JP4303027B2 - Apparatus and method for converting lexical data to data - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an apparatus and a method for converting a lexical phrase including a plurality of words into another data including one piece of information specified by the plurality of words.
[0002]
[Prior art]
For example, a bank may require a CIF analysis process for a so-called telegram (CIF means a customer information file). Specifically, in the processing of messages between banks or within banks, for example, a plurality of words in the “GETRONICS FOODs CO. LTD 1-2-34 AKASAKA” in the message as indicated by reference numeral 30 in FIG. It is necessary to convert the combination of the words “GETRONICS”, “FOODS”, and “AKASAKA” into the customer code “123-45678” indicated by reference numeral 32 in FIG.
[0003]
Conventionally, this conversion processing is performed as follows. That is, a customer code table of combinations of a plurality of words and corresponding customer codes is stored in the storage device in advance. Next, a plurality of words to be converted are extracted from the input data, and the extracted plurality of words and a plurality of words in the customer code table are subjected to character string comparison. (See Patent Document 1 for such an example.)
[0004]
[Patent Document 1]
JP 2002-56005 A
[0005]
[Problems to be solved by the invention]
However, since such character string comparison is performed in units of 1 byte, particularly when a large amount of data is targeted, the processing cannot be performed at high speed due to the search time required for the processing. There was a problem.
[0006]
Accordingly, an object of the present invention is to convert a lexical word including a plurality of words into another data including one piece of information specified by the plurality of words at a high speed and to perform processing by a computer that receives the input of the data. It is to do.
[0007]
[Means for Solving the Problems]
  The above object is based on a lexical phrase including a plurality of words according to one aspect of the present invention.,The multiple wordsPreset identification for the object represented byIn a device that converts to other data containing information, multiple wordsIn advanceRegistrationWasBasic dictionary,as well as,Words registered in the basic word dictionaryKey word selected as a key and a combination of the key word and other wordsPair,The identification set in advance for the object represented by the set of wordsInformation and,ButAssociationIn advanceRegistered name dictionaryStorage means for storingBefore a lexical that contains the wordsIdentificationA processing engine for converting into another data including information, the processing engine comprising the basic word dictionaryIs stored in memory,conversionSubjectBreak the lexical into wordsRememberLetAboveBasic dictionaryInDisassemblyIndividualwordIs stored in memoryaddressBy acquiring each, a combination of memory addresses representing the lexical terms to be converted is obtained.Get and beforeReportSolutionShiTheIndividualwordThe name dictionaryKeyRegistered as a wordExtract wordIn addition, among the word pairs registered in the name dictionary, for each word group including at least the extracted word, each of the individual word words constituting the word set in the basic word dictionary stored in the memory Memory that stores wordsaddressEachAcquired,A memory representing at least the word set including the extracted word by replacing individual words constituting the word set including at least the extracted word with the acquired memory address, respectively. A name dictionary converted into a combination of addresses is stored in a memory, and a memory address combination that is registered in the name dictionary stored in the memory and that represents the set of words including the extracted word is included in the memory Individual memory in address combinationAddress is,TransformationSubjectLexicalIndividual memory in a combination of memory addresses representingSame as one of the addressesMemory addresscombinationTheSelected,The conversion targetLexicalThe identification registered in the name dictionary in association with the selected combination of memory addressesSolved by a device that converts to another data containing information.
[0008]
  According to one aspect of the apparatus of the present invention, the processing engine is configured to individually configure each word set in the basic word dictionary stored in the memory with respect to all word sets registered in the name dictionary. A memory address combination in which each word set is replaced with the acquired memory address, and each set of words is replaced with the acquired memory address. It is preferable to store the name dictionary converted into each in a memory.
[0009]
  Further, according to one aspect of the apparatus of the present invention, the processing engine is registered in the basic word dictionary with respect to the basic word dictionary stored in the memory when the name dictionary is stored in the memory. Among words that are registered as the key words in the name dictionary, a set of words in which the words are combined with other words as the key words in the name dictionary stored in the memory is stored. A memory address is added, and among the decomposed individual words, a word to which the memory address is added in the basic word dictionary stored in the memory is used as the key word in the name dictionary. It is preferable to extract as a registered word.
