KR20010098277A - Merchandise recommendation apparatus and method for electronic transaction - Google Patents

Abstract

The present invention relates to a product recommendation device and method for electronic commerce, and in particular, a number of intelligent agents based on the expert knowledge of the product when the purchaser who does not have the expertise is difficult to determine the product easily due to the characteristics of the purchased product The objective is to analyze and recommend the best products and sellers that meet the buyer's requirements or preferences from various viewpoints. For this purpose, the present invention provides a buyer by searching the database and knowledge base for products that meet the purchasing conditions from different perspectives such as practical, performance, aesthetic, and consumer's recommendation. The first to fourth product recommendation agents 210 to 240, a shopping mall search agent 250 for presenting a shopping mall site suitable for a purchase condition by searching the Internet shopping mall provided with the recommended products provided above, and the agents Coordinating unit 260 for enabling communication with each other and the shopping mall site searched in real time by the shopping mall search agent 250 are provided in order of being the closest and cheapest in the purchase area to be connected to the corresponding shopping mall site if desired by the buyer. It consists of a user interface unit 270 to be. Therefore, the present invention allows the buyer to select one of the products by presenting the results of analysis from different perspectives to the buyer based on the conditions of the desired product from the buyer in the e-commerce.

Description

MERCHANDISE RECOMMENDATION APPARATUS AND METHOD FOR ELECTRONIC TRANSACTION}

The present invention relates to electronic commerce, and more particularly, to an apparatus and method for recommending a product for electronic commerce.

As e-commerce using communication networks is becoming more common, buyers can purchase goods from sellers all over the world, and sellers can also sell goods to buyers around the world.

For example, in the case of electronic commerce using the Internet, buyers can easily provide and view product information through a web browser using Hyper-Text Mark-up Language (; HTML) or Virtual Reality Modeling Language (; VRML). Check the product information and purchase the desired product.

Therefore, as the electronic commerce using the Internet becomes more common, both buyers and sellers have a great choice.

However, in e-commerce using the Internet, it is almost impossible for a buyer to navigate and negotiate a number of suppliers around the world and to purchase goods or to process inquiries about products from numerous buyers around the world. Do.

In other words, if a buyer purchases a communication product, he or she needs to visit the vast sea of information called the Internet without any help, and the time and economic cost spent on it will be enormous.

In particular, it is almost impossible for a general shopper to search numerous sales sites and products on the Internet in order to purchase the least expensive product in an internet shopping mall.

Therefore, as a service to relieve the efforts of the buyer in recent years, a comparison shopping service (Comparison Shopping Service) has been proposed.

In the comparative purchase service, if a buyer enters a desired model name or name of a product, the shopping mall that sells the product is searched to display the price of the product.

In addition, services that recommend products and retailers that fit buyers' preferences and automate purchases and sales are presented to make it easier to purchase products. Computer software for these services is usually called an e-commerce user interface agent. .

The current agent technology in the field of user interface manages the purchase trend of buyers with a user profile to recommend products similar to those previously purchased. Currently, the US Internet bookstores 'Amazon', 'CD NOW', etc. It is widely used as a one-to-one service.

This recommendation engine is mainly a form of recommending a product or service by analyzing a form of customer consumption.

However, the conventional comparative purchase service is also possible only if the buyer knows exactly the model name of the desired item.

That is, in the case of the conventional comparative purchase service, even if there is a new product or a product having improved performance, the purchaser does not know the model name, and there is a disadvantage in that the comparison purchase service cannot be used.

In addition, the conventional agent technology simply analyzes the purchase trend of the buyer when the buyer wants to purchase a specific product, and recommends a specific product. Due to this, there is still a disadvantage in that it is difficult for the buyer to purchase the optimal product.

Accordingly, the present invention analyzes the optimal product and the seller that meets the buyer's requirements from various viewpoints based on expert knowledge of the product using a plurality of intelligent user interface agents to improve the disadvantages of the prior art and recommends it to the buyer. An object of the present invention is to provide an apparatus and method for recommending products for electronic commerce.

1 is a block diagram of a network for an embodiment of the present invention.

2 is a block diagram of an apparatus for an embodiment of the present invention.

3 is a block diagram of an agent when using a rule-based knowledge base and an inference engine in an embodiment of the invention.

