JP5132430B2 - Information processing apparatus, information processing method, and program for generating first and last name candidates - Google Patents

Description

  The present invention relates to a first name surname search, and more particularly to an information processing apparatus, an information processing method, and a program for generating first name surname candidates described in Chinese characters.

  In recent years, international human activities have become active due to advances in transportation and network technology. When an individual or a company conducts various activities across multiple countries, the person's first and last name and the name of the company are described and various procedures are performed. As international activities increase, it may be necessary to perform various searches using the information in countries other than the countries where individuals have processed. As an example, in the case of international money laundering, etc., the identity of the person who created, remittance, withdrawal, etc. in the bank account etc. is shared in multiple countries, and different cultures The case where it is necessary to recognize the first and last name corresponding to the category without losing the identity of the first and last name can be given.

  Last name search is not always necessary for international action. For example, in Japan, it is necessary to determine whether a bank account, health insurance, etc. indicate the same person. It is necessary to generate with high accuracy and efficiency.

  First and last names written in Indo-European languages including English, Cyrillic, Greek letters, etc. are written in phonetic characters such as alphabets, so it is suitable for searching using character strings, It is relatively easy to include similar search using character strings. Note that a method of comparing names by comparing the names of the names described in alphabets and calculating the similarity score of the names is used for name search. As a system / method for searching for a person name written in alphabet, for example, an automated person name for searching for a person name written in alphabet is described in US Pat. No. 6,963,871B1 (Patent Document 1). List search systems.

  In addition, the URI specified in http://publibfp.boulder.ibm.com/epubs/pdf/c1912860.pdf (Non-Patent Document 1) uses the similarity of person names in alphabets to search for names. A Global Name Analytics (GNA) system is also disclosed.

  By the way, in the cultural sphere using so-called ideographs, which can be exemplified as CJK (Chinese, Japanese, Korean), due to the existence of typographical characters and the difference in phonetic characteristics, There are also problems that are different from first name search. In the cultural sphere referred to as CJK, ideographic characters are used, and there are a plurality of pronunciation characteristics for the displayed characters. In addition, when the entered first and last names in kanji notation are input to the information processing apparatus in a different or abbreviated form, the comparison of the kanji code verbs causes a miss hit and decreases the search accuracy. In addition, when searching for first and last names using ideographic characters, not only the difference in phonetic name, but also transcription mistakes, misprints, and misrecognitions due to OCR are included. It is also necessary to perform a similar search in consideration of the similar shape and the like, and there are problems to be noted as compared with the surname / first name similar search in the Indo-European language area, and there is a problem in terms of accuracy. In other words, in a cultural area that includes ideographs, it is necessary to determine the similarity of the same-character and different sounds.

  Furthermore, even ideographic characters are accompanied by pronunciation, so the first and last names described in, for example, written interviews may be described as incorrect listening. In such a case, it is not possible to generate a similar candidate for a surname with high accuracy only by searching for only surnames with the same character and different sound.

  So far, techniques for searching for first and last names for kanji strings, for example, Japanese are known. An information search apparatus that performs document search by dividing a document written in Chinese characters, such as Japanese, into words (tokens) of a set unit is described in, for example, Japanese Patent Application Laid-Open No. 2004-206473 (Patent Document 2). ing. Furthermore, an information search device technique for searching by assigning different weights using surnames and first names among the names described in kanji is also described in, for example, Japanese Patent Application Laid-Open No. 2004-295797 (Patent Document 3). . Japanese Laid-Open Patent Publication No. 7-230472 (Patent Document 4) describes a personal name misreading correction method. Patent Document 4 describes a process of continuously referring to both a personal name-to-name / kanji correspondence table and a name-to-name / kanji correspondence table, and passing all the readings or kanji obtained by the table search to the search program.

  The information retrieval disclosed in Patent Documents 1 to 3 and Non-Patent Document 1 can provide sufficient accuracy and searchability to retrieve personal names described in alphabets or other Roman codes. Further, in Patent Documents 2 and 3, it is possible to search for a document in which one character is described by a Kanji character defined by a Kanji code, extract a person name from the document, and reflect it in weighting.

  However, surnames written in kanji such as kanji and katakana have various pronunciation characteristics, there are also different fonts, etc., and they were written in kanji from first and last names written in alphabetic strings such as roman letters It is difficult to generate first and last names accurately because of the diversity of phonetic characteristics of kanji.