[0010]
  Moreover, according to one form of the apparatus of this invention, the said processing engine was made to memorize | store in the said memory only about the word group containing the said extracted word among the said word groups registered into the said name dictionary. A memory in which individual words constituting the set of words are stored in the basic word dictionary;addressEachAcquired,A memory address that represents the set of words only for the set of words that includes the extracted word by replacing each individual word that comprises the set of words that includes the extracted word with the acquired memory address. Among the name dictionaries converted into the combination of the above, the word set including the extracted word and converted into the memory address combination and the identification information registered in the name dictionary in association with the word set are stored in the memory It is preferable to make it.
[0011]
  Moreover, according to one form of the apparatus of this invention, the name dictionary memorize | stored in the said memory | storage means includes the said key word, the group of the word which combined the said key word with the other word, and the group of the said word. A nuclear name dictionary registered in advance in association with the set code; a code registered in the nuclear name dictionary; another word further combined with a set of words associated with the code; and the code And a full name dictionary registered in advance in association with identification information set in advance for a target represented by a set of words in which another set of words is combined with a set of words associated with the processing engine. Is a word set in the basic word dictionary stored in the memory for at least a set of words including the extracted word among the set of words registered in the nuclear name dictionary. Each of the memory addresses storing the individual words constituting the word, and replacing each of the words constituting the word set including at least the extracted word with the obtained memory address, at least A nuclear name dictionary obtained by converting the set of words including the extracted word into a combination of memory addresses representing the set of words is stored in a memory, and among the other words registered in the full name dictionary A memory address at which the other word is stored in the basic word dictionary stored in the memory for the other word associated with a code set in a set of words including at least the extracted word. Acquire the previous word associated with the code set in the word set including at least the extracted word A full name dictionary replaced with the acquired memory address is stored in a memory, and a combination of memory addresses representing the set of words including the extracted word registered in the nuclear name dictionary stored in the memory Among them, after selecting a memory address combination in which each memory address in the memory address combination is the same as any one of the individual memory addresses in the memory address combination representing the conversion target token, The code registered in the nuclear name dictionary in association with the selected combination of memory addresses is extracted, and the other code registered in association with the extracted code in the full name dictionary stored in the memory Among memory addresses representing words, the memory address Select a memory address that is the same as any of the remaining memory addresses excluding the combination of the selected memory addresses from the combination of memory addresses representing the target lexical, and select the lexical to be converted It is preferable to convert into another data including the identification information registered in the full name dictionary in association with a memory address.
[0012]
  Moreover, according to one form of the apparatus of this invention, the name dictionary memorize | stored in the said memory | storage means is a part of word which comprises the said word group among the word groups registered into the said name dictionary. It is preferable that a set of a plurality of words having different objects and having the same object to be represented is associated with the same information as the identification information.
[0013]
  The above-described problem is a basic word dictionary in which a plurality of words are registered in advance in a method of converting a lexical word including a plurality of words into another data including identification information set in advance for an object represented by the plurality of words, and The key word selected as a key among the words registered in the basic word dictionary, a set of words in which the key word is combined with other words, and a target represented by the set of words The basic word dictionary is stored in a memory by a computer having storage means for storing a name dictionary registered in advance in association with the identification information, and the words to be converted are decomposed into words and stored in the memory. By acquiring each memory address where the decomposed individual words are stored in the basic word dictionary, a combination of memory addresses representing the lexical terms to be converted is acquired. While extracting the word registered as the key word in the name dictionary among the decomposed individual words, the word including at least the extracted word from the set of words registered in the name dictionary For each set, a memory address in which each word constituting the word set is stored in the basic word dictionary stored in the memory is acquired, and at least the word set including the extracted word is formed. By replacing each individual word to be replaced with the obtained memory address, a name dictionary obtained by converting at least the word set including the extracted word into a memory address combination representing the word set is stored in the memory. A memory address representing the set of words including the extracted word registered in the name dictionary stored in the memory. A memory address combination in which each memory address in the combination of memory addresses is the same as any one of the individual memory addresses in the combination of memory addresses representing the word to be converted This is solved by a method of selecting and converting the lexical object to be converted into another data including the identification information registered in the name dictionary in association with the selected combination of memory addresses.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
Preferred embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a diagram showing a basic configuration of an apparatus for converting a lexical word into data according to a preferred embodiment of the present invention. In FIG. 1, 10 is a processing engine composed of an arbitrary data processing device such as a main frame computer, a personal computer, and a microprocessor, 12 is a main memory, 14 is a basic word dictionary, and 16 is a nucleus. A name dictionary and 18 a full name dictionary, respectively. The basic word dictionary 14, the nuclear name dictionary 16, and the full name dictionary 18 are stored in a hard disk (not shown) such as a magnetic disk, but are not limited thereto, and may be stored in any other type of storage device. Can be stored. The data processor that functions as the processing engine 10 and the hard disk that stores the main memory 12, the basic word dictionary 14, the nuclear name dictionary 16, and the full name dictionary 18 are an ordinary data bus or the like (not shown). Are connected to each other.
