KR20020030545A - Automatic answer and search method - based on artificial intelligence and natural languane process technology - for natural and sentencial questions. - Google Patents

Abstract

PURPOSE: A method for automatically providing a solution and a search for a natural language query based on the AI(Artificial Intelligence) and the natural language processing technology is provided to recognize the natural language query and generate an accurate response corresponding to the natural language query through a knowledge database. CONSTITUTION: An AI response search site(300) constructs a web site database(400) by searching the web sites connecting to an Internet network and connects with a user(100) through a company providing the both-direction communication or a relay tool(200). If the user carries out the natural language search or query in the AI response search web site, the AI web site converts the natural language search or query into a computer language by searching a Korean analysis database1(500). A process providing the natural language response extracts the answer by searching an encyclopedia database(700) with the converted query and sends the answer to the user. The process providing the related site sends the result to the user by searching a web site database(500) with the converted query.

Description

Automatic answer and search method-based on artificial intelligence and natural languane process technology-for natural and sentencial questions.}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for providing automatic answers and search for natural sentence sentences, and more particularly, to a method for accurately and quickly finding desired information in a website search engine.

As the Internet grows, it's often difficult to know where to get the information you need. In this case, an Internet search site is used. Such Internet search sites include Yahoo (www.yahoo.com), Lycos (www.lycos.com), and Hanmir (www.hanmir.com).

Internet Search Sites displays a list of websites that are most closely related to the user's key words. In addition, the user may search for a website that is listed in the category by searching for the desired category in the directory provided by the search site without inputting a subject.

An example of results retrieved from an existing search site is shown in FIG. As shown in FIG. 1, websites related to a specific word are displayed in a predetermined order with respect to the subject word having a specific word. According to the conventional search method, if only a word is searched, a reasonable result is provided. However, the search ability is limited.

First, the homonym can't be distinguished, resulting in erratic results. For example, the eye and the snow or the abdomen and the pear are not distinguished as illustrated in FIG. 2.

You cannot search for the second sentence type. For example, if you want to know "When King Sejong was born?", You should put "Sejong and When and Birth" into the search box using the existing search method. none. Recently, there are search sites that represent sentence-type search engines. For example, Empas (www.empas.co.kr) and Naver (www.naver.co.kr). However, they are not based on natural language processing technology by artificial intelligence, but simply recognize words in sentences and extract them in order of sites where the most similar number of words are found in comparison with the existing website database. It is merely to omit the operator process as provided, the search results are not significantly different from the existing word search sites as illustrated in FIG. 5 illustrates a search and answer result when a query of "When was King Sejong born?" Was entered in the site of the present invention.

Third, it is inevitable to escape manual work in building a search site. Each site has its own rules regarding the order of sorting and listing the search results on Internet search sites. However, if you look at the results of many search sites, it may be that the site that you are looking for is not very relevant or has nothing to do with it, or that the site you are looking for is very far behind, and it actually takes a long time to reach the site that most users want. there was. In order to solve this problem, many existing search sites use a method of preparing and displaying a list that is considered most appropriate for a user's input of a lot of keywords. However, this method is inconvenient because it is not only limited to the key words that users input a lot, but also such a list must be prepared in advance on the corresponding search site. There is also a burden of constantly updating these lists, as the Internet is a changing area.

Fourth, there is a limit to finding and showing the information you want only on Internet sites. Originally, the starting point of the existing Internet search sites started from a collection of personal favorite sites, and the current search sites are a form of corporateization and enlargement of these collections of favorites. Therefore, there is a problem that the collection of these Internet sites itself is not a database optimized for searching for desired information from the start. Even if there is a site with the right answer for the question you want, you have to work on finding the answer in that site, so you cannot overcome the inefficiency in information retrieval and also have the information and type of information you want. The site containing may not exist.

As a result, the limitations of existing search sites can be roughly derived from two big problems, which is a problem in the search method that the word-centered search cannot overcome the limitations rather than the first sentence. The reason is that when people think alone or talk to others, they use sentences and not words. For example, you think, "What did I do yesterday?" And not "I and yesterday and what." The second problem is the search database source itself, which is not a well-organized knowledge database, but a simple bundle of Internet sites, no matter how large that bundle is. From these two major problems, all the limitations of existing search sites emerge.

