Exhibition buyer recommendation method and system

Abstract

The invention relates to an exhibition buyer recommendation method and system, and the method comprises the steps: filtering data information of a new user through judging whether the buyer is the new user or not; and then historical behavior data in buyer data of the old user is combined with exhibit information data of each exhibit to analyze an interest relationship between each exhibit and each buyer, and deep binding is carried out on the interest relationship between the buyer and the exhibit, so that a recommended buyer set of each exhibit is obtained. And the recommendation buyer set of the exhibitor is put into the buyer recommendation list, so that the buyer wanting to recommend can be pushed to the exhibitor. Besides, when the total number of the recommended buyers in the buyer recommendation list is smaller than the recommended buyer number threshold value, the supplementary recommendation list is generated and supplemented into the buyer recommendation list, and the problem that the number of the recommended buyers in the recommended buyer list is too small possibly due to insufficient buyer information is solved.

Description

Technical Field

[0001] The present application relates to the field of intelligent exhibition technology, and in particular to a method and system for recommending exhibition buyers. Background Art

[0002] An exhibition is a comprehensive activity that focuses on a specific theme and displays products, technologies, services or cultural achievements to achieve information exchange, trade cooperation and brand promotion. The core feature of an exhibition is that it gathers a large number of exhibitors and visitors in a short period of time to form an efficient interactive platform.

[0003] The traditional way of conducting exhibitions is that the platform recruits exhibitors to participate in the exhibition. Exhibitors bring physical or virtual products to the exhibition venue to display their products in the form of stalls. The exhibitors' products are also called exhibits. In recent years, the interaction between exhibitors and buyers is generally that exhibitors provide scannable QR codes. Buyers scan the exhibitors' QR codes to learn about the exhibitors' basic information and detailed information about the exhibits when visiting the exhibition, thereby generating substantive behaviors such as inquiries or transactions.

[0004] However, traditional solutions lack a personalized recommendation method for exhibitors. The general practice of traditional recommendation solutions is that the platform creates a comprehensive promotional webpage for the exhibition, creates user profile data based on the user's visit cookie, and then recommends relevant exhibits to users on the homepage of the webpage by analyzing the user's profile data.

[0005] However, traditional recommendation programs can only recommend exhibits to buyers, resulting in the interaction between exhibitors and buyers being dominated by buyers and overly dependent on buyers' active behavior. The reality is that many buyers who participate in the exhibition do not know which exhibitors are participating in the exhibition, have poor motivation, and often do not want to or cannot make effective interactions. Therefore, there is currently a lack of exhibitor-led interaction methods between buyers and exhibitors, and it is impossible to proactively recommend buyers suitable for exhibitors to different exhibitors based on the different characteristics of the exhibitors. Summary of the invention

[0006] Based on this, it is necessary to provide an exhibition buyer recommendation method to address the problem that the traditional way of interaction between exhibitors and buyers lacks a personalized exhibition buyer recommendation method exclusively provided to exhibitors.

[0007] In one aspect, the present application provides a method for recommending buyers for an exhibition, the method comprising: Get buyer data of all buyers; Pre-process the buyer data of all buyers; Select a buyer and obtain the pre-processed buyer data of the buyer; Determine whether the buyer is a new user based on the pre-processed buyer data; If the buyer is not a new user, extract the historical behavior data from the pre-processed buyer data; Returning to the step of selecting a buyer and obtaining the pre-processed buyer data of the buyer until all buyers have been selected; Analyze the interest relationship between each exhibitor and each buyer based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and obtain the recommended buyer set of each exhibitor based on the interest relationship between each exhibitor and each buyer; When an exhibition buyer recommendation request of an exhibitor is obtained, the exhibitor is taken as a pending exhibitor, and a recommended buyer set of the pending exhibitor is obtained; Create a buyer recommendation list of exhibitors to be processed; Place the recommended buyers in the recommended buyer set of the exhibitors to be processed into the buyer recommendation list and publish the buyer recommendation list on the buyer recommendation interface of the exhibitors to be processed; Determine whether the total number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold; If the total number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold, a supplementary recommendation list is generated; The supplementary recommendation list will be added to the buyer recommendation interface of the exhibitors to be processed.

[0008] On the other hand, the present application also provides an exhibition buyer recommendation system, including: A server, used to execute the exhibition buyer recommendation method mentioned above; An exhibitor terminal connected to the server; The buyer's terminal is connected to the server.

[0009] The present application relates to a method and system for recommending buyers for exhibitions. First, by determining whether the buyer is a new user, the data information of the new user is filtered out. Then, the historical behavior data in the buyer data of the old user is used in combination with the exhibit information data of each exhibitor to analyze the interest relationship between each exhibitor and each buyer, so as to deeply bind the interest relationship between the buyer and the exhibitor, thereby obtaining a set of recommended buyers for each exhibitor. The recommended buyer set of the exhibitor is placed in the buyer recommendation list to push the buyers that the exhibitor wants to recommend. In addition, when the total number of recommended buyers in the buyer recommendation list is less than the threshold number of recommended buyers, a supplementary recommendation list is generated and added to the buyer recommendation list, which overcomes the problem that insufficient buyer information may lead to too few recommended buyers in the recommended buyer list. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a flowchart of S100 to S510 in the exhibition buyer recommendation method provided in an embodiment of the present application.

[0011] Figure 2 This is a flowchart of steps S710 to S900 in the method for recommending exhibition buyers according to an embodiment of the present application.

[0012] Figure 3 A schematic diagram of the structure of an exhibition buyer recommendation system provided in one embodiment of the present application.

[0013] Reference numerals: Server - 100; Exhibitor terminal - 200; Buyer terminal - 300. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solutions and advantages of this application more clear, the following is a further detailed description of this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application. The acquisition, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations.

[0015] The present application provides a method for recommending buyers for an exhibition. It should be noted that the method for recommending buyers for an exhibition provided by the present application is applicable to any type of exhibition.

[0016] In addition, the exhibition buyer recommendation method provided in this application does not limit its execution subject. Optionally, the execution subject of the exhibition buyer recommendation method provided in this application can be a server. The server is connected to the exhibitor terminal in communication, and the server pushes the recommended buyer to the exhibitor terminal. The exhibition terminal can be a mobile terminal, which can be implemented through a mobile phone APP.

