KR20020071418A - Information searching system and method thereof - Google Patents

Abstract

PURPOSE: An information retrieval system and method is provided to give a word code to a search word according to a rule, to search for the word code in a database, to separate retrieved information into information identical to a word code of a search word, information having a main component word code identical to a main component word code of a search word, and information having a component word code identical to a component word code of a search word, to weight the retrieved information according to a matching value, and to display the retrieved information according to a weighted order. CONSTITUTION: The method comprises steps of determining if an input search word is a compound noun consisting of more than two words(S8100-S8110), in a case that the input search word is a single word, converting the search word into a corresponding word code, searching for information in a database based on the word code, and retrieving information identical to the search word code or information including word codes most similar to the search word code(S8130-S8140), giving a serial number to the retrieved information(S8150), in a case that the input search word is a compound noun, determining if a subject word is separable from the search word(S8120), in a case that the subject word is separable from the search word, converting the search word into a word code, searching for information in a database based on the word code, and retrieving information identical to the search word code or information including word codes most similar to the search word code(S8160), giving a serial number to the retrieved information(S8170), giving a matching value to the retrieved information according as the retrieved information has a word code identical to the search word code, has a main component word code identical to a main component word code of the search word or a component word code identical to a component word code of the search word, determining a weighting order of the retrieved information based on the matching value, in a case that the search word has more than two words and a subject word is separable from the search word, separating the subject word from general words, giving different matching values according as the retrieved information has the subject word or the general word, and determining a weighting order of the retrieved information based on the matching value.

Description

Information retrieval system and its method {INFORMATION SEARCHING SYSTEM AND METHOD THEREOF}

The present invention relates to an information retrieval method, and more particularly, to a method of retrieving information by using a word code to calculate a match value between a search word and the retrieved information and to rank the retrieved information.

In recent years, information exchange through the Internet has been rapidly increasing, and accordingly, various search engines have been developed to quickly and accurately find desired information on the Internet.

However, current search engines search only the information that matches the input word, so if the Internet users do not know the word that matches the information they are looking for, they may not be able to easily find the desired information. You will not be able to find it quickly and accurately.

In particular, when the number of searched information is large, it is more difficult to determine which information most matches the search word.

Therefore, it is an object of the present invention to provide an information retrieval system and method that can easily find information that most closely matches a search word among desired information found quickly and accurately.

In addition, another object of the present invention is to provide a searched information efficiently by ranking the searched information according to the degree matching the search word.

1 is a block diagram of an information retrieval system according to an embodiment of the present invention.

2 to 6 are flowcharts illustrating a method of ranking the retrieved information according to an embodiment of the present invention.

In order to achieve the above technical problem, the information retrieval method according to the characteristics of the present invention is a method in which all words representing information are classified into a basic word and a compound word, the search term is encoded according to a set rule, and the coded search term is In the method of searching the database based on the same or most matching information to the word code, and ranking the searched information, the information comprising the same information as the word code of the search word, the main constituent word code of the search word The information which has a word code and the information which has a constituent word code of a search word as a constituent word code is distinguished, and each gives a different matching value, and ranks each said retrieved information.

In the processing apparatus, when a search word is input, the search word is composed of two or more words, and when the main word can be distinguished from the search word, the main word and the general word of the search word are distinguished from each other so that the main air and the general word are different from each other. By assigning a matching value, each of the retrieved information is ranked.

Therefore, it is possible to distinguish the information having the same or most matched word code as the word code of the main word of the search word and the information having the component word code that matches the component word code of the general word of the search word. To give a value.

On the other hand, information that has the same main word code as the main word code of the search word but does not have any of the main word codes of the search word, and that the main word code of the search word but does not have the main word code of the query, All information has the same priority.

In addition, it has priority in order of the same information as the search word, information containing the search word, information that is the same as the main word of the search word, and has some word codes of the general word of the search word. However, all of the same information has the same priority, and the next priority has the same order as the main word of the search word or the same information as the general word of the search word.

Therefore, the ranking of the retrieved information is listed in the order of the magnitude of the matching value.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to those skilled in the art.

The present invention is to provide a ranking of the searched results in the concept search using the meaning of the word.

In general, describing a word describes the meaning of the word in detail, and can be encoded according to a unified rule through the meaning of the word. At this time, most of the words can be described as a basic word determined by a certain number, it is possible to generate a word code of the word by encoding the basic word with a code having a certain number of digits. Therefore, a word code is a list of meanings of words as basic word codes.