It is an object of the present invention to provide a method for recognizing natural language sentence queries, generating accurate answers corresponding thereto through an organized knowledge database, and providing such answers to users.

Another object of the present invention is to provide a directory display ranking method for recognizing natural sentence sentences and showing the exact sites corresponding thereto in order.

Another object of the present invention is to provide such a precise search and answer function to build a center of all gathering of human knowledge so that people can be connected anywhere, anytime, through any medium such as personal computer, telephone or mobile communication and interactive communication as well as internet. By making it available, it aims to provide a way to enjoy free access to information, acquisition rights and usage rights.

It is another object of the present invention to provide a method for achieving a more accurate, efficient, fast service and cost reduction by processing the guidance and response system provided by a government office or a business in a manual or hierarchical manner using the present invention. The purpose.

1 is an exemplary view of a search result after inputting a main word as a word in a conventional website search engine.

2 is an exemplary diagram of a search result after inputting a main word as a homonym word in an existing website search engine.

FIG. 3 is an exemplary view of a search result when inputting key words using an operator as an existing website search engine "Sejong and when and birth".

4 is an exemplary view of a search result after inputting "When was King Sejong born?" In the existing sentence type search engine.

Figure 5 is an exemplary view of the search results and answers that came out after entering the query "When was King Sejong born?" In the search site of the present invention.

6 is a system configuration diagram of a search engine to which the present invention is applied.

7 is a flowchart showing a process of extracting an optimal answer for a natural language sentence queried by a user from an Internet search and answer website according to the present invention.

8 is a key detailed flowchart of searching for an optimal answer with a computer language query.

9 is a flowchart illustrating a process of extracting an optimal site for a natural language sentence queried by a user from an Internet search website according to the present invention.

10 is a key detailed flowchart for searching for an optimal site with a query in computer language.

11 is a procedure diagram illustrating a method in which the artificial intelligence search answering method is used in combination with other communication means in addition to the Internet.

12 is a flowchart illustrating a procedure performed between an internet user, an artificial intelligence search and answer website that provides the present invention, and an operator and an advertiser such as other Internet and mobile communication.

FIG. 13 is a system configuration diagram showing how an artificial intelligence search and answer method is constructed and used in an automatic response and guidance system such as a government office or a business.

In order to achieve this purpose, an Internet user accesses a website and inputs a natural sentence query. The input natural language query is converted into a programming language through a Korean language analyzer linked with a Korean language analysis server. The query in this programming language is used to extract the best answer through the best answer interlocked with the first encyclopedia database server. The search results are converted back to answers in natural language through the Korean language analyzer and then displayed to the user to provide sentence type query results. Second, the optimal site is extracted through the optimal site searcher linked with the internet site database server. Such a search result is displayed to the user to provide a sentence type search result.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 6 is a schematic diagram of a system of a search engine to which the present invention is applied. FIG. 6 shows a web database, a Korean analyzer database 1 for converting natural language into computer language, and a Korean analyzer for converting computer salmon into natural language. This is a schematic diagram showing the connection with Database 2, the encyclopedia database.

In FIG. 6, the artificial intelligence answer search site 300 constructs a website database 400 by searching for websites connected to the Internet network.

Such a search site is connected to a user by a business or relay means 200 which performs two-way communication. When the user 100 enters the AI answer search site 300 and makes a sentence search or query, the AI answer search site 300 searches the Korean analysis database 1 500 and converts it into a computer language.

Here, the process of providing natural language answers is to search the encyclopedia database 700 with the query converted to computer language, extract the answer, and then go through the Korean analysis database 2 (600) which once again converts the computer language to natural language answers. The answer is sent to the user (100). On the other hand, the process of providing the relevant site searches the website database 500 with the query converted to computer language and transmits the result to the user 100.

7 is a system structural diagram showing a process of extracting an optimal answer for a natural language sentence queried by a user from an Internet search website according to the present invention.

When the user inputs a query in a natural language sentence (71), the artificial intelligence answer search server 300 searches the Korean analysis database 1 (500) and converts it into a computer language form using a Korean analyzer (72). The AI answer search server 300 searches the encyclopedia database 400 with the computer language query to search for an optimal answer (73). With this answer, the artificial intelligence answer search server 300 searches the Korean analysis database 2 (600) to generate a natural language answer (74). The generated natural language answer is displayed to the user (75).