  In addition, when the first and last names described in kanji such as kanji and katakana are input to the information processing apparatus, the first and last names that are similar in appearance may be erroneously input. In addition, when using OCR or the like as input processing, the described kanji is erroneously converted into a similar kanji by OCR, and as a result, it may be erroneously input to the information processing apparatus. The techniques of the above-mentioned patent documents and non-patent documents do not efficiently generate first and last names derived from first and last names described in Kanji, including the possibility of transcription mistakes, misprints, and misrecognitions.

  In this regard, although it is possible in Patent Document 4 to correct the reading of first and last names and correct kanji, Patent Document 4 is not intended to search for first and last names themselves. Further, Patent Document 4 does not execute the process of generating the maximum likelihood kanji estimated from the alphabet characters, and it is an object to generate surname candidates with similar homonyms and similar pronunciation characteristics. is not.

  Also, especially for Japanese surnames, there are about 150,000 known surnames, and it is known that there are more than 300,000 to 400,000 surnames, and their phonetic characteristics are diverse. , Searching first and last names in a batch also has a problem in processing efficiency. In addition, when searching for first and last names at the same time, the generation of possible first and last name combinations is limited for first and last names that may include transcription errors or misprints of the input kanji strings, or misrecognition of OCR. There was also a problem of end.

US Pat. No. 6,963,871B1 JP 2004-206473 A JP 2004-295797 A JP-A-7-230472 http://publibfp.boulder.ibm.com/epubs/pdf/c1912860.pdf

  That is, an information processing device that receives input of first and last names written in kanji so far, performs a first and last name search with high accuracy, and generates a first and last name candidate of kanji that may correspond to the first and last name entered from the search result, information A processing method and program were needed. In addition, from the viewpoint of international cooperation such as money laundering, when searching for first and last names in alphabetical letters in the Indo-European culture area, when searching for first and last names expressed in alphabetical characters with Chinese characters, There was a need to improve search integration and generate high-precision kanji surname candidates.

  The present invention has been made in view of the above-mentioned problems of the prior art, and it is efficient for surnames written in Japanese, Korean, and Chinese, including homonyms and similar pronunciation characteristics. A similar name search is performed, and a first name surname that may arise from the first name surname is generated with high accuracy and high efficiency. In the present invention, from the first name and last name entered in the inputted kanji string, the first and last name candidates expressed automatically in alphabetic strings described in Indo-European languages such as alphabets are automatically generated by referring to the kanji first name-roman name surname conversion dictionary To do. In the automatic generation, an identifier for identifying the first name and the last name in the generated alphabet string is generated, and the last name search and the first name search sequence identifying the CJK cultural area and the first and last name sequence of the Indo-European language area and Name search is possible.

  The generated alphabetic character string is divided into search words for searching for first and last names and used for two types of similar searches. The first similar search is a Kanji-specific homonym search, and uses a surname search word and a first name search word directly to obtain a Kanji surname candidate and first name candidate to obtain a Kanji surname-alphabet. Search the first name surname conversion dictionary. The second similarity search searches the alphabet surname dictionary for surname search terms and surname search terms. The search of the alphabet first name surname dictionary is a similarity search in terms of so-called pronunciation similarity based on differences in pronunciation characteristics, such as differences in vowel and consonant types, word lengths, and the like.

  After that, the information processing apparatus receives the search result, and uses the search result as a search word for reverse-searching the kanji surname-roman name surname conversion dictionary to generate surname candidates and name candidates with similar pronunciation characteristics. In the present invention, the term “similar reverse search” means a search process in which a Chinese character corresponding to an alphabetic character string is reversely searched from an alphabetic character string. Further, the term “reverse search” means a process of reversely retrieving a Chinese character from an alphabetic character string.

  In the present invention, since the search is performed by separating the first name and the last name, it is possible to perform the search by absorbing the difference in the cultural sphere, and it is possible to perform the first name search integrated with the first name search in the Indo-European language area. In addition, first and last name candidates having possible first name and last name combinations that are expected to be generated from the first name and last name entered using the search results may be generated. Furthermore, in the present invention, it is possible to perform a wide range of similar searches especially for Japanese names with various pronunciation characteristics.

  The information processing apparatus of the present invention can be configured as a server connected to a network, and the present invention provides a computer-executable information processing method and program for executing the above-described processing.

  As described above, according to the present invention, the first and last name input described in kanji is received, the first and last name search is performed with high accuracy, and the same character allophone and pronunciation similar viewpoint corresponding to the first and last name input from the search result Information processing apparatus, information processing method, and program can be provided that can generate possible first name and surname candidates of kanji characters and that can be integrated with first name surname search in the Indo-European area.

  Hereinafter, the present invention will be described with reference to embodiments shown in the drawings, but the present invention is not limited to the embodiments described below.