4 is a block diagram of an agent when using neural networks in an embodiment of the invention.

Figure 5 is an exemplary view of a screen for the selection of product conditions and the display of recommended products in an embodiment of the present invention.

Figure 6 is an exemplary view of a screen for the final selection of goods in the embodiment of the present invention.

7 is an exemplary view showing a list of shopping mall sites for a product finally selected in the embodiment of the present invention.

8 is an operational flowchart for an embodiment of the present invention.

9 is a flowchart illustrating an operation using a rule-based knowledge base and an inference engine in an embodiment of the present invention.

10 is a flowchart illustrating operations in the case of using a neural network in an embodiment of the present invention.

Explanation of symbols on the main parts of the drawings

210 to 240: product recommendation agent 250: shopping mall search agent

260: adjustment unit 270: user interface unit

The present invention provides a device for providing an e-commerce service to any buyer on the Internet in order to achieve the above object, the practical point of view such as price, performance point of view, aesthetic point of view, evaluation point of view by recommendation of other consumers, etc. A shopping mall that collects information on shopping malls, such as price and location, by searching a plurality of product recommendation agents that search for and recommend products that meet the purchase conditions from different perspectives, and an internet shopping mall including the recommended products in real time. And a user interface unit which provides a shopping mall site searched by the shopping mall search agent in real time in the order of being nearest to the purchase area and in the cheapest order, and connects to the corresponding shopping mall site desired by the buyer.

The plurality of product recommendation agents may be configured to periodically search for product information on the Internet and store it in a knowledge base.

The plurality of product recommendation agents are configured to have a rule-based reasoning engine or neural network to recommend products that are optimal for a product condition desired by a buyer.

In addition, the present invention provides a method for providing an e-commerce service to any buyer on the Internet by providing a plurality of product recommendation agent and shopping mall search agent in order to achieve the above object, when a client (client) is connected to the purchase Requesting input of a region and a type of a desired product, presenting to the client various conditions to be satisfied by the desired product from a plurality of product recommendation agents, and selecting the product conditions as described above. Each time, the plurality of product recommendation agents satisfy the selection conditions from different points of view, such as a practical point of view such as price, performance point of view, appearance (product appearance) point of view, and evaluation based on votes of other consumers. To dynamically search for the best product Presenting a plurality of searched products to a client; when one of the plurality of products presented above is selected, a shopping mall search agent searching a shopping mall site including the selected products in real time; and Presenting to the client in order of the nearest and cheapest shopping mall sites in the purchase area, and connecting one client to the corresponding shopping mall site when one of the shopping mall sites presented above is selected.

Hereinafter, the present invention will be described in detail with reference to the drawings.

In the network for the embodiment of the present invention, as shown in the configuration diagram of FIG. 1, the client-side computer 110, 120 is connected to the product recommendation server 130 through the Internet 140 to use the Web service. Configure to

The standard protocol used by the client-side computer (110) 120 and the product recommendation server 130 is HTTP, the program using the HTTP is a client browser (100, 120) has a web browser and product recommendation server 130 ) Has a web server.

Therefore, when the client-side computer 110, 120 is connected to the product recommendation server 130 through the Internet 140, the product recommendation server 130 provides a service.

2 is a block diagram of a device for recommending a product in an embodiment of the present invention, as shown in FIG. A first product recommendation agent 210 for presenting a product list, and a second product recommendation agent 220 for searching for a product from a performance point of view and storing the product in the knowledge base 221 and presenting a list of products suitable for a purchase condition of a buyer. And a third product recommendation agent 230 that searches for products from an aesthetic point of view, stores them in the third knowledge base 231, and presents a list of products suitable for the purchase conditions of the buyer, and the evaluation values based on the recommendation of other consumers. A fourth product recommendation agent 240 for searching for a product in the fourth knowledge base 241 and presenting a list of products suitable for a purchase condition of a buyer, A shopping mall search agent 250 for real-time searching of a shopping mall having a product and storing it in a fifth knowledge base 251 and presenting a list of shopping mall sites selling recommended products, and the first to fourth product recommendation agents 210 -240) and the agent necessary for agent communication for multiple simultaneous connections between the shopping mall search agent 250 and the Internet 140, the first to fourth product recommendation agents 210 to 240, and the shopping mall search agent 250 Coordinator 260 including an Agent Name Service (ANS), message routing, mediation, and matchmaking between agents, and a residential area when a client-side computer is connected through the Internet. Requesting input of a desired product type, presenting a condition to be satisfied by the product at the time, and analyzing the product recommendation agent 210 to 240 in response to the selection condition described above; The shopping mall site provided by the shopping mall search agent 250 in real time with respect to the purchase decision for the provided recommended product is provided in order of the shopping mall sites that are the closest to the purchase region and are the cheapest. Configures the user interface unit 270 to notify the shopping mall search agent 250 to connect to the selected shopping mall site.