[0017] like Figure 1 and Figure 2 As shown, in one embodiment of the present application, the method for recommending exhibition buyers includes: S100, obtaining buyer data of all buyers.

[0018] S200, pre-processing the buyer data of all buyers.

[0019] S300, selecting a buyer and obtaining the pre-processed buyer data of the buyer.

[0020] S400: Determine whether the buyer is a new user based on the pre-processed buyer data.

[0021] S510: If the buyer is not a new user, extract historical behavior data from the pre-processed buyer data.

[0022] S600, returning to S300, that is, returning to selecting a buyer and obtaining the pre-processed buyer data of the buyer, until all buyers are selected.

[0023] S710, analyzing the interest relationship between each exhibitor and each buyer based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and obtaining a recommended buyer set for each exhibitor based on the interest relationship between each exhibitor and each buyer.

[0024] S810: When an exhibition buyer recommendation request of an exhibitor is obtained, the exhibitor is taken as an exhibitor to be processed, and a recommended buyer set of the exhibitor to be processed is obtained.

[0025] S820, creating a buyer recommendation list of exhibitors to be processed.

[0026] S830: placing the recommended buyers in the recommended buyer set of the exhibitor to be processed into a buyer recommendation list and publishing the buyer recommendation list on the buyer recommendation interface of the exhibitor to be processed.

[0027] S840, determining whether the total number of recommended buyers in the buyer recommendation list is less than a threshold value of the number of recommended buyers.

[0028] S851: If the total number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold, a supplementary recommendation list is generated.

[0029] S900, adding the supplementary recommendation list to the buyer recommendation interface of the exhibitors to be processed.

[0030] Specifically, S100 to S600 can be understood as the process of early data acquisition, data sorting and data processing, and S710 to S900 can be understood as the actual application process of exhibitors using this method to push recommended buyers.

[0031] The buyer's buyer data comes from the buyer's terminal. The server creates a comprehensive exhibition publishing platform, which is embedded with exhibition web pages. Users create accounts in the exhibition publishing platform and log in to the account to browse the exhibition web pages, place orders in the exhibition publishing platform, and other behaviors, which can generate a large amount of buyer data.

[0032] In S200, the buyer data of all buyers are preprocessed. The preprocessing may include converting unstructured data into structured data for computer processing. For example, text data uses regular expressions to extract fields, NLP entity recognition, and LLM (Large Language Models, LLM) information extraction methods. For example, a table file or a log file is converted using pandas to convert the nested structure. Pandas is a tool based on NumPy (Numerical Python, an open source numerical computing extension framework). The tool is created to solve data analysis tasks. The table file can be a participation intention form for the next exhibition issued by the organizer to the exhibitors after the exhibition ends. Image files or PDF files can use OCR (Optical Character Recognition) to extract text. Image files can be chats or transaction screenshots between exhibitors and buyers, and PDF files can be transaction contracts between exhibitors and buyers. Audio and video files can be processed using speech-to-text software, such as ASR software (Automatic Speech Recognition, ASR, intelligent speech recognition). Audio and video files can be recordings of calls between buyers and exhibitors or buyers and exhibition organizers. All of the above data can be converted into data that can be stored and processed by computers through preprocessing.

[0033] S400 determines whether the buyer is a new user based on the pre-processed buyer data, and there are multiple implementation methods. The purpose of S400 is to divert and screen buyers of new and old users to facilitate the establishment and calculation of interest relationships between subsequent exhibitors and buyers. Since the buyer may be a new user, the new user does not have a lot of historical behavior data. Too little historical behavior data is not conducive to the establishment of interest relationships between subsequent exhibitors and buyers, and may cause inaccurate recommended buyer sets and lead to low accuracy of the recommendation algorithm. Therefore, it is necessary to divert and screen buyers of new and old users by executing S400 to improve the accuracy of the subsequent recommendation algorithm.

[0034] Optionally, S400 may include: S410, reading the buyer's account registration time in the pre-processed buyer data, and determining whether the difference between the buyer's account registration time and the current system time is less than a preset time difference threshold.

[0035] When the difference between the buyer's account registration time and the current system time is less than the preset time difference threshold, the buyer is determined to be a new user. When the difference between the buyer's account registration time and the current system time is greater than or equal to the preset time difference threshold, the buyer is determined to be an old user. For example, the preset time difference threshold is set to 1 month. If the difference between the buyer's account registration time and the current system time is less than 1 month, the buyer is determined to be a new user. On the contrary, if the difference between the buyer's account registration time and the current system time is greater than or equal to 1 month, the buyer is determined to be an old user.

[0036] Optionally, S400 may include: S420, reading the buyer's exhibition website browsing history record, exhibitor website browsing history record, exhibitor website browsing history record, exhibit sub-page browsing history record, transaction history record, and invitation history record from the pre-processed buyer data.

[0037] S430, determining whether the following conditions are met at the same time: 1) The buyer’s browsing history of the exhibition website is less than the first preset number.

[0038] 2) The buyer’s browsing history of the exhibitor’s website is less than the second preset number.

[0039] 3) The buyer’s history of browsing the exhibit sub-pages of the exhibitor’s website is less than the third preset number.

[0040] 4) The buyer’s historical transaction record is less than the fourth preset amount.

[0041] 5) The buyer’s historical invitation record is less than the fifth preset number.

[0042] If all the above conditions are met, the buyer is determined to be a new user. If at least one of the above conditions is not met, the buyer is determined to be an old user.

[0043] In an optional implementation, the first preset number, the second preset number, the third preset number, the fourth preset number and the fifth preset number can be set to be exactly the same, for example, all set to 3. Records can be segmented by the unit "strip", and of course, they can also be segmented by a fixed string length or total number of characters. How to define "a record" is a common technology and is not the focus of protection of this application.

[0044] Condition 1), condition 2) and condition 3) are the historical browsing behaviors of the user, and condition 4) and condition 5) are the historical substantive behaviors of the user. Browsing behaviors can be considered as relatively minor actions of interest, and substantive behaviors can be considered as relatively serious actions of interest. Therefore, in an optional embodiment, corresponding weight factors can be set for the first preset number, the second preset number, the third preset number, the fourth preset number and the fifth preset number, respectively, to reflect the importance of different actions of interest.

