KR20010092922A - Method for Searching Similar Websites - Google Patents

Abstract

PURPOSE: A method for searching a similar web site is provided to enable a user to search a web site similar to the web site with the user was familiar by enabling a web server to register URLs of web sites and search words and compare a URL the user inputs with an arbitrary search word, thereby enabling the web server to estimate the degree of similarity. CONSTITUTION: A user wants to search a web site similar to a specific web site. The user inputs a URL of the site(10). A web server stores search words including rankings and weight values and makes the first search word group of the search words(12). The user inputs an arbitrary search word to an input window(14). The search word the user inputs is compared with the first search word group. It is judged whether the search term is repeated(16). In case that there's the repeated search word, the search word in the first search word group is removed(18). The search word inputted from the user is inserted into a relevant ranking(20). The second search word group is formed(22). A candidate web site is selected(24). Weight values are given to search words in the second search word group(26).

Description

How to search similar websites {Method for Searching Similar Websites}

The present invention relates to a method for searching a similar website, and more particularly, a user inputs a uniform resource locator (URL) of a specific website and inputs any desired search term along with a ranking, so that the occurrence frequency and weight of the search terms are displayed. It relates to a search method of a similar website that presents a search result reflecting.

The Internet connects numerous computers around the world to communicate based on TCP / IP, and countless websites have been established to provide various contents. Because these websites have their own unique information and organization, they can provide useful information and access to a wide variety of content in all areas, including politics, economy, society and culture.

Therefore, the Internet is also referred to as the sea of information, and the use of the Internet well is recognized as a result of personal development and success of a company. Moreover, e-commerce is transforming all existing commerce practices and developing Internet-based financial techniques that are evolving in the direction of all transactions online.

As such, it is not easy to find and access a website that a user wants among numerous websites connected to the Internet. Websites have been formed and provide services to help Internet users find their website. Among them, www.yahoo.com, www.altavista.com, www.lycos.com are famous worldwide. Www.simmani.com, www.naver.com, www.empas.com are widely used.

These search engines search and present related websites based on search terms entered by users in unique ways. For example, a yahoo search engine is run by visiting a website directly, an expert determines an index term, registers it, compares the registered index term with a user-entered search term, and finds a matching website.

In addition, recently developed search engines adopt a natural language search method by inputting a natural language such as "Where can I buy a television?"

However, these search engine methods have their own advantages and disadvantages, and in some cases a particular search engine can find the site you want, but it only presents a large number of lists, and because these lists are associated with too many websites. There are many cases where it does not help. In addition, even if you are looking for a website similar to the website you are familiar with, these conventional search engines do not provide this function. In practice, however, users are familiar with well-known websites and often want to compare similar websites to get the information they want.

On the other hand, the performance of search engines depends on the exact presentation of the desired website address, so even if the number of search results is too large, the searcher may not be satisfied with the search results. In the case of searching for a site similar to the site, the general search method similar to that of the conventional search engine could not obtain satisfactory results.

Accordingly, it is an object of the present invention to provide a search method through the Internet that effectively searches and presents a web site similar to a web site that is familiar to the user.

1 is a view showing a structure of a registered search word and a weight of a registered website according to the present invention,

2 is a view showing the structure of a user input window of a search website to which the method of the present invention is applied;

3A and 3B are flowcharts showing the flow of a method of searching for a similar website desired by a user according to the method of the present invention.

As described above, an object of the present invention is to provide a method for searching a similar website in a website built on a web server connected to the Internet, the method including: storing a URL of a registered website and a plurality of search terms representing the website; ; A user inputting a URL of a homepage; A user inputting an arbitrary search term; Adding a second search term input by the user to a first search term group of the URL input by the user, and forming a second search term group in which the search term is limited to a predetermined number; Comparing the search terms of the websites registered in the web server with the second search term group, and selecting a predetermined number of search candidate websites in ascending order of search terms; Comparing the search terms registered with the websites of the selected search candidates with the second search term group, and presenting a list of the websites sorted in order of high similarity based on the usage of users as a search result; There is provided a method comprising a.