Since the first to fourth product recommendation agents 210 to 240 and the shopping mall search agent 250 are heterogeneous with each other, it is necessary to use an agent communication language, KQML, and define an ontology used between agents. .

Referring to the operation process and the effect of the embodiment of the present invention configured as described above are as follows.

The present invention is an intelligent product recommendation system for recommending an optimal product by applying artificial intelligence technology to the Internet, and belongs to the field of user interface internet agent technology combining the internet and artificial intelligence.

First, in the present invention, a product recommendation agent 210 representing a practical point of view, a product recommendation agent 220 representing a point of importance of performance, a product recommendation agent 230 representing aesthetic point of view, and other consumers are selected. The representative product recommendation agent 240 periodically collects the product information from the set point of view from the Internet 140 and stores it in each of the knowledge bases 211 to 214.

Thereafter, when the client-side computer 110 or 120 is connected through the Internet, the user interface unit 270 displays an input request message of a product type desired to purchase on the screen of the client-side computer 110 or 120.

At this time, when the purchaser views the screen of the client-side computer 110 or 120 and inputs a product type to purchase, the user interface unit 270 may be a knowledge base 211 connected to the buyer's residence area and product recommendation agents 210 to 240. Messages such as product selection conditions, buyer preferences, and the like stored on the screen of the client-side computer 110 or 120 are displayed.

Accordingly, when the buyer selects a condition and a taste that the product to be purchased should satisfy while viewing the screen of the client-side computer 110 or 120, the user interface unit 270 sets the selection condition at that time to the product recommendation agents 210 to 240. Enter a value to dynamically recommend the product that matches the product conditions selected by the buyer.

In this case, when the product recommendation agent 210 to 240 recommends a suitable product using the inference engine or neural network of the expert system, the user interface unit 270 displays this on the screen of the client side computer 110 or 120.

Accordingly, the purchaser selects a product to be purchased from among the products recommended by the respective product recommendation agents 210 to 240 while viewing the screen of the client-side computer 110 or 120.

When the conditions of the product to be purchased above are entered, the product recommendation agent 210 to 240 dynamically recommends a list of products meeting the selection conditions up to that point, so that the buyer can stop entering the conditions at any time and select one of the currently recommended list of products. You can choose.

When the buyer selects a product to purchase, the user interface unit 270 inputs the product to the shopping mall search agent 250, and the shopping mall search agent 250 sells the selected product in response to the purchaser's input. Search in real time on the Internet or from the knowledge base 251 and notify the user interface 270 to present it to the buyer.

In this case, the user interface 270 displays a list of shopping mall sites searched and recommended by the shopping mall search agent 250 on the screen of the client-side computer 110 or 120.

Therefore, when the purchaser selects a shopping mall site where various conditions are considered appropriate while viewing the screen of the client-side computer 110 or 120, the user interface unit 270 notifies the shopping mall search agent 250 of the selected shopping mall site. The shopping mall search agent 250 directly connects the client-side computer 110 or 120 to the shopping mall site so that the buyer can purchase the desired product.

That is, the above operation is performed in the same process as the flowchart of FIG. 8, which will be briefly described with reference to the exemplary diagrams of FIGS. 5 to 7.

In this case, a case where the product which the buyer wants to purchase is a high speed communication network will be described with an example.

When the buyer is connected for e-commerce, a screen such as the example of FIG. 5 is displayed on the screen of the client-side computer 110 or 120 to select a product condition to purchase, and the product recommendation agent 210 to 240 corresponds to the selection. ) Will dynamically display a list of recommended products.

Thereafter, when the buyer selects a product to be purchased among the recommended products, a screen such as the example of FIG. 6 is displayed on the screen of the client-side computer 110 or 120 to determine whether the selection is a final selection. .

