KR101997578B1 - System and method for artificial intelligence electronic commerce - Google Patents

Abstract

The present invention discloses an electronic commerce system. More specifically, the present invention relates to an artificial intelligence electronic commerce system and method that implements automation by applying artificial intelligence techniques to a company's market research, search field, negotiation and bidding.
According to an embodiment of the present invention, by artificial intelligence automating the supplier market research and search field that takes the most time among the purchasing business performed by companies, and the negotiation and supplier winning step that requires the most fairness and objectivity, In addition to improving business efficiency and increasing the chances of securing excellent suppliers and supply targets through fairness, it is expected to secure reliability through high price competitiveness and transparency.

Description

Artificial intelligence e-commerce system and method {SYSTEM AND METHOD FOR ARTIFICIAL INTELLIGENCE ELECTRONIC COMMERCE}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic commerce system, and more particularly, to an artificial intelligence electronic commerce system and method for implementing automation by applying an artificial intelligence technique to a business such as market research, search field, negotiation, and auction.

Artificial intelligence technology aims to enable computers to communicate with intellectual actions such as recognition and judgment similar to humans, and has a very long history. Since the 1950s, when the first discussion of artificial intelligence was made, artificial intelligence remained at the research level for a long time and did not receive much attention, but in the 1990s, there was a significant attempt in the application of the industry using artificial intelligence.

At that time, rule-based expert systems became mainstream. This expert system is a system that builds and operates a knowledge base that stores expert knowledge, called domain experts, in the system by using rules and methods. need.

Therefore, in order to construct such a knowledge base, an effort was required for an expert system development engineer to collect expert knowledge through interview with the expert or to learn the knowledge in the relevant field and input it in the form of rules and methods.

Accordingly, according to the conventional method, not only can the expert's knowledge be applied to the system completely, but also it cannot be expected to continue to develop. For this reason, the artificial intelligence field based on expert systems has reached the dark age and has reached the present day.

Recently, however, a breakthrough theoretical development has been made in the field of neural networks, and interest in artificial intelligence has increased again. The neural network theory has been mainly used as a back propagation multi layer perceptron method as a field of artificial intelligence.

Conventionally, the multilayer perceptron method is a neural network having one or more hidden layers between one input layer and one output layer, and has been used in relatively simple fields such as image recognition and classification by characteristics. In this case, the neural network is called a multi-layer perceptron, but when a large number of hidden layers are actually applied, an overfit problem occurs and the learning rate is lowered. Thus, one or two hidden layers are generally used. It consists of a shallow neural network, which makes it impossible to learn complex input patterns.

However, by overcoming these limitations, a deep learning neural network has been developed that can combine complex unsupervised learning and supervised learning to learn complex input patterns with multi-layered hidden layers. Techniques for processing input patterns will be developed. Such a deep neural network and an existing rule base system may be appropriately combined to form a desired system.

The rule base system executes the defined matters of the execution unit when the conditions are compared and satisfied. Here, the execution unit may call another rule or perform a command for an action. In addition, forward reasoning to search the target in all directions while executing rules that satisfy the condition and backward reasoning to search for the cause of a certain result are applied, resulting in smooth results such as priority processing and contention resolution strategy. Techniques for doing this may be provided.

On the other hand, the online electronic purchasing system using the Internet is a field of electronic commerce has been introduced and operated by many institutions and companies at home and abroad since the early 2000s.

1 is a diagram schematically illustrating a purchasing process of a conventional online electronic purchasing system.

Referring to Figure 1, the purchase process of the conventional electronic purchasing system is a purchase request step (S10), investigation step (S11), planning step (S12), purchase approval step (S13), purchase notification step (S14), negotiation step ( S15), the successful bidding step (S16), the contracting step (S17), the receipt and inspection step (S18), and the payment step (S19) is divided into 10 steps are generally progressed sequentially.

According to the above-described process, a number of personnel must be involved in each process, thereby increasing the work efficiency due to the heavy work of the personnel. In addition, there is a possibility of violating fairness and transparency, which are important purposes of electronic purchasing, and improvement is required.

The present invention has been made to solve the above-described problems, the present invention is a system of autonomous processing of some of the purchasing process directly handled by a number of persons, the machine learning the results through the machine learning reflected in the process Therefore, there is a problem in providing an AI electronic commerce system and method for improving performance.

