KR100938983B1 - System and method for producting attribute item extraction - Google Patents

Abstract

The present invention relates to a system for extracting evaluation items for each product and an extraction method.

More specifically, the client is connected to the network and transmits the evaluation information for each product, and receives the evaluation contents from the client through the network, extracts the candidate nouns through the correction process, searches the candidate nouns on the network, and then It relates to a product-specific evaluation item extraction system, characterized in that it comprises a server for extracting the evaluation item.

The present invention has the effect of providing accurate product evaluation contents to the user, and also has the effect of shortening the evaluation time searching time according to the user's product.

In addition, by accumulating and storing the evaluation items extracted as described above in a database, and providing them as statistical data to manufacturers manufacturing the corresponding products, the corresponding manufacturers can be utilized as necessary data for the development of products in accordance with consumers' preferences.

Evaluation item, review, network, online shopping mall, home shopping, client

Description

System and method for producting attribute item extraction}

The present invention relates to a system for extracting evaluation items for each product and an extraction method.

More specifically, to receive the evaluation information of the corresponding product from the client through a series of correction process to extract the candidate noun, search the candidate noun on the network and extract the corresponding evaluation items to provide accurate product evaluation content It relates to a system for extracting evaluation items for each product and a method of extraction.

Recently, due to the development of communication means, it has become very important to match the preferences of users who use home shopping or shopping malls as electronic commerce such as home shopping or shopping malls on the Internet or a communication network is activated.

To this end, operators of home shopping or shopping malls want to collect more opinions from users who have purchased products, that is, actual buyers. In addition, before using a home shopping or a shopping mall, the user performs the actual product purchase by referring to the product reviews left by the actual buyers who purchased the corresponding product. This is because the reviews left by real buyers feel more accurate and reliable for users who want to purchase a product.

For these reasons, the evaluation contents of the product are important for the home shopping or shopping mall operator and the user who wants to purchase the product.

If you take a closer look at product reviews, you'll find a review of simple clicks that directly access the product, the quality of the product, the design, and the delivery service. It can be divided into three forms.

The number of simple clicks and the closed product review have some limitations in the information providing part of the product, so that detailed information delivery of the product is difficult.

In the end, open product reviews that leave user experiences or evaluations in text form can actually be helpful to users or operators.

However, in such open reviews, the reviews are listed in the order of registered time, so that the user needs to read the entire registered open reviews in order to grasp specific information desired by the user or the overall mood of actual buyers.

In particular, there is a difficult part in finding information desired by the user, and there is a problem in that it takes a lot of time as the user has to read the entire product review.

The present invention is to solve the problems described above, to provide a user with accurate product evaluation content for the corresponding product, and provides a system for evaluating the evaluation items for each evaluation item according to the product extraction method and extraction method There is a purpose.

In addition, by accumulating and storing the evaluation items extracted as described above in a database, and providing them as statistical data to the manufacturers who manufacture the products, the manufacturers evaluate the products for use as data necessary for the development of products more suited to consumers' preferences. The purpose is to provide an item extraction system and extraction method.

In order to achieve the above object, the product-specific evaluation item extraction system according to the present invention is connected to the network and transmits the evaluation information for each product through the network and receives the evaluation content from the client, through the correction process candidate nouns And extracting the candidate nouns from the network by searching for the candidate nouns, and extracting the evaluation items based on the evaluation items.

According to an aspect of the present invention, there is provided a system for evaluating items for item evaluation, comprising: an evaluation content collection unit for receiving evaluation information for each product from the client via a network, and a candidate noun extraction unit for extracting candidate nouns through a correction process in the evaluation content; And an operation server including an evaluation item extracting unit for searching for extracted nouns on a network and extracting evaluation items corresponding to the searched results.

Evaluation item extraction system for each product according to an aspect of the present invention further comprises a review output unit for generating and displaying a review output screen for displaying the evaluation item extracted from the evaluation item extraction unit and the user's review information according to the evaluation item; Contains an operational server.

