Product recommendation method and device, equipment, medium and product

Abstract

The invention provides a product recommendation method and device, equipment, a medium and a product, and can be applied to the field of artificial intelligence. The product recommendation method comprises the steps of obtaining product resource data of preferred products of a target object and candidate resource data of candidate products, wherein candidate text data in the candidate resource data is used for describing product operation activities of the candidate products in a preset historical period; processing the product resource data and the candidate resource data based on a correlation algorithm to obtain a correlation degree; determining an intermediate product which does not meet a preset correlation degree threshold value from the candidate products based on the correlation degree; performing semantic understanding on the candidate text data by using a large language model to obtain a correlation influence relationship between the preference product and the intermediate product; and under the condition that the correlation influence relationship meets a preset negative correlation condition, determining a target product from the intermediate products, and pushing the target product and the preference product to the target object.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, in particular to a product recommendation method, device, equipment, medium and product. BACKGROUND

[0002] At present, an important way for major platforms to recommend products to customers is to give recommended products on the platform website. These recommended products are mostly products with better past click rates, but past data cannot guarantee that the products match the customers' own will. Therefore, there is an urgent need for a product recommendation method that can improve user satisfaction. SUMMARY

[0003] In view of the above problems, the present application provides a product recommendation method, device, equipment, medium and product.

[0004] According to a first aspect of the present application, a product recommendation method is provided, comprising: obtaining product resource data of a preferred product of a target object, candidate resource data of a candidate product, candidate text data in the candidate resource data being used to describe product operation activities of the candidate product in a preset historical period, the preset historical period being determined based on an interaction operation for the preferred product; processing the product resource data and the candidate resource data based on a correlation algorithm to obtain a correlation degree between the preferred product and the candidate product; determining an intermediate product that does not satisfy a preset correlation degree threshold from the candidate product based on the correlation degree; performing semantic understanding on the candidate text data using a large language model to obtain a correlation influence relationship between the preferred product and the intermediate product, the correlation influence relationship representing a click tendency state of the candidate text data on the intermediate product; determining a target product from the intermediate product in a case where the correlation influence relationship satisfies a preset negative correlation condition, and pushing the target product and the preferred product to the target object.

[0005] According to an embodiment of the present application, the correlation degree between the preferred product and the candidate product is obtained by processing the product resource data and the candidate resource data based on the correlation algorithm, comprising: determining a product click rate mean and a product click rate standard deviation of the preset historical period based on the product resource data; determining a candidate click rate mean and a candidate click rate standard deviation of the preset historical period based on the candidate resource data; determining a product click rate difference between a daily click rate of the preferred product in the preset historical period and the product click rate mean; determining a candidate click rate difference between a daily click rate of the candidate product in the preset historical period and the candidate click rate mean; and determining the correlation degree between the preferred product and the candidate product according to the product click rate difference, the candidate click rate difference, the product click rate standard deviation and the candidate click rate standard deviation.

[0006] According to an embodiment of the present application, the target product is determined based on the correlation influence value, including: processing the correlation degree and the correlation influence value by using a time series model to obtain a predicted correlation degree between the preferred product and the intermediate product; and determining the intermediate product as the target product when the predicted correlation degree is less than a preset value.

[0007] According to an embodiment of the present application, the predicted correlation degree between the preferred product and the intermediate product is obtained by processing the correlation degree and the correlation influence value by using a time series model, including: multiplying the correlation degree and the correlation influence value between the preferred product and the intermediate product to obtain a total correlation value; and processing the total correlation value by using a time series model to obtain the predicted correlation degree between the preferred product and the intermediate product.

[0008] According to an embodiment of the present application, the preferred product of the target object is obtained by further including: receiving an answer list returned by the target object for a preset question list, the preset question list including at least one question for mining product preference information; performing semantic analysis on the answer list to extract a preference feature vector related to a product attribute; matching the preference feature vector with each product feature vector in a candidate product feature library to obtain a matching score of each candidate product; and determining the preferred product of the target object from the candidate products according to the matching score.

[0009] According to an embodiment of the present application, the product recommendation method further includes: determining a benchmark product from the candidate products, the benchmark product having a click rate indicator satisfying a preset click rate threshold, and the benchmark product attribute of the benchmark product being the same as a product attribute of at least one of the target product and the preferred product; calculating a product similarity between the benchmark product and each of the target product and the preferred product based on benchmark product resource data of the benchmark product; and recommending the target product and the preferred product to the target object when the product similarity is greater than a preset product similarity.