Accordingly, when the purchaser finally selects the recommended product on the screen as illustrated in FIG. 6, the shopping mall search agent 250 searches the shopping mall site in real time on the Internet or from the knowledge base 251. Similarly, the shopping mall sites are displayed on the screen in the order in which they are close to the area where the buyer lives (or the area in which the buyer wants to buy), so that the buyer can purchase the desired product.

In addition, although the message for inputting the buyer's region of residence is displayed when the client-side computer 110 or 120 is connected, the buyer may input the region in which the buyer wants to purchase the product.

Meanwhile, the product recommendation agent 210 to 240 and the shopping mall search agent 250 may be implemented in two ways for the search / recommendation of the shopping mall site in response to the search / recommendation of the condition desired by the buyer and the selection of the recommended product at that time. Can be.

One is to use a rule-based knowledge base and a backward-chaining and forward-chaining inference engine.

In this case, the rules are in the form of "IF condition THEN results" and are a widely used tool for the implementation of traditional expert systems.

Here, backward reasoning is a way of starting reasoning on a hypothesis and inferring the facts that support it, and forward inference provides all the facts related to the problem you are trying to reason about, and the inference engine It is a way to find a matching rule by matching the condition part of the rules in the knowledge base, to execute the rule, and to add the conclusion part of the rule as a new fact that is found to proceed with inference.

The other is to use a multi-layer perceptron technique, which is one of neural network models.

When the user's preferences or various conditions are assigned to the input nodes, this method internally calculates the value of each output node and determines the communication product best suited to the user whose product corresponds to the output node with the highest value. It is.

However, since the multi-layer perceptron can be learned, if data such as product performance in the region is added by the service user after the initial learning using the sample data, new data can be easily added to the knowledge base through learning again. .

First, when using a rule-based knowledge base and an inference engine, each agent 210 to 250 in FIG. 2 is connected to the coordinator 260 and the user interface 270, as shown in the block diagram of FIG. The message processing unit 310 converts the transmission / reception message using the TCP / IP protocol, checks the adequacy of the converted message, and passes only valid messages, conditions for product selection, rules for product recommendation, and the Internet 140. Knowledge base 330 for storing a list for each product (or vendor) retrieved from the information storage system; and product selection conditions stored in the knowledge base 330 during the e-commerce process to the message processor 310 and correspond to the provided conditions. Analyze a buyer's input condition input through the message processing unit 310 to find a rule and from the list of products stored in the knowledge base 330 using the rule After determining the most suitable product, each of the inference engine unit 320 is configured to transmit the determined product list to the message processor 310.

The knowledge base 330 is the same as each knowledge base 211 to 215 connected to each agent 210 to 250 in FIG.

Referring to the operation flowchart of FIG. 9, the operation of the technology having the rule-based knowledge base and the inference engine in the above-described embodiment of the present invention will be described as follows.

The product recommendation agent 210 to 240 may connect a message to the user interface unit 270 in order to input a request for input of a residence area or a region where the buyer wants to purchase or a product type to purchase. The data is transmitted and displayed on the screen of the client side computer 110 or 120.

In this case, when the product type data inputted by the purchaser to the client-side computer 110 or 120 is input through the user interface 270, the message processor 310 may determine a product selection condition stored in the knowledge base 330. The data is transmitted to the user interface 270 to be displayed on the screen of the side computer 110 or 120.

Accordingly, when a purchaser inputs a product condition to purchase on the client side computer 110 or 120, the purchaser inputs the product condition to the message processing unit 310 through the user interface 270.

At this time, the inference engine 320 stores the product conditions desired by the buyer in the working memory therein and compares them with the condition parts of the rules stored in the knowledge base 330.

Accordingly, when a rule is found in which a product condition stored in an internal working memory and a condition part of a rule stored in the knowledge base 330 match, the inference engine unit 320 uses the corresponding rule to determine the knowledge base 330. It finds the best product among the items stored in and adds the best product at that time to the internal working memory.

Thereafter, the inference engine unit 320 determines whether a new condition of a product desired by the buyer is input.

If a new condition of a product desired by the buyer is input to the working memory, the inference engine 320 performs the same process as described above to find a rule, and finds the optimal product desired by the buyer using the rule to the internal working memory. Add.