In the product-specific evaluation item extraction system according to an aspect of the present invention, a candidate searcher counts the number of searches in the evaluation contents, searches the first search unit and the candidate noun on the network, and transmits the number of search results. And an evaluation item extraction unit including a second search unit for transmitting, a third search unit for searching the candidate names and category names of the corresponding products on a network, and transmitting the number of search results.

According to an aspect of the present invention, the evaluation item extraction system for each product divides the number of search results output from the third search unit by the number of search results output through the second search unit, and then multiplies the number of searches received from the first search unit. And an evaluation item extracting unit further including an operation unit for sorting in descending order and extracting candidate nouns corresponding to the evaluation items.

The evaluation item extraction system for each product according to a feature of the present invention includes a candidate noun extraction unit including a correction unit that performs at least one check of repeated characters, dependent nouns, emoticons, special symbols, brand names, and spacing so as to conform to grammar. .

In order to achieve the above object, the method for extracting the evaluation items for each product according to the present invention is a candidate for extracting candidate nouns through an evaluation content collection step of receiving an evaluation content of a corresponding product from a client and a process of correcting the evaluation content. A noun extracting step and an evaluation item extracting step of performing a target search for the candidate noun on the network and extracting an evaluation item corresponding to the search result.

According to an aspect of the present invention, the method for extracting evaluation items for each product further includes a review output step of displaying the extracted evaluation item and a user's review according to the evaluation item.

In the method for extracting evaluation items for each item according to an aspect of the present invention, the target search includes a first process of counting the number of searches of candidate nouns in the evaluation contents and transmitting the result and searching the candidate nouns on the network for the number of search results. And a third process of transmitting a number of search results by searching the network for the candidate nouns and the category names of the corresponding products, and transmitting at least one of the first process, the second process, and the third process. The method further includes extracting an evaluation item based on the process.

According to an aspect of the present invention, a method for extracting evaluation items for each product is obtained by dividing the number of search results received through the third process by the number of search results received through the second process, and then multiplying the number of searches received through the first process. And extracting the evaluation item for sorting the descending order and extracting candidate nouns corresponding to the descending order.

According to an aspect of the present invention, a method for extracting an evaluation item for each product includes a correction process of removing repeated characters so that the evaluation content conforms to a grammar.

According to an aspect of the present invention, a method for extracting a valuation item for each product further includes a correction process of removing at least one of a noun, an emoticon, a special symbol, and a brand name in the evaluation content.

According to an aspect of the present invention, a method for extracting an evaluation item for each product includes a correction process for confirming a spacing.

As described above, the inventors' evaluation item extraction system and extraction method according to the present invention can provide accurate product evaluation contents to the user, and can shorten the search time of the evaluation contents according to the products.

In addition, as described above, the present inventors 'evaluation item extraction system and extraction method accumulates and stores the extracted evaluation items in a database, and provides them as statistical data to a manufacturer who manufactures the corresponding product, so that the corresponding manufacturer more suitable products to consumers' preferences. It can be used as a material for development.

Examples of the evaluation item extraction system and extraction method for each product according to the present invention can be applied in various ways, hereinafter with reference to the accompanying drawings will be described an embodiment.

1 is a block diagram of an evaluation item extraction system for each product according to an embodiment of the present invention.

As shown in FIG. 1, the evaluation item extraction system for each product includes a client 100, a network 200, and an operation server 300.

The client 100 transmits the product-specific evaluation contents to the operation server 300 via the network 200. In this case, the client 100 may include a server managed by a user PC or a shopping mall operator.

The network 200 may be various types of communication networks including wired and wireless Internet, wide area network (WAN), local area network (LAN), and the like.

The operation server 300 receives the evaluation contents from the client 100 via the network 200, extracts candidate nouns through a correction process, and retrieves the candidate nouns on the network 200. Extract evaluation items. In this case, the operation server 300 includes an evaluation contents collection unit 320, a candidate noun extracting unit 340, an evaluation item extracting unit 360, and a product review output unit 380.