The method consists of a search site built on a server connected to the Internet. That is, users access the search site of the present invention through the Internet and use a desired search function. Therefore, the search site of the present invention has its own URL, and users can access the search site of the present invention through the Internet using this URL.

It is desirable to use a server with sufficient computing power to connect to the Internet on a high-speed dedicated line to accommodate a large number of users and to provide fast search results.

Websites are registered on the web server, and these registered websites are searched. Therefore, it is desirable that the web server registers and stores all website addresses existing on the Internet as much as possible. The web server of the present invention stores the search terms for the website together with the addresses of these registered websites. The search word may be selected and registered by a person, particularly an expert in a corresponding field. However, more preferably, the search term is selected by tracking the usage status of users who use the website.

That is, there are a number of websites that provide online bookmarks that store a list of websites that Internet users prefer. When storing such bookmarks online, users enter a brief description of the bookmark. For example, the Korea Broadcasting Corporation's "www.kbs.co.kr" will be different for each person who bookmarks it, but kbs, broadcasting, media, Korea, journalist, news, drama, sports, etc. will be added. Thus, the words used by the users for the actual corresponding website are collected in statistics, and the search terms for the corresponding website are determined and registered in the order of the most used words. Since the registered search word is more likely to correspond to the input of the searcher who wants to search the corresponding website, more accurate search can be performed.

1 shows an example of registering a search word while registering a website as described above. Here, each search term is set with a weight. This weight usually consists of the concept of a percentage. If the website "www.kbs.co.kr" is a percentage of Internet users' bookmarks based on their search query rankings and their frequency, the word "kbs" comes in at 35%. If the word was ranked second at 25% and the word "Korea" was ranked third at 10% ......., according to this ranking and percentage, the website "www.kbs.co.kr" is shown in FIG. As shown in the search terms (ksb, broadcasting, Korea, the press, news .....) and the weight (3.5, 2.5, 1, 0.9, 0.7 .....) for the search terms are registered. In FIG. 1, a case in which the sum of weights is 10 is described. However, the sum may be 1 or the sum may be 100.

Similarly, a survey of the frequency found in the description of bookmarks for the website "www.kipo.go.kr" shows that 42% of the word "Patent Office", 23% of the word "invention", and "word" Is 11%, and the word "utility model" has a frequency of 9% ..., the website has a search term and its weight as shown in FIG.

In addition, various search terms may be registered as a result of descriptions of bookmarks of many users, for example, a website called "www.microsoft.com". Here, it can be seen that the weight of the first or second search word is less than the above two examples. A website with a high weight of one or two search terms has a large number of users who recognize the website by the word, and this type of website is easily accessible by the first or second search term. You can search for. On the contrary, a website having a small difference in weight between search terms is that many users recognize the website in various forms, and thus the weight of the search term has a more important meaning.

As such, the web server of the present invention stores URLs of websites, search terms of the URLs, and weights of the search terms at the same time. There are generally two or more search terms in one URL, and in general, several or more search terms are registered with weights.

As shown in Figure 2, the URLs of the websites, their search words and weights are stored in the form of a database (DB) in the web server of the present invention. In addition, the search website according to the method of the present invention built in the web server is provided with a plurality of URL input window and a search term input window for the user to input the search word arbitrarily. In addition, each arbitrary search word input window is provided with a window for specifying a ranking, and an up button and a down button are installed so that the user can determine the ranking of the search word by himself.

In FIG. 2, five input windows are shown to input five search terms, and a ranking is set to 3 as an initial value. However, the search term input window may be provided more or less as needed. The initial value of can also be set differently.

A user who wishes to search for a website similar to that for a particular website first enters the URL of that particular website. In addition, a plurality of arbitrary search terms may be input by one or two or more, and a desired rank may be designated for each search term.

For this number of searched websites, the candidate websites should be reranked in the order that is closest to the particular website entered by the user.

To this end, a new search term group is formed by combining registered search terms (first search term group) of a URL input by the user and arbitrary search terms (first to fifth search terms) input by the user. The new query group is formed by first inserting search terms into the ranking of the first query group by respecting any search terms entered by the user and the ranking, because the search terms of the first search query group are ranked in the order of high weight. do. After the insertion, the search terms in the ranking naturally move to the next rank, and the weight is fixed to the corresponding rank regardless of the search word inserted or moved.