[0010] The second aspect of the present application provides a product recommendation device, comprising: an acquisition module configured to acquire product resource data of a preferred product of a target object, candidate resource data of a candidate product, candidate text data in the candidate resource data being used to describe product operation activities of the candidate product in a preset historical period, the preset historical period being determined based on interaction operations for the preferred product; a correlation determination module configured to process the product resource data and the candidate resource data based on a correlation algorithm to obtain a correlation degree between the preferred product and the candidate product; an intermediate product determination module configured to determine, based on the correlation degree, an intermediate product that does not satisfy a preset correlation degree threshold from the candidate product; a semantic understanding module configured to perform semantic understanding on the candidate text data by using a large language model to obtain a correlation influence relationship between the preferred product and the intermediate product, the correlation influence relationship representing a click tendency state of the candidate text data on the intermediate product; and a recommendation module configured to determine a target product from the intermediate product in a case where the correlation influence relationship satisfies a preset negative correlation condition, and push the target product and the preferred product to the target object.

[0011] The third aspect of the present application provides an electronic device, comprising: one or more processors; a memory configured to store one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method.

[0012] The fourth aspect of the present application further provides a computer-readable storage medium having a computer program or instructions stored thereon, wherein the computer program or instructions are executed by a processor to implement the steps of the method.

[0013] The fifth aspect of the present application further provides a computer program product comprising a computer program or instructions, wherein the computer program or instructions are executed by a processor to implement the steps of the method.

[0014] According to the product recommendation method, device, equipment, medium and product provided by the present application, product resource data of a preferred product of a target object and candidate resource data of a candidate product can be acquired, candidate text data in the candidate resource data being used to describe product operation activities of the candidate product in a preset historical period, the correlation degree between the preferred product and the candidate product can be quantified by processing the product resource data and the candidate resource data based on a correlation algorithm, an intermediate product that does not satisfy a preset correlation degree threshold can be screened from the candidate product based on the correlation degree, the click tendency state of the candidate text data on the intermediate product can be obtained by performing semantic understanding on the candidate text data by using a large language model, a target product can be determined from the intermediate product in a case where the correlation influence relationship satisfies a preset negative correlation condition, and the target product and the preferred product are pushed to the target object, so that users can obtain matching content in a high-interest scenario or a low-interest scenario, and user satisfaction is improved. BRIEF DESCRIPTION OF DRAWINGS

[0015] The above and other objects, features and advantages of the present application will become more apparent from the following description when taken in conjunction with the accompanying drawings, in which:

[0016] Figure 1 An application scenario diagram of a product recommendation method, apparatus, device, medium and program product according to embodiments of the present application is shown;

[0017] Figure 2 A flowchart of a product recommendation method according to embodiments of the present application is shown;

[0018] Figure 3 A flowchart of a preferred product determination method according to embodiments of the present application is shown;

[0019] Figure 4 A structural block diagram of a product recommendation apparatus according to embodiments of the present application is shown; and

[0020] Figure 5 A block diagram of an electronic device suitable for implementing a product recommendation method according to embodiments of the present application is shown. DETAILED DESCRIPTION

[0021] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. It should be understood, however, that the description which follows is merely exemplary and is not intended to limit the scope of the application. In the following detailed description of embodiments of the present application, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that one or more embodiments of the present application can be practiced without these specific details. In other instances, well-known structures and functions have not been described in detail in order to avoid obscuring aspects of the present application.

[0022] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the term "includes" and tautological expressions thereof, such as "including," "includes," "include," "contains," "containing," and so forth, mean the term "comprises," unless otherwise noted.

[0023] All terms used herein (including technical and scientific terms) have the meanings commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein are to be interpreted as having a meaning that is consistent with the understanding of those terms by those in the relevant art, and not in an overly idealized or overly formal sense.

[0024] In the case of using expressions similar to "at least one of A, B, and C, etc.", it is generally to be interpreted in the meaning that the skilled person would normally understand the expression (for example, "a system having at least one of A, B, and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having both A and B, a system having both A and C, a system having both B and C, and / or a system having A, B, and C, etc.).

[0025] The way of recommending products in the related art is relatively single, mainly through the platform homepage to display the products with better performance to attract users. The products displayed in this way are often difficult for customers to understand the specific situation of the recommended products, which will lead customers to be unwilling to click on the products because they do not understand the platform-recommended products, and the platform-recommended products may be based on the interests of the platform itself, and the platform enjoys the advantage of information, which further aggravates the emotion of customers unwilling to click or purchase. In addition, the sudden increase or decrease of user click rate will cause the volatility of the platform page recommended product click rate to be too large. The sudden increase is often not the accurate satisfaction of user real demand, which may be caused by short-term operation activities, and such clicks will soon subside. The sudden decrease usually means that users are wrongly divided into an "interest blind group", which leads to the original matching content being submerged, and users cannot see what they want to see. Both of them increase the variance of click rate, and reduce the deep satisfaction indicators such as average stay time, completion rate, and revisit rate. In the long run, this "high volatility and low retention" cycle will dilute the user's trust in the platform. Therefore, the click rate curve with violent fluctuations all reflects that the matching degree between the recommended products and the users is low, which reduces the user satisfaction.

