. Reference numeral 9 denotes a dictionary reference unit that refers to the dictionary for the Kana notation stored in the Kana character array unit 6 and searches for the Kanji to be converted.

Reference numeral 10 is an address kanji dictionary, which is a dictionary in which kana notation and kanji notation of prefectures, municipalities, etc. are compared. Reference numeral 11 denotes a kana-kanji dictionary, which is a dictionary in which kana and kanji expressions for reading the kana are compared. 1
Reference numeral 2 refers to the address kanji dictionary 10 in the dictionary reference section,
Reference numeral 13 is a process of referring to the kanji dictionary for kanji characters 11. Reference numeral 14 is kanji address data converted into kanji as a result of the dictionary reference processing. Reference numeral 15 is a correction screen for displaying the Kanji address data 14. Reference numeral 16 indicates a correction result.

The operation of the basic configuration of FIG. 1 will be described later (see the section of [Operation]). FIG. 2 shows the basic configuration (2) of the present invention.
The figure shows the Kana conversion method for Kana corporation name notation. In the figure, 21 is an input form, and 22 is an input field for a name and a corporate name. Reference numeral 23 is a character reading unit, and 25 is a Kanji conversion unit.

Reference numeral 26 is a kana character array portion, and 27 is a kana character array of kana notation of a name or a corporate name. Reference numeral 29 is a dictionary reference unit. Reference numeral 30 is a general name conversion dictionary, in which the kana notation and the kanji notation of the general name in a corporation name such as "stock corporation" are compared. Reference numeral 31 is a surname conversion dictionary, in which the kana notation and the kanji notation of the surname in the name are contrasted. Reference numeral 32 denotes a name conversion dictionary, which compares the kana notation of the name in the name with the kanji notation.

33 is a general name conversion dictionary reference process,
Reference numeral 34 is a family name conversion dictionary reference processing, and 35 is a first name conversion dictionary reference processing. 36 is converted Kanji output data, 37 is a correction screen, and 38 is a correction result. The operation of the basic configuration of FIG. 2 will be described later (see [Operation] section).

[0028]

The operation of the basic configuration of FIG. 1 will be described with reference to FIG.
FIG. 3 is a flow of kanji conversion of kana address notation according to the present invention. The flow will be described according to the numbers shown.

The data (kana characters) entered in the S1 input form is optically read. S2 Interpret the address field data (kana address notation). S3 Judge whether the name is written or not. If there is a description of the person's name, proceed to S4. If no name is entered, proceed to S5.

If there is a description of the S4 name, the character string (Kana character string) of the part of the name and the candidate character are used to access the dictionary for the name (direction dictionary) to perform knowledge processing (the following collation processing). )
I do. First, the base words (general names such as "apartment" and "sou", which will be described later) are collated.

Next, the expanded words (proper nouns such as “Kou” and “Yamada”, which will be described later) are collated. Correct kana character strings. For example, if the character written on the input form is "apartment" and "o" cannot be read in the reading process and it is recognized as "ahaat", it is matched with "apartment" in the matching process. ，
Kana character strings are corrected.

From the S5 address segmentation (dividing words by prefecture name, city / gun name, etc. into the input form), first level (prefecture name), second level (city group name), third level (Town name), 4th level (chome address)
Up to, the character data (kana character array) and candidate characters are used to process the address and kanji dictionary.

S6 Kana-Kanji conversion processing is performed. S7: The combination data of the kana character representation of the reading of the address and the name and the converted kanji data is generated. S8 A frame indicating the address display area is displayed on the correction screen (frame WRITE).

S9 The converted Kanji data is displayed on the correction screen (Kanji WRITE). S10 Display alphanumeric data and kana character data on the correction screen (WRITE of ANK data). S12 The address data converted into Kanji is output to the user.