Since the new search word group formed by combining the first search word group of the registered URL and any search word input by the user is large in number, the number of search terms needs to be limited to a certain number. In other words, the search terms below 10 or 15 ranks are very low in weight and thus have little effect on the results of the search.

Accordingly, the search term group of the second group is formed by limiting the ranking to about 10, 15, or 20.

The second search word group thus formed includes 10 to 20 search terms, and compared with the registered search terms of all websites registered in the web server based on the search terms, and the search terms are returned in order of increasing order. Select candidate websites to be presented. The number of candidate websites listed at this time can be limited to a certain number, such as 50 to 200.

Since the selected candidate websites are simply determined in ascending order of search terms that match the second search term group, the similarity does not necessarily match the similar websites desired by the actual user. Therefore, the similarity is measured by reflecting the usage status of searchers and general users, and the candidate websites should be reordered in the order of higher similarity.

To this end, the search terms of the second search term group and the candidate websites are compared to separate the matching search terms, and the weighted value of the second search term group and the registered search terms of the candidate websites are compared with the separated search terms. Multiply. The sum obtained by adding the multiplication results to each matched search term is the score of the candidate website. Therefore, scores are calculated in this manner for each candidate website, and the scores are sorted in the order of high score, and presented to the user as a search result.

At this time, the weight given to each search word of the second search word group is determined by one of the following three methods.

First, a first search term group is a search term that a web server has for a URL input by a user and has a weight as itself. Therefore, when the user inputs any of the search terms entered into the first group search word in the above-described manner, the search terms are changed in rank, but the weights for each rank are maintained without changing the rank, so that the weight for each rank in the second search term group is maintained. Shall be.

In the second method, the web server has search terms and weights of the search terms for all registered websites. Therefore, the arithmetic average of the weights for each of the search terms is calculated, and these average values are applied to all the second search term groups with the general weights of the ranks.

The first method is to reflect the weights recognized by the users for the website as it is, but if the weight variation by rank is severe according to the uniqueness of the website, applying the weight as it is may cause a distortion of the weight. The second method generally applies fixed weights to the search term ranking of search terms. Unlike the first method, it does not reflect the specificity of the corresponding website.

The third method is to proportionally average the average value obtained by averaging the intrinsic weight of the website obtained by the first method and the weight according to the ranking of search terms of the websites generally obtained by the second method. In this case, the proportional average means a value obtained by dividing two weights by two, or by respecting one of the weights more, for example, by multiplying the intrinsic weight by two, adding the average weight, and dividing by three. This third method reflects both the specificity of the website and the ranking of general search terms, which is most preferable for the method of the present invention. In this case, the question of which weight factor to respect for the proportional average is applied can be experimentally determined by comprehensively evaluating the actual search results.

As described above, weighting is applied to the second search term group in any one of three methods, and then the same search terms are extracted by comparing the search terms of the candidate websites and the second group search terms. For the extracted search terms, a value obtained by multiplying the weight given by the second search term group and the weight given by the candidate website is obtained. The weights of the same search terms are multiplied, and the sum of all the search terms is the score of the corresponding website. Therefore, by rearranging candidate websites in order of high score, the results of searching similar websites are presented.

The method of the present invention is described with reference to FIGS. 3A and 3B. First, the user wishes to search for a website similar to the specific website he wants. Thus, the user enters the URL (10).

In the present invention, since the web server stores the search terms including the rank and the weight for all possible websites based on the description of the online bookmark, the registered search terms are the first search term group (12).

The user inputs an arbitrary search word which is considered to be important with the URL in the input window, and also inputs the ranking of the search word (14).

The search term input by the user is compared with the first search term group to determine whether there is a duplicate search term (16). If there are no duplicate search terms, arbitrary search terms for the first group search term group are inserted in the ranking input by the user, and the search terms in the corresponding ranking are naturally shifted to the next rank by the insertion.