[0026] In order to at least partially solve the technical problems existing in the related art, the embodiments of the present application provide a product recommendation method, obtaining product resource data of a preferred product of a target object, candidate resource data of a candidate product, the candidate text data in the candidate resource data being used to describe product operation activities of the candidate product in a preset historical period, the preset historical period being determined based on an interaction operation for the preferred product; processing the product resource data and the candidate resource data based on a correlation algorithm to obtain a correlation degree between the preferred product and the candidate product; determining an intermediate product that does not satisfy a preset correlation degree threshold from the candidate product based on the correlation degree; performing semantic understanding on the candidate text data by using a large language model to obtain a correlation influence relationship between the preferred product and the intermediate product, the correlation influence relationship representing a click tendency state of the candidate text data on the intermediate product; in the case that the correlation influence relationship satisfies a preset negative correlation condition, determining a target product from the intermediate product, and pushing the target product and the preferred product to the target object.

[0027] It should be noted that the product recommendation method and the product recommendation device provided in the present application can be used in the field of artificial intelligence, and can also be used in any field other than the field of artificial intelligence, such as project resource management, and therefore the application field of the product recommendation method and the product recommendation device provided in the present application is not limited.

[0028] In the technical solutions of the present application, the user information (including but not limited to user personal information, user image information, user device information such as location information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved are information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for user to choose authorization or refusal.

[0029] In the scenario of using personal information for automatic decision-making, the method, device and system provided in the embodiments of the present application all provide corresponding operation portal for the user to choose to agree or refuse the automatic decision-making result; if the user chooses to refuse, the expert decision-making process is entered. The expression "automatic decision-making" here refers to the activity of automatically analyzing, evaluating the behavior habits, interests and hobbies or economic, health, credit status of a person through a computer program, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by personnel who are engaged in a certain field, have special experience, knowledge and skills, and reach a certain professional level.

[0030] Figure 1 An application scenario diagram of the product recommendation method, device, equipment, medium and program product according to the embodiments of the present application is shown.

[0031] As shown in Figure 1 According to the application scenario 100 of this embodiment, the first terminal device 101, the second terminal device 102, the third terminal device 103, the network 104 and the server 105 can be included. The network 104 is a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0032] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be installed with various communication terminal applications, such as shopping applications, web browser applications, search applications, instant messaging tools, mailbox terminals, social platform software, etc. (only as examples).

[0033] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smartphones, tablet computers, laptop computers, desktop computers, etc.

[0034] The server 105 can be a server providing various services, such as a background management server supporting websites browsed by the user using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (only as examples). The background management server can analyze and process received user requests, etc., and feed back the processing results (such as web pages, information, or data, etc. obtained or generated according to user requests) to the terminal device.

[0035] It should be noted that the product recommendation method provided by the embodiments of the present application can generally be executed by the server 105. Correspondingly, the product recommendation device provided by the embodiments of the present application can generally be arranged in the server 105. The product recommendation method provided by the embodiments of the present application can also be executed by a server or a server cluster different from the server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or the server 105. Correspondingly, the product recommendation device provided by the embodiments of the present application can also be arranged in a server or a server cluster different from the server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or the server 105.

[0036] It should be understood that Figure 1 The number of terminal devices, networks, and servers in

[0037] The product recommendation method according to the embodiments of the present application will be described in detail below based on the scenario described above. Figure 1 Figures 2-3 The product recommendation method according to the embodiments of the present application will be described in detail below based on the scenario described above.

[0038] Figure 2 A flowchart of the product recommendation method according to the embodiments of the present application is shown.

[0039] As Figure 2 ​As shown, the product recommendation method of this embodiment includes operation S210 to operation S250.

[0040] In operation S210, product resource data of a preferred product of a target object, candidate resource data of a candidate product are acquired, candidate text data in the candidate resource data is used to describe product operation activities of the candidate product in a preset historical period.

[0041] In operation S220, the product resource data and the candidate resource data are processed based on a correlation algorithm to obtain a correlation degree between the preferred product and the candidate product.

[0042] In operation S230, based on the correlation degree, an intermediate product that does not satisfy a preset correlation degree threshold is determined from the candidate product.

[0043] In operation S240, the candidate text data is subjected to semantic understanding by using a large language model to obtain a correlation influence relationship between the preferred product and the intermediate product.

[0044] In operation S250, in a case where the correlation influence relationship satisfies a preset negative correlation condition, a target product is determined from the intermediate product, and the target product and the preferred product are pushed to the target object.