  FIG. 1 shows an embodiment of an information processing system 100 including an information processing apparatus 130 of the present embodiment. The information processing system 100 includes a network 120 and a number of clients 112 connected to the network 120 and operated by a user to access the information processing apparatus 130 via the network 120.

  The information processing apparatus 130 can be configured from a server machine, and can be configured as a web server, an SNS server, or a dedicated server for searching for first and last names. In the present embodiment, the network 120 preferably includes a network such as the Internet, but may include a WAN (Wide Area Network), a LAN (Local Area Network), and the like in addition to the Internet.

  Further, the information processing apparatus 130 manages a plurality of databases in order to generate first and last names written in kanji. The first database 140 is a Kanji surname / Romaji surname database. In the first database, the information processing apparatus 130 receives a first and last name written in kanji and executes a process of converting a kanji string constituting the first and last name into an alphabetic character string. The second database 150 is an alphabet surname database. The second database 150 performs a similarity search using pronunciation similarities of kanji surnames converted to alphabetic characters, and searches for alphabetic strings of surnames and first names according to the degree of matching with the input alphabetic character strings. As a result, the information is returned to the information processing apparatus 130.

  In the present embodiment, the kanji means a character used in a language described using a typical kanji referred to as a so-called CJK, such as Japanese, Chinese, or Korean.

  In addition, the alphabet character string is a character string that is described in Japanese using Hebon's formula, ceremonial formula, etc., and that uses alphabetic characters that reproduce the pronunciation of kanji, and is referred to as Roman characters hereinafter. In the present embodiment, the present invention can also be applied to Chinese and Korean, as well as to an alphabet character string that reproduces the pronunciation of ideographic characters such as Chinese characters in alphabet. This embodiment can be applied to first and last names in the Kanji culture area designated by CJK. However, in this embodiment, for convenience of explanation, the first and last names will be converted into Roman characters and further converted into first and last names. .

  The above-described information processing apparatus 130 is implemented with a CISC architecture microprocessor (MPU) such as PENTIUM (registered trademark), a PENTIUM (registered trademark) compatible chip, or a RISC architecture microprocessor such as POWER PC (registered trademark). can do. The information processing apparatus 130 is controlled by an operating system such as WINDOWS (registered trademark) 200X, UNIX (registered trademark), or LINUX (registered trademark). The information processing apparatus 130 is a server program such as CGI, servlet, APACHE, or IIS, which is implemented using a programming language such as C ++, JAVA (registered trademark), JAVA (registered trademark) BEANS, PERL, or RUBY. To process the request from the client 112.

Client 112 can be implemented using a personal computer or workstation, etc., and its microprocessor may include any single-core or multi-core processor known so far. The client 112 is controlled by an operating system such as WINDOWS (registered trademark), UNIX (registered trademark), LINUX (registered trademark), or MAC OS. The client 112 includes Internet Explorer (trademark), Mozilla, Opera, Netscape.
Use an HTTP transaction infrastructure using browser software such as Navigator (trademark), or a CORBA (Common
(Object Resource Broker Architecture) or the like is accessed on the information processing device 130 using a transaction basis, and processing such as request issuance and information acquisition is performed.

  FIG. 2 is a functional block diagram showing the software module 200 of the information processing apparatus 130 of this embodiment. The software module 200 of the information processing apparatus 130 reads an execution format object stored on the hard disk into an execution space such as a RAM mounted on the information processing apparatus 130, and the microprocessor executes the object as a function unit. Realized on an information processing apparatus.

  As shown in FIG. 2, the information processing apparatus 130 is configured to receive first and last names written in kanji sent from the network 120 and include a NIC (network interface card) or the like in the OSI basic reference model. To the network adapter 210 that provides physical layer and data link layer functions. The first and last name received by the network adapter 210 is sent to each software module group configured as a server program including CGI, servlet, and the like.

  The full name is first sent to the alphabet character string generation unit 220. The detailed configuration and processing of the alphabet character string generation unit 220 is described in Japanese Patent Application No. 2008-117538 (Athony Docket number: JP9080054). The outline of the processing will be described below.

  The alphabet character string generation unit 220 includes a character normalization unit. The character normalization unit normalizes the kanji strings that make up the first and last names. In the present embodiment, the term “normalization” executes a process of unifying the different fonts into a font that can be used by the information processing apparatus 130 for subsequent processing. For example, in the case of the Japanese name “Saito”, the allo-style is unified with the characters “Sai” and “Sai” in the Kanji “Saito” and “Sai”. Means processing. Further, in the case of “wisteria”, it means a process of unifying the “rattan” variant into “wisteria”.