[0045] For example, the weight of the first preset number is set to 0.1, which means that the user has the least interest in browsing the exhibition history records. At this time, the user is still simply browsing the exhibition official website and is still browsing the exhibition official website. The weight of the second preset number is set to 0.15, which means that the user has a relatively low interest in browsing the exhibitor website history records. At this time, the user has already clicked on the official website page of the exhibitor he likes. The weight of the third preset number is 0.2, which means that the user has a medium interest in browsing the exhibit sub-page history records of the exhibitor website. At this time, the user has changed from browsing the official website to being interested in the exhibitors' exhibits and starts clicking on the exhibit sub-pages to browse.

[0046] The weight of the fourth preset number is set to 0.3, which means that the user has a strong interest in browsing historical transaction records, and the user has actually purchased exhibits at this time. The weight of the fifth preset number is set to 0.25, which means that the user has a strong interest in historical invitation records, and the user actively initiates exhibition invitations at this time, and has a strong desire to participate in the exhibition. Of course, historical invitation records can include multiple types such as the invitation was initiated but not subsequently cancelled and the user attended, the invitation was initiated but not subsequently cancelled and the user did not attend, and the invitation was initiated but subsequently cancelled, and different weight factors can be set for each type.

[0047] You can set a total number, for example, 10, and then define the values ​​of the first preset number, the second preset number, the third preset number, the fourth preset number, and the fifth preset number according to 10 multiplied by different weight factors. Continuing with the above example, the first preset number is 10 multiplied by 0.1, which is 1; the second preset number is 10 multiplied by 0.15, which is 1.5; the third preset number is 10 multiplied by 0.2, which is 2; the fourth preset value is 10 multiplied by 0.3, which is 3; and the fifth preset value is 10 multiplied by 0.25, which is 2.5. Of course, this example is just for the convenience of understanding. In actual operation, you should try to make each preset number calculated in the end an integer for easy calculation.

[0048] In S830, the buyer recommendation interface may be a page in the exhibitor's mobile terminal APP, and the buyer recommendation interface may be popped up in response to a request to pop up the page. The buyer recommendation interface may display a buyer recommendation list, and the buyer recommendation list may be browsed by a drop-down component or a slide-down component.

[0049] In this embodiment, the application relates to a method and system for recommending buyers for exhibitions. First, by determining whether the buyer is a new user, the data information of the new user is filtered out. Next, the historical behavior data in the buyer data of the old user is used in combination with the exhibit information data of each exhibitor to analyze the interest relationship between each exhibitor and each buyer, and the interest relationship between the buyer and the exhibitor is deeply bound, thereby obtaining a set of recommended buyers for each exhibitor. The recommended buyer set of the exhibitor is placed in the buyer recommendation list to push the buyers that the exhibitor wants to recommend. In addition, when the total number of recommended buyers in the buyer recommendation list is less than the threshold number of recommended buyers, a supplementary recommendation list is generated and added to the buyer recommendation list, which overcomes the problem that insufficient buyer information may lead to too few recommended buyers in the recommended buyer list.

[0050] In one embodiment of the present application, S510 includes, that is, extracting the historical behavior data from the pre-processed buyer data, including: S511, establish a historical behavior data table.

[0051] S512, extracting historical behavior data from the pre-processed buyer data and placing it into a historical behavior data table. The historical behavior data includes one or more of buyer browsing exhibition website history, buyer browsing exhibitor website history, buyer and exhibitor historical transaction records, and buyer and exhibitor historical invitation records.

[0052] Specifically, the historical behavior data may also include the historical records of buyers browsing the exhibit subpages of the exhibitor's website. The historical behavior data may also include the historical invitation records between buyers and exhibition organizers.

[0053] In this embodiment, by placing the historical behavior data in the pre-processed buyer data into the historical behavior data table, the interest-related data between exhibitors and buyers can be uniformly collected and stored, which is convenient for subsequent analysis and processing.

[0054] In one embodiment of the present application, S710 includes, that is, analyzing the interest relationship between each exhibitor and each buyer based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and obtaining the recommended buyer set of each exhibitor based on the interest relationship between each exhibitor and each buyer, including: S711, counting all buyers who are interested in at least one exhibitor, and defining the buyers who are interested in at least one exhibitor as interested buyers.

[0055] S712, counting all exhibitors that have generated interest in at least one buyer, and defining the exhibitors that have generated interest in at least one buyer as interested exhibitors.

[0056] S713, creating a list of interested buyers, obtaining the ID of each interested buyer, and placing the IDs of all interested buyers into the list of interested buyers.

[0057] S714, randomly select two different interested buyers from the interested buyers list and define them as the first interested buyer and the second interested buyer.

[0058] S715 , counting the number of exhibitors of interest generated by the first interested buyer, the number of exhibitors of interest generated by the second interested buyer, and the number of exhibitors of interest generated by both the first interested buyer and the second interested buyer.

[0059] S716, calculating the product of the arithmetic square root of the number of exhibitors of interest generated by the first interested buyer and the arithmetic square root of the number of exhibitors of interest generated by the second interested buyer.

[0060] S717, obtaining the quotient of the number of exhibitors that both the first interested buyer and the second interested buyer are interested in and the product, and using the quotient as the coupling relationship degree between the first interested buyer and the second interested buyer.

[0061] S718, repeatedly executing the steps of randomly selecting two different interested buyers from the list of interested buyers, defining them as the first interested buyer and the second interested buyer, to obtaining the quotient of the number of interested exhibitors that both the first interested buyer and the second interested buyer are interested in and the product, and taking the quotient as the coupling relationship degree between the first interested buyer and the second interested buyer, that is, repeatedly executing S714 to S717 until the coupling relationship degree between every two different interested buyers in the list of interested buyers is obtained.

[0062] S719, based on the coupling relationship between every two different interested buyers in the interested buyers list, obtain a recommended buyer set for each interested exhibitor.

[0063] Specifically, the purpose of S711 and S712 is to perform preliminary filtering on the data, remove meaningless data, filter out buyers who have no interest in any exhibitor, and improve the efficiency of subsequent processing. Buyers who have no interest in any exhibitor are buyers who are not interested in any exhibitor. For example, some buyers who just pass by the exhibition and stroll around, do not show any interest in the exhibits, and do not have any subsequent transaction behavior. The definition rules of buyers who are interested in exhibitors can be customized by the exhibitor's custom selection and filtering of the corresponding historical behavior data. For example, exhibitor A believes that buyers who have generated transaction behavior are considered to be interested in exhibitors, so the server can store the screening rules of exhibitor A, and when executing S711, the buyers who have transaction behavior records with exhibitor A are screened out as buyers who are interested in exhibitor A. In this way, different exhibitors can be customized to filter the interested buyers corresponding to different exhibitors, and then in S713, the interested buyers of different exhibitors are placed in the list of interested buyers.