The evaluation contents collection unit 320 receives evaluation information for each product from the client 100 via the network 200.

The candidate noun extractor 340 extracts a candidate noun through a correction process from the evaluation contents, and includes a correction unit 342.

The correction unit 342 corrects the evaluation contents to match the grammar.

The evaluation item extractor 360 searches for the candidate noun on the network 200 and extracts the evaluation item corresponding to the candidate noun. The first search unit 362, the second search unit 364, and the And a three search unit 366.

The first search unit 362 counts the number of times the candidate noun is searched among the evaluation contents, and outputs the result.

The second search unit 364 searches the candidate nouns on the network and outputs the number of search results.

The third search unit 366 searches the network 200 for the candidate noun and the category name of the corresponding product, and outputs the number of search results.

In this case, the evaluation item extractor 360 may further include an operation unit 368 in addition to the first search unit 362, the second search unit 364, and the third search unit 366.

The operation unit 368 divides the number of search results output from the third search unit 366 by the number of search results output from the second search unit 364 and then divides the number of searches output from the first search unit 362. The multiplied values are sorted in descending order to extract candidate nouns corresponding to the evaluation items.

In addition, the review output unit 380 displays the evaluation items extracted from the evaluation item extraction unit and the reviews according to the evaluation items for the user to see.

Hereinafter, an extraction method of the evaluation item extraction system for each product will be described with a flowchart.

2 is a flowchart illustrating a method of extracting a valuation item for each product according to an embodiment of the present invention.

As shown in FIG. 2, the product evaluation contents are received from the client (S410), and after correcting the received evaluation contents, candidate nouns are extracted from the evaluation contents (S420). After the candidate noun is searched through the network, the content of the evaluation item is extracted based on the frequency of the search result (S430), and the evaluation item and a review according to each evaluation item are displayed (S440).

Looking at the method of extracting the evaluation items for each product in more detail, as shown in Figure 3, receives the product evaluation content from the client (S410). As described above, after receiving the product from the actual purchaser who first purchased the product, the user may receive a testimonial or evaluation of the product in text form. In this case, the evaluation content is made through a network, the network may be a communication method of various forms such as wired and wireless Internet or WAN, LAN.

It is checked whether the same character or sentence is repeated among the received evaluation contents (S421). Since the evaluation content is not used only neat sentences as in newspaper articles or books, various problems such as errors in spelling, grammar, and spacing, unnecessary emoticons, special characters, typos, and repetition of the same sentence may occur. .

In particular, problems such as repeated use of the same letter or sentence may be confusing to the user who wants to obtain accurate information about the product through the evaluation content, and thus is an essential element to be removed.

The method devised to solve this problem is as follows.

A text of a certain length is extracted from the sentence of the evaluation contents and set as a pattern. By counting how many times the set pattern is repeated in the evaluation contents, it is determined whether a repeated character is generated, and when the repeated character is generated, the corresponding repeated character is removed from the evaluation contents (S423).

However, even if only repeated characters are removed from the evaluation content as described above, even if the spacing of sentences is not correct in the evaluation content, the accurate evaluation content is not completed.

In order to prevent this, if the repeated character or repeated sentence is removed from the evaluation contents, or if the repeated character or repeated sentence does not occur in the evaluation contents, check and correct the spacing of the evaluation contents (S425).

The present invention provides a method for checking a sentence that is not spaced by using a space ratio average per sentence for a more accurate checking and correction of spacing. Statistically, the average void ratio of news data is 25.99%, and the average void ratio of product evaluation contents is 20.89%. Statistically, when compared with the average space ratio of news data that requires strict spacing rules, the average space ratio of the product evaluation contents is low, but it is more than 20% so that some space is achieved. Can be.