In order to solve the above problems, the artificial intelligence electronic commerce system according to an embodiment of the present invention, a purchase receiving unit for receiving a purchase request for goods from a user terminal, and provides one or more purchase templates to the user terminal, An environment setting unit for setting the range of the web crawler by receiving the above web URL, an information collecting unit storing the sale information collected by the web crawler for the web URL, and the sale information input from the user terminal. A request generation unit for generating and posting a request form including an evaluation on the web; and receiving and collecting a plurality of pieces of receipt information on the request from a plurality of other terminals, and matching one or more pieces of receipt information suitable for the set purchase rule. A search unit for generating a and one of the extracted reception information by the user terminal Surface, and a query unit and the result information stored in the knowledge database to provide the result information corresponding to the selected information received may include a learning engine using a machine learning data.

The environment setting unit may include a template module for providing a plurality of purchasing templates including a UI for receiving at least one of a keyword, a price, a unit, an origin, a supplier, a delivery date, and a category for the article, and the data input in the purchase template. It may include a crawler module for providing a web crawler for searching and extracting data in the web URL based on the.

The information collecting unit may convert the collected atypical sales information into standardized data.

The search unit may compare the sale information with the purchase template and transmit a negotiation request message to the corresponding company terminal according to a purchase rule if there is a difference.

In the purchase rule, one or more purchase conditions for a corresponding item purchase and contents of a message corresponding to the purchase condition may be defined.

In addition, an electronic commerce method by an artificial intelligence electronic commerce system according to another embodiment of the present invention for solving the above problems, the step of receiving a purchase request for goods from a user terminal, and the user terminal Providing at least one purchase template, receiving at least one web URL, setting a range of a web crawler, storing sales information collected by the web crawler for the web URL, Generating and posting a request form including an evaluation of inputted sales information on the web, receiving and collecting a plurality of pieces of receipt information on the request from a plurality of other terminals, and matching one or more suitable rules with a set purchase rule. Generating reception information and any one of the extracted reception information by the user terminal; If is selected, the method may include generating result information corresponding to the selected reception information and storing the result information in a knowledge database.

Providing one or more purchase templates to the user terminal, and setting the range of the web crawler by receiving one or more web URLs from the user terminal, includes a keyword, a price, a unit, a place of origin, a supplier, a delivery date, and a category. The method may include providing a plurality of purchase templates including a UI for receiving at least one, and searching and extracting data in the web URL based on data input in the purchase template through the web crawler.

After collecting the sales information by the web crawler, the method may further include converting the collected atypical sales information into standardized data.

The step of extracting one or more pieces of receipt information suitable for comparison with the set purchase rule may include comparing the sale information and the purchase template, and if there is a difference as a result of the comparison, transmitting a negotiation request message to the corresponding company according to the purchase rule. It may include a step.

The purchase rule may define at least one purchase condition for a corresponding item purchase and message contents corresponding to the purchase condition.

According to the artificial intelligence electronic commerce system and method according to an embodiment of the present invention, the supplier market research and exploration field that takes the most time among purchasing business performed by a company, negotiation and supplier bidding that require the most fairness and objectivity By automating the steps by artificial intelligence, it not only increases the opportunity to improve work efficiency and secure excellent suppliers and supply targets through fairness, but also can secure reliability through high price competitiveness and transparency. It works.

1 is a diagram schematically illustrating a purchasing process of a conventional online electronic purchasing system.
2 is a view showing the overall flow of the purchase process of the artificial intelligence electronic commerce system according to an embodiment of the present invention.
3 is a view showing the structure of an artificial intelligence electronic commerce system according to an embodiment of the present invention.
4 is a diagram illustrating a structure of a learning engine of an artificial intelligence electronic commerce system according to an embodiment of the present invention.
5 is a diagram illustrating an artificial intelligence electronic commerce method according to an embodiment of the present invention.
6 is a diagram illustrating the artificial intelligence e-commerce method according to an embodiment of the present invention for each purchase process.
7A and 7B are diagrams illustrating an example of a purchase template setting screen and a purchase rule setting screen used for negotiation in an artificial intelligence electronic commerce system according to an embodiment of the present invention.

Prior to the description, when any part of the specification to "include" or "include" any component, unless otherwise stated, it may further include other components, not to exclude other components. It means to be. In addition, terms such as "unit" and "... module" described in the specification mean a unit for processing at least one function or operation, which means hardware, software, or hardware. And software.