For patents on methods of retrieving information using the above word codes, reference may be made to the applicant's patent application No. 2000-69722 or the like.

When a basic concept that can describe "world words" is set and a combination of these basic concepts is used to describe "world words", the basic concept here becomes the basic word of the present invention. Therefore, a word code is represented by a combination of codes of basic words, and each basic word code corresponds to one meaning.

In an embodiment of the present invention, basically, all words representing information are divided into a skeleton word, that is, a basic word and a compound word, and each word is encoded into a basic word to generate a corresponding word code.

After all the information is coded and stored according to these coding rules, if you search the information with the word code, you can get the result of searching the information with meaning. .

However, in order to support natural language search, it must be applicable to sentences, and to be used for sentence search, information should be searched considering the role of each word in each sentence.

Since the word code of the present invention conceptually represents a sentence or a word, the most important thing in the sentence is the main word. Therefore, in the present invention, a separate role code is assigned only to the main word.

The analysis of the main words in a sentence can be known through survey analysis, morphological analysis of words and meaning analysis by linking each word, or analysis of the position of words. This method is based on common theory made in linguistics.

A subject word is a word that acts as a subject in a sentence. Of course, if the search word does not form a sentence, but is a phrase composed of several words, the word that is finally modified is the main word.

In addition, the automatic analysis by the program which selects a main word can utilize the conventional various word processor programs produced by this theory. In fact, the theory of the role analysis of each word in a sentence is used in a translation program.

Also, in sentence search, it is not necessary to replace every word in a sentence with a word code. In other words, only some important words such as nouns, adjectives, and verbs in a sentence need to be replaced with word codes. This is because it is much more efficient to search conceptually when searching for information, and even if only essential words are searched in a sentence, the concept search can be sufficiently satisfied.

In the present invention, since the subject is searched by distinguishing the subject from the search word, it is necessary to distinguish the subject. That is, when there is a main word in a sentence, a role code "S (main control)" representing the main word is assigned to the word code of the main word.

An example of creating a word code is: The number of digits of the basic word code constituting the word code is made constant, so that the search comparison on the program can be facilitated. For example, if the word code is "nmamkpo-fstelolor", except for the "n" code in the first place of the noun part of speech, the remaining basic word codes are "ma, mk, po, -f, st, el." , ol, or "are composed of all two digits to facilitate mutual comparison between word codes.

It also locates each component word code within the word code, making it easier to find the most matching information. That is, the basic word codes of the modifiers that function as modifiers are located immediately after the main constituent word codes of the subject words receiving the modifiers, and the basic word codes that are adverb functions are located after "-".

For example, the term "valve" refers to the organ "co," which controls the flow of blood (bl, blood) in the "ha, heart" (-i, in). (or, organ) ”to create the word code“ menor = coblfl-ha ”. As in this word code, the word code that acts as a verb or descriptive function is given a code of“ = ", Identify word codes that act as descriptive functions. Accordingly, when comparing word codes, two digits are simply compared to prevent the distinction between word codes having a descriptive function.

1 shows a structure of an information retrieval system according to an embodiment of the present invention.

As illustrated in FIG. 1, the information retrieval system 10 (hereinafter, referred to as an information retrieval server for convenience of description) according to an embodiment of the present invention may include an input unit 11 for inputting words or sentences corresponding to information to be searched for. A central processing unit 22 for subdividing words or sentences (hereinafter referred to as search terms) input through the input unit 11 into basic words and then encoding them, and searching for corresponding information based on the encoded search terms; A display unit for displaying a search word input through the database 23 and the input unit 11 in which a plurality of pieces of information subdivided into basic words and coded are stored, and a result searched by the central processing unit 22 ( 14).

As shown in FIG. 1, the information retrieval server 10 may be connected to a network (wired / wireless network, future network, etc.), for example, the Internet 20, and may be connected to the information input device 30 through the Internet 20. Accordingly, the apparatus further includes an interface unit 15 for transmitting and receiving data to and from the information input device 30 under the control of the central processing unit 22.

The information retrieval server 10 constructs a database 23 by subdividing and coding a plurality of pieces of information according to a set rule, and from the information input device 30 through the interface unit 15 based on the database 23. Information corresponding to the search word input through the input search word or the input unit 11 is searched and the result is provided to the user's information input device 30 or displayed on the display unit 14.