On the basis of the average number of characters in the product evaluation content, which is 22 characters, the spaces with poor spacing are determined at the line ratio of 20%. In this case, when the number of characters in the evaluation content is equal to or larger than 22 characters, the space ratio is determined based on a 20% line, and when the number of characters in the evaluation contents is less than 22 characters, the space ratio standard is reduced every time one character decreases. Decrease spacing by 0.5% to identify bad sentences. In addition, if the total number of characters in the evaluation is less than 6 characters, it should not be discriminated as a spacing sentence without including spacing.

As internet usage becomes more familiar, online shopping malls are used from children to adults, and actual buyers of products also include various age groups.

In particular, they tend to express their thoughts through emoticons or special symbols, especially among young people in their teens and 20s, and this is also applied to product evaluation contents. Therefore, in order to be accurate, contents such as emoji or special symbols including ambiguity must be removed.

Therefore, as described above, after the process of removing the repeated characters and checking the spacing is completed, the candidate noun is output after removing at least one of the dependent noun, the emoticon, the special symbol, and the brand name. (S427).

In addition, the brand name removal method removes various brand names based on the brand dictionary.

The various correction processes described above are just one embodiment for implementing the present invention, and the correction process of the evaluation contents may be generated in various ways.

The candidate nouns output through the correction process as described above are searched for by using a network. In this case, the target search may be performed through various processes.

The first process counts how many times the candidate noun is searched in the product evaluation contents and outputs a search frequency (S431).

In the second process, the candidate noun is searched on the network and the search result number is output (S433).

In the third process, the candidate noun and the category name of the corresponding product are searched on the network, and the number of search results is output (S435). At this time, by searching the candidate noun and the category name of the product on the network together, it is possible to determine a candidate noun with more accurate and professional contents.

The candidate nouns corresponding to the number of search results and the number of search results output through the first, second, and third processes are extracted as product evaluation items.

However, the number of search results output through the first process may include a lot of noise, which is far from accurate evaluation items, and the number of search results output through the third process may be somewhat professional, resulting in a small number of search results. have.

Accordingly, in order to extract the evaluation items more accurately, the number of search results and the number of search results outputted through the first, second, and third processes are calculated with each other, and candidate nouns corresponding to the result values are extracted as evaluation items (S437). .

For example, after dividing the number of search results output through the third process by the number of search results output through the second process, the result value multiplied by the number of search results output through the first process is arranged in descending order, and the result value Candidate nouns corresponding to can be viewed and extracted as final evaluation items. In this case, more accurate and professional evaluation items can be extracted.

Subsequently, the evaluation item and the corresponding product review are output (S439). In other words, the product-specific evaluation items that users want to know and product reviews according to the evaluation items are output together, so that the user's convenience can be increased.

Hereinafter, the product review output screen will be described in detail with reference to FIG. 4.

The review output screen illustrated in FIG. 4 illustrates an example of a review output screen of the digital camera.

As shown in FIG. 4, the evaluation items extracted through the extraction method are displayed, and the product reviews according to each evaluation item are divided into the positive and negative reviews. At this time, in the positive and negative ratings for each evaluation item, the user satisfaction for each evaluation item is displayed in the form of a graph. In this case, the graph used may be represented by a bar graph according to a percentage, as shown in FIG. 4, and may be represented by other application graphs.

Therefore, the user can recognize at a glance the evaluation items according to each evaluation item in addition to the evaluation items which are evaluation criteria for the corresponding product.

The present invention, which provides an evaluation item extraction system and an extraction method for each product, has the advantage of not only providing accurate product evaluation contents to a user, but also shortening a search time of evaluation contents according to a corresponding product.

As described above, the evaluation item extraction system and extraction method for each product according to the present invention have been described. Such a technical configuration of the present invention will be understood by those skilled in the art that the present invention can be implemented in other specific forms without changing the technical spirit or essential features of the present invention.

Therefore, the above-described embodiments are to be understood in all respects as illustrative and not restrictive, and the scope of the present invention is indicated by the following claims rather than the foregoing description, and the meanings of the claims and All changes or modifications derived from the scope and the equivalent concept should be construed as being included in the scope of the present invention.