[0045] According to the embodiments of the present application, the preferred product can include a preferred short video type, a preferred graphic post type, a preferred advertisement type, a preferred commodity category, etc., but is not limited thereto, and the embodiments of the present application do not limit the same. The candidate product can be a different product of the same category as the preferred product, for example, the preferred product is a preferred advertisement type, and the candidate product can include other advertisement types different from the preferred advertisement type. The product resource data can include click rate data related to the preferred product, a fluctuation amplitude related to the click rate data within a preset historical period, etc., but is not limited thereto, and the embodiments of the present application do not limit the same. The preset historical period is determined based on an interaction operation for the preferred product. The candidate text data in the candidate resource data can be used to describe product operation activities of the candidate product in the preset historical period. For example, in the case where the preferred product is a preferred commodity, the candidate text data can be a full-reduction activity related to the commodity, etc.

[0046] According to the embodiments of the present application, the product resource data and the candidate resource data are processed based on a correlation algorithm to obtain a correlation degree between the preferred product and the candidate product. The correlation degree can represent a rising-falling amplitude relationship between the preferred product and the candidate product. Based on the correlation degree, an intermediate product that does not satisfy a preset correlation degree threshold can be determined from the candidate product. The intermediate product can represent another product related to the preferred product, but the click rate of which changes in the opposite direction. When the click rate of the preferred product increases, the click rate of the intermediate product decreases.

[0047] According to an embodiment of the present application, the correlation influence relationship represents the click tendency state of the candidate text data on the intermediate product. Specifically, the click tendency state can enhance or weaken the click probability or willingness degree of the product on the target object. The influence of the candidate text data on the product is analyzed by the large language model. First, the product operation activity information in the preset historical period is collected, and the product operation activity information is converted into corresponding numerical features by the large language model. The correlation influence relationship between the preferred product and the intermediate product is scored according to the numerical features. The score can be set between -1 and 1. If it is less than 0, it means that the hedge between the two products will be reversed. If it is greater than 0, it means that the hedge condition is unchanged.

[0048] According to an embodiment of the present application, in the case that the correlation influence relationship meets the preset negative correlation condition, the target product is determined from the intermediate product, and the target product and the preferred product are pushed to the target object.

[0049] For example, regarding the advertisement click rate hedge recommendation, if the click rate fluctuates greatly in the page, the user click rate increases or decreases sharply, and the user loss risk will increase. Therefore, the product recommendation method in the present application can be used to recommend the preferred product and the target product as a combination to the target object. The combination of high click rate and low click rate products can ensure that users can obtain matching content in high interest scenarios or low interest scenarios, thereby improving user satisfaction.

[0050] According to an embodiment of the present application, according to the product recommendation method, device, equipment, medium and product provided by the present application, the product resource data of the preferred product and the candidate resource data of the candidate product can be obtained. The candidate text data in the candidate resource data is used to describe the product operation activity of the candidate product in the preset historical period. By processing the product resource data and the candidate resource data by using the correlation algorithm, the correlation degree between the preferred product and the candidate product can be quantified. Based on the correlation degree, the intermediate product that does not meet the preset correlation degree threshold can be selected from the candidate product. By using the large language model to understand the semantics of the candidate text data, the click tendency state of the candidate text data on the intermediate product can be obtained. In the case that the correlation influence relationship meets the preset negative correlation condition, the target product is determined from the intermediate product, and the target product and the preferred product are pushed to the target object. This makes users obtain matching content in high interest scenarios or low interest scenarios, thereby improving user satisfaction.

[0051] According to an embodiment of the present application, the product resource data and the candidate resource data are processed based on a correlation algorithm to obtain a correlation degree between the preferred product and the candidate product, including: determining a product click rate mean value and a product click rate standard deviation of a preset historical period based on the product resource data; determining a candidate click rate mean value and a candidate click rate standard deviation of the preset historical period based on the candidate resource data; determining a product click rate difference value between a daily click rate of the preferred product and the product click rate mean value of the preset historical period; determining a candidate click rate difference value between a daily click rate of the candidate product and the candidate click rate mean value of the preset historical period; and determining the correlation degree between the preferred product and the candidate product according to the product click rate difference value, the candidate click rate difference value, the product click rate standard deviation and the candidate click rate standard deviation.

[0052] According to an embodiment of the present application, the product resource data can include click rate data of the preferred product in a preset historical period. The candidate resource data can include candidate click rate data of the candidate product in the preset historical period. For example, the preset historical period can be one month, and the daily click rate mean value of the preferred product and the product click rate standard deviation of the preset historical period can be determined based on the obtained click rate data of the preferred product. The daily click rate mean value of the candidate product and the candidate click rate standard deviation of the preset historical period can be determined based on the obtained click rate data of the candidate product. The product click rate difference value between the daily click rate of the preferred product and the product click rate mean value of each day in one month is determined, and the candidate click rate difference value between the daily click rate of the candidate product and the candidate click rate mean value of each day in one month is determined. The correlation degree between the preferred product and the candidate product is determined according to the product click rate difference value, the candidate click rate difference value, the product click rate standard deviation and the candidate click rate standard deviation, as shown in formula (1).