This process is repeatedly performed whenever a new condition of a product desired by the buyer is input into the working memory.

Thereafter, when there is no input of a new condition in the working memory, the inference engine 320 transmits the product list stored in the working memory to the message processor 310 to display to the buyer.

In this case, the message processing unit 310 converts the commodity message recommended by the inference engine unit 320 and transmits the converted content to the user interface unit 270 to display on the screen of the client-side computer 110 or 120. .

Therefore, when the purchaser selects a product to be purchased while viewing the screen of the client-side computer 110 or 120, the message processor 310 reads the details of the product selected by the buyer and transmits the details to the user interface unit 270 to transmit the client-side computer ( 110 or 120), and asks whether the recommended product is finally selected.

Thereafter, when the purchaser finally selects a product to be purchased while viewing the screen of the client-side computer 110 or 120, the user interface unit 270 transmits the selected product name to the shopping mall search agent 250.

In this case, the shopping mall search agent 250 searches for a shopping mall site having a product desired by a buyer in real time on the Internet through the adjusting unit 260 and stores the shopping mall site in the working memory of the inference engine unit 320.

In addition, the inference engine unit 320 stores the product name finally selected by the purchaser, the living area of the buyer (or an area desired to purchase), etc. in an internal working memory and compares the condition part of the rule stored in the knowledge base 330. We will find the corresponding rule that matches the result.

Thereafter, the inference engine unit 320 infers the shopping mall sites stored in the working memory using the rules found above in order of being closest to the buyer's desired area and in the lowest price, and a message of the list of inferred shopping mall sites at that time. The processor 310 transmits the data to the user interface 270.

Accordingly, when the buyer selects a shopping mall site from which the shopping mall site is desired to purchase a product among the shopping mall sites displayed on the screen of the client-side computer 110 or 120, the shopping mall search agent 250 automatically connects the buyer to the shopping mall site. Make the product available for purchase.

In the above description, an operation of finally recommending the shopping mall site through the inference engine 320 after searching the shopping mall site in real time when the product to be purchased is finally selected, is described in the knowledge base 330. It can also be configured to search and provide to buyers.

In addition, in the case of using the multi-layer perceptron of the neural network according to the embodiment of the present invention, each agent 210 to 250 of FIG. 2 has an adjustment unit 260 and a user interface unit 270 as shown in the block diagram of FIG. It is connected to and inputs the buyer's input condition to the input node and calculates it by using the multi-layer perceptron theory to determine the best product for the product corresponding to the node with the largest output node value, and then the message list of the determined product. Each consists of a neural network 420 transmitted to the processing unit 410.

The neural network 420 is composed of a multi-layered perceptron consisting of an input layer 421, a hidden layer 422, and an output layer 423, and all knowledge for product selection is stored in a weighted form of the perceptron.

The operation of the technique using the neural network in the embodiment of the present invention configured as described above will be described with reference to the operation flowchart of FIG.

The product recommendation agent 210 to 240 may connect a message to the user interface unit 270 to request the input of the region where the message processor 410 wants to purchase or the region in which the purchaser wants to purchase and the type of the product to be purchased. The data is transmitted and displayed on the screen of the client side computer 110 or 120.

In this case, when the product type data inputted by the purchaser to the client-side computer 110 or 120 is input through the user interface unit 270, the message processing unit 410 may apply the product selection condition stored in the database to the client-side computer 110. Or it is transmitted to the user interface unit 270 to be displayed on the screen (120).

Accordingly, when a purchaser inputs a product condition to purchase on the client side computer 110 or 120, the purchaser inputs a product condition to the message processing unit 410 through the user interface unit 270, and the message processing unit 410 is a neural network 420. To a real value between '0' and '1'.

In this case, the neural network 420 calculates a product condition desired by the purchaser by inputting it into a multi-layer perceptron composed of the input layer 421, the hidden layer 422, and the output layer 423. The product corresponding to the output node of the value is determined as the optimal product and transmitted to the message processor 410.

This process is repeated every time the buyer enters a new condition of the desired product.

Accordingly, if the message processing unit 410 transmits to the product user interface unit 270 recommended by the product recommendation agents 210 to 240, the user interface unit 270 may recommend the product recommended by the product recommendation agents 210 to 240. The list is displayed on the screen of the client-side computer 110 or 120.