  After normalizing the first and last names, the alphabet character string generation unit 220 calls the morpheme analysis unit to search the morpheme dictionary 280, and generates an alphabet character string corresponding to the normalized kanji character string. The morpheme dictionary 280 can be implemented as a table set configured in the first database 140. The morpheme dictionary 280 includes a table for registering kanji (columns) and alphabetic character strings corresponding to kanji (columns) in order to execute the above-described processing. In addition, the morpheme dictionary 280 registers, for each morpheme, associated words such as first name, last name, “san”, and “dono”.

  The combination supplement creation unit receives the combination information generated by the morpheme analysis unit, and analyzes the connection for the attribute identifiers attached to the surname and first name. In the analysis of attribute identifiers, attribute identifiers attached to morphemes corresponding to consecutive last names and first names are extracted and determined as connection identifiers from the sequence. The connection identifier refers to, for example, an attribute identifier attached to each morpheme, and the sequence of morphemes is a surname-surname (SG), surname-surname (SS), surname-surname (GS), surname-surname (GG). Are assigned different score values and are also used to give an indication of the validity of the combination candidates.

  The combination candidate creation unit is a type in which a connection identifier characterizing the connection of morphemes is inserted between morphemes, and includes a record including a field of (morpheme-connection identifier-morpheme-connection identifier,...). A candidate list is created for all generated morpheme sequences. The result of the processing described above is configured as a combination candidate list. The combination candidate list is generated such that a morpheme and a connection identifier for a specific first and last name candidate constitute one record. The combination candidate creation unit identifies the cultural area to which the first and last names belong from the generated first and last name candidates, and the combination of the first and last names is, for example, CJK in Japan or Korea Or a cultural weight that identifies China.

  Thereafter, the information processing apparatus 130 uses the connection identifier generated as described above as an identifier for identifying the last name and the first name, and adds the connection identifier to an appropriate portion of the alphabet character string as described above. This identifier is used to identify where in the alphabet string sequence is a surname and where is a first name, and for subsequent search of surname and first name and for the alphabet surname / name dictionary 290-2. Used to specify last name search terms and first name search terms for searching. In the present embodiment, the connection identifier is described as using a value such as SG indicating the sequence of first and last names. However, as long as it is possible to distinguish between the last name sequence and the first name sequence such as a blank character, Any identifier can be used.

  Moreover, the alphabet character string generation unit 220 includes a score calculation unit, and the score calculation unit acquires, for a specific record in the surname candidate list, the value of the field whose record value is not NULL, and the field value The score values assigned for are obtained and the score values obtained for the non-NULL field values of the processing record are summed.

  Further, the alphabet string generation unit 220 extracts the first and last name candidate having the smallest total score value from the first and last name candidate list as the most likely combination candidate, and the kanji character has the closest pronunciation characteristic for the most likely combination candidate. With reference to the notation conversion dictionary for registering alphabetical notation, the alphabetical notation registered for the most likely first name surname is assigned to the surname and first name, and is output as the most likely alphabetical notation for the surname surname candidate.

  Further, each module in FIG. 2 will be described. The alphabet character string generation unit 220 sends the surname and first name to the similar reverse search unit 240. The similar reverse search unit 240 directly receives the first and last names output from the alphabetic character string generation unit 220 and performs a similar reverse search on the first name conversion dictionary 290-1. This similar reverse search is executed in order to make it possible to search for first and last names of the same character in the generated combination candidate alphabetic character string. Note that the similar reverse search for homophones is preferably used to search for so-called homonymous names for names with particularly diverse pronunciation characteristics. The searched surname of the same character and different sound is sent to the alphabet character string search unit 230, added as a candidate name of the same character and different sound, and used as a search word of the alphabet character string search unit 230. In another embodiment, the similar reverse search for the same character and different sound can be performed only for the first name, and the surname is directly sent to the alphabet string search unit 230 to perform only the phonetic similar search. You may let them.

  The similar reverse search unit 240 sends the searched first and last names of the same character and different sounds to the alphabet character string search unit 230 together with the name received from the alphabet character string generation unit 220. After that, the alphabet character string search unit 230 sets the received surname as a surname search word, and sets a name including a first name with the same character as a first name search word. Thereafter, the alphabet character string search unit 230 performs a search of the alphabet first name surname dictionary 290-2 to search for an alphabet character string related to a surname that matches or is similar to the alphabet character string to be searched.