[0064] Of course, a unified rule for screening interested buyers that all exhibitors abide by can also be set, for example: as long as there is at least one record of the buyer's browsing history of the exhibitor's website, the buyer's historical transaction record with the exhibitor, and the buyer's historical invitation record with the exhibitor, the buyer can be determined as an interested buyer and the interested buyer can be placed in the interested buyer list in S713.

[0065] The definition rules and screening methods for interested exhibitors are the same as those for interested buyers and will not be elaborated here.

[0066] S714 to S717 are methods for calculating the coupling relationship between two different interested buyers in the interested buyer list. The following specifically describes how to perform S714 to S717, and we use an example to illustrate.

[0067] For example, for the sake of simple description, there are 5 interested exhibitors, namely A, B, C, D, and E. There are 1, 2, 3, 4, and 5 interested buyers.

[0068] Buyers who are interested in Exhibitor A: Interested Buyer 1, Interested Buyer 2.

[0069] Buyers who are interested in exhibitor B: Interested Buyer 2, Interested Buyer 3, Interested Buyer 5.

[0070] Buyers who are interested in exhibitor C: Interested buyer 1, Interested buyer 3, Interested buyer 5.

[0071] Buyers who are interested in exhibitor D: Interested Buyer 2, Interested Buyer 5.

[0072] Buyers who are interested in exhibitor E: Interested Buyer 2, Interested Buyer 4.

[0073] Then, assuming that interested buyer 1 is selected as the first interested buyer and interested buyer 2 is selected as the second interested buyer, calculating the coupling relationship between the first interested buyer and the second interested buyer is converted into calculating the coupling relationship between interested buyer 1 and interested buyer 2.

[0074] The specific calculation method is to first calculate the arithmetic square root of the number of exhibitors interested by interested buyer 1. It can be seen that the exhibitors interested by interested buyer 1 are exhibitor A and exhibitor B. Then the number of exhibitors interested by interested buyer 1 is 2, and the arithmetic square root of 2 is .

[0075] Similarly, the exhibitors that interested buyer 2 is interested in are exhibitor A, exhibitor B, exhibitor D, and exhibitor E. Then the number of exhibitors that interested buyer 2 is interested in is 4. The square root of 4 is 2. The product of square root 2 and 2 is .

[0076] Next, calculate the number of interested exhibitors that are of interest to both interested buyer 1 and interested buyer 2. If there is only interested exhibitor A, then the number is 1.

[0077] Finally, divide by 1 , the quotient is Therefore, the coupling relationship between interested buyer 1 and interested buyer 2 is .

[0078] S718 needs to calculate the coupling relationship degree between each pair of interested buyers in the list of interested buyers to ensure the adequacy of the data. Therefore, A, B, C, D, and E need to calculate the coupling relationship degree between A and B, the coupling relationship degree between A and C, the coupling relationship degree between A and D, the coupling relationship degree between A and E, the coupling relationship degree between B and C, the coupling relationship degree between B and D, the coupling relationship degree between B and E, the coupling relationship degree between C and D, the coupling relationship degree between C and E, and the coupling relationship degree between D and E, which is a total of 4+3+2+1=10 coupling relationship degrees.

[0079] In this embodiment, by calculating the coupling relationship between every two different interested buyers in the interested buyers list, unfamiliar buyers with high coupling relationship with the exhibitors who have interacted with the exhibitors can be recommended to the exhibitors based on the buyers' historical behavior data.

[0080] In an embodiment of the present application, S810 includes obtaining a recommended buyer set for each interested exhibitor based on the coupling relationship between every two different interested buyers in the interested buyer list, including: S811, select an exhibitor of interest.

[0081] S812: Create a buyer interest set for the interested exhibitor, obtain all interested buyers who are interested in the interested exhibitor, and place all interested buyers who are interested in the interested exhibitor into the buyer interest set.

[0082] S813, selecting an interested buyer from the buyer interest set.

[0083] S814, create a buyer association set of the interested buyer, obtain M interested buyers with the largest coupling relationship with the interested buyer, and place all the M interested buyers into the buyer association set. M is a positive integer.

[0084] S815, returning an interested buyer in the selected buyer interest set, that is, returning to S813, until the buyer association set of each interested buyer in the buyer interest set is obtained.

[0085] S816, obtaining the union of the buyer association sets of each interested buyer in the buyer interest set, removing each element that appears in the buyer interest set from the union, and using the final set obtained as the recommended buyer set for the interested exhibitor.

[0086] S817, returning to the step of selecting an exhibitor of interest, that is, returning to S811, until a set of recommended buyers for each exhibitor of interest is obtained.

[0087] Specifically, this embodiment introduces the specific steps of S810, that is, how to generate a recommended buyer set for each interested exhibitor.

[0088] Continuing with the above example, if we want to get the recommended buyers set of exhibitor A, we first need to get the buyers' interest set of exhibitor A. We see that the buyers who are interested in exhibitor A are buyer 1 and buyer 2, so the buyers' interest set of exhibitor A is set A = {buyer 1, buyer 2}.

[0089] Next, we obtain the buyer association set of interested buyer 1 and the buyer association set of interested buyer 2. Before obtaining, we need to set M, which is the upper limit. The larger the M is, the greater the computing power required for calculation and the more complex the calculation is, but the accuracy of the recommended buyer set is higher. The smaller the M is, the smaller the computing power required for calculation and the simpler the calculation is, but the accuracy of the recommended buyer set is lower. Therefore, we can customize the settings in the actual application process. Here we set M to 3. The coupling relationship between interested buyer 1 and interested buyer 2 is , the coupling relationship between interested buyer 1 and interested buyer 3 is , the coupling relationship between interested buyer 1 and interested buyer 4 is 0, and the coupling relationship between interested buyer 1 and interested buyer 5 is Then the top three buyers with the highest coupling degree with interested buyer 1 are interested buyer 3, interested buyer 5, and interested buyer 2. The buyer association set of interested buyer 1 is set B1 = {interested buyer 3, interested buyer 5, interested buyer 2}.