1 is a block diagram of a product evaluation item extraction system according to an embodiment of the present invention,

2 is a flowchart illustrating a method of extracting a valuation item for each product according to an embodiment of the present invention.

3 is a flowchart illustrating a method of extracting evaluation items for respective products according to an exemplary embodiment of FIG. 2.

4 is a view showing a review output screen according to the product-specific evaluation item extraction system according to an embodiment of the present invention.

*** Explanation of symbols for main parts of drawing ***

100: Client 200: Network

300: operation server 320: evaluation content collection unit

340: candidate noun extraction unit 342: correction unit

360: evaluation item extraction unit 362: first search unit

364: second search unit 366: third search unit

368: calculation unit 380: product review output unit

Claims (13)

A client for a user connected to a network to transmit evaluation information for each product; Receiving evaluation contents from the client through the network, extracting candidate nouns from the received evaluation contents through a correction process, searching the candidate nouns on the network, and extracting evaluation items based on the extracted nouns. It includes an operation server for displaying review information according to the evaluation items, The operation server includes an evaluation contents collection unit for receiving evaluation information for each product from the client via a network; A candidate noun extracting unit for extracting a candidate noun through a correction process in the evaluation contents; An evaluation item extraction unit for searching the extracted candidate nouns on a network and extracting evaluation items in response to the searched result; The operation server may further include a review output unit for generating and displaying a review information screen for displaying the review items extracted from the evaluation item extracting unit and the review information of the user according to the evaluation item, The evaluation item extracting unit includes: a first searching unit which counts the number of searches in the evaluation contents of the candidate nouns and outputs a result; A second search unit for searching the candidate nouns on a network and outputting a number of search results; And a third search unit for searching the candidate noun and the category name of the corresponding product on a network, and outputting a number of search results. delete delete delete The method of claim 1, The evaluation item extractor divides the number of search results output from the third search unit by the number of search results output from the second search unit, and then sorts the value multiplied by the number of search results output from the first search unit in descending order to correspond thereto. Evaluation item extraction system for each product, characterized in that it further comprises a calculation unit for extracting the candidate noun as an evaluation item. The method of claim 5, The candidate noun extracting unit extracts at least one of the repeated characters, dependent nouns, emoticons, special symbols, brand names, and the evaluation unit for each item extraction system characterized in that it comprises a correction unit for performing a space check. An evaluation content collection step of receiving evaluation content of the corresponding product from a client; A candidate noun extraction step of extracting a candidate noun from the received evaluation contents through a correction process; An evaluation item extraction step of performing a subject search on the candidate noun on a network and extracting an evaluation item corresponding to a search result; And And a review output step of displaying the extracted evaluation item and user's review information according to the evaluation item. The target search may include a first step of counting a number of searches of candidate nouns in the evaluation content and outputting a result; A second step of searching for the candidate noun on a network and outputting a number of search results; And a third process of searching for the category names of the candidate nouns and the corresponding products on a network, and outputting the number of search results, based on at least one of the first, second, and third processes. Evaluation method extraction method for each product, characterized in that the process of extracting. delete delete The method of claim 7, wherein In the extracting of the evaluation item, the number of search results output through the third process is divided by the number of search results output through the second process, and then the values multiplied by the number of search results output through the first process are sorted in descending order. The method for extracting the evaluation item for each product, characterized in that it further comprises extracting a candidate noun. The method of claim 10, The correction process is a method for extracting the evaluation item for each product, characterized in that to remove the repeated character so that the evaluation content to match the grammar. The method of claim 10, The method of extracting the evaluation item for each product, characterized in that further comprising the step of removing at least one of the dependent nouns, emoticons, special symbols, brand names in the evaluation content. The method of claim 10, The method of extracting the evaluation item for each product, characterized in that the correction process is to check the spacing.

Images