  Similar alphabetic character strings are determined by configuring each search word and using the specified characteristic values of the search word, for example, the concordance rate of consonants or vowels of similar pronunciations, the concordance rate of surname or first name character strings, etc. can do. These characteristic values are collectively referred to as the pronunciation characteristics, and in this embodiment, the similarity of the pronunciation characteristics is referred to as the pronunciation similarity. Note that the pronunciation characteristic similarity determination process is not directly taken up by the present invention, so a more detailed description is omitted.

  The search by the alphabet character string search unit 230 is mainly performed to execute a pronunciation similarity search. In this embodiment, the alphabet character string search unit 230 of this embodiment can use a Global Name Recognition system manufactured by International Business Machines Corporation, which implements Global Name Analytics.

  In addition, the alphabet character string search unit 230 is shown as being configured as a module of the information processing apparatus 130 in the embodiment shown in FIG. However, in other embodiments, the alphabet character string search unit 230 can be functionally separated as a separate server device or the like.

  In the following, the similarities search for pronunciation similarities and similar sounds of different characters will be described in detail. The search by the alphabet string search unit 230 returns, for example, the alphabet strings “takahashi” and “takanashi” that have similar pronunciation characteristics, and can extract names with similar pronunciation characteristics such as “Takahashi” and “Takanashi”, respectively. And As for the name, the search result of the alphabet string search unit 230 returns search results such as “Takeshi” and “Takashi”. In the similar reverse search, “Ken” and “Takeshi” with the same sound and different characters are used. , “Take” etc. are extracted. As described above, the similar reverse search unit 240 can extract similar surnames including surname candidates and surname candidates registered as pronunciation similar to those other than the alphabetic character string search unit 230 used as search terms. And expand the range of first and last name candidate extraction.

  On the other hand, the homonym / sound search is executed when the similar reverse search unit 240 receives an alphabetic character string corresponding to a name from the alphabetic character string generation unit 220 and is applied particularly to generation of name candidates having various pronunciation characteristics. Is done. For example, if the name is input with “恭子”, the similar reverse search unit 240 first obtains “kyouko” and reversely references the surname conversion dictionary 290-1 to search for the presence of other Chinese characters having the same sound. . For “Isuko” = “kyouko” illustrated, “yasuko” is also registered, and “yasuko” is returned as a search result. The similar reverse search unit 240 can expand the name candidates by sending the values of “kyouko” and “yasuko” to the alphabetic character string search unit 230, and performing a pronunciation similar search for the names of the same character and different sounds. It becomes possible.

  The similar reverse search unit 240 receives the search result generated by the alphabetic character string search unit 230, and performs a reverse search for the first name and last name conversion dictionary 290-1 again for a first name, and generates a surname candidate and a first name candidate in the corresponding kanji string. .

  The similar reverse search unit 240 sends the search result to the Chinese character string output unit 250. The kanji string output unit 250 generates the output of the similar reverse search unit 240 as an output result 260 in the illustrated embodiment, such as kanji (string), alphabetical notation, and corresponding reading. The output result 260 shown in FIG. 2 is created and output as a table 270 in which kanji strings corresponding to name candidates, alphabetical notation, and corresponding readings are recorded.

  In addition, the kanji string output unit 250 receives the surname candidate and the first name candidate generated by the alphabet character string generation unit 220, creates a union, and creates a first name surname candidate in association with each surname candidate and first name candidate. To do. The created first and last name candidates are ranked using, for example, a generic ratio of first name and last name, and can be listed in descending order relative to the rank.

  FIG. 3 shows an embodiment of conversion tables 300 and 310 included in the first and last name conversion dictionary 290-1 of the present embodiment. The conversion table 300 is a conversion table used for name conversion. The conversion table 310 is a conversion table 310 used for last name conversion. In the conversion tables 300 and 310, alphabetical expressions corresponding to Chinese character expressions are registered in association with each other. When converting kanji into alphabetic notation, the alphabet character string changing / generating unit 220 acquires an alphabet character string using the conversion tables 300 and 310, and generates first and last name candidates.

  To explain using the embodiment shown in FIG. 3, for example, if the similar reverse search unit 240 acquires “Takeshi” when performing a reverse search for a name, the name conversion table 300 is dereferenced and “ The Chinese character notation registered with “Takeshi” is reversely searched, “Ken” and “Tsuyoshi” are acquired, and the acquired “Ken” and “Tsuyoshi” are sent to the alphabet string search unit 230. Note that the conversion table 310 shown in FIG. 3 is used to reversely convert surnames, and is used to generate corresponding surname candidates using the pronunciation similarity search result of the alphabet string search unit 230.