[0053] (1)

[0054] wherein x and y respectively represent the preferred product and the candidate product, represents the click rate of the preferred product x in one month, represents the average click rate of the preferred product x in the month. represents the click rate of the candidate product y in one month, represents the average click rate of the candidate product y in the month.

[0055] According to an embodiment of the present application, the numerical value of the correlation degree can be set between -1 and 1, 1 representing strong positive correlation and -1 representing strong negative correlation.

[0056] According to an embodiment of the present application, the correlation degree between the preferred product and the candidate product is determined by the click rate data of the preferred product and the click rate data of the candidate product, the risk of the target object information cocoon is reduced, the user can obtain matching content in a high-interest or low-interest scenario, and user satisfaction is improved.

[0057] According to an embodiment of the present application, the target product is determined from the intermediate product, including: using a time series model to process the correlation degree and the correlation influence value to obtain a predicted correlation between the preferred product and the intermediate product; and in a case where the predicted correlation is less than a preset value, determining the target product.

[0058] According to an embodiment of the present application, the time series model can include a Long Short-Term Memory (LSTM), a Support Vector Regression (SVR), and a Multi-Layer Perceptron (MLP), but is not limited thereto. The predicted correlation between the preferred product and the intermediate product can be obtained by using the time series model to process the correlation degree and the correlation influence value. In a case where the predicted correlation is less than a preset value, the intermediate product can be determined as the target product.

[0059] According to an embodiment of the present application, the time series model is used to process the correlation degree and the correlation influence value, the predicted correlation between the preferred product and the intermediate product can be predicted, and in a case where the predicted correlation is less than a preset value, the target product that hedges the preferred product can be determined. By combining the target product and the preferred product in a page, the click volatility can be reduced.

[0060] According to an embodiment of the present application, the time series model is used to process the correlation degree and the correlation influence value, the predicted correlation between the preferred product and the intermediate product is obtained, including: multiplying the correlation degree and the correlation influence value between the preferred product and the intermediate product to obtain a total correlation value; and using the time series model to process the total correlation value to obtain the predicted correlation between the preferred product and the intermediate product.

[0061] According to the embodiment of the present application, the degree of correlation between the preferred product and the intermediate product can be multiplied by the correlation influence value to obtain a total correlation value. The total correlation value is used as the input of the time series model, and finally the predicted correlation between the preferred product and the intermediate product is obtained. The predicted correlation can be used as the historical data input when the time series model is iterated next time. When the time series model is first run, the cell state and hidden information of the last time period can be set to 0. The calculation steps of the time series model can be divided into forget gate, input gate, candidate cell state, update cell state, output gate and current hidden information, and the specific formulas are as follows.

[0062] f_t =σ(W_f[h_t-1, x_t] + b_f)(2)

[0063] i_t = σ(W_i[h_t-1, x_t] + b_i)(3)

[0064] g_t = tanh(W_g[h_t-1, x_t] + b_g)(4)

[0065] C_t = f_t ⊙ C_t-1 + i_t ⊙ g_t(5)

[0066] h_t = f_t * tanh(C_t)(6)

[0067] Wherein, W_f represents the forget gate weight matrix, W_i represents the input gate weight matrix, W_g represents the candidate cell state weight matrix, h_t-1 represents the hidden information at t-1 time, x_t represents the input at t time, i.e. the total correlation value, b_f represents the forget gate bias term, and b_* of the remaining steps represents the respective weight matrix, ⊙ represents Hadamard product, σ represents sigmoid function, C_t-1 represents the cell state at t-1 time, f_t represents the output of the forget gate at t time, i_t represents the output of the input gate, g_t represents the candidate cell state at t time, and C_t represents the cell state at t time.

[0068] According to the embodiment of the present application, the time series model can predict the correlation decay between the preferred product and the intermediate product in advance, arrange the page product in advance, improve the user experience, and reduce the click rate fluctuation.

[0069] Figure 3 A flowchart of a preferred product determination method according to an embodiment of the present application is shown.

[0070] As Figure 3 shown, the preferred product determination method of this embodiment includes operations S310-S340.

[0071] At operation S310, a list of answers returned by the target object for a preset list of questions is received.

[0072] At operation S320, the list of answers is semantically analyzed to extract a preference feature vector related to product attributes.

[0073] At operation S330, the preference feature vector is matched with each product feature vector in a candidate product feature library to obtain a matching score of each candidate product.

[0074] At operation S340, a preferred product of the target object is determined from the candidate products according to the matching score.

[0075] According to an embodiment of the present application, the target object can enter a question collection page before entering a product recommendation page, and the question collection page includes a preset list of questions, which can include the target object's interest preference type, a preset product spending limit, product type preference, and the like, but is not limited thereto. Embodiments of the present disclosure do not limit the specific content of the preset list of questions.