Therefore, when the buyer selects a product to purchase while looking at the screen of the client-side computer (110 or 120), the message processing unit 410 reads the details of the product selected by the buyer from a separate database and transmits it to the user interface unit 270. It is displayed on the screen of the client side computer 110 or 120 to ask whether the recommended product is finally selected.

Thereafter, when the purchaser finally selects a product to be purchased while viewing the screen of the client-side computer 110 or 120, the user interface unit 270 transmits the selected product name to the shopping mall search agent 250.

In this case, the shopping mall search agent 250 searches for a shopping mall site having a product desired by a buyer in real time on the Internet through the message processing unit 410, and stores the data in a separate database.

In addition, the neural network 420 calculates the product name selected by the buyer, the buyer's residence area (or a desired area of purchase), etc. using the multi-layer perceptron, and the closest and most expensive price in the buyer's desired area based on the calculation result. The list of shopping mall sites is transmitted to the user interface unit 270 through the message processing unit 410 in a low order.

Accordingly, when the buyer selects a shopping mall site from which the shopping mall site is desired to purchase a product among the shopping mall sites displayed on the screen of the client-side computer 110 or 120, the shopping mall search agent 250 automatically connects the buyer to the shopping mall site. Make the product available for purchase.

As described in detail above, the present invention does not recommend products by analyzing past purchase trends of buyers, but rather from various viewpoints such as practical aspects, aesthetic aspects, performance aspects, and evaluation values by other buyer recommendations. It is effective in recommending the buyer to purchase the best products.

That is, the present invention can be effectively used when it is difficult to determine which of a plurality of products to purchase without the help of experts due to the nature of the product.

In addition, the present invention is effective than the conventional comparison purchase service that can compare the price to know the product model in advance because it determines the model of the product by receiving the requirements from the buyer. Therefore, the present invention can solve the problems of the conventional product recommendation system.

Claims (5)

In a device that provides an e-commerce service to any buyer on the Internet, search for a product that satisfies the purchase conditions from different points of view, such as practicality, performance, appearance, and evaluation by other buyers. A multi-product recommendation agent, a shopping mall search agent that searches a shopping mall having a product desired by a consumer, and presents a shopping mall site suitable for a purchase condition, and a product selected at the time of requesting a type of a desired product when a client is connected Presenting a condition to be satisfied, and searching each product from a plurality of product recommendation agents in response to a buyer's selection condition, and presenting each of the recommended product names. Between shopping malls searched in real time And a user interface unit for presenting the data to the shopping mall search agent so as to connect to the shopping mall site if the buyer wants to present the order in the nearest and cheapest order in the purchase area. The method of claim 1, wherein the plurality of product recommendation agents periodically search for product information on the Internet and store the information in a knowledge base, and the shopping mall search agent periodically searches for a shopping mall site on the Internet and store the information in a knowledge base. Product recommendation device for e-commerce. The product recommendation agent of claim 1 or 2, wherein the plurality of product recommendation agents and the shopping mall search agent have a rule-based reasoning engine or a neural network configured to recommend products that are optimal for a product condition desired by a buyer. Product recommendation device for e-commerce, characterized in that. The apparatus of claim 1, wherein the shopping mall search agent is configured to search a shopping mall site having a product desired by a buyer in real time on the Internet or from information stored in a knowledge base. A method of providing an e-commerce service to any buyer on the Internet by using the apparatus of claim 1, comprising: a first step of requesting input of a purchase area and a desired brand name when a client is connected; A second step of presenting to the client various conditions that a product should satisfy from a plurality of product recommendation agents, and each time the product conditions presented above are selected, the plurality of product recommendation agents are practical, performance and appearance. A third step of dynamically searching for an optimal product that satisfies the selection criteria from different perspectives, such as the general aspect (appearance of the product), the evaluation value based on the recommendation of other buyers, and the multiple searched from the different perspectives A fourth step of presenting the goods of the client to the client; A fifth step in which a shopping mall search agent searches for a shopping mall site including the selected product in real time on the Internet or in a knowledge base when the one is selected, and the lowest and lowest shopping mall site order of the shopping mall site searched above; And a sixth step of presenting to the client, and a seventh step of connecting the client to the corresponding shopping mall site when one of the shopping mall sites presented above is selected.