Accordingly, the database 23 of the information retrieval server 10 includes an operation database 131 for storing data necessary for Internet site operation and system operation, and a word in which a plurality of pieces of information are subdivided and encoded into basic words. Database 132.

In addition, the central processing unit 22 subdivides the site operator 121 operating the site and the system and the plurality of pieces of information input through the input unit 11 based on the data stored in the operation database 131 into basic words. A data processor 122 and a data processor 122 for segmenting and encoding the data input through the input unit 11 or the interface unit 15, that is, a search word into a basic word. The data search unit 123 searches the word database 132 based on the search word processed by the to find information corresponding to the search data.

A computer is used as the information input device 30 that can be connected to the information retrieval server 10 according to an embodiment of the present invention, and other communication devices that may be connected to the Internet 20 may be used.

First, the words or sentences constituting the information in the information retrieval system having such a structure will be described.

In the present invention, the search word is subdivided into basic words, and each basic word is encoded into a code consisting of alphabets or Arabic numerals. Here, the search word may be a single word or may be a phrase, phrase, or sentence consisting of two or more words. The coded search word is called a word code of the search word, and the two-digit code constituting the word code is called a component word code.

The component word codes are further divided into main component word codes and component word codes. The component word codes that play a main role in the word codes are the main component word codes, and the other component word codes are the component word codes. Becomes

For example, in the sentence "Engine technology is becoming more electronic day by day in the 2000s", when coded, "Nmamkpo-fstelolor technology (nkn-iscinan) in the 2000s is becoming more (nel) by day (vbc)" Yes, you can. " Since the subject of this sentence is "description", the role code of the main word "S", is given to the descriptive word code.

Here, “engine” means “mk, make” (-f, from) steam (el, electric) or (or) oil (ol). ), The basic word "ma, machine", and if you select and code only important words, the engine's word code can be represented as "nmamkpo-fstelolor". Here, the first code "n" indicates that the word "engine" is a noun, and after the code representing the part of speech, the code "ma" of the main constituent word "machine", which is modified, is placed, followed by the " "Mk", which is the code of "making", is placed, and "po", which is the code of "power", is placed next to the object. And the basic word code "fstelolor" which functions as an adverb is placed after "-". Each word is represented as a code with two digits, and the word "or" at the end of the word code indicates that the "stelol" has a logical OR connection with each other.

"Technology" can be represented by the basic words "kn, know" in the science (sc, science) and in (in) industry. This can be represented by the word code "nkn-iscinan". Here, "n" at the beginning of the word code indicates that the part of the word "description" is a noun, and "an" at the end of the word code indicates that "scin" has a logical AND connection with each other.

Therefore, when the sentence is changed to a word code, it may be expressed as "nmamkpo-fstelolor technology (nkn-iscinanS) is electronicized (vbc) every day in the year 2000 (nyr)."

The informational statement also states that "Clinton (npr) in the United States is busy (vli) in the White House (nhoofpr-ius)." Let's say At this time, "Clinton" is a proper noun (C), "President" is the subject (S). Therefore, if the sentence is coded, it can be represented by the word code of "us Clinton (C) nprS nhoofpr-ius vli".

As described above, in the present invention, the encoding of a sentence selects only important words having a meaning in the sentence, replaces the word with a word code, and distinguishes a main word. In this case, use the same punctuation mark used in the sentence to distinguish between the sentence and the sentence.

For reference, in the above, since Clinton is a proper noun, a code "C" indicating a proper noun is given, so that the proper noun can be directly used without being converted into a word code. Alternatively, the word Clinton can be given a word code that means "00th President of the United States," or a code given to Clinton itself.

On the other hand, the analysis of the role of words in a sentence can be found through survey analysis, morphological analysis of words and meaning analysis by linking each word, or location analysis of words. This method is a common theory made in linguistics. By

In addition, automatic analysis by a program can utilize various conventional word processor programs produced by this theory. In fact, the theory of the role analysis of each word in a sentence is used in a translation program.

Next, a method of ranking the retrieved information based on a database in which a plurality of pieces of information are coded in this way will be described.

2A-2E illustrate the flow of a method of ranking the retrieved information according to the present invention.

As shown in FIG. 2A, when a search word to be searched through the input unit 11 or the interface unit 15 is input, the data processing unit 122 of the central processing unit 12 may first enter a word of the input search word. It is determined whether the words consist of two or more words (S8100 to S8110).