[0076] The list of answers of the target object for the preset list of questions can be semantically analyzed to extract a preference feature vector related to product attributes. The preference feature vector can be matched with each product feature vector in a candidate product feature library using a matching algorithm to obtain a matching score of the target object and each candidate product. The matching algorithm can include a cosine similarity algorithm, a Pearson correlation coefficient, and the like. Embodiments of the present disclosure do not limit the matching algorithm. The candidate product with a matching score greater than a preset matching score can be determined as the preferred product of the target object.

[0077] According to an embodiment of the present application, by matching the preference feature vector corresponding to the list of answers with the candidate product feature library, the matching accuracy can be improved, and the process of obtaining the preferred product includes question and answer interaction, semantic analysis, and vector matching, supporting a real-time recommendation scenario of answering questions and outputting results at the same time, which can respond in real time and improve processing efficiency.

[0078] According to an embodiment of the present application, the product recommendation method further includes: determining a reference product from the candidate products, the reference product having a click rate index satisfying a preset click rate threshold, and the reference product attribute of the reference product being the same as the product attribute of at least one of the target product and the preferred product; calculating the product similarity between the reference product and the target product and the preferred product based on the reference product resource data of the reference product; and recommending the target product and the preferred product to the target object when the product similarity is greater than a preset product similarity.

[0079] According to the embodiment of the present application, the click rate index of the benchmark product meeting the preset click rate threshold is determined from the candidate products, and the benchmark product attribute of the benchmark product is the same as the product attribute of at least one of the target product and the preferred product. The similarity between the product with the same attribute as the benchmark product and the benchmark product is calculated, the similarity of the two is analyzed, and in the case that the product similarity is greater than the preset product similarity, the target product and the preferred product are recommended to the target object, which can further improve the stability of the page click rate.

[0080] According to the embodiment of the present application, by analyzing the product components of the page with high click rate, recording the products with high click rate, and comparing the similarity with the products recommended to the target object, the user experience is further improved through horizontal comparison.

[0081] Based on the above product recommendation method, the present application also provides a product recommendation device. The following will be combined Figure 4 to describe the device in detail.

[0082] Figure 4 The structure block diagram of the product recommendation device according to the embodiment of the present application is shown.

[0083] As Figure 4 shown, the product recommendation device 400 of the embodiment includes an acquisition module 410, a relevance determination module 420, an intermediate product determination module 430, a semantic understanding module 440, and a recommendation module 450.

[0084] The acquisition module 410 is configured to acquire product resource data of a preferred product of a target object and candidate resource data of a candidate product, wherein the candidate text data in the candidate resource data is used to describe product operation activities of the candidate product in a preset historical period, and the preset historical period is determined based on an interaction operation for the preferred product.

[0085] The relevance determination module 420 is configured to process the product resource data and the candidate resource data based on a relevance algorithm to obtain a relevance degree between the preferred product and the candidate product.

[0086] The intermediate product determination module 430 is configured to determine an intermediate product that does not meet a preset relevance degree threshold from the candidate product based on the relevance degree.

[0087] The semantic understanding module 440 is configured to perform semantic understanding on the candidate text data by using a large language model to obtain a relevance influence relationship between the preferred product and the intermediate product, wherein the relevance influence relationship represents a click tendency state of the candidate text data on the intermediate product.

[0088] The recommendation module 450 is configured to determine a target product from the intermediate product when the correlation influence relationship meets the preset negative correlation condition, and push the target product and the preferred product to the target object.

[0089] According to the product recommendation method, device, equipment, medium and product provided by the embodiments of the present application, the product resource data of the preferred product of the target object and the candidate resource data of the candidate product can be obtained, the candidate text data in the candidate resource data is used to describe the product operation activities of the candidate product in a preset historical period, the correlation algorithm is used to process the product resource data and the candidate resource data, the correlation degree between the preferred product and the candidate product can be quantified, based on the correlation degree, the intermediate product that does not meet the preset correlation degree threshold can be screened from the candidate product, the semantic understanding of the candidate text data can be obtained by using the large language model, the click tendency state of the candidate text data to the intermediate product can be obtained, the target product is determined from the intermediate product when the correlation influence relationship meets the preset negative correlation condition, the target product and the preferred product are pushed to the target object, so that the user can obtain matching content in a high interest scene or a low interest scene, and the user experience is improved.

[0090] According to the embodiments of the present application, the correlation determination module 420 includes a product click rate unit, a candidate click rate unit, a first determination unit, a second determination unit and a correlation degree unit.

[0091] The product click rate unit is configured to determine a product click rate mean and a product click rate standard deviation of the product in the preset historical period based on the product resource data.

[0092] The candidate click rate unit is configured to determine a candidate click rate mean and a candidate click rate standard deviation of the candidate product in the preset historical period based on the candidate resource data.

[0093] The first determination unit is configured to determine a product click rate difference between a daily click rate of the preferred product in the preset historical period and the product click rate mean.