[0090] Similarly, the coupling relationship between interested buyer 2 and interested buyer 1 is , the coupling relationship between interested buyer 2 and interested buyer 3 is , the coupling relationship between interested buyer 2 and interested buyer 4 is , the coupling relationship between interested buyer 2 and interested buyer 5 is Then the top three buyers with the highest coupling degree with interested buyer 1 are interested buyer 5, interested buyer 4, and interested buyer 3. The buyer association set of interested buyer 2 is set B2 = {interested buyer 5, interested buyer 4, interested buyer 3}.

[0091] Take the union of set B1 and set B2, and get set C = {interested buyer 2, interested buyer 3, interested buyer 4, interested buyer 5}, and the buyer interest set is set A = {interested buyer 1, interested buyer 2}. Set C needs to remove every element in set A. The purpose of removal is to remove interested buyers who are already interested in exhibitor A to avoid repeated recommendations. Therefore, interested buyer 2 is removed from set C, and finally the recommended buyer set of interested exhibitor A is set D = {interested buyer 3, interested buyer 4, interested buyer 5}.

[0092] In this embodiment, by creating a buyer interest set of the interested exhibitor, using the buyer interest set of the interested exhibitor as a starting point to find interested buyers with a large coupling relationship with each interested buyer in the buyer interest set, and including them in the buyer association set corresponding to the interested buyers, potential buyers suitable for recommendation to the interested exhibitor can be found divergently. Next, the union of the buyer association sets of each interested buyer in the buyer interest set is obtained, and each element that appears in the buyer interest set is removed from the union, and the final set obtained is used as the recommended buyer set of the interested exhibitor, which can avoid repeated recommendation of interested buyers who are interested in the exhibitor.

[0093] In one embodiment of the present application, after S710, after analyzing the interest relationship between each exhibitor and each buyer based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and obtaining the recommended buyer set of each exhibitor based on the interest relationship between each exhibitor and each buyer, the method further includes: S720, calculating the interest level of each interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor.

[0094] Specifically, in this embodiment, after obtaining the recommended buyer set, the interest level of each recommended buyer in the recommended buyer set is calculated.

[0095] For example, when the total number of recommended buyers in the recommended buyer set is too large and the buyer recommendation interface is not fully displayed, the interest can be sorted from large to small, and only the top N recommended buyers with the greatest interest can be displayed in the buyer recommendation interface, that is, the buyer recommendation list needs to be deleted. N is a positive integer.

[0096] In addition, interest has many extended applications. For example, there are some situations where the interest is too high and far exceeds the set interest threshold. This is because some popular exhibits may attract buyers. At this time, recommending these buyers to exhibitors may result in some buyers who really know and like the exhibits not being recommended. At this time, it is necessary to delete these buyers with too high interest from the buyer recommendation list.

[0097] In this embodiment, the interest level of each interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor is calculated, which provides an indicator for measuring the interest level and provides data basis for subsequent more complex processing.

[0098] In one embodiment of the present application, S720 includes, that is, calculating the interest level of each interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor, including: S721, select an exhibitor of interest, and obtain a buyer interest set and a recommended buyer set of the exhibitor of interest.

[0099] S722, selecting an interested buyer from the recommended buyer set as a recommended buyer.

[0100] S723, selecting an interested buyer from the buyer interest set.

[0101] S724, calculating the interest intensity value of the interested buyer for the interested exhibitor.

[0102] S725, calculating the interest index between the interested buyer and the recommended buyer. The interest index is the product of the coupling relationship between the interested buyer and the recommended buyer multiplied by the interest intensity value of the interested buyer for the interested exhibitor.

[0103] S726, returning to select an interested buyer in the buyer interest set, that is, returning to S723, until all interested buyers in the buyer interest set are selected, and a plurality of interest indexes are obtained.

[0104] S727, all interest indexes are accumulated, and the accumulated result is used as the interest degree of the recommended buyer to generate interest in the interested exhibitor.

[0105] S728, returning one of the interested buyers in the recommended buyer set as the recommended buyer, that is, returning to S722, until the interest level of the interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor is obtained.

[0106] S729, returning to the step of selecting an exhibitor of interest, obtaining the buyer interest set and the recommended buyer set of the exhibitor of interest, that is, returning to S721, until the interest level of each exhibitor of interest generated by each recommended buyer in the recommended buyer set of the exhibitor of interest is obtained.

[0107] Specifically, the above example is continued for explanation. As mentioned above, the recommended buyer set of interested exhibitor A is finally obtained as set D = {interested buyer 3, interested buyer 4, interested buyer 5}. The buyer interest set is set A = {interested buyer 1, interested buyer 2}. The buyer association set of interested buyer 1 is set B1 = {interested buyer 3, interested buyer 5, interested buyer 2}. The buyer association set of interested buyer 2 is set B2 = {interested buyer 5, interested buyer 4, interested buyer 3}.

[0108] The recommended buyers in set D are interested buyer 3, interested buyer 4, and interested buyer 5.

[0109] Then the interest of interested exhibitor A in interested buyer 3 = the interest index of interested buyer 3 and interested buyer 1 + the interest index of interested buyer 3 and interested buyer 2.

[0110] Then the interest of exhibitor A in buyer 4 = the interest index of buyer 4 and buyer 1 + the interest index of buyer 4 and buyer 2. The coupling relationship between buyer 1 and buyer 4 is 0, so the interest index of buyer 4 and buyer 1 is 0, so exhibitor A's interest in buyer 4 = the interest index of buyer 4 and buyer 2.

[0111] Then the interest of interested exhibitor A in interested buyer 5 = the interest index of interested buyer 5 and interested buyer 1 + the interest index of interested buyer 5 and interested buyer 2.

[0112] In this embodiment, an interest index is introduced to calculate the exhibitor's interest in each recommended buyer in the recommended buyer set. The interest index is the coupling relationship between the interested buyer and the recommended buyer multiplied by the interest intensity value of the interested buyer in the interested exhibitor, so that the final interest degree includes both the coupling factor and the real interest intensity value, thereby avoiding the situation where some popular exhibits are attracted by buyers, resulting in some buyers who really understand and like the exhibits not being recommended.

[0113] In one embodiment of the present application, S724 includes, that is, calculating the interest intensity value of the interested buyer for the interested exhibitor, including: S724a, retrieve historical behavior data of interested buyers.