If the input search word is one word, the data processor 122 changes the search word to a corresponding word code, and then the data search unit 123 searches the word database 132 based on the word code and corresponds to the corresponding word code. Retrieve information. At this time, the information search is to search for information having a word code that is the same as or equal to the word code of the search word (S8130 to S8140).

Searching for information having a word code that is the same as or most frequently matched with a word code of a search word means to search for information having a word code of the search word as it is or having a word code having only a part of the word code of the search word.

If the information is found, the searched information is listed and numbered from 1 to n times, respectively. (S8150)

On the other hand, if the input search word consists of two or more words, it is determined whether the main word can be distinguished from the search word (S8120).

The algorithm for distinguishing the main words is performed by a program determined by the central processing unit 22.

If the search word is made up of sentences, the subject is the main word. If the search word is not made up of sentences, the word at the position where the expression is received is the main word. If the words are not formed in a sentence and the two or three words are arranged, the main word may be selected according to the position, but the main word may not be selected. May ask something on the screen, and may assume that there is no topic in the search word.

For example, if the search term is "nmamkpo-fstelolor technology" (nkn-iscinanS), the main word may be "technology". However, it can also be said to have no subject word. The distinction and selection of these subjects may be ambiguous, in which case they may be based on random probabilistic selection.

If the input search word is two or more words, and the main word can be distinguished, the data processing unit 122 changes the search word to a corresponding word code, and then the data search unit 123 performs a word database based on the word code. 132) to search for corresponding information. At this time, the information search is to search for information having a word code that is the same as or most frequently matches the word code of the search word (S8160).

If the information is found, the searched information is listed and numbered from 1 to n times, respectively. (S8170)

2B is a flowchart for ranking the searched information when there is one search word.

The ranking algorithm is performed in the database 23 in which various kinds of information are stored and in the central processing unit 22 in which the algorithm is programmed.

From the information numbered from 1st to nth, information having n order is selected. (S8200)

Then, it is determined whether the information n is information having the same word code as the word code of the search word. (S8210) If the information has the same word code, the coincidence value Cn of the corresponding information n described above is Cn = 210. (S8220)

For example, if the search word is "engine (nmamkpo-fstelolor)", if the information having the same word code as the word code of the engine, the corresponding value of the information is 210.

If it is not the same information as the word code of the search word, it is determined whether the information has the same main constituent word code as the main constituent word code of the search word (S8230).

If the search term is "engine (nmamkpo-fstelolor)", the main constituent word code is "ma", where ma is the base word code representing the machine. That is, it is determined whether the information has a word code having "ma" as the main constituent word code.

If the information does not have the main constituent word code of the search term, that is, in the example of the above search term, if the information does not have "ma" as the main constituent word code, then it calculates a match value according to the degree to which the constituent word code matches, and Vn. In step S8240, the coincidence value Cn of the information is referred to as Vn.

This section describes how to calculate the match value of a constituent word code. If the information contains all the constituent word codes of the search word, the match value is set to 100. Therefore, if there are k configuration word codes, if one configuration word code matches, the matching value Vn becomes 100 / k. Therefore, if the two constituent word codes match, the coincidence value Vn according to the coincidence of the constituent word codes is doubled to 100 / k. In other words, if there are four constituent word codes and two of them match, the formula for Vn becomes " Vn = 100/4 × 2 ", so that the value of Vn becomes 50.

On the other hand, in the case of information having the main constituent word code of the search word as the main constituent word code, it is determined whether the constituent word code of the search word is present (S8250).

If the information n does not have a constituent word code of the search word, the coincidence value Cn of the information n is Cn = 100 (S8270).

That is, if the search term is "engine (nmamkpo-fstelolor)", the searched information has "ma" as the main constituent word code, but does not have any other constituent word codes "mk, po, st, el, ol". The coincidence value Cn of the corresponding information n is 100.

At this time, in the engine word code, except for the main constituent word code, it is "mk, po, -f, st, el, ol, or". However, since "-f" is a word code meaning "from" and "or" is a word code indicating a logical OR, such a stop word is omitted from the constituent word code. Thus, in the word code of "nmamkpo-fstelolor", the constituent word code is "mk ,, po, st, el, ol".

Further, if the information has a word code that has the main word code of the search word and has the word code of the word of the search word, the matching value Vn of the word of the structure word is obtained according to the degree to which the word of the word of the word matches. At this time, the coincidence value Cn of the n information is as follows.