[0022]
FIG. 2 shows a state in which words (hereinafter, also referred to as “basic words”) registered in advance in the basic word dictionary 14 are symbolized on the main memory 12 using memory addresses as symbols, that is, expanded in memory. Note that the word or basic word in this specification includes not only common nouns, proper nouns, and abbreviations, but also any set of symbols having a certain meaning. As shown in FIG. 2, an example of the basic word dictionary 14 includes, as items, keys, parts of speech, name attributes, and code attributes, but the basic word dictionary of the present invention is an item for registering basic words. It is sufficient to include at least a certain key, and the other items are not limited to those described above. The basic word dictionary 14 preferably has a structure that allows a new basic word to be registered when it includes a basic word that is not registered in the lexical phrase to be converted, and can be deleted when it is no longer used in the registered basic word. . Before starting the conversion process, the processing engine 10 symbolizes or expands the basic words registered in the basic word dictionary 14 on the main memory 12 using the memory address as a symbol, as shown in FIG. . That is, a memory address is assigned as an entry point for each registered content. Specifically, the basic word “AKASAKA” in the key column stores information at a location of memory address 100 on the main memory 12, and the basic word “BANK” in the key column is called memory address 101. Information is stored in place, and so on. When the basic word is expanded on the memory, an item “name pattern” for storing a memory address is added to each basic word, as will be described later, and the memory is expanded. If the basic word included in the lexical phrase to be converted is known in advance, only the basic word to be used may be developed on the main memory 12, and the conversion processing speed may be slow depending on the application. However, a part of the basic words registered in the basic word dictionary 14 may be expanded in the memory, and the unexpanded basic words may be additionally expanded in the memory when necessary for the conversion process.
[0023]
FIG. 3 shows a state in which a nuclear name and a full name registered in advance in the nuclear name dictionary 16 and the full name dictionary 18 are symbolized on the main memory 12 using a memory address as a symbol, that is, expanded in the memory. As shown in FIG. 3, the items in the nuclear name dictionary 16 are composed of keys, name patterns, and codes. In the name pattern item of the nuclear name dictionary 16, two basic words included in a combination of basic words that may be converted among basic words registered in the basic word dictionary 14 are registered in advance. ing. Specifically, “GETRONICS” and “FOODS” are in the first line of the nuclear name dictionary 16, “GETRONICS” and “SHOKUHIN” are in the second line, and “GETRONICS” and “GOTRONICS” are in the third line. “BANK” is registered as a character string. The basic word common to these name patterns is “GETRONICS”, and this basic word is registered in the key item of the nuclear name dictionary 16. In the code, a symbol for representing the association with each name pattern is registered. The name patterns “GETRONICS FOODS” and “GETRONICS SHOKUHIN” have the same meaning and are preferably assigned the same symbol “# GETRO #”, but may be different. The nuclear name dictionary 16 can be deleted when a lexical phrase to be converted includes a combination including a basic word that is not registered, a combination including a new basic word is registered, and the combination that includes a new basic word is not used. It is preferable that it is made so.
[0024]
The items of the full name dictionary 18 are also composed of keys, name patterns, and codes as shown in FIG. The name pattern item of the full name dictionary 18 includes symbols corresponding to combinations of basic words indicated in the name pattern of the nuclear name dictionary 16 among combinations of basic words that may be converted, and combinations thereof. Are registered in advance as a pair. Specifically, “# GETRO #” and “AKASAKA” are registered in advance in the first line of the name pattern of the full name dictionary 18, and “# GETRO #” and “OSAKA” are registered in advance in the second line. Has been. Since the symbol common to these name patterns is “# GETRO #”, the symbol is registered in the key of the full name dictionary 18. In the code of the full name dictionary 18, target data after conversion corresponding to the name pattern, in this case, a customer code is registered. Specifically, since the customer code of both “GETRONICS FOODS AKASAKA” and “GETRONICS SHOKUHIN AKASAKA” is “123-45678”, the customer code is in the first line of the code of the full name dictionary 18, and Since the customer codes of “GETRONICS FOODS OSAKA” and “GETRONICS SHOKUHIN OSAKA” are “101-23456”, the customer codes are respectively registered in the second line of the code of the full name dictionary 18. The full name dictionary 18 can be deleted when there is a combination including a basic word that is not registered in a lexical phrase to be converted, a combination including a new basic word is registered, and the combination that includes a new basic word is not used. It is preferable that it is made so.