[0114] S724b, obtaining the behavior value of each behavior dimension of the interested buyer towards the interested exhibitor based on the historical behavior data of the interested buyer.

[0115] S724c, select a behavioral dimension.

[0116] S724d, calculating the product of the behavior value of the behavior dimension and the weight value of the behavior dimension.

[0117] S724e, returning to the selection of a behavior dimension, that is, returning to S724c, until the product of the behavior value of each behavior dimension and the weight of the behavior dimension is obtained.

[0118] S724f, accumulate all products and use the obtained sum as the weighted sum.

[0119] S724g, dividing the weighted sum by the total number of behavior dimensions to obtain a quotient as the interest intensity value of the interested buyer for the interested exhibitor.

[0120] Specifically, this embodiment introduces the steps of how to calculate the interest intensity value of the interested buyer for the interested exhibitor. Multiple behavioral dimensions are introduced here. Similar to the analysis of historical behaviors mentioned above, we can also assign weights to some historical behavioral operations of buyers on exhibitors by dimension. For example, the dimension of buyers browsing the exhibitor's website can be weighted, the dimension of buyers browsing the exhibitor's exhibit sub-page can be weighted, the dimension of buyers purchasing a product of the exhibitor can be weighted, the dimension of buyers evaluating and promoting the exhibitor's products can be weighted, and the dimension of buyers and exhibitors having subsequent in-depth negotiations can be weighted. These behaviors can be recorded through the exhibitor terminal and the buyer terminal, and then uploaded to the server. The server assigns weights to each behavioral dimension and calculates the behavioral value, and finally obtains the interest intensity value of the interested buyer for the interested exhibitor.

[0121] In this embodiment, by taking the weighted sum of the product of the behavior value of each behavior dimension and the weight of the behavior dimension as the interest intensity value of the interested buyer for the interested exhibitor, the influence of different behaviors of buyers on the interest intensity value of the interested buyers for the interested exhibitors is fully considered, which is more in line with the actual situation and avoids the situation where the buyer's behavior is untrue.

[0122] In one embodiment of the present application, before S900, that is, before adding the supplementary recommendation list to the buyer recommendation interface of the exhibitors to be processed, the method further includes: S861, Get the layout of the upcoming exhibition.

[0123] S862, obtaining the number of recommended buyers in the buyer recommendation list of each exhibitor.

[0124] S863, balance the relationship between customer flow and the number of recommended buyers based on the number of recommended buyers in each exhibitor's buyer recommendation list and the location of each exhibitor's booth in the layout of the upcoming exhibition.

[0125] S864, determining whether the booth location of the exhibitor to be processed is a location with sparse passenger flow.

[0126] S865: If the booth location of the exhibitor to be processed is a location with sparse customer traffic, at least one new user buyer is placed in the supplementary recommendation list of the exhibitor to be processed according to the balancing result.

[0127] Specifically, this embodiment can establish a deep learning model and train the deep learning model. The trained model can have the ability to input the layout of the upcoming exhibition, the location of the exhibitor's booth, the total number of buyers participating in the upcoming exhibition, and the time of the upcoming exhibition, and the model outputs the passenger flow of each exhibitor's booth location. The unit of passenger flow can be defined as person / minute, which represents the number of buyers passing by a circle formed with the booth as the center and a preset distance (such as 2 meters) as the radius per minute.

[0128] A simple training process can be: first align the layout of the upcoming exhibition with the two-dimensional coordinate system, and after setting the coordinate origin, the two-dimensional coordinates of the physical center of each exhibitor's booth can be obtained. Next, based on the total number of buyers participating in the upcoming exhibition and the direction of the flow of people, a scatter plot of the flow of people at each time node is drawn. The scatter plot of the flow of people path includes the possible location coordinates of each buyer at each time node, and the predicted walking path of each buyer (combined with the exhibitors of interest corresponding to the buyers provided in the aforementioned content, and the buyers' historical behavior data). Each buyer is presented as a coordinate point on the flow of people path scatter plot, and the model can further calculate the number of coordinate points in the circle formed by each exhibitor's booth as the center and a preset distance (such as 2 meters) as the radius at each time node. In the calculation process, time factors can also be incorporated. For example, in the first half of the exhibition, the total number of coordinate points tends to be large, the degree of decay is small, the mobility of representatives is high (the mobility of people who change booths to browse is high), and the number of people leaving the exhibition is small. Conversely, in the second half of the exhibition, the total number of coordinate points tends to be small, the degree of decay is large, the mobility of representatives is low (the mobility of people who change booths to browse is low), and the number of people leaving the exhibition is large.

[0129] A complex training process can be: obtain the static attributes of the booth through the exhibition information module, such as the exhibitor's industry category, booth area, and exhibitor popularity. Obtain booth coordinate data, the distance between the booth coordinates and each aisle, and other data through the venue construction data module. Obtain dynamic time series data of crowd flow through infrared sensors or other sensors installed at the exhibition venue. The dynamic time series data of crowd flow includes the hourly flow of people and the density of people flow in other booths around the booth. Obtain activity process events and activity duration through the emergency module, such as the duration and start time of the host's opening remarks, the duration and start time of the lottery event, and the duration and start time of the main stage guest speech. Obtain external environmental data through the external environmental factor module. The external environmental data can include weather data, traffic congestion data, and data from competing exhibitions at the same time. The external environmental factor module can be connected to the interface of a third-party platform to obtain. Use the above data as training data to conduct comprehensive training on the model. The training steps that can be performed during the training process may include: 1) Space-time coding: Convert the booth coordinates into a polar coordinate system (relative to the origin coordinates of the main entrance) and add temporal periodic coding.

[0130] 2) Graph structure construction: Using booths as nodes, a weighted graph is constructed (edge ​​weight = walking time between booths).

[0131] 3) Use kernel density estimation (KDE) to generate real-time crowd flow heat map.

[0132] 4) Event Embedding: Pre-train Embedding for activity types (such as Transformer encoder).