[0094] The second determination unit is configured to determine a candidate click rate difference between a daily click rate of the candidate product in the preset historical period and the candidate click rate mean.

[0095] The correlation degree unit is configured to determine a correlation degree between the preferred product and the candidate product according to the product click rate difference, the candidate click rate difference, the product click rate standard deviation and the candidate click rate standard deviation.

[0096] According to the embodiments of the present application, the recommendation module 450 includes a predicted correlation unit and a third determination unit.

[0097] a prediction correlation unit, configured to process the correlation degree and the correlation influence value by using a time series model to obtain a prediction correlation between the preference product and the intermediate product.

[0098] a third determination unit, configured to determine the intermediate product as the target product when the prediction correlation is less than a preset value.

[0099] According to an embodiment provided in the present application, the prediction correlation unit comprises:

[0100] a total correlation value sub-unit, configured to multiply the correlation degree and the correlation influence value between the preference product and the intermediate product to obtain a total correlation value.

[0101] a prediction correlation sub-unit, configured to process the total correlation value by using a time series model to obtain a prediction correlation between the preference product and the intermediate product.

[0102] According to an embodiment provided in the present application, the obtaining module comprises a receiving unit, an analyzing unit, a matching unit and a preference product determination unit.

[0103] The receiving unit is configured to receive an answer list returned by a target object for a preset question list, wherein the preset question list comprises at least one question for mining product preference information.

[0104] The analyzing unit is configured to perform semantic analysis on the answer list to extract a preference feature vector related to product attributes.

[0105] The matching unit is configured to match the preference feature vector with each product feature vector in a candidate product feature library to obtain a matching score of each candidate product.

[0106] The preference product determination unit is configured to determine a preference product of the target object from the candidate products according to the matching score.

[0107] According to an embodiment provided in the present application, the product recommendation device 400 of the embodiment further comprises a reference product determination module, a product similarity module and a pushing module.

[0108] The reference product determination module is configured to determine a reference product with a click rate index satisfying a preset click rate threshold from candidate products, wherein a reference product attribute of the reference product is the same as a product attribute of at least one product of the target product and the preference product.

[0109] The product similarity module is configured to calculate a product similarity between the reference product and each of the target product and the preference product based on reference product resource data of the reference product.

[0110] The pushing module is configured to recommend the target product and the preferred product to the target object when the product similarity is greater than a preset product similarity.

[0111] According to an embodiment of the present application, any of the modules of the obtaining module 410, the relevance determining module 420, the intermediate product determining module 430, the semantic understanding module 440 and the recommending module 450 can be combined in one module, or any of the modules can be split into multiple modules. Alternatively, at least part of the function of one or more of the modules can be combined with at least part of the function of other modules, and implemented in one module. According to an embodiment of the present application, at least one of the obtaining module 410, the relevance determining module 420, the intermediate product determining module 430, the semantic understanding module 440 and the recommending module 450 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on board, a system on package, an application specific integrated circuit (ASIC), or any other reasonable manner of integrating or packaging a circuit, etc. or implemented by hardware or firmware, or implemented in any one of software, hardware and firmware or in a proper combination of any of them. Alternatively, at least one of the obtaining module 410, the relevance determining module 420, the intermediate product determining module 430, the semantic understanding module 440 and the recommending module 450 can be at least partially implemented as a computer program module which can perform corresponding functions when the computer program module is run.

[0112] Figure 5 A block diagram of an electronic device suitable for implementing the product recommendation method according to an embodiment of the present application is shown.

[0113] As shown in Figure 5 The electronic device 500 according to an embodiment of the present application includes a processor 501 which can perform various appropriate actions and processes according to programs stored in a read only memory (ROM) 502 or loaded from a storage portion 508 to a random access memory (RAM) 503. The processor 501 can include, for example, a general purpose microprocessor (e.g. a CPU), an instruction set processor and / or a related chipset and / or a special purpose microprocessor (e.g. an application specific integrated circuit (ASIC)), etc. The processor 501 can also include an on-board memory for cache use. The processor 501 can include a single processing unit or multiple processing units for performing different actions of the method processes according to embodiments of the present application.

[0114] In the RAM 503, various programs and data required for the operation of the electronic device 500 are stored. The processor 501, the ROM 502, and the RAM 503 are connected to each other via the bus 504. The processor 501 performs various operations of the method flow according to the embodiments of the present application by executing the programs in the ROM 502 and / or the RAM 503. It should be noted that the programs can also be stored in one or more memories other than the ROM 502 and the RAM 503. The processor 501 can also perform various operations of the method flow according to the embodiments of the present application by executing the programs stored in the one or more memories.