In addition, the term "embodiment" herein is meant to serve as an example, example or illustration, but the subject matter of the invention is not limited by such an example. In addition, other similar terms such as "comprising", "including", and "having" are used, but when used in the claims they are open transition words that do not exclude any additional or other components. It is used generically in a similar manner to the term "comprising" as.

The various techniques described herein may be implemented with hardware or software, or where appropriate, with a combination of both. As used herein, terms such as "Unit", "Module" and "System" likewise refer to computer-related entities, i.e., hardware, combinations of hardware and software, software or Can be treated as equivalent to the software at runtime. For example, a program module may be composed of one component equivalent or a combination of two or more components. In addition, in the present invention, both the application program and the hardware executed in the server or the terminal may be configured in a module unit, and may be recorded in one physical memory or distributed between two or more memories and recording media.

In the artificial intelligence e-commerce system and method according to an embodiment of the present invention, the steps of investigation, planning, announcement, and negotiation, which are directly performed by a person in charge according to the conventional multiple purchasing process (FIG. 1), are based on a knowledge database. By configuring the intelligent e-commerce system to automatically process, it can be reduced to the stage of purchase request, successful bidder review and approval, contract, goods receipt and inspection, payment to enable the operation.

To this end, according to the embodiment of the present invention, when the artificial intelligence electronic commerce system receives a purchase template such as a predetermined purchase request purchase rule and evaluation factor from the user, that is, the user purchases a purchase requirement. When registered in the system, based on the purchase template, the system collects and analyzes the information on the online supply site and searches for the supplier.

In this case, if the search target is found as a search result, the auto negotiation module compares and analyzes the supply conditions and shows the analysis results in order of priority according to the results. By reflecting on the base, it is possible to improve the accuracy of future autonomy based on the knowledge database.

2 is a view showing the overall flow of the purchase process of the artificial intelligence electronic commerce system according to an embodiment of the present invention.

2, in the artificial intelligence e-commerce system according to an embodiment of the present invention, if the person in charge is connected to the system through his terminal and inputs information on the goods to purchase using a purchase template having a certain format (S20), the artificial intelligence electronic commerce system proceeds with the autonomous electronic purchasing step for the article using the purchase template (S21). At this time, the system can be seen as a step (S21-1) to process the agent by referring to the knowledge database (DB5) built therein.

In particular, in steps S21 and S21-1, as a self-purchasing electronic purchasing step, a web crawler collects sales information necessary for a product purchase decision on a corresponding web site according to a set purchase environment, and converts the data into standardized data having a predetermined form. Done.

Here, as the collected data is standardized, the search means enables direct comparison with the purchase template and determines suitability. The system summarizes the appropriate sales information and presents the received information.

After that, if the person in charge reviews the received information presented by the system and approves the contract for a particular company and goods (S22), the system sends the approval to the terminal of the company to receive and inspect the goods (S24). ).

After that, the performance evaluation of the purchase process is performed (S25). In this case, in step S23 to step S25, the person in charge can receive a guidance by referring to the previously built knowledge database (S25-1), thereby increasing the work efficiency of each purchasing process.

Next, the system applies the performance evaluation result to the artificial intelligence technique to perform machine learning (S26), and the learning results are accumulated in the knowledge database DB5 (S26-1).

 According to the steps S26 and S26-1, the artificial intelligence electronic commerce system of the present invention reflects the learning result accumulated in the above-described agency processing step (S21-1), thereby implementing the autonomous purchase function.

The above-described knowledge database may have a hybrid structure in which a production rule of "If ~ then ~ else ~" and a multi-layer neural network model are combined.

The role of the rule in this structure is to be used to describe logical processes, such as to do something about a certain condition. For example, if the type of the product is hiking boots and the price is 100,000 won to 150,000 won, it can be set as the list of purchases.

In addition, the neural network learns a pattern with respect to a value of a factor referenced by a specific rule, and learns that the value of the factor is increased based on the learned information. can do. For example, if you want to deal with "If a hiking product is your favorite in the 30s, put it in the purchase list", you will learn the preference pattern by age for the "hiking shoes product". The preference can be derived by prediction.

The values for these factors are tuned by the neural network learning and used as condition values to draw correct conclusions in the rule part. In addition, the value of the corresponding factor may be changed as a result of the logical rule. For example, "If a product has a purchase proportion of less than 10% in the 30s, reduce the factor value of the preference ratio of the 30s by 5" can be processed.