[0025]
In this example, a combination of two basic words is used as the name pattern of the nuclear name dictionary 16 and the full name dictionary 18, but when it is acceptable that the processing speed is somewhat slow, three or more Combinations may be used. In this example, the nuclear name dictionary 16 and the full name dictionary 18 and the two-stage name dictionary are used. However, only the nuclear name dictionary 16 or two or more full name dictionaries 18 may be used depending on the application. Good.
[0026]
Prior to starting the conversion process, the processing engine 10 converts the name pattern registered in the nuclear name dictionary 16 into a symbol, that is, a basic word memory of the developed nuclear name dictionary 16 as shown in FIG. Referring to the address, the memory address is symbolized or expanded in the main memory 12 as a symbol. At that time, those with the same key in the nuclear name dictionary 16 are grouped into one group and expanded in memory. Specifically, “GETRONICS”, “FOODS”, “SHOKUHIN”, and “BANK” in the first to third lines of the name pattern of the nuclear name dictionary 16 are the basic word dictionary expanded in the main memory 12. With reference to the 14 basic words and the corresponding memory addresses, “107”, “106”, “112”, and “101” are assigned as indicated by 34 in FIG. Then, any memory address that is not used, for example, 2000, is acquired by symbolizing three cases registered as “GETRONICS” in the key of the nuclear name dictionary 16 as a name pattern. Specifically, since the first to third lines of the name pattern of the nuclear name dictionary 16 have the same key “GETRONICS”, the memory address 107 corresponding to the name pattern “GETRONICS” of the first line As an entry point, an arbitrary memory address not used, for example, 2000 is assigned.
[0027]
Next, the codes of the nuclear name dictionary 16 registered with “# GETRO #” and “# GETROBK #” are symbolized. That is, any memory addresses that are not used for “# GETRO #” in the first and second lines of the code of the nuclear name dictionary 16 and “# GETROBK #” in the third line, for example, “500” and “501”. Each number is assigned. However, Nos. 500 and 510 only have an area for storing a memory address, and “# GETRO #” and “# GETROBK #” are not stored. In the first line of the number 2000 on the main memory 12, “107” and “106” and “500” are stored in association with the first line of the nuclear name dictionary 16, In the second row of No. 2000 on the main memory 12, “No. 107”, “No. 112” and “No. 500” are stored in association with the second row of the nuclear name dictionary 16, In the third line of No. 2000 on the main memory 12, “No. 107” and “No. 101” and “No. 501” are stored so as to correspond to the third line of the nuclear name dictionary 16. . Further, in order to link the memory address 2000 of the name pattern having the basic word “GETRONICS” in the nuclear name dictionary 16 as a key to the symbolized basic word “GETRONICS”, the basic word expanded in the memory is used. “2000” is stored in the “name pattern” storage area of the memory address 107 on the dictionary 14.
[0028]
Next, before starting the conversion process, the processing engine 10 converts the name pattern registered in the full name dictionary 18 into a symbolic name, that is, a basic word of the nuclear name dictionary 16 that has been expanded in memory, as shown in FIG. , And the memory address assigned to the code of the nuclear name dictionary 16, the memory address is symbolized or expanded as a symbol on the main memory 12. At that time, those with the same key in the full name dictionary 18 are grouped into one group and expanded in memory. Specifically, the memory address of the symbolized nuclear name dictionary is expanded (that is, linked) to the symbol of the full name dictionary 18, and therefore is in the first and second lines of the name pattern of the full name dictionary 18. Since “# GETRO #” is assigned the number 500 first, the memory address number is assigned as shown by 36 in FIG. In “AKASAKA” and “OSAKA”, “100” and “111” are shown in FIG. 3 with reference to the basic words of the basic word dictionary 14 expanded in the main memory 12 and the corresponding memory addresses. Of 36. Since the first and second lines of the name pattern of the full name dictionary 18 have the same key “# GETRO #”, the memory address corresponding to the name pattern “# GETRO #” of the first line As an entry point of No. 500, a memory address that is not used, for example, No. 8000 is allocated. Next, in order to expand (ie, link) the memory address 8000 obtained by symbolizing the full name dictionary 18 into the symbol name of the nuclear name dictionary 16, the number 8000 is stored in the storage area of the memory address 500. Thus, in the first row of the memory address 8000, “# 500” and “# 100” are stored in association with the converted target data, that is, the customer code “123-4567”, and in the second row, “# 500” and “# 100” are stored. “# 500” and “# 111” are stored in association with the converted target data, that is, the customer code “101-23564”.