[0133] In an optional embodiment, the model may use a neural network model, which may be specifically a multimodal spatiotemporal graph neural network (MSTGNN): class MSTGNN(nn.Module): def __init__(self): super().__init__() # Spatial modeling branch self.gcn = GraphConv(in_feat=64, out_feat=128) # Time modeling branch self.tcn = TemporalConv(input_dim=64, output_dim=128) # Multimodal Fusion self.cross_attn = MultiheadAttention(embed_dim=256, num_heads=8) def forward(self, graph_data, time_series, static_features): spatial_feat = self.gcn(graph_data) # spatial dependency temporal_feat = self.tcn(time_series) # Time dynamics fused = torch.cat([spatial_feat, temporal_feat], dim=-1) # Cross-modal attention fusion output, _ = self.cross_attn(fused, fused, fused) return output During the training process of the loss function, Huber Loss can be used in combination with auxiliary loss. The auxiliary loss uses contrast loss to control and constrain the consistency of the prediction results of similar booths.

[0134] loss = huber_loss(pred, true) + 0.3 * contrastive_loss(anchor,positive, negative) After calculating the customer flow of each exhibitor's booth location through the model of this embodiment, the model also has the following two capabilities: 1) Evaluate whether the booth location of each exhibitor is a location with sparse passenger flow. A passenger flow threshold can be preset for the model, such as 20 people / minute. When the passenger flow is greater than or equal to 20 people / minute, the booth location is considered to be a location with dense passenger flow. When the passenger flow is less than 20 people / minute, the booth location is considered to be a location with sparse passenger flow.

[0135] 2) It can balance the relationship between customer flow and the number of recommended buyers. For example, a simple implementation method is that the customer flow threshold is 20 people / minute, and the actual exhibitor customer flow is 10 people / minute. The model can combine historical exhibition data to calculate that the average customer flow of each booth is 5 people / minute. Then multiply 5 by 3 to place 15 new user buyers in the supplementary recommendation list of exhibitors to be processed.

[0136] In this embodiment, the relationship between the number of recommended buyers and the number of recommended buyers is balanced based on the number of recommended buyers in the buyer recommendation list of each exhibitor and the booth location of each exhibitor in the layout diagram of the upcoming exhibition. At least one new user buyer is placed in the supplementary recommendation list of the exhibitor to be processed based on the balancing result, so as to achieve recommendation compensation for exhibitors with bad locations and avoid the problem of stampedes caused by booths with good locations being too popular.

[0137] In one embodiment of the present application, after S840, that is, after determining whether the number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold, the method further includes: S852a: If the number of recommended buyers in the buyer recommendation list is greater than or equal to the recommended buyer threshold, a layout diagram of the upcoming exhibition is obtained.

[0138] S852b, obtaining the number of recommended buyers in the buyer recommendation list of each exhibitor.

[0139] S852c, balancing the relationship between customer flow and the number of recommended buyers based on the number of recommended buyers in each exhibitor's buyer recommendation list and the location of each exhibitor's booth in the layout of the upcoming exhibition.

[0140] S852d, determining whether the booth location of the exhibitor to be processed is a location with dense customer flow.

[0141] S852e, if the booth location of the exhibitor to be processed is a location with dense customer traffic, delete at least one recommended buyer from the buyer recommendation list of the exhibitor to be processed according to the balancing result. When deleting the recommended buyers, priority is given to deleting the recommended buyers with low interest.

[0142] Specifically, in this embodiment, when the booth location of the exhibitor to be processed is a location with dense customer traffic, the recommended buyers can be deleted without affecting the recommended buyers with higher interest.

[0143] In this embodiment, a balance is achieved between the passenger flow and the number of recommended buyers, while not affecting the recommendation of buyers with high interest to exhibitors.

[0144] This application also provides an exhibition buyer recommendation system.

[0145] For the sake of brevity, all terminals and devices appearing in this application are numbered in the embodiments of the exhibition buyer recommendation system, and terminals and devices with the same names mentioned in the aforementioned exhibition buyer recommendation method are not numbered.

[0146] like Figure 3As shown, in one embodiment of the present application, the exhibition buyer recommendation system includes a server 100, an exhibitor terminal 200 and a buyer terminal 300. The server 100 is used to execute the exhibition buyer recommendation method mentioned in any one of the embodiments mentioned in the foregoing content. The exhibitor terminal 200 is connected to the server 100. The buyer terminal 300 is connected to the server 100.

[0147] The technical features of the above-described embodiments may be arbitrarily combined, and the execution order of the method steps is not limited. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0148] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for recommending buyers for an exhibition, characterized in that: The method comprises: Get buyer data of all buyers; Pre-process the buyer data of all buyers; Select a buyer and obtain the pre-processed buyer data of the buyer; Determine whether the buyer is a new user based on the pre-processed buyer data; If the buyer is not a new user, extract the historical behavior data from the pre-processed buyer data; Returning to the step of selecting a buyer and obtaining the pre-processed buyer data of the buyer until all buyers have been selected; Analyze the interest relationship between each exhibitor and each buyer based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and obtain the recommended buyer set of each exhibitor based on the interest relationship between each exhibitor and each buyer; When an exhibition buyer recommendation request of an exhibitor is obtained, the exhibitor is taken as a pending exhibitor, and a recommended buyer set of the pending exhibitor is obtained; Create a buyer recommendation list of exhibitors to be processed; Place the recommended buyers in the recommended buyer set of the exhibitors to be processed into the buyer recommendation list and publish the buyer recommendation list on the buyer recommendation interface of the exhibitors to be processed; Determine whether the total number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold; If the total number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold, a supplementary recommendation list is generated; The supplementary recommendation list will be added to the buyer recommendation interface of the exhibitors to be processed.

2. The method for recommending exhibition buyers according to claim 1, characterized in that: The extracting of historical behavior data from the pre-processed buyer data includes: Establish historical behavior data table; The historical behavior data extracted from the pre-processed buyer data is placed in a historical behavior data table; the historical behavior data includes one or more of the buyer's browsing history records of the exhibition website, the buyer's browsing history records of the exhibitor website, the buyer's and exhibitor's historical transaction records, and the buyer's and exhibitor's historical invitation records.