If there is a duplicate search word, it is unreasonable to doublely weight the same search word and calculate a score. Therefore, the search word input by the user is inserted into the corresponding ranking and the search word originally owned by the first search word group is deleted. This avoids duplication of search terms (18, 20).

When the search word group is formed in this way, the lower rank search terms are deleted because the weight is low and does not significantly affect the search results. Accordingly, the second search word group is limited to about 10 to 20 search terms in order of high weight (22).

Candidate websites are selected in order of the number of matching search terms by comparing the search terms of all websites registered in the web server based on the second search term group. The number of candidate websites is limited to 50 to 200, preferably 100 (24).

The second search term group has the ranking of the search term only and has no weight, but weights the second search terms in any one of the three methods described above (26).

When weighted, the search terms of the second search term group and the candidate websites are compared to find the same search terms (28).

For the same search word, a value obtained by multiplying the weight of the second search word group by the weight of the candidate website is obtained. The value obtained by adding up the values obtained by multiplying the weights of both of the search terms thus obtained is the score of the website (30).

Thus, the candidate websites are presented as a result of similar websites in accordance with the method of the present invention by rearranging them in order of high score.

The search method of the present invention will be described by way of example.

First, if the user wants to search a website similar to "www.microsoft.com", the user accesses the website of the present invention, enters "www.microsoft.com" in the URL input box, and the user himself Suppose you entered Linux (Rank 1), Dinosaur (Rank 4), OS (Rank 7), and President (Rank 9) as an arbitrary search term and ran the search. Accordingly, the first search term group registered in the web server and the new search term group added with the arbitrary term entered by the user are as follows.

The first group of keywords: ① Windows (2.2), ② Bill Gates (2.1), ③ OS (1.9), ④ Software (1.8), ⑤ Litigation (1.1), ⑥ Monopoly ((0.9), ⑦ Linux (0.8), ⑧ Venture (0.7), ⑨ Icon (0.6), ⑩ United States (0.5), ⑪ Thought Speed (0.4), ⑫ Rich (0.3), ⑬ Dinosaur (0.2), Dream (0.1) .........

New query groups: ① Linux (2.2) , ② Windows (2.1), ③ Bill Gates (1.9), ④ Dinosaur (1.8) , ⑤ Software (1.1), ⑥ Litigation ((0.9), ⑦ OS (0.8) , ⑧ Exclusive (0.7), ⑨ President (0.6), ⑩ Venture (0.5), ⑪ Icon (0.4), ⑫ United States (0.3), ⑬ Thought Speed (0.2), ⑭ Rich (0.1) ⑮ Dream (0.09) ..... .....

As shown here, any search word entered by the user is inserted into the entered ranking and the search word in the ranking moves to the next rank. If the search word entered by the user exists in the first search word group (Linux, dinosaur, OS), the word is moved to the corresponding rank, and if the search word entered by the user does not exist (President), the word is inserted as the entered rank 9 as it is. In this way, even when the search terms are moved or inserted, the weights registered in the first search term group are not moved at all and are fixed to the ranking (the first weight setting method is applied).

Since the number of new search word groups formed is large, it is limited to a certain number. For example, if the search terms are limited to ten, the second group search term group is as follows.

Second query group: ① Linux, ② Windows, ③ Bill Gates, ④ Dinosaur, ⑤ Software, ⑥ Litigation, ⑦ OS, ⑧ Monopoly, ⑨ President, ⑩ Venture

A candidate website having a large number of search terms matching the search terms is searched with a second search term group limited to 10 search terms, and a predetermined number of candidate websites is selected, for example, 100 in order of increasing number of matching search terms.

Suppose 100 candidate websites were retrieved as follows.

(1) www.xxx.com: ① Linux (2.4), ② Electronic Money (1.9), ③ OS (1.1), ④ Screen (1), ⑤ Download (0.9), ⑥ Monopoly (0.8), ⑦ Shopping Mall (0.7 ), ⑧ DOS (0.6), ⑨ shopping card (0.5), ⑩ credit card (0.4), ⑪ software (0.3), ⑫ venture (0.2), ⑬ Windows (0.1), ⑭ flower (0.09), ⑮ birthday (0.08) ) ....