[0029]
When there are two or more full name dictionaries 18, the code of the nuclear name dictionary 16 (in this example, “# GETRO #” or “# GETRO #” or “ Symbols that allow each name pattern to be identified by symbols similar to “# GETROBK #”) are registered. In the memory expansion of the intermediate full name dictionary, the symbolization of the name pattern is the same as the storage state at the number 8000 in the full name dictionary 18, but “123-45678” and “101” in the storage area of the number 8000. In the storage area corresponding to “−23564”, the memory address given to the symbol of the intermediate full name dictionary is stored.
[0030]
Next, input data conversion processing will be described with reference to FIGS. 1 to 3 and FIGS. 4 and 5. 4 and 5 are diagrams for explaining a process in which data input to the conversion device shown in FIG. 1 is converted. The memory expansion in FIG. 5 is the same as the memory expansion shown in FIG. 3, but in order to facilitate understanding of the explanation, the memory expansion of all the basic words described in the basic word dictionary 14 shown in FIG. 2 is shown. ing.
[0031]
Here, it is assumed that the basic word dictionary 14, the nuclear name dictionary 16, and the full name dictionary 18 are symbolized on the main memory 12 as described above. Assume that data indicated by reference numeral 40 in FIG. 4 is input. The processing engine 10 breaks the input data 40 into words as shown in step 42. Next, the processing engine 10 acquires a memory address corresponding to the decomposed word with reference to the basic word dictionary 14a expanded in the memory on the main memory 12 shown in FIG. Although a binary search is preferable for this acquisition method, any acquisition method may be used in the present invention. The memory address corresponding to the basic word circled in the basic word dictionary 14a of FIG. 5 is acquired.
[0032]
Next, in step 44, the processing engine 10 converts a word for which a memory address can be obtained from the decomposed words into a memory address for which the word can be obtained. If <1-234> does not exist in the basic word dictionary 14a, it is left as it is.
[0033]
In step 46, the processing engine 10 uses the basic word as a key, here, the memory address “107” of “GETRONICS” as a key, and other memory addresses, that is, “107” and “106”, A search is made as to whether any pair of “No. 104” and “No. 100” is in the nuclear name dictionary 16a expanded in the memory shown in FIG. The memory address “500” of the matched code is acquired. Specifically, the processing engine 10 reads the number 2000 stored in the storage area of the “name pattern” of the memory address 107 in the basic word dictionary 14a expanded in memory, and the memory is expanded based on the number 2000. It is checked whether there is a combination of “107” and “106”, “104”, or “100” among the memory address pairs stored in the 2000 of the nuclear name dictionary 16a. . In this example, the combination of “No. 107” and “No. 106” matches (see the combination marked with a circle in step 46 in FIG. 4 and the circled line in the nuclear name dictionary 16a in FIG. 5). , “500” is acquired, and the combination of “107” and “106” is converted to “500”.
[0034]
In step 48, the processing engine 10 subsequently uses the memory address “500” of the key symbol as a key and a combination with other memory addresses, here “500” and “100”. Whether or not the combination exists in the full name dictionary 18a expanded in the memory shown in FIG. 5 is searched, and if it matches, the matched code in the full name dictionary 18 is acquired. Specifically, the processing engine 10 reads the memory address 8000 stored at the memory address 500 in the main memory 12, and stores the number 8000 in the full name dictionary 18a expanded in memory based on the memory number 8000. Is checked whether there is a set of “number 500” and “number 100” in the set of memory addresses stored in. In this example, the combination of “No. 500” and “No. 100” match (see the combination circled in step 48 in FIG. 4 and the circled line in the full name dictionary 18a in FIG. 5). , “123-45678” on the main memory 12 is acquired, and the combination of “500” and “100” is converted to “123-45678”. As a result, “GETRONICS FOODS AKASAKA” in the input data, that is, the lexical phrase, is converted into customer code “123-23564” which is desired data.