Due to the characteristics of neural networks, the accuracy of the system increases as the learning progresses for a large number of learning data, and as the amount of information fed back increases, the accuracy improves.

In such a hybrid AI structure, the life cycle of a rule may be changed according to verification of an execution result of the rule. That is, the importance of a rule may change according to a contention resolution strategy between rules, and addition, modification, and deletion of rules may be determined according to this importance. This process can be implemented in a manual manner by a person and in an automatic manner by a system.

Therefore, the artificial intelligence electronic commerce system according to an embodiment of the present invention is to automate the investigation process, information search and negotiation process for the purchase through the artificial intelligence in the existing goods purchase process. In this case, the survey process, the search and the negotiation process is the part that requires the most human resources, the large number of parts are automated by the present system can be expected to save the human and time resources.

In the purchase request step (S20), the purchase template and the negotiation rule (Rule) are used when setting the purchase environment, which can be reused after registration, and can be used by modifying only the changed part. Detailed description of the purchase template and the negotiation rule will be described later.

Hereinafter, with reference to the drawings will be described the structure of an artificial intelligence electronic commerce system according to an embodiment of the present invention.

3 is a view showing the structure of an artificial intelligence electronic commerce system according to an embodiment of the present invention.

Referring to FIG. 3, the AI electronic commerce system 200 according to an embodiment of the present invention may include one or more purchase templates in a purchase reception unit 210 and a user terminal that receive a purchase request for goods from the user terminal 100. And an environment setting unit 220 configured to receive one or more web URLs and set a range of a web crawler, an information collection unit 230 for storing sales information collected by the web crawler for the web URLs, A request generation unit 240 for generating a request including an evaluation of the sales information input from the user terminal 100 and posting it on the web, receiving and collecting a plurality of received information on the request from a plurality of other terminals; When the search unit (S250) generates one or more pieces of receipt information suitable for the set purchase rule, and one of the extracted receipt information is selected by the user terminal, An inquiry unit 260 for providing result information corresponding to the selected reception information may include a learning engine 270 using the result information stored in the knowledge database as machine learning data.

The purchase reception unit 210 receives a purchase request for a predetermined article from the terminal 100 of a user who is a person in charge of a job and thus starts a purchase process.

The environment setting unit 220 provides the user terminal 100 with a purchase template and a rule for proceeding the purchase process and receives and registers a setting value of a web crawler for information search.

To this end, the environment setting unit 220 includes a template module 221 which provides a plurality of purchase templates including a UI for receiving at least one of a keyword, a price, a unit, a country of origin, a supplier, a delivery date, and a category for an article; A crawler module 222 may be provided to provide a web crawler that searches for and extracts data in the web URL based on data input in the purchase template.

The information collecting unit 230 collects web site information of a web URL designated by a web crawler, and serves to format the collected unstructured data. In order to perform this function, the information collecting unit 230 may refer to the machine learning result accumulated in the knowledge database 280.

In addition, the information collecting unit 230 determines whether the atypical data is compatible with the purchase template, and as a result, provides the appropriate sales information to the user terminal 100 and stores the result as sales information. This procedure can be performed repeatedly for the target sales website.

When the request generator 240 checks the sales information and inputs a review result and evaluation of the sales information through its terminal, the request generator 240 generates a request corresponding to this and announces the request on the web.

In addition, the request generation unit 240 may notify the proposal request (RFP) or the quotation request (RFQ) on the web, depending on the setting. As the request is notified, a proposal or a quotation may be received through a terminal from a company that has confirmed the request, and the search unit 250 searches for information included in the purchase template of the corresponding work and according to the degree of matching with the aforementioned proposal or quotation. As a result of the evaluation, one or more pieces of reception information are generated.

If the request is set to not be announced on the web, the request generator 240 compares the collected sales information with the purchase template to determine whether the conditions are met, and if there is a difference, the user terminal and the corresponding company according to the rule. The terminal can be connected to each other via a messenger such as chat, text or e-mail to control the transmission of the message for the negotiation request between the user and the company.

When a response to the message is transmitted from the company, the user can directly contact the company to proceed with the purchase negotiation. The above-described process may be repeatedly performed on the target sales information.

In addition, the search unit 250 provides the generated one or more pieces of receipt information to the user terminal, thereby allowing the user to determine the final successful bidder.