Cn = 100 + Vn (S8260)

That is, if the search term is "engine (nmamkpo-fstelolor"), the word code of the searched information has "ma" as the main constituent word code, and some of the other constituent word codes "mk ,, po, st, el, ol" Or if it has all, the coincidence value Vn according to the degree of coincidence of the constituent word codes is obtained, and then 100 is added to be the coincidence value Cn of the information n.

With respect to the information finally retrieved in the above manner, when all the matching values are found (S8275), the ranking of the retrieved information is listed according to the size of the matching value Cn. (S8280)

If the rankings are listed in this way, the information is listed in the order that most closely matches the searched information.

That is, the word code that is identical to the word code of the search word, the information having the word code including the word code of the search word, the word code and the main constituent word code of the search word are the same, and the constituent word codes are prioritized in order of the same information. You lose.

Information that has the same main word code as the main word code of the search word but does not have any of the main word codes of the search word, and the constituent word code of the search word as the constituent word code All information has the same priority. This is because the coincidence value is equal to 100 in both cases.

The next priority is information having the main constituent word code of the search word as the main constituent word code, and information having a part of the constituent word code of the search word as the constituent word code.

2C and 2D are flowcharts for ranking the searched information when there are two or more words in a search word and a main word cannot be distinguished from the search word.

Here, when there are two or more search terms, this means determination within the same sentence. For example, if the search word is "engine technology", the information having the same word code as the search word means that the words "engine" and "technology" must be included in the same sentence. That is, when a sentence having a word of "engine" and a sentence having a word of "technology" are different, it is not the same information as a search word. In other words, if you have the words "engine" and "technology" in one sentence A, then the information with the sentence A is the same information as the search term.

First, from the searched information given any order from 1 to n, information having n order is selected, and it is determined whether the selected information has one or more word codes identical to the word code of the search word (S8300 to S8310).

For example, if the number of input words is two, it means information having one to two word codes of the input words. In other words, if the search term is "nmamkpo-fstelolor technology" (nkn-iscinanS), the word code of the searched information has one or all of the word codes of the word code of "engine" and the word code of "technology". If it is.

If the number of word codes equal to the word code of the input word possessed by the information with n order is k, then the coincidence value Cn of the n information becomes Cn = k X 210 (S8320).

Then, it is determined whether one or more pieces of information having the main constituent word code of the search word as the main constituent word code among the word codes except for the same word code as the word code of the search word of the n information (S8330).

For example, if the search term is "nmamkpo-fstelolor technology" (nkn-iscinanS), then in the searched information, in the word code where the word code is not the same as the word code of "engine" and "technology", In the above example, it is determined whether there is more than one word code having "ma" and "kn" as the main constituent word codes.

If n information has one or more major component word codes, and the number of the major component word codes is j, the matching value of n information is as follows.

Cn = Cn + j X 100 (S8340)

That is, the number of coincidences of n information is obtained by adding the coincidence value in step 8320 to the coincidence value in consideration of the number of principal component word codes matching the principal component word code of the search word.

In addition, if there is a constituent word code that matches the constituent word code of the search word, the coincidence degree Vn considering the constituent word code is calculated to calculate a match value. The coincidence value at this time is as follows.

Cn = Cn + Vn (S8345)

That is, the value of Vn is added to the coincidence value obtained in step 8340. In the above, the degree of coincidence in consideration of the coincidence of the constituent word codes is the same as the example of the description of FIG. 2B.

In the example of the above search word, the searched information has "ma" and "Kn" as main constituent word codes, and all or part of "mk, po, st, el, ol, sc, in" as constituent word codes. If you have.

Among the word codes that are not the same as the word code of the search word of the information, if there is no word code having the main constituent word code of the search word as the main constituent word code, the coincidence degree Vn of the constituent word codes is obtained. Therefore, the coincidence value Cn of the information n becomes as follows.

Cn = Cn + Vn (S8350)

That is, the value obtained by adding the coincidence degree Vn obtained in step 8350 to the match value obtained in step 8320 becomes the match value Cn of the information n.

When n information does not have one or more word codes identical to the word code of the search word, a flowchart for ranking the searched information is shown in FIG. 2D.