3. The method for recommending exhibition buyers according to claim 2, characterized in that: The interest relationship between each exhibitor and each buyer is analyzed based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and the recommended buyer set of each exhibitor is obtained based on the interest relationship between each exhibitor and each buyer, including: Count all buyers who are interested in at least one exhibitor and define them as interested buyers; Count all exhibitors that have attracted the interest of at least one buyer, and define exhibitors that have attracted the interest of at least one buyer as interested exhibitors; Create a list of interested buyers, obtain the ID of each interested buyer, and put all the IDs of interested buyers into the list of interested buyers; Randomly select two different interested buyers from the list of interested buyers, defined as the first interested buyer and the second interested buyer; Count the number of exhibitors that are of interest to the first interested buyers, the number of exhibitors that are of interest to the second interested buyers, and the number of exhibitors that are of interest to both the first interested buyers and the second interested buyers; Calculate the product of the arithmetic square root of the number of exhibitors interested by the first interested buyer and the arithmetic square root of the number of exhibitors interested by the second interested buyer; Obtain the quotient of the number of exhibitors that both the first interested buyer and the second interested buyer are interested in and the product, and use the quotient as the coupling relationship degree between the first interested buyer and the second interested buyer; Repeat the steps of randomly selecting two different interested buyers from the list of interested buyers, defining them as the first interested buyer and the second interested buyer, until obtaining the quotient of the number of interested exhibitors that both the first interested buyer and the second interested buyer are interested in and the product, and using the quotient as the coupling relationship degree between the first interested buyer and the second interested buyer, until the coupling relationship degree between every two different interested buyers in the list of interested buyers is obtained; Based on the coupling relationship between every two different interested buyers in the interested buyers list, a set of recommended buyers for each interested exhibitor is obtained.

4. The method for recommending exhibition buyers according to claim 3, characterized in that: The recommended buyer set for each interested exhibitor is obtained based on the coupling relationship between every two different interested buyers in the interested buyer list, including: Select an exhibitor of interest; Create a buyer interest set for the interested exhibitor, obtain all interested buyers who are interested in the interested exhibitor, and place all interested buyers who are interested in the interested exhibitor into the buyer interest set; Select an interested buyer from the buyer interest set; Create a buyer association set for the interested buyer, obtain M interested buyers with the largest coupling relationship with the interested buyer, and place all M interested buyers into the buyer association set; M is a positive integer; Returning an interested buyer in the selected buyer interest set until a buyer association set of each interested buyer in the buyer interest set is obtained; Obtain the union of the buyer association sets of each interested buyer in the buyer interest set, remove each element that appears in the buyer interest set from the union, and use the final set obtained as the recommended buyer set for the interested exhibitor; Return to the step of selecting an exhibitor of interest until a set of recommended buyers for each exhibitor of interest is obtained.

5. The method for recommending exhibition buyers according to claim 4, characterized in that: After analyzing the interest relationship between each exhibitor and each buyer based on the historical behavior data of each buyer and the exhibit information data of each exhibitor, and obtaining the recommended buyer set of each exhibitor based on the interest relationship between each exhibitor and each buyer, the method further includes: The interest level of each interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor is calculated.

6. The method for recommending exhibition buyers according to claim 5, characterized in that: The step of calculating the interest level of each interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor includes: Select an exhibitor of interest and obtain the buyer interest set and recommended buyer set of the exhibitor of interest; Select an interested buyer from the recommended buyer set as the recommended buyer; Select an interested buyer from the buyer interest set; Calculate the interest intensity value of the interested buyer to the interested exhibitor; Calculating the interest index between the interested buyer and the recommended buyer; the interest index is the product of the coupling relationship between the interested buyer and the recommended buyer multiplied by the interest intensity value of the interested buyer for the interested exhibitor; Return and select an interested buyer in the buyer interest set until all interested buyers in the buyer interest set are selected, and obtain multiple interest indexes; All interest indexes are accumulated, and the accumulated result is used as the interest degree of the recommended buyers for the interested exhibitors; Returning an interested buyer in the selected recommended buyer set as a recommended buyer, until the interest level of the interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor is obtained; Return to the step of selecting an interested exhibitor, and obtain the buyer interest set and the recommended buyer set of the interested exhibitor, until the interest level of each interested exhibitor generated by each recommended buyer in the recommended buyer set of the interested exhibitor is obtained.

7. The method for recommending exhibition buyers according to claim 6, characterized in that: The calculation of the interest intensity value of the interested buyer to the interested exhibitor includes: Retrieve historical behavior data of interested buyers; Based on the historical behavior data of interested buyers, obtain the behavior values ​​of interested buyers on each behavior dimension of the interested exhibitors; Select a behavioral dimension; Calculate the product of the behavior value of the behavior dimension and the weight value of the behavior dimension; Returning to the step of selecting a behavior dimension until the product of the behavior value of each behavior dimension and the weight of the text dimension is obtained; Accumulate all products and use the sum as the weighted sum; The quotient obtained by dividing the weighted sum by the total number of behavioral dimensions is used as the interest intensity value of the interested buyer for the interested exhibitor.

8. The method for recommending exhibition buyers according to claim 7, characterized in that: Before adding the supplementary recommendation list to the buyer recommendation interface of the exhibitors to be processed, the method further includes: Get layouts of upcoming trade shows; Get the number of recommended buyers in each exhibitor's buyer recommendation list; Balance the relationship between customer flow and the number of recommended buyers based on the number of recommended buyers in each exhibitor's buyer recommendation list and the location of each exhibitor's booth in the layout of the upcoming exhibition; Determine whether the booth location of the exhibitor to be processed is a location with sparse customer traffic; If the booth location of the exhibitor to be processed is a location with sparse customer traffic, at least one new user buyer is placed in the supplementary recommendation list of the exhibitor to be processed according to the balancing result.

9. The method for recommending exhibition buyers according to claim 8, characterized in that: After determining whether the number of recommended buyers in the buyer recommendation list is less than the recommended buyer number threshold, the method further includes: If the number of recommended buyers in the buyer recommendation list is greater than or equal to the recommended buyer number threshold, the layout of the upcoming exhibition is obtained; Get the number of recommended buyers in each exhibitor's buyer recommendation list; Balance the relationship between customer flow and the number of recommended buyers based on the number of recommended buyers in each exhibitor's buyer recommendation list and the location of each exhibitor's booth in the layout of the upcoming exhibition; Determine whether the booth location of the exhibitor to be processed is a location with dense customer flow; If the booth location of the exhibitor to be processed is a location with dense customer traffic, at least one recommended buyer will be deleted from the buyer recommendation list of the exhibitor to be processed based on the balancing result; when deleting recommended buyers, priority will be given to deleting recommended buyers with low interest.

10. An exhibition buyer recommendation system, characterized in that: include: A server, configured to execute the exhibition buyer recommendation method according to any one of claims 1 to 9; An exhibitor terminal connected to the server; The buyer's terminal is connected to the server.

Images