[0035]
The processing blocks described in the processing engine 10 in FIG. 1 and the processing steps in FIG. 4 are the steps 42 and 44 in FIG. 4 in the word recognition block 20 in FIG. 1, and the step 46 in FIG. 4 corresponds to the full name recognition block 24, respectively.
[0036]
The apparatus and method for converting a lexical word into data according to the present invention includes a conventional spelling correction using, for example, a spelling pattern dictionary when an input error is entered in input data, for example, “GETRONICS” is entered as “GETROMICS”. A function may be provided, and when the input word is spelled continuously, a conventional collocation processing function such as using a collocation dictionary may be provided.
[0037]
Furthermore, in the apparatus and method for converting lexical data of the present invention into data, the name “GETRONICS FOODS CO. LTD” is entered from the input data 30 as indicated by reference numeral 32 as shown in FIG. A function to extract may be included.
[0038]
FIG. 6 is a diagram for explaining the difference between word comparison by symbolization of the present invention and conventional character string comparison. For example, a case will be described where the input data “GETRONICS FOODs” is searched for a match from three sets of “GETRONICS BANK”, “GETRONICS ELECTRONICS”, and “GETRONICS FOODS”. In the present invention, as shown in FIG. 6 (a), these three sets 60 are symbolized as memory addresses as described in the above embodiment, and the memory addresses are set as shown in 62. Convert. The total number of words in the converted set is 6 words. Moreover, these six words are numbers because they are memory addresses. Accordingly, since the numbers of the two words of the input data converted into the memory address and the numbers of these six words are compared in units of words, the comparison can be performed very quickly. On the other hand, in the conventional character string comparison, as shown in FIG. 6B, since a total of 47 characters are compared in units of characters, the comparison speed has to be slow. The comparison method based on symbolization of the present invention is essentially faster in processing speed than the conventional character string comparison method even when the search target is small. However, the processing speed is increased when the data to be searched becomes enormous, such as in banking operations. The difference becomes remarkable, and the processing can be performed at an extremely high speed as compared with the conventional character string comparison. In the comparison method using symbolization according to the present invention, it is necessary to expand the dictionary data in the memory. However, this processing does not affect the performance of the comparison processing after the startup for the initial processing at the time of system startup. Absent.
[0039]
Next, modifications of the above-described embodiment will be described below. The description of the same configuration and operation as in the above embodiment will be omitted, and only the differences will be described. Before receiving the input data, the processing engine 10 symbolizes the basic word dictionary 14 on the main memory 12 with the memory address as a symbol, but the main memory 12 in advance for the nuclear name dictionary 16 and the full name dictionary 18. Do not symbolize above. It is not necessary to provide a storage area for the “name pattern” as shown in FIG. 3 in the basic word dictionary 14 expanded in memory.
[0040]
Next, the processing engine 10 receives the input data and performs the processing up to step 44 in FIG. Next, the processing engine 10 extracts a key word from words included in the input data, and searches for a set including the extracted word in the item “key” in the nuclear name dictionary 16 (see FIG. 3). ) With reference to the basic word dictionary 14a (FIG. 5) expanded in the memory on the main memory 12, the memory address is symbolized as a symbol. For example, when the input data 40 shown in FIG. 4 is input, “GETRONICS” is extracted as the key word, and a group including “GETRONICS” in the key item of the nuclear name dictionary 16 is shown in FIG. 3 (or FIG. 5). Symbolized as indicated by memory address 2000 on the main memory 12. Here, the processing engine 10 associates each row of the nuclear name dictionary 16 in FIG. 3 with each row indicated by the memory address 2000 using any conventional technique. Therefore, the memory addresses “500” and “501” need not be stored.
[0041]
The processing engine 10 performs processing similar to step 46 in FIG. However, the processing engine 10 identifies the first set of the matched set, that is, the first row of the memory address 2000 in the example shown in FIGS. 4 and 5, and the first row of the nuclear name dictionary 16 associated with the first row. The code “# GETRO #” (see FIG. 3) is extracted.