Here, we explain how to find the best answer with a query in natural language.

8 shows a key flow chart for searching for an optimal answer with a computer language query.

When the user inputs a query in a natural language sentence (81), the user inputs an answer value (82) and initializes all the input values.

The next step is to analyze and reform 84 the AI. Details of this step are given in 84-1 through 84-8. The first step is to convert these questions into a form that can be handled (84-1). Next, convert the original question into a standard question (84-2), then extract the keywords (84-3), then parse the sentence (84-4), and then specially constructed In operation 84-5, the forms of words and sentences are analyzed using the Korean analysis database 1 500. After the above process, if the query word is a word (84-6), if it is a phrase rather than a complete sentence (84-7), and if it is a sentence form (84-8), it is processed separately. The next step is step 85 of AI data retrieval and data classification. This search and classification results in the eviction 86 of the answers from the encyclopedia database 700. The next step is the AI answer inference 87. Inferring the answer evicted from the encyclopedia database 700 or if no answer is found in the encyclopedia database 700, analyze other related sites or databases 400 and infer the nearest answer from the analysis. . The next step is to generate an answer in the form of a natural sentence 88. This will work in conjunction with the Korean analysis database 2 (600). The next step is to finally output 89 to the user.

The above description only describes the schematic part of the flowchart of the program implemented by the database design and the program design work.

9 is a system structure diagram showing a process of extracting an optimal site for a natural language sentence queried by a user from an Internet search website of the present invention.

When the user inputs a query in a natural language sentence (91), the artificial intelligence search response server 300 searches the Korean analysis database 1 (500) and converts it into a computer language form using a Korean analyzer (92). With the computer language query as described above, the artificial intelligence search response server 300 searches the website database 400 to search for the optimal site (93). The retrieved optimal sites are extracted 94 in order of best matching the query and displayed 95 to the user.

Here, we explain how to search the best site using computer language query.

Fig. 10 shows a key flow chart of searching for the best site with a query in computer language.

When the user inputs a query in a natural language sentence (101), this answer value is input (102) and the step 103 of initializing all the input values are passed. The next step is to analyze and reform 104 the AI. Details of this step are given in 104-1 through 104-8. The first step is to convert these questions into a form that can handle them (104-1). Next is a step 104-2 of converting the original question into a question of a standard form, a step 104-3 of extracting keywords, a step 104-4 of sentence syntax analysis, and a specially constructed Analyzing the shapes of words and sentences using the Korean analysis database 1 (500) (104-5). After the above process, if the query is a word (104-6), if it is a phrase (104-7) rather than a complete sentence (104-7), and if it is a sentence (104-8), it is processed separately. The next step is step 105 of AI data retrieval and data classification. This search and classification extracts the answers 106 from the website database 400. The next step is the AI answer inference 107. Inferring the nearest sites from the website database. The next step is to finally output 108 to the user.

The above description only describes the schematic part of the flowchart of the program implemented by the database design and the program design work.

11 is a procedure diagram illustrating a method in which the artificial intelligence answer search method is used in combination with other communication means in addition to the Internet. The artificial intelligence search and answer server 300 is connected to the PC user 100, the mobile communication user 800, the telephone user 900, etc. by the business or relay means 200 that makes two-way communication. Through this structure, AI search service is available not only to PC and Internet users but also to all users who have two-way communication means. Since the connection and information processing methods of the mobile communication and the telephone service provider are the same as the connection and information processing between other general Internet servers and the communication service providers, detailed description thereof will be omitted.

FIG. 12 is a flowchart illustrating a procedure performed between an internet user and an artificial intelligence search and answer website providing the present invention, and other Internet and mobile communication providers and advertisers, and the artificial intelligence search and answer internet and mobile communication. It can be combined with a business model, such as receiving a fee on a fee and providing it as a fee or as a means of advertising and membership, recorded on a storage medium such as a CD and sold in the form of program software. First, the user 100 recognizes the necessity of use and makes an application for use in the artificial intelligence search answer site 300 (①) and the artificial intelligence search answer site 300 provides a service thereto (②). We pay the fee for use (③). Next, the user 100 requests an artificial intelligence search and answer service to a communication provider such as an internet site and a mobile communication that he or she is using (④). The communication service provider 400 such as mobile communication requests the service connection to the artificial intelligence search and answer site 300 (⑤), thereby providing a connection service (⑥) and receives a fee (⑦). The advertiser 400 recognizes the advertising effect (⑧) and requests the advertisement placement (⑨) to control the advertising accordingly (⑩) to collect advertising charges (⑪).