(2) www.yyy.com: ① Thought speed (3.1), ② Pictures (2.1), ③ Windows (1.9), ④ Courier (1.1), ⑤ Linux (0.9), ⑥ Bill Gates (0.8), ⑦ Postcard ( 0.7), ⑧ lawsuits (0.5), ⑨ auctions (0.4), ⑩ reverse auctions (0.3), ⑪ ancient documents (0.2), ⑫ giveaways (0.1), ⑬ newspapers (0.1), ⑭ magazines (0.1) ...

(3) www.zzz.com: ① Linux (4.2), ② Bill Gates (2.1), ③ Windows (1.9), ④ Paint Shop (0.9), ⑤ Photoshop (0.8), ⑥ Download (0.7), ⑦ Credit Card (0.7), ⑧ Upgrade (0.6), ⑨ Shareware (0.5), ⑩ Freeware (0.5), ⑪ Upload (0.4), ⑫ Game (0.4), ⑬ StarCraft (0.3), ⑭ Three Kingdoms (0.3) ... .....

(4) www .....

.................

100 www .....

Here, a weight is assigned to the second search term group. The weight is applied to the rank-specific intrinsic weight of the first "www.microsoft.com" described above (the same as the rank-specific weight of the first corresponding website).

As a way of evaluating the similarity, the first search terms of www.xxx.com [① Linux (2.4), ② Electronic money (1.9), ③ OS (1.1), ④ Screen (1), ⑤ Download (0.9), ⑥ Exclusive (0.8), ⑦ Shopping Mall (0.7), ⑧ Dos (0.6), ⑨ Shopping Card (0.5), ⑩ Credit Card (0.4), ⑪ Software (0.3), ⑫ Venture (0.2), ⑬ Windows (0.1), ⑭ Flowers (0.09), ⑮ Birthdays (0.08) ....] and the second search term group [① Linux (2.2), ② Windows (2.1), ③ Bill Gates (1.9), ④ Dinosaur (1.8), ⑤ Software (1.1), ⑥ Litigation ((0.9), ⑦ OS (0.8), ⑧ Monopoly (0.7), ⑨ President (0.6), and ⑩ Venture (0.5)] to find common search terms: Linux, OS, Monopoly, The six search terms of software, venture, and Windows matched, and after separating the common search terms, multiplying the weights given in the second group of search terms and the registered weights of www.xxx.com for these terms, Sum of multiplied values The system calculates the score of www.xxx.com and calculates the score of www.xxx.com as follows.

(1) www.xxx.com: Linux (2.4 × 2.2) + OS (1.1 × 0.8) + Proprietary (0.8 × 0.7) + Software (0.3 × 1.1) + Venture (0.2 × 0.5) + Windows (0.1 × 1.1) = 7.26

The second www.yyy.com also matched the second term group with four terms: Windows, Linux, Bill Gates, and Litigation. Therefore, the score of this website is as follows.

(2) www.yyy.com: Windows (1.9 × 2.2) + Linux (0.9 × 2.2) + Bill Gates (0.8 × 1.9) + Litigation (0.5 × 0.9) = 8.13

www.zzz.com matched the second query group with three queries: Linux, Bill Gates, and Windows. Therefore, the score of this website is as follows.

(3) www.zzz.com: Linux (4.2 × 2.4) + Bill Gates (2.1 × 1.9) + Windows (1.9 × 0.1) = 14.26

In this way, the matched search terms for the candidate websites were separated, multiplied by the weight of the website and the weight given by the second query group, and the total score was calculated. As a result, the rank was 14.26, 8.13. , 7.26 ......., the result list of this search would be:

(1) www.zzz.com

(2) www.yyy.com

(3) www.xxx.com

.........

This list is presented to the user who performed the search as a result of the search.

Previously, similar websites were searched by applying the unique weight of each ranking of the first website to the second group query as it is. However, if the second method described above is selected, the weight is the same according to the ranking of the search terms in any search. The fixed weight, for example, has a weight such as 3.2 for the first search word, 2.8 for the second search word, 1.9 for the third search word, and 1.1 ..... for the fourth search word. If the weighted search terms are used to calculate the scores of the candidate websites, the search results of the candidate websites may be different from the results using the first weight.