The basic structure of FIG. 2 will be described with reference to FIG. FIG. 4 shows a flow of kanji conversion of kana corporation name according to the present invention. In the figure, S1 and S2 are processes of the character reading unit in FIG. S3 to S8 are processes of the Kanji conversion unit in FIG. S10 to S11 are processes on the correction screen in FIG.

S1 Read the data entered in the input form. S2 Interpret the name field data. Since the S3 name field is written in word units of the family name, given name, and common name, the number of written words is determined.

If the number of words is 1, the process proceeds to S4. If there are two or more, the process proceeds to S5. S4 If there is one word, collate in the order of surname>general> name (refer to the surname conversion dictionary, general name conversion dictionary, and name conversion dictionary in this order). After matching, the process proceeds to S6. If there are two or more S5, the front (the third word counted from the end of the word array) is compared in the order of general> surname.

Next, the second word counted from the end is collated by surname> general. If the last word is the last name, the last word is first name>
In general, if the last word is general, surname> general, and if the last word cannot be read, the order is first name>surname> general. S6 It is judged whether there is a character array in the dictionary. If there is a character array, the process proceeds to S7, and if not, the process proceeds to S11.

If there is a character array in the S7 dictionary, the recognized kana characters are corrected and the kanji conversion process is performed. S8 Kana data (an array of converted Kanji characters) corresponding to the Kana data (Kana character array for reading the name and corporate name) is generated.

S9 A frame indicating the display area of the name and the corporation name is displayed on the correction screen (frame WRITE). S10 The converted Kanji is displayed on the correction screen (WRITE of Kanji data). S11 Display alphanumeric characters and kana characters on the correction screen (WRITE of ANK data).

S12 The converted name and corporate name (name / corporate data) are output to the user. In the present invention, each dictionary can be registered not only by the manufacturer of the OCR system but also by the user.

[0042]

EXAMPLE FIG. 5 shows an example of conversion of kana addresses into Chinese characters according to the present invention. In the figure, 50 is an input form, 51 is an address field in the input form, and 52 is an OCR device, which corresponds to the character reading unit in the configuration of FIG. Reference numeral 53 is an address knowledge program, which performs Kanji conversion processing on the read character string and the candidate character string. 54 is an address dictionary and a name dictionary.

In the address field 51, an address is entered by dividing it into prefecture names, city / gun names, etc. In the figure, the ward name, town name, street address, and direction name are separated by spaces and entered in intervals. The OCR device 52 is
The data in all fields entered in the input form is read, and in the Kanji conversion process for Kana address notation, the Kana characters entered in the address field are recognized and the read character string and candidate character string are output.

The address knowledge processing program 53 refers to the dictionary for the read character string and the candidate character string, and performs Kanji conversion for the address and the name. First, access the address dictionary and name dictionary for kana notation of addresses and kana notation of names. Next, Kanji conversion is performed based on the dictionary, and if necessary, the read Kana character string is corrected. For example, regarding "Yamamoto Apartment 2F",
Convert the kanji into "Yamamoto Apartment 2F" by referring to the person dictionary.

The converted Kanji data is formed into a correction screen and output data (following FIG. 6). FIG. 6 shows a kanji address notation kanji conversion embodiment of the present invention (continuation of FIG. 5). In the figure, 60 is a correction screen, 61 is a screen output of the conversion result, 62
Is another display on the correction screen. 63 is output data.

On the correction screen 60, the read kana character string and the converted kanji character string are displayed. The record information in the structure of the output data 63 is a combination of the read kana character string and the kanji character string. Figure 7
Shows an embodiment of a method for referring to a person dictionary according to the present invention.

In the figure, 70 is an example of inputting a person's name, 71 is a person's dictionary, 72 is a base word in the person's dictionary, and the general name part such as apartment, building, 1F, dormitory, etc. is written in kana. It is a contrast of the official notation. Reference numeral 73 is an extended word, which is used to compare kana notation and kanji notation for names such as Kotobuki (Kotobuki) and Yamada (Yamada).