  Moreover, the first name surname conversion dictionary 290-1 includes an attention character list 320 as shown in FIG. 3. The attention character list 320 is referred to only in the similar reverse search for acquiring the same character abnormal sound. When the searched surname candidate and first name candidate are included in the attention character list 320, the similar reverse search unit 240 creates an additional last name candidate and first name candidate by replacing the character corresponding to the attention character with the attention character. , The search result is added and passed to the kanji string output unit 250. The surname candidate and the first name candidate replaced with the attention character are added to an appropriate record of the output result 260 and output as a table 270 or the like.

  FIG. 4 shows a flowchart of the first and last name search method of the present embodiment. The process of FIG. 4 starts from step S400, and in step S401, a kanji character corresponding to the first and last name is acquired. In step S402, a similarity search process is executed. The similarity search process in step S402 executes a similarity search for both first-name and second-name candidates of the same character and different sound in the present embodiment.

  In step S403, it is determined whether or not the first and last names to be processed remain. If the first and last names remain in the request queue, for example (yes), the process branches to step S401 and is processed until there are no unprocessed first and last names. repeat. On the other hand, if it is determined in step S403 that there are no surnames to be processed (no), the process branches to step S404, a result output is generated, and the process ends.

  When generating the result output, the kanji such as the original kanji normalized by the alphabet character string generation unit 220 is stored, and the kanji corresponding to the surname or first name in the kanji before normalization for the result output is stored. May be displayed, or the first and last names including Chinese characters before and after normalization may be displayed in parallel to improve the recognition of the search result.

  FIG. 5 shows a flowchart of first and last name search processing including similar reverse search processing of the present embodiment. The process of FIG. 5 starts from step S500, and in step S501, an alphabet character string is generated from kanji, and a search term corresponding to the first name and last name is set with reference to the first name and last name sequence identifier.

  In step S502, the similar reverse search unit 240 performs a first name surname similar reverse search (same character abnormal sound) using the first name search word and the first name search word directly, and the same character abnormal sound corresponding to the first name acquired in step S503. Candidates are generated and sent to the alphabet string search unit 230 along with name search terms. After that, in step S504, the alphabet surname / name dictionary 290-2 is searched for surname candidates and surname candidates including pronunciation similarities, and the end of the search is awaited in step S505. If it is determined in step S505 that the search is complete (yes), first and last name candidates are output in step S506. On the other hand, if the search does not end (no), the process branches to step S504 until the search ends, and the search process is continued until there are no surname candidates and first name candidates.

  On the other hand, in step S506, the Chinese character string output unit 250 receives the results of each search and generates a search result. In step S507, as a search result, a table 270 for receiving and outputting the results generated by the same character abnormal sound search and pronunciation similarity search is created, and the process ends in step S508. In the creation of the table 270, a union is created so as to eliminate duplication of surname candidates and first name candidates that are duplicated in the result of each similar search. Thereafter, by registering the first name candidate in association with the last name candidate, it is possible to generate a search result in a table type or a list type. At this time, it is possible to make it easier to recognize the matching of the results by replacing or sorting the surnames or surnames including the characters before character normalization so that they are arranged at the top.

  In the process shown in FIG. 5, a search process using a surname / name sequence identifier for identifying a surname and a first name is adopted. The reason for this is that, as in Japanese, when the number of surnames and the number of first names are significantly different, the search process is separated and streamlined, and the alphabet string search unit 230 offsets the differences in cultural spheres. This is to directly execute the global first name search together with the first and last name in the Indo-European area by directly using the result of the automatic first name combination processing from the kanji.

  FIG. 6 shows an embodiment of a name candidate 600 generated by last name generation by first name surname search of this embodiment. In the embodiment illustrated in FIG. 6, “lion” is input as the name candidate 610. The alphabet character string generation unit 220 acquires the input name 610, determines the cultural sphere, and generates an alphabet character string of “kyoko” 620. On the other hand, the alphabet character string search unit 230 that has received “kyoko” 620 performs reverse search of the similar reverse search unit 240 and acquires an alphabet character string of “yasuko” 630 that has the same character and sound as “kyoko” 620. The alphabet string search unit 230 also performs a pronunciation similarity search for “kyouko” 620 and “yasuko” 630, and returns alphabet string strings such as “kiyoko” 640 and “yatsuko” 650 as search results.

  The similar reverse search unit 240 performs similar reverse conversion for each of “kyoko” 620, “yasuko” 630, “kiyoko” 640, and “yatsuko” 650, and sets name candidates 670 such as “Kyoko” and “Kyoko”. Name candidate sets 660 such as “Yasuko” and “Yasuko”, name candidate sets 680 such as “Kiseko” and “Keiko”, and name candidate sets 690 such as “Yatsuko” and “Yatsuko” are generated. These first name candidate sets are registered in the first name field of the table 270, and are registered as possible first and last name candidates in combination with the last name candidate sets.