[0042]
The processing engine 10 symbolizes a set including “# GETRO #” in the key item of the full name dictionary 18 as indicated by the memory address 8000 in the main memory 12 in FIG. 3 (or FIG. 5). . However, “500” need not be stored. Next, the processing engine 10 performs processing similar to step 48 in FIG. When the memory address “500” is not used, the processing engine 10 selects the memory address that has not been processed in the previous steps in the input data among the rows of the memory address 8000, in this example “ The line including “100” is identified and converted to the target customer code “123-45678”. In this modification, the conversion processing speed is slower than in the previous embodiment, but the capacity of the main memory 12 may be small.
[0043]
【The invention's effect】
Since the present invention is configured and operates as described above, the search time can be significantly reduced by eliminating the search processing in units of 1 byte required in the conventional character string comparison. Can be processed at a high speed by converting the lexical word including the word into another data including one piece of information specified by the plurality of words, and receiving the input of the data.
[Brief description of the drawings]
FIG. 1 is a diagram showing a basic configuration of an apparatus for converting a lexical word into data according to a preferred embodiment of the present invention.
2 shows a state in which words registered in advance in the basic word dictionary 1 of FIG. 1 are symbolized on the main memory 12 with memory addresses as symbols, that is, expanded in memory.
3 is a diagram that symbolizes each name and full name registered in advance in each of the nuclear name dictionary 16 and the full name dictionary 18 of FIG. 1 on the main memory 12 with a memory address as a symbol. Indicates the expanded state of memory.
FIG. 4 is a part of a diagram for explaining a process in which data input to the conversion device shown in FIG. 1 is converted; The memory expansion in FIG. 5 is the same as the memory expansion shown in FIG. 3, but in order to facilitate understanding of the explanation, the memory expansion of all the basic words described in the basic word dictionary 14 shown in FIG. 2 is shown. ing.
FIG. 5 is a part of a diagram for explaining a process in which data input to the conversion device shown in FIG. 1 is converted; The memory expansion in FIG. 5 is the same as the memory expansion shown in FIG. 3. However, in order to facilitate understanding of the explanation, all the basic words in the basic word dictionary 14 shown in FIG. It is shown.
FIG. 6 is a diagram for explaining a difference between word comparison by symbolization of the present invention and conventional character string comparison;
[Explanation of symbols]
10 Processing engine
12 Main memory
14 basic word dictionary
16 Nuclear name dictionary
18 Full name dictionary

Claims (7)

Lexical including a plurality of words, the device for converting the other data including the preset identification information to the target represented by the plurality of words,
A basic word dictionary in which a plurality of words are registered in advance, a key word selected as a key among the words registered in the basic word dictionary , and a set of words in which the key word is combined with other words ; storage means for storing the identification information previously set to a subject set of words representing the pre Me registered name dictionary in association with the,
And a processing engine that converts the lexical including the plurality of words into another data including the pre-Symbol identification information,
The processing engine is
Storing the basic word dictionary in a memory;
Break down the lexical terms to be converted into words,
The notes that re individual words that the decomposition in the basic dictionary that has been stored in the to obtain each memory address stored, acquires a combination of a memory address representing the lexical of the conversion target,
Extracts a word that is an the key word to the name dictionary of individual words that solution before Symbol min,
Among the word sets registered in the name dictionary, for each word set including at least the extracted word, individual words constituting the word set are stored in the basic word dictionary stored in the memory. It is respectively obtains the memory address is, by replacing each memory address individual words and the acquired constituting a set of words including words that at least the extraction, the words with words that at least the extraction A name dictionary obtained by converting a set of words into a combination of memory addresses representing the set of words is stored in a memory;
The memory is registered in the name dictionary stored, among the combinations of the memory address representing the set of words including words that the extracted, the individual memory addresses in combination with the memory address, the conversion It selects a combination of memory address to be the same as either of the individual memory addresses in combination with a memory address representing the target lexical,
An apparatus for converting the word to be converted into another data including the identification information registered in the name dictionary in association with the selected combination of memory addresses . The processing engine uses, for all word pairs registered in the name dictionary, memory addresses at which individual words constituting the word set are stored in the basic word dictionary stored in the memory. Each name is acquired and each individual word is replaced with the acquired memory address, thereby storing a name dictionary obtained by converting each set of words into a combination of memory addresses representing the set of words. The apparatus of claim 1. When the processing engine stores the name dictionary in a memory, the processing engine stores the basic word dictionary stored in the memory as the key word in the name dictionary among the words registered in the basic word dictionary. A memory address in which a set of words in which the word is combined with another word as the key word in the name dictionary stored in the memory is added to the registered word, and the decomposition is performed. 