In addition, if the third weighting method is applied to the second group query but is simply divided by 2 by adding both weights, the first search term is (3.2 + 2.2) /2=2.7, and the second search term is (2.8 + 2.1). ) /2=2.45, the third search term is weighted as (1.9 + 1.9) /2=1.9, and the fourth search term is (1.1 + 1.8) /2=1.45 .... When the score of the candidate website is calculated using the weights thus assigned, another result is output.

In the method of the present invention, the registration of the website uses a method of adding and updating a database by tracking the usage status of users who use the online bookmark, and if necessary, by using the expert group by the web server operator of the present invention. can do.

In addition, as a method of determining the weight of the registered website, the percentage is calculated by accumulating the user's description of the online bookmark. However, in addition to the online bookmark, users executing a search using the search method according to the present invention input a URL. It is preferable that any search word additionally inputted thereon is used as an input and reflected in the weight. By continuously updating the weights of the registered search terms in accordance with the user's usage situation, more accurate search results can be obtained.

The method for searching a similar website according to the present invention searches a website most similar to the website desired by the user and presents the list. The present invention can effectively search for similar websites desired by a user by outputting a result by applying a combination of weights reflecting an evaluation of a search word of users using the web and a search word input by the searcher. In addition, the quality of the search result according to the method of the present invention can be improved by continuously updating the search word weights of the registered websites to reflect the usage forms of the web users.

Although the present invention has been described in detail above based on the preferred embodiments, it should be understood that substitutions, conditions or modifications apparent to those skilled in the art within the spirit of the present invention and the appended claims fall within the scope of the present invention. .

Claims (9)

In a method for searching a similar website from a website built on a web server connected to the Internet, Storing a URL of a registered website and a plurality of search terms representing the website; A user inputting a URL of a homepage; A user inputting an arbitrary search term; Adding a second search term input by the user to a first search term group of the URL input by the user, and forming a second search term group in which the search term is limited to a predetermined number; Comparing the search terms of the websites registered in the web server with the second search term group, and selecting a predetermined number of search candidate websites in ascending order of search terms; Comparing the search terms registered with the websites of the selected search candidates with the second search term group, and presenting a list of the websites sorted in order of high similarity based on the usage of users as a search result; How to include. The method of claim 1, wherein a weight is assigned to each search term of the registered website, and the selection and weight of the search term are determined by evaluating the utilization of users who use the website. 3. The method of claim 2, wherein the utilization of the users is determined based on a description of the user's online bookmarks for the registered website. The method of claim 2, wherein the user inputs an arbitrary search term by assigning a desired rank, and the arbitrary search terms are added to the first search term group in the given rank, and by the addition, the search terms of the first search term group are ranked next. Ranking fluctuation, when the added random search word already exists in the first search word group, the search word that already exists in the first search word group is deleted to avoid duplication of the same search word. The method of claim 4, wherein the weight for each of the search terms of the second search term group is given an average weight for each rank of the registered search terms. The method of claim 4, wherein the weight of each of the search terms of the second search term group maintains a weight unique to each ranking given to the first search term group. The method of claim 4, wherein the weight of each of the search terms of the second search term group is a value obtained by proportionally averaging average weights of the rankings of the registered search terms and inherent weights of the ranks assigned to the first search term group. Characterized in that the weighting. The method of claim 5, wherein the evaluation of the similarity based on the usage of the users comprises: comparing the second search term group with registered search terms of the candidate websites, and extracting the same search term; For each extracted search term, a value obtained by multiplying a weight given by the second search term group with a weight given to a registered search term of the candidate website is obtained for the search term, and obtaining the multiplied value for each extracted search term, and multiplying these multiplied values Wherein the sum of all the sums is the score of the candidate website, and the candidate websites are arranged in order of high score. The method of claim 2, wherein the registered search terms and weights are changed by feeding back any search terms and weights inputted by users executing the search with respect to the registered search terms and weights of the registered website. Way.