  FIG. 7 shows an embodiment of a surname candidate 700 generated by a surname search by a surname search according to this embodiment. The alphabet character string generation unit 220 receives “Takahashi” as the surname candidate 710 and sends the alphabet character string “takahashi” 720 to the alphabet character string search unit 230. The alphabet string search unit 230 performs a search for “takahashi” 720 and also acquires “takanashi” 730 by a pronunciation similar search, and returns it as a search result.

  The similar reverse search unit 240 executes searches for “takahashi” 720 and “takanashi” 730, respectively, and generates surname candidate sets 740 and 750, respectively. These surname candidate sets 740 and 750 are described in the surname field of the table 270, and can be output as the table 270 together with the corresponding alphabetic notation. As a result, first and last names that may be derived from a kanji string such as kanji are output. The kanji list can be added to the surname conversion dictionary 290-1, for example, “Ming”, “朋”, “Help”, “肋”, and similar Kanji characters are registered. When either one is searched, it is possible to perform processing for adding to the output data as a reference first and last name. The attention kanji list is a preferred embodiment for raising the possibility of reading failure when the input is performed by OCR or the like.

  FIG. 8 shows an embodiment of a graphical user interface (GUI) showing the output result of the alphabet string search unit 230 in this embodiment. In the GUI 800 according to the embodiment shown in FIG. 8, the search result of last name (SN) and Ming (GN) is displayed in the display window 810. In the embodiment of FIG. 8, in the display window 810, search results for surnames (SurName: SN) are listed in the upper row, and search results for first names (GivenName: GN) are listed in the lower row. In the embodiment shown in FIG. 8, GN = hideki and SN = matsu are input, and search results similar to the input first and last names are displayed as SN and GN.

  The search result shown in FIG. 8 is sent to the similar reverse search unit 240, and the first and last names expressed in kanji are reverse searched. In the embodiment shown in FIG. 8, 15 surnames and surnames are displayed for both surnames and surnames corresponding to the pronunciation similarity of surnames. Correspondingly, the type of generated first and last names can be adjusted, such as passing first and last range of first and last names.

  As described above, according to the present embodiment, the first and last name search including the names having various reading possibilities is performed by separating the first name and the last name in the alphabet string using the first and last names obtained in kanji. Therefore, it is possible to execute a high-accuracy first-name search regardless of the difference in the kanji font style. Furthermore, in the present embodiment, it is possible to include the identifier of the first and last name sequence corresponding to the cultural sphere corresponding to the difference in the cultural sphere, etc., and to allow the alphabetic character string search unit 230 to execute the first and last name search. Without causing problems such as searching for data corresponding to first names, it can be integrated into a global first name search system, and more accurate and efficient search can be performed.

  Furthermore, by performing a first and last name search separated by first name and last name, it is possible to perform a parallel search corresponding to the number of last name types and the number of first name types. Since many names can be generated rather than performing a name search, it is possible to generate first and last names in a kanji string including the possibility of transcription mistakes, misprints, and misrecognitions.

  The functions of this embodiment are described in an object-oriented programming language such as C ++, Java (registered trademark), Java (registered trademark) Beans, Java (registered trademark) Applet, Java (registered trademark) Script, Perl, and Ruby. The program can be realized by a program executable by the apparatus, and the program can be stored in a device-readable recording medium such as a hard disk device, CD-ROM, MO, flexible disk, EEPROM, EPROM, and distributed. It can be transmitted over the network in a possible format.

  Although the present embodiment has been described so far, the present invention is not limited to the above-described embodiment, and other embodiments, additions, changes, deletions, and the like can be conceived by those skilled in the art. It can be changed, and any aspect is within the scope of the present invention as long as the effects and effects of the present invention are exhibited.

The functional block diagram of the information processing system of this embodiment. The functional block diagram which showed the software module of the information processing apparatus of this embodiment. The figure which showed embodiment of the registration table content of the full name conversion dictionary of this embodiment. The flowchart of the information processing method for performing the full name search of this embodiment. The flowchart of embodiment of the similarity search process including the similarity reverse search of this embodiment. The figure which showed embodiment of the name candidate produced | generated by the surname search by the surname search of this embodiment. The figure which showed embodiment of the surname candidate produced | generated by the surname search by the surname search of this embodiment. The figure which showed embodiment of the graphical user interface (GUI) which shows the output result of an alphabet character string search part in this embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Information processing system, 112 ... Client, 120 ... Network, 130 ... Information processing apparatus, 140 ... Kanji surname-Romaji surname database, 150 ... Alphabet surname database, 200 ... Software module, 210 ... Network adapter, 220 ... Alphabet Character string generation unit, 230 ... Alphabetic character string search unit, 240 ... Similar reverse search unit, 250 ... Kanji string output unit, 260 ... Output result, 270 ... Table, 280 ... Morphological dictionary, 290-1 ... Last name conversion dictionary, 290 -2 ... Alphabetic first name surname dictionary, 300 ... Conversion table, 310 ... Conversion table, 320 ... Caution character list