In the present invention, the negotiation procedure is to search the proposals or quotations provided by the company and the purchase rules and compare the contents, extract the predetermined companies that meet the conditions to generate the receipt information and provide it to the user, the user The final winning bidder is selected with reference to the reception information.

Specifically, the negotiation process includes data on the price, unit, country of origin, supplier, delivery date, etc. of proposals and quotations received from the company, and the company's sales, years of establishment, number of employees, etc. for a certain period of time. The search unit 250 compares whether the contents meet the conditions defined in the purchase rule, and extracts several companies and provides them to the user terminal as reception information.

The inquiry unit 260 may provide an inquiry function for the final successful bidder, that is, a result of negotiation and determination, that is, the final successful bidder, to the user terminal 100 regarding the purchases for which negotiation is completed.

When the final successful bidder is selected based on the sales negotiation result information, the learning engine 270 performs machine learning on the selection result and stores the learning result in the knowledge database 280. The knowledge database 280 may be used to help select a more suitable target by using the predictive function of artificial intelligence in the search and negotiation of future sales information.

In addition, the artificial intelligence e-commerce system 200 of the present invention, the machine learning data by the learning engine 270, registration information such as a web URL registered in a web crawler, purchase registration information generated by a purchase template, It may include a knowledge database 280 that stores a variety of information for driving the system, such as sales information and requests.

According to this structure, the artificial intelligence e-commerce system 200 according to an embodiment of the present invention is not directly performing a complex purchase process of 10 steps when the fencer performs a task such as purchasing goods in a company, data collection, The data can be processed through the simpler and more accurate purchasing process by referring to the data learned and extracted through artificial intelligence in request generation.

That is, the present invention collects sales information for various companies through a web crawler, when a user posts a request with reference to the sales information, the companies confirm the request and provide a proposal or a quotation, and the system provides the proposal. Or, it determines whether the contents of the quotation conforms to the purchase rule and summarizes them to a certain number of companies to provide the reception information to the user. Then, the user selects any one company with reference to the received information, characterized in that the purchase process is completed.

In particular, the artificial intelligence e-commerce system 200 according to an embodiment of the present invention performs a learning, evaluation, and prediction through a learning engine based on neural network learning techniques, such as survey processing, search, negotiation processing, etc. The procedure can be automated by the system. Hereinafter, a learning engine according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. Such a learning engine may be implemented in a plurality of program modules and recorded on a recording medium that can be read and written by a computing device, and may be executed in the form of a program process by a processor of an artificial intelligence electronic commerce system.

4 is a diagram illustrating a structure of a learning engine of an artificial intelligence electronic commerce system according to an embodiment of the present invention.

Referring to FIG. 4, the learning engine 270 of the artificial intelligence electronic commerce system of the present invention may include a plurality of processes, and each process may include one or more subprocesses.

Such a process includes a preprocessing process 271 for preprocessing the training data or the evaluation data, a work process 273 for performing the learning, evaluation, and prediction of the preprocessed data, and a process controller for progressing the learning by controlling the work process. 275, a neural network standard interface 277 connected to a user terminal through an information communication network and performing data input / output, a post-processing process 279 for converting the results derived by the work process 273 into the form of data; Knowledge database 280.

When the learning and evaluation process is requested, the learning engine 270 performs machine learning with reference to the learning data and the evaluation data of the knowledge database, and stores the result in the knowledge database again.

In detail, the preprocessing process 271 of the learning engine 270 serves to normalize the sales information collected by the web crawler in various formats such as HTTP and PHP. The preprocessing process 271 may generate and update training data of the knowledge database 280.

The work process 273 is a process of performing the learning functions (a), the evaluation functions (b), and the prediction functions (c), which are the main functions provided by the learning engine 270, and the learning functions (a) are evaluation variables for the data. After selecting the input variables, defining the properties of each variable, defining the network parameters, the optimal answer that can be obtained through data analysis by the learning technique is derived.

In addition, the evaluation function (b) is a function for evaluating whether the answer derived by data analysis is appropriate, and the prediction function (c) is a function for finding a target variable through input variables by data learning.

To implement this function, the work process may refer to learning data, evaluation data, parameters, and neural network data stored in the knowledge database 280.

The process controller 275 performs signal control between each process executed by the learning engine 270, and provides a learning result to the neural network standard interface in response to a request from the neural network standard interface 277.

The neural network standard interface 277 connects the learning engine 270 to an external terminal and a database (not shown) through an information communication network in a wired or wireless manner.