Determines whether one or more main word codes that are identical to the main word code of a search term are multiplied, multiplied by a value of 100 by the number of matching main word codes, and taking into account the degree of match of the constituent word codes of the search word. The coincidence value of the information n is obtained. If the main search word code of the search word is not provided, the coincidence degree of the search word of the search word is calculated, and then a matching value of n information is obtained. (S8360 to S8380)

Since the processes of steps 8360 to 8380 are the same as those of step 8330 to 8350, detailed descriptions thereof will be omitted.

As described above, all matching values of the searched information given any order from 1 to n are obtained (S8355), and then the ranks of the searched information are listed according to the order of the magnitude of the matching value (Cn) of each information. (S8390)

When the searched information is listed according to the rank of the match value, the searched information is listed in the order that most closely matches the searched information, starting with information having the same word code as the search word.

FIG. 2E is a flowchart for ranking the searched information when there are two or more words in a search word and subject words can be distinguished.

When the main word is selected in the search word, information having a word code that is the same as or most frequently matches the word code of the main word is searched for (S8400).

For example, when a search word is “Nwawofi” (nti-obeenan) of the United States (nusS), if a role code is assigned and coded, it may be represented by a word code of “nwawofiA nti-obeenanA nusS”.

At this time, since the main word is the United States, the main word is to search for information having the word code equal or most identical to the word code " us "

Here, the word code of the first, fi world, wo, war, wa is "nwawofi", and the word code of the period is "time (ti) of (- o) a beginning (be) and (an) end (en) ”to be“ nti-obeenan ”. Of course, the last "an" means "and", and the first single-digit code represents the part of speech of the word.

In another example, if the search term is "motor of a pump in a distillation tower", the word code of the search term is "ntwmk (gs-flq) (lq-fgs) or nma = pomvlqgsor nmamkmv-fpoS". Since the main word is "motor", the word code of the motor is given "S" which is a code representing the main word.

In addition, to help the convenience of description in the present invention, other words except for the main term in the search term will be referred to as "general words". Therefore, if the search term is "motor of a pump in a distillation tower" as described above, the general words are "distillation tower" and "pump".

In this case, step 8400 of FIG. 6 is to search for information having a word code equal to or most identical to "nmamkmv-fpoS" among the search word word codes.

On the other hand, "distillation tower" is "to make liquid (lq) into gas (gas, gs) (to make (mk)) or gas into liquid (tower, tw)", if you make a word code based on this, It becomes "ntwmk (gs-flq) (lq-fgs) or". Of course, in the chemical industry where the "distillation tower" is used, "distillation tower" is the basic process, so in the chemical industry "distillation" is the basic word. Thus, the word code of the basic word "distillation (ds)" in the chemical industry can be used to produce the word code of the "distillation tower". In this case, the word code for "distillation tower" is "cindstw". "Ci" before the word is an area code indicating a chemical industry, and "n" is a noun code.

Thus, there are two word codes of the distillation column: the word code "ntwmk (gs-flq) (lq-fgs) or" made with word meanings and the word code "cindstw" made as words in the chemical industry.

Also, the word meaning of "pump" is a machine (ma) that moves liquid (lq) or gas (gas, gs) with power (machine, ma). "nma = pomvlqgsor" In this word code, the base word code of "po" is placed before the base word code of "mv" because "po" is responsible for modifying "mv". That is, the word "po" and the word "mv" are combined to serve to modify the word "ma".

And the word meaning of "motor" is "make (mk) machine (ma)," which is a movement (mv) with power (po) of electricity (el, etc.). Therefore, the word code of "motor" becomes "nmamkmv-fpo". The word "mv" is placed after the word "mk" in the word because the word "mv" serves as the object of the word "mk" and does not act as a modifier. " A word code that represents a noun.

The searched information is listed, a random number is assigned to each searched information from 1 to n times, and the matching value Cn for the main word of the search word of n information is obtained using the algorithm of FIG. 2B. (S8410 to S8420) )

That is, if the input word is "the motor of the pump in the distillation tower", the information retrieved in step 8400 is to search for information with the main word of the word code that is the same or at least partly match the word code of the motor "nmamkmv-fpo". Therefore, to obtain a coincidence value with respect to the main term of n information, the coincidence value of the word code "nmamkmv-fpo" and the main term of n information is calculated | required. Since the information retrieved in step 8400 is a search for information having a word code that at least partially matches the word code of "nmamkmv-fpo", it is possible to obtain a matching value for the main word.