Reference numeral 74 shows an embodiment of converting Kana characters into Kanji characters. To enter the Kana name notation on the input form, separate the Kana address notation with parentheses or spaces. It is recognized by the entered delimiter that it is the name notation.

The name dictionary 71 is constructed by distinguishing general names such as apartments and buildings from unique names such as "Kou" and "Yamada". The general name is formed as a base word 72 by comparing the kana notation and the regular notation in the kana dictionary 71, and the proper name is formed as an expanded word by comparing the kana notation and the kanji notation.

The conversion of kana characters into kanji characters will be described with reference to an example of converting kana characters into kanji characters. Read "Yamamoto Apartment 2F". From the read character string, refer to the base word,
Decide on an "apartment".

Similarly, referring to the base dictionary for “2F”, “2F” is determined and “apartment 2F” is determined. With reference to the expansion word for "Yamamoto", "Yamamoto" is determined and "Yamamoto apartment 2F" is determined.

FIG. 8 shows a kanji character name conversion embodiment of the kana corporation name notation according to the present invention. In the figure, 80 is an input form, and 81 is a name field in the input form. 82 is OCR
The device corresponds to the character reading unit in FIG.

83 is a name knowledge processing program,
Kana name or Kana corporation name notation is converted to Kanji. Reference numeral 84 is a dictionary of names and corporate names. In the present invention, the corporate name is divided into words in the name field by separating the words with spaces.

The OCR device 82 reads the Kana corporation name notation in the name field and outputs the read character string and the candidate character string. The name knowledge processing program uses the name and corporate dictionary 84 for the read character string and the candidate character string.
, And refers to the name and corporate dictionary 84 for each word such as “Yamadownyu”, for example. next,
As a result of referring to the dictionary, "Yamada Transport" is determined, and the kana character array is also corrected if there is any correction. Similarly, for other words, Kanji is determined and Kana characters are corrected.

The result of the conversion is displayed on the correction screen and is configured as output data. FIG. 9 is a continuation of FIG. 8 showing conversion of kana corporation name into kanji according to the present invention. In the figure, 90 is a correction screen, 91 is a display of the corporation name on the correction screen, and 92 is another display. 93 is a structure of output data.

On the correction screen 90, the kanji character string resulting from the kanji conversion of the kana corporation name and the corrected kana character string are displayed. The record in the structure of the output data 93 is composed of a combination of the read kana character string and the converted kanji character string. FIG. 10 shows an embodiment of a dictionary reference process in the conversion of kana notation corporation name into kanji according to the present invention.

In the figure, reference numeral 100 denotes a surname conversion dictionary in which the kana notation and the kanji notation of the surname word are compared. Reference numeral 101 is a name conversion dictionary in which kana notation and kanji notation of a name are compared. Reference numeral 102 is a general name conversion dictionary, in which the kana notation and the kanji notation of general words are compared.

The general words stored in the general name dictionary are XX, YY Construction Co., Ltd., President and CEO, etc. Reference numeral 103 shows an example of the recognized kana character array. In the kanji conversion of kana corporation name according to the present invention, the order of referring to the surname dictionary, the name dictionary, and the general name dictionary is determined for the words entered in the name field according to the order in which the words are entered.

That is, if there is one word, collation is performed in the order of surname>general> first name (the dictionary is referred to in the order of surname conversion dictionary, general name conversion dictionary, name conversion dictionary). If there are two or more, the front (up to the third word from the end of the word array)
For, check the order of general> surname. Next, the second word counted from the end is matched by surname> general. If the last word is the last name, the last word is first name> general, if the last word is general, last name>
In general, if the previous word could not be read, the order is first> last name> general.

According to the above collation rules, "Yama Down You"
About "Yamada Transport" is decided by referring to the general name dictionary. For "Honsha", refer to the general name dictionary,
"Head office" is decided. For "Higashiyama", refer to the surname conversion dictionary to determine "Higashiyama". As for "Kenji", "Kenji" is decided by referring to the name conversion dictionary because the whole name is "Last name".