The system manager may access the learning engine 270 through the terminal to change settings related to driving, reset parameters, and add or update learning data and evaluation data.

The post-processing process 279 serves to output data according to the learning result derived by the work process 273 in a report form as valid information by the system administrator.

Hereinafter, an artificial intelligence electronic commerce method according to an embodiment of the present invention will be described with reference to the accompanying drawings.

5 is a diagram illustrating an artificial intelligence electronic commerce method according to an embodiment of the present invention. In the following description, the subject of each step may be the above-mentioned artificial intelligence e-commerce system of FIG. 3 and each component constituting the system, unless otherwise indicated.

Referring to Figure 5, the artificial intelligence electronic commerce method according to an embodiment of the present invention, the step of receiving a purchase request for goods from the user terminal (S100), providing one or more purchase templates to the user terminal, one or more web Setting the range of the web crawler by receiving the URL (S110), storing the sales information collected by the web crawler for the web URL (S120), and evaluating the sales information input from the user terminal. Generating and posting a request form on the web (S130), receiving and collecting a plurality of pieces of receipt information on the request form from a plurality of other terminals, and generating one or more pieces of receipt information suitable by comparing with a set purchase rule ( S140, if one of the extracted acceptance information is selected by the user terminal, generating result information corresponding to the selected acceptance information (S150). In operation S160, the result information may be stored in the knowledge database.

First, in step S100, if a person in charge of purchasing a company, that is, a user, receives a purchase request for an item from an AI electronic commerce system using his terminal, the system receives the request and starts a purchase process.

Next, in step S110 provides a purchase template and rules for the progress of the purchase process to the user terminal, and then stores the purchase template input from the user terminal, and registers the web URL input from the user terminal in the system to the web crawler Set the range.

Next, the sales information is collected by the web crawler, and the result is stored (S120). At this time, the system formalizes the collected unstructured data.

Next, in step S130, it is determined whether the atypical data is compatible with the purchase template, and if appropriate, the evaluation is performed on the sales information, and the result is stored as the sales information. This step S130 may be performed repeatedly for the target sales website.

When the user who confirms the sales information inputs the result of the review and evaluation of the sales information through his terminal, the system may generate a proposal request or a quotation request according to the input items and notify the user on the web.

The proposal request or quotation request may be generated by inputting a result of reviewing the sales information and evaluation contents in a text form on a form provided by the system.

At this time, the request may be announced or not announced according to the setting, if notified according to the step S140 receives a proposal or a quotation as a reply to the request, and search the target based on the purchase template to negotiate the procedure Proceed to generate one or more receipt information.

If the request is not announced at step S130, the collected sales information is compared with the purchase template to determine whether the conditions are met, and if there is a difference, negotiation with the corresponding company through chat, text or e-mail according to the purchase rule is made. Send a message for the request. Accordingly, when a response to the message is transmitted from the corresponding company, the negotiation is continued according to the purchase template conformance as described above.

Subsequently, in step S150, the user selects one of the successful bidders by referring to the reception information generated by the system through the terminal, and the system generates the result information.

In operation S160, when the final successful bidder is selected, the system stores the result information in the knowledge database. This can be used as learning data for the learning engine later.

Hereinafter, each step of the artificial intelligence e-commerce method of the present invention shown in FIG. 4 will be described in more detail by the purchase process with reference to the accompanying drawings.

6 is a diagram illustrating the artificial intelligence e-commerce method according to an embodiment of the present invention for each purchase process.

Referring to FIG. 6, after receiving a purchase request, a user receives a purchase environment setting (S200), and stores and manages a site address to be searched by a web crawler as Web crawler registration information DB1. In addition, a purchase template that is a criterion to purchase is input and stored as purchase registration information DB2. The stored purchase template is then used as reference information when the automatic electronic purchasing system processes the survey process and the sales negotiation by the web crawler.

Next, the system performs the information survey according to the purchase of goods depending on whether or not to conduct the investigation (S201). When the investigation is performed, the information is collected by accessing the site with the URL registered by the web crawler (S202), and the unstructured data collected by crawling the sales site is standardized (S203). The system determines the suitability with the purchase template based on the standardized data with reference to the purchase registration information DB2 (S204 and S205). If it is determined to be suitable, the system evaluates the sales information and registers the result as sales registration. It stores as information DB4 (S206). The stored sales information DB3 may later be used as basic information for creating a proposal request (RFP) or quotation request (RFQ). If it is not suitable at the time of determining the suitability (S205), or if there is more information of another web site, the above process is repeated for the target sales site (S202).