Fig. 13 shows how an artificial intelligence search and answer method is constructed and used in an automatic response and guidance system such as a public office or a business. Internet user 100, mobile communication user 800, telephone user 900, directly connected to the public office or business (1100) requiring a response and guidance desired by the complaint person (1000)-for example Auto answer and search method is located in the middle of internet, telecommunication company, telephone company, information desk and so on. Since the connection and information processing method of the public offices and businesses that provide and apply the solution to such government offices and businesses is the same as the connection and information processing of other general system builders and public offices and businesses, detailed description thereof will be omitted.

Although described in detail using the preferred embodiment of the present invention, the scope of the present invention is not limited to the specific embodiment, it should be interpreted by the appended claims. In addition, those of ordinary skill in the art should understand that many modifications and variations are possible without departing from the scope of the present invention.

As described above, according to the present invention, it is possible to search a website in sentence form, thereby avoiding false search results by homonyms or synonyms, and to search with more complicated ideas or contents. It is possible to obtain a search result closer to the information desired by the user.

In addition, according to the present invention, it is possible to exclude the use of the operator used in the existing word search because it can be searched by natural sentence rather than search the website search by simple words, thereby increasing the convenience of website search. .

In addition, according to the present invention, instead of simply extracting sites containing related information from a bundle of websites and searching for desired information in the site again, a direct answer can be directly found using a systematically constructed knowledge database. By doing so, it is possible to save time and increase accuracy in information retrieval.

In addition, the present invention has the effect of ensuring the free information acquisition and use rights by allowing a deep and direct retrieval of such information anytime, anywhere through a PC, a terminal, a mobile communication, two-way communication, and the like.

In addition, the present invention has the effect of ensuring a profit model, such as paid for the fee on the Internet and mobile communication, can be used as advertising and membership means, and recorded on a storage medium such as a CD and sold in the form of program software. There is.

In addition, the present invention has the effect of achieving a more accurate, efficient and quick service and cost savings by processing the guidance and response system provided by a public office or a business in a manual or hierarchical manner using the present invention.

Claims (5)

The technique of providing a solution by analyzing a sentence in a natural language is literally a technical method of providing a direct answer to a query in a natural language. Switching to a program language through a Korean language analyzer linked with a natural language database 1 that converts input of a query sentence in natural language into a computer language; Extracting the optimal answer by searching the encyclopedia database with the query converted to the programming language with the query converted to the programming language, Generating the answers in natural language sentences through the Korean analyzer database 2 which converts the extracted answers into computer language natural answers; Displaying the generated answer to the user sentence-type natural language answer generating method characterized in that it comprises a In a technical method of providing a related site by recognizing a sentence in natural language, The step of converting the query sentence input into natural language into the computer language through the Korean language analyzer linked with the natural language database 1 and converting the program language into a program language. Extracting them, A sentence type natural language search method comprising the steps of displaying the extracted optimal sites in the order that best matches the query. In various applications of the artificial intelligence search and answer site of claim 1, Intelligent search answer service method of any time and place, characterized in that it comprises the step of connecting the information processing process with other two-way communication means, such as another Internet site or mobile communication In the business motel of the artificial intelligence search and answer site of claim 1 or 2, Connecting the information processing process with other two-way communication means such as another internet site or mobile communication to provide the same service and collect the fee accordingly; Increasing the user's need to obtain advertisements from advertisers to control the advertisements and collect the advertisement charges, and to make the service available offline through software storage media such as CDs, Profit structure forming method characterized in that it comprises the step of obtaining In the business motel of the artificial intelligence search and answer site of claim 1 or 2, the automatic answering and retrieval method based on the artificial intelligence and natural language processing of their response and guidance system in response to requests from public offices and businesses requiring automatic response and guidance. Reconstructing the database to meet the needs of the customer, and creating a revenue structure by connecting the new system to the information desk for Internet, mobile communication, other telephone service providers or direct visitors. Way.