If the character string to be converted does not exist in the dictionary referred to first, the dictionary of the next rank is referred to. 104 is the result of converting the recognized kana character array 103 into kanji.

[0062]

According to the present invention, even when a name is written in an address, the kanji conversion is automatically performed in the kanji conversion of the kana address notation, so that the process of correcting the name in the address conversion is unnecessary. The recognition rate is improved.

In addition, the entry of the Kana corporation name notation on the form,
Since it is not necessary to be aware of the difference between the full name and the corporate name, the entry work becomes easier.

[Brief description of drawings]

FIG. 1 is a diagram showing a basic configuration (1) of the present invention.

FIG. 2 is a diagram showing a basic configuration (2) of the present invention.

FIG. 3 is a diagram showing a flow of kanji conversion of kana address notation according to the present invention.

FIG. 4 is a diagram showing a kanji conversion flow for kana corporation name notation according to the present invention.

FIG. 5 is a diagram showing a kanji character conversion example of kana address notation according to the present invention.

FIG. 6 is a diagram (continuation of FIG. 5) showing an example of converting Kana to Kana addresses according to the present invention.

FIG. 7 is a diagram showing an embodiment of a name dictionary reference process of the present invention.

FIG. 8 is a diagram showing an embodiment of converting kana to kana corporation name according to the present invention.

FIG. 9 is a diagram showing a kanji character conversion example for kana corporation name notation (sequel to FIG. 8) according to the present invention.

FIG. 10 is a diagram showing an example of a dictionary reference process in kana conversion of kana corporation name according to the present invention.

FIG. 11 is a diagram showing a kanji conversion system for kana address notation in a conventional OCR system.

FIG. 12 is a diagram showing a kanji conversion process for kana name notation and kana corporation name notation in a conventional OCR system.

[Explanation of symbols]

 1: Input form 2: Address field 3: Character reading part 5: Address kanji conversion part 6: Kana character arrangement part 7: Address 8: Name 9: Dictionary reference part 10: Address kanji conversion dictionary 11: Name / kanji dictionary 12 : Address Kanji dictionary reference 13: Name kanji dictionary reference 14: Kanji address data 15: Correction screen 16: Correction result

Claims (2)

[Claims] 1. A character reading section (3) for optically reading and recognizing a kana character array of kana address notation and kana name notation entered in an input form (1) in a kanji conversion processing method of kana address notation. , Kana notation of elements that represent addresses on administrative divisions such as place names and street addresses, and the Kanji notation of addresses that store corresponding Kanji notations (10) and kana notation of person names that are auxiliary address indications such as buildings An address kanji dictionary is created by creating a kana character array from the kana notation read by the character reading part (3) and the kanji notation stored in the kanji notation storing the kanji notation.
(10) and kana dictionary written in the input form (1) with the kanji conversion part (5) for converting to kanji addresses and kanji patterns by referring to the kanji dictionary (11) A kanji conversion processing method for addresses, which is characterized by converting each name notation into kanji and outputting it. 2. A character reading section (23) for optically reading and recognizing a kana character array of a kana name or a kana corporation name entered in an input form (21) in a kana conversion method of kana corporation name notation. And a surname conversion dictionary (31) that stores the kanji notation corresponding to the surname of the name, and a name conversion dictionary (32) that stores the kanji notation corresponding to the kana notation of the name.
And a general name conversion dictionary (30) that stores the kanji notation corresponding to the kana notation of the general name of the corporation.
Create a kana character array from the kana name notation or kana corporation name notation read in (23) and refer to the above dictionaries to convert the kana name notation or kana corporation name notation to kanji. A kanji character conversion processing method for kana corporation name notation, characterized by converting the kana corporation name notation described in the entry area common to the name or corporation name in the input form (21) to output.