If the survey is not carried out or the survey is completed, the survey results are inquired (S207), and at this time, the results of the previous evaluation at the time of the proposal (proposal request) or RFP (quote request) are generated (S209). At this time, when the announcement (S208), if the proposal or quotation is received after the request is posted on the website, evaluation and analysis in accordance with the criteria of the purchase rules, and then received the receipt information that is the result of extracting the appropriate company sales registration information (DB4) Save as (S210).

If no notification is made, the collected sales information is compared with the purchase template to determine whether the conditions are met, and if there is a difference, a request for negotiation is sent to the corresponding company through a messenger such as chat, text and e-mail according to the purchase rules. . If a response to the negotiation message is received, it is determined whether the negotiation is continued by determining whether the purchase template is matched (S211). The above steps are repeated for the target sales information.

In operation S210, the plurality of pieces of reception information selected by the system are reviewed by the user as a negotiation procedure to select a final successful bidder (S212).

Thereafter, the result of the successful bid according to step S212 may be reflected in the machine learning (S213) and stored as an electronic purchasing knowledge database DB5 and provided as a report. The knowledge database DB5 may be used for the prediction function of the learning engine at the time of searching and negotiating sales information for other goods purchase later (S211), and may contribute to performance improvement in selecting a more suitable company.

Hereinafter, the technical idea of the present invention will be described with reference to the drawings by way of example for providing a screen provided to an artificial intelligence electronic commerce system according to an embodiment of the present invention.

7A and 7B are diagrams illustrating an example of a purchase template setting screen and a purchase rule setting screen used for negotiation in an artificial intelligence electronic commerce system according to an embodiment of the present invention.

Referring to FIG. 7A, after receiving a purchase request from a user terminal, the artificial intelligence electronic commerce system provides a screen for setting a purchase environment, and the environment setting window w100 receives a setting input field for receiving a setting value as a purchase template (w110). And it may include a condition input field (w120) for receiving an item for the item conditions.

As the setting value input to the setting input field w110, as illustrated, there are a keyword, a price, a unit, an origin, a supplier, a delivery date, a category, and a supplier size. In addition, in the condition input field w120, delivery information about an actual item may be input. The crawler of the artificial intelligence e-commerce system of the present invention searches the information on the designated website to collect sales information meeting the conditions of the purchase template, and the collected information is analyzed through an analysis function to determine whether to process an appropriate process. Will be decided.

In addition, referring to FIG. 7B, a purchase rule setting screen is illustrated. In the artificial intelligence electronic commerce system according to the present invention, a purchase rule set when a portion that does not meet the condition for verifying the sales information analyzed by the real collection is generated. Negotiations are conducted by (Rule), and these negotiations are conducted by inquiring negotiations with the company through e-mail or chat. The purchase rule may define one or more purchase conditions for the corresponding item purchase and the content of the message corresponding to the purchase condition, and when the system receives a response from the company, the system may use the AI natural language processing function to match the negotiation contents. To confirm. The purchase rule setting window w200 for this may include a check box w210 for selecting a command for one or more rules (rule1 to rule3).

Many details are set forth in the foregoing description but should be construed as illustrative of preferred embodiments rather than to limit the scope of the invention. Therefore, the invention should not be defined by the described embodiments, but should be defined by the claims and their equivalents.

100: user terminal 200: artificial intelligence electronic commerce system
210: purchase unit 220: environment setting unit
221: template module 222: crawler module
230: information collecting unit 240: request generation unit
250: search unit 260: search unit
270: Learning Engine 280: Knowledge Database

Claims (10)