Since the flowchart of FIG. 2B is an algorithm for obtaining a coincidence value when there is only one search word, the algorithm of FIG. 2B can be applied to a method for obtaining a coincidence value for one main word.

If the match value Cn of the searched n-information word is obtained, and the number of general words is C in the search word, the match value Cn becomes Cn = Cn x C (S8430).

That is, the value obtained by multiplying the match value obtained in step 8420 by the number of general words of the search word becomes the match value in step 8430. In other words, if the match value of the searched n information for the main word is 210 and the general words of the search word are two as described above, the match value of the search word n becomes 420 (210 × 2 = 420) in step 8430.

Then, the word code of the general word of the search word and the word code of the general word of n information are compared to determine whether there is a matching component word code (S8440 to 8850).

For example, if the searched information is "liquid and gas moving motor", the word code of the general term "distillation tower pump" of the search term is compared with the word code "nlq nga nmv" of the general word "liquid and gas movement". This is to determine if there is a matching component word code.

If there is a matching component word code, use the algorithms of Figs. The degree of coincidence Vn between the general word of the search term and the general word of n information is obtained (S8460).

According to the example of the search word, the degree of correspondence between the word code of the general word "liquid and gas movement" of the searched information and the word code of the general word "amplifier tower pump" of the search word is obtained.

Then, the coincidence value Cn of the information n is

Cn = Cn + Vn (S8470)

That is, the value obtained by adding the coincidence value Vn obtained in step 8460 to the coincidence value Cn obtained in step 8430 becomes the final match value Cn of the n information.

If there is no component word code that matches between the general word of the search word and the general word of the searched information, the matching value of the n information becomes the matching value Cn determined in step 8430.

As described above, all the matching values of the searched information for the search word having two or more word codes and the main word are obtained (S8475), and the ranking of the information is listed according to the size of the matching value Cn. (S8480)

Since the larger the match value, the closer the information is to the search word, if the rank of the information is listed according to the size of the match value, the same effect as the order of the information close to the search word is obtained.

According to the above algorithm, the priority of the retrieved information becomes as follows.

Priority in order of the same information as the search term, information containing the search term, information that is the same as the subject term of the search term, and contains some of the word codes of the general term of the query term, followed by only the subject term of the query term or only the general term of the term term. All of the same information has the same priority, and the next priority has the same order of information as the main word of the search word or part of the same information as the general word of the search word.

The invention is susceptible to various modifications and implementations without departing from the scope of the following claims.

As described above, according to an embodiment of the present invention, in a search system and method for searching for information using a concept, it is possible to more easily determine the rank of the searched information by calculating a match value between the search word and the searched information. Provide the effect.

By ranking the information in this way, it is possible to select the same or most matching information from the searched information in the searched information, it is possible to further improve the effectiveness of the search.

Claims (6)

In this way, all words representing information are classified into basic words and compound words, and a search term is encoded according to a set rule, and a database in which a search target to be searched is stored based on the coded search terms is the same as the word code. In the method for searching the most matching information and ranking the retrieved information, Distinguishing the same information as the word code of the search word, information having the main constituent word code of the search word as the main constituent word code, and information having the constituent word code of the search word as the constituent word code and assigning different matching values to each one; And ranking each of the retrieved information using the given match value. In the step of giving a matching value of claim 1, When the word of the input search word consists of two or more, and the main word can be distinguished from the search word, And distinguishing the main word and the general word of the search word, assigning different matching values to the main word and the general word, and ranking the searched information. In the step of giving a matching value of claim 1, Distinguish different match values by distinguishing between information having a word code identical or most identical to a word code of a main word of the search word and information having a component word code matching a component word code of a general word of the search word. An information retrieval method characterized by the provision. In the step of giving a matching value of claim 1, Information that has the same main word code as the main word code of the search word but does not have any of the main word codes of the search word, and does not have the main word code of the search word but constitutes the word code of the search word The information retrieval method, characterized in that the priorities are all equal to each other. In the step of giving the coincidence value of claim 1, Priorities in the order of the same information as the search word, information containing the search word, information that is the same as the main word of the search word and has a part of a word code of a general word of the search word. The same information is the same as the main word or only the general word of the search word, and the same priority is given. The next priority has the same priority as the main word of the search word or the same information as the general word of the search word. Information retrieval method. The method according to any one of claims 1 to 6, And order the rank of the searched information according to the order of the magnitude of the matching value.