Claims (9)

An information processing apparatus for generating a first name surname similar to a first name surname expressed in kanji, wherein the information processing apparatus includes:
Receiving the first and last name, searching the first database and generating an alphabetic character string corresponding to the first and last name together with an identifier for identifying the first name and last name;
Receiving the alphabet string to identify the first name and last name, separating the last name search word and the first name search word and searching the second database, the last name search word and the first name including the homonymous character given by the alphabet string An alphabet string search unit that obtains pronunciation-like search results for the search terms;
Returning the first character search word of the same character and different sounds written in a Chinese character string using the first name search word and the first name search word of the alphabet character string to the alphabet character string search unit, obtaining the search result and A similar reverse search unit that generates a surname candidate and a first name candidate by performing a similar reverse search of the first database;
An information processing apparatus, comprising: a kanji string output unit that outputs a first name surname candidate similar to the first surname from the generated surname candidate and the first name candidate.   The first database includes a kanji surname / alphabet surname / name conversion dictionary, and the similar reverse search unit performs a reverse search of the first database by directly using the surname search word and the first name search word, and the same character abnormal sound. The information processing apparatus according to claim 1, wherein each of the search terms is acquired.   The second database includes an alphabet first name surname dictionary, and the alphabet string search unit performs a phonetic similarity search on the alphabet first name surname dictionary using the first name surname search word and the first name search term of the same character and different sounds. The information processing apparatus according to claim 2.   The kanji string output unit generates a union from the last name candidate and the first name candidate generated by the similar search for the same-character different sound and the similar search for the phonetic similarity, and outputs the union as the last name candidate. The information processing apparatus described in 1. An information processing method for generating a first name surname similar to a first name surname expressed in Chinese characters, the information processing method comprising:
Receiving the first and last name and searching a first database to generate an alphabetic string corresponding to the first and last name together with an identifier identifying the first and last name;
Receiving the alphabet string to identify the first name and last name, separating the last name search word and the first name search word and searching the second database, the last name search word and the first name including the homonymous character given by the alphabet string Obtaining pronunciation-like search results for the search terms;
Each of the search terms including the same name / same name search term and the first name search word written in a kanji string using the first name search term and the first name search term of the alphabet character string, the alphabet character string search unit And generating a surname candidate and a first name candidate by obtaining the search result and performing a similar reverse search on the first database;
A step of outputting a first and last name candidate similar to the first and last name from the generated last name candidate and the first name candidate.   The first database includes a kanji surname / alphabet surname / name conversion dictionary, and the similar reverse search unit performs a reverse search of the first database by directly using the surname search word and the first name search word, and the same character abnormal sound. The information processing method according to claim 5, further comprising a step of searching for each of the search terms.   The second database includes an alphabet first name surname dictionary, and the alphabet character string search unit performs a phonetic similarity search on the alphabet first name surname dictionary using the first name surname search word and the first name search term of the same character and different sound. The information processing method according to claim 6, comprising:   The step of outputting a first and last name candidate similar to the first and last name generates a union from the first name surname candidate and the first name candidate generated by the similar search for the same character and different sounds and the similar search for phonetic similarity, The information processing method according to claim 7, comprising an output step. An apparatus-executable program for generating first and last name candidates similar to first and last names expressed in Chinese characters, the program comprising:
Means for receiving the first and last name and searching a first database to generate an alphabetic string corresponding to the first and last name together with an identifier identifying the first and last name;
Receiving the alphabet string to identify the first name and last name, separating the last name search word and the first name search word and searching the second database, the last name search word and the first name including the homonymous character given by the alphabet string Means for obtaining pronunciation-like search results for the search terms;
Each of the search terms including the same name / same name search term and the first name search word written in a kanji string using the first name search term and the first name search term of the alphabet character string, the alphabet character string search unit And generating a surname candidate and a first name candidate by obtaining the search result and performing a similar reverse search of the first database;
A program which functions as means for outputting a first name surname candidate similar to the first name surname from the generated surname candidate and the first name candidate.

Images