A purchase reception unit which receives a purchase request for an item from a user terminal;
An environment setting unit configured to provide one or more purchase templates including purchase conditions and evaluation items to the user terminal, and receive one or more web URLs and set a range of a web crawler;
An information collection unit for storing the sales information collected by the web crawler for the web URL and converting the collected atypical sales information into standardized data;
A request generation unit for determining whether the formal data and the purchase template are suitable and storing as sale information, if appropriate, generating a proposal or quotation request including an evaluation of the sale information input from the user terminal and posting it on the web. ;
From a plurality of other terminals, a predetermined company is extracted by comparing the data on the proposal and the company included in the proposal or the quotation according to the confirmation of the proposal or the request for quotation with the set purchase rule, and one or more pieces of the received information. A search unit for generating and providing the user terminal;
An inquiry unit which provides result information corresponding to the selected acceptance information when any one of the extracted reception information is selected by the user terminal; And
A learning engine using the result information stored in the knowledge database as machine learning data,
The environment setting unit,
A template module for providing a plurality of purchasing templates including a UI for receiving at least one of a keyword, a price, a unit, a country of origin, a supplier, a delivery date, and a category for the article; And
A crawler module for providing a web crawler for searching and extracting data in the web URL based on data input in the purchase template;
The search unit,
If there is a difference by comparing the sales information and the purchase template, the negotiation request message is transmitted to the corresponding terminal according to the purchase rule, and if the response is received, the same decision whether to continue the negotiation by determining whether the purchase template conformity,
The purchase rule,
One or more purchase conditions for the purchase of the goods and the contents of the message corresponding to the purchase conditions are defined,
The learning engine,
A preprocessing process of preprocessing the training data or the evaluation data through normalization of the sales information collected by the web crawler;
Learning function to select an evaluation variable and input variable for preprocessed data and define variable attributes and network parameters to derive the answer, evaluation function to evaluate the derived answer is appropriate, and prediction function to find target variable through input variable Working process to perform;
A neural network standard interface connected to an external terminal and a database by wire or wireless through an information communication network;
A process controller performing inter-process signal control and providing a learning result to the neural network standard interface according to a request; And
And a post-processing process for outputting a report on data according to the learning result derived by the work process,
The learning engine performs machine learning on the learning result derived by the work process and the selection result of the final successful bidder and stores it in the knowledge database.
And the result information stored in the knowledge database is used as learning data of the learning engine to improve the accuracy of autonomous performance. delete delete delete delete As an electronic commerce method by an artificial intelligence electronic commerce system,
Receiving a purchase request for an item from a user terminal;
Providing at least one purchase template to the user terminal and receiving at least one web URL to set a range of a web crawler;
Storing the sales information collected by the web crawler for the web URL and converting the collected atypical sales information into standardized data;
Determining the suitability of the formalized data with the purchase template and storing it as sale information if appropriate;
Generating and posting on the web a proposal or quotation request including an evaluation of sales information input from the user terminal;
From a plurality of other terminals, a predetermined company is extracted by comparing the data on the proposal and the company included in the proposal or the quotation according to the confirmation of the proposal or the request for quotation with the set purchase rule, and one or more pieces of the received information. Generating and providing the same to the user terminal;
Generating any result information corresponding to the selected acceptance information if any one of the extracted reception information is selected by the user terminal; And
Storing the result information in a knowledge database;
Providing one or more purchase templates to the user terminal and receiving one or more web URLs from the user terminal and setting the range of the web crawler,
Providing a plurality of purchasing templates including a UI for receiving at least one of a keyword, a price, a unit, an origin, a supplier, a delivery date, and a category for the article; And
Searching for and extracting data in the web URL based on data entered in a purchase template through the web crawler,
Extracting one or more pieces of receipt information suitable in comparison with the set purchase rule,
Comparing the sales information and the purchase template;
If there is a difference as a result of the comparison, transmitting a negotiation request message to the corresponding terminal according to the purchase rule; And
If the response to the message includes the step of determining whether to continue negotiation by determining whether the purchase template conforms,
The purchase rule,
One or more purchase conditions for the purchase of the goods and the contents of the message corresponding to the purchase conditions are defined,
The learning engine
A preprocessing process of preprocessing the training data or the evaluation data through normalization of the sales information collected by the web crawler;
Learning function to select an evaluation variable and input variable for preprocessed data and define variable attributes and network parameters to derive the answer, evaluation function to evaluate the derived answer is appropriate, and prediction function to find target variable through input variable Working process to perform;
A neural network standard interface connected to an external terminal and a database by wire or wireless through an information communication network;
A process controller performing inter-process signal control and providing a learning result to the neural network standard interface according to a request; And
And a post-processing process for outputting a report on data according to the learning result derived by the work process,
The learning engine performs machine learning on the learning result derived by the work process and the selection result of the final successful bidder and stores it in the knowledge database.
And the result information stored in the knowledge database is used as learning data of the learning engine to improve the accuracy of autonomous performance. delete delete delete delete

Images