3. The word to which the memory address is added in the basic word dictionary stored in the memory among individual words is extracted as a word registered as the key word in the name dictionary. apparatus. The processing engine is configured to configure each word set in the basic word dictionary stored in the memory only for a word set including the extracted word among the word sets registered in the name dictionary. Each of the memory addresses in which the words are stored are obtained, and the extracted words are included by substituting each of the words constituting the set of words including the extracted words with the obtained memory addresses. Among the name dictionaries converted into a combination of memory addresses representing the word set only for the word set, the word set including the extracted word and converted into the memory address combination, and the word set The apparatus according to claim 1, wherein the identification information associated and registered in the name dictionary is stored in a memory. The name dictionary stored in the storage means is pre-registered in association with the key word, a set of words obtained by combining the key word with another word, and a code set for the set of words. Nuclear name dictionary, a code registered in the nuclear name dictionary, another word further combined with a set of words associated with the code, and another word further combined with the set of words associated with the code Is composed of a full name dictionary registered in advance in association with identification information set in advance for the target represented by a set of words
The processing engine is
Among the word sets registered in the nuclear name dictionary, for each word set including at least the extracted word, individual words constituting the word set in the basic word dictionary stored in the memory are Each of the stored memory addresses is acquired, and each of the words constituting the word set including at least the extracted word is replaced with each of the acquired memory addresses, thereby including at least the extracted word. Storing in the memory a nuclear name dictionary obtained by converting a set of words into a combination of memory addresses representing the set of words;
Of the other words registered in the full name dictionary, the basic word stored in the memory for the other word associated with a code set in a set of words including at least the extracted word A memory address where the another word is stored in the dictionary is acquired, and the other word associated with a code set in the set of words including at least the extracted word is set to the acquired memory address. Store the replaced full name dictionary in memory,
Of the memory address combinations that are registered in the nuclear name dictionary stored in the memory and that represent the set of words including the extracted word, individual memory addresses in the combination of the memory addresses are After selecting a memory address combination that is the same as any of the individual memory addresses in the memory address combination representing the lexical to be converted,
Extracting the code registered in the nuclear name dictionary in association with the selected memory address combination,
Of the memory addresses representing the other words registered in association with the extracted code in the full name dictionary stored in the memory, the memory address represents the lexical word to be converted. Selecting a memory address that is the same as any of the remaining memory addresses excluding the selected memory address combination from the combination of:
The lexical phrase to be converted is converted to another data including the identification information registered in the full name dictionary in association with the selected memory address. apparatus. The name dictionary stored in the storage means includes a plurality of words that are part of a word set registered in the name dictionary and in which some of the words constituting the word set are different and the objects to be expressed are the same. The apparatus according to any one of claims 1 to 5, wherein a set is associated with the same information as the identification information. In a method for converting a lexical phrase including a plurality of words into another data including identification information set in advance for an object represented by the plurality of words,
A basic word dictionary in which a plurality of words are registered in advance, a key word selected as a key among the words registered in the basic word dictionary, and a set of words in which the key word is combined with other words; By means of a computer comprising storage means for storing a name dictionary registered in advance in association with the identification information set in advance on the target represented by the set of words,
Storing the basic word dictionary in a memory;
Break down the words to be converted into words,
By obtaining a memory address at which each of the decomposed individual words is stored in the basic word dictionary stored in the memory, a combination of memory addresses representing the lexical terms to be converted is obtained,
While extracting the word registered as the key word in the name dictionary among the decomposed individual words,
Among the word sets registered in the name dictionary, for each word set including at least the extracted word, individual words constituting the word set are stored in the basic word dictionary stored in the memory. Each of the stored memory addresses and replacing each individual word constituting the set of words including at least the extracted word with the acquired memory address, thereby including at least the extracted word A name dictionary obtained by converting a set of words into a combination of memory addresses representing the set of words is stored in a memory;
Of the memory address combinations that are registered in the name dictionary stored in the memory and that represent the set of words including the extracted word, individual memory addresses in the memory address combination are Select a memory address combination that is the same as any of the individual memory addresses in the memory address combination that represents the lexical of interest,
A method of performing a process of converting the lexical object to be converted into another data including the identification information registered in the name dictionary in association with the selected combination of memory addresses.

Images