[0115] According to the embodiments of the present application, the electronic device 500 can further include an input / output (I / O) interface 505, which is also connected to the bus 504. The electronic device 500 can further include one or more of the following components connected to the input / output (I / O) interface 505: an input part 506 including a keyboard, a mouse, and the like; an output part 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), and the like, and a speaker, and the like; a storage part 508 including a hard disk, and the like; and a communication part 509 including a network interface card such as a LAN card, a modem, and the like. The communication part 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the input / output (I / O) interface 505 as necessary. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 510 as necessary, so that a computer program read therefrom is installed in the storage part 508 as necessary.

[0116] The present application also provides a computer readable storage medium, which can be included in the device / apparatus / system described in the above embodiments; or can exist separately without being assembled into the device / apparatus / system. The above computer readable storage medium carries one or more programs, when the one or more programs are executed, the product recommendation method according to the embodiments of the present application is implemented.

[0117] According to an embodiment of the present application, the computer readable storage medium can be a non-transitory computer readable storage medium, for example, can include but not limited to: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this application, a computer readable storage medium can be any tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present application, the computer readable storage medium can include one or more of the above-described ROM 502 and / or RAM 503 and / or a memory other than the ROM 502 and the RAM 503.

[0118] Embodiments of the present application also include a computer program product, which includes a computer program containing program codes for executing the methods shown in the flowcharts. When the computer program product is run in a computer system, the program codes are used to make the computer system implement the product recommendation method provided by the embodiments of the present application.

[0119] The above-described functions defined in the system / device / apparatus of the embodiments of the present application are performed when the computer program is executed by the processor 501. According to an embodiment of the present application, the above-described system, device, module, unit, etc. can be implemented by computer program modules.

[0120] In one embodiment, the computer program can rely on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program can also be transmitted, distributed, and downloaded in the form of a signal on a network medium, and be downloaded and installed through the communication part 509, and / or installed from the detachable medium 511. The program codes contained in the computer program can be transmitted by any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

[0121] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the detachable medium 511. When the computer program is executed by the processor 501, the above-described functions defined in the system of the embodiments of the present application are performed. According to an embodiment of the present application, the above-described system, device, apparatus, module, unit, etc. can be implemented by computer program modules.

[0122] According to embodiments of the present application, program code for implementing the computer programs provided by embodiments of the present application can be written in any combination of one or more programming languages, and can be implemented in a high-level procedural and / or object-oriented programming language, and / or in assembly / machine language. Programming languages include, but are not limited to, Java, C++, python, "C", or the like. Program code can execute entirely on a user's computing device, partly on the user's device, as a stand-alone software package, partly on a remote computing device, or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computing device, such as through the Internet using an Internet Service Provider.

[0123] The computer program instructions can also be loaded onto a computer or other programmable information processing apparatus to cause a series of operations to be performed on the computer or other programmable information processing apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable information processing apparatus implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0124] Those skilled in the art will understand that features recited in the various embodiments of the present application can be combined and / or integrated in various combinations and / or permutations, even if such combinations and / or permutations are not expressly noted in the present application. In particular, features recited in the various embodiments of the present application can be combined and / or integrated in various combinations and / or permutations without departing from the spirit and teachings of the present application. All such combinations and / or integrations are within the scope of the present application.

Claims

1. A product recommendation method, characterized in that, The method includes: Acquire product resource data of the target object's preferred products and candidate resource data of candidate products. The candidate text data in the candidate resource data is used to describe the product operation activities of the candidate products in a preset historical period. The preset historical period is determined based on the interactive operations for the preferred products. The product resource data and the candidate resource data are processed based on a correlation algorithm to obtain the degree of correlation between the preferred product and the candidate product. Based on the degree of relevance, intermediate products that do not meet the preset relevance threshold are determined from the candidate products; The candidate text data is semantically understood using a large language model to obtain the correlation influence relationship between the preferred product and the intermediate product. The correlation influence relationship represents the click tendency of the candidate text data towards the intermediate product. If the correlation relationship meets the preset negative correlation condition, the target product is determined from the intermediate products, and the target product and the preferred product are pushed to the target object.

2. The method according to claim 1, characterized in that, The process of processing the product resource data and the candidate resource data based on the correlation algorithm to obtain the degree of correlation between the preferred product and the candidate product includes: Based on the product resource data, determine the mean product click-through rate and the standard deviation of product click-through rate for the preset historical period. Based on the candidate resource data, determine the mean and standard deviation of the candidate click-through rate for the preset historical time period; Determine the difference between the daily click-through rate of the preferred product and the average click-through rate of the product during the preset historical period; Determine the difference between the daily click-through rate of the candidate products and the average click-through rate of the candidate products during the preset historical period; The correlation between the preferred product and the candidate product is determined based on the product click-through rate difference, the candidate click-through rate difference, the product click-through rate standard deviation, and the candidate click-through rate standard deviation.

3. The method according to claim 1, characterized in that, The step of determining the target product from the intermediate products includes: The correlation degree and the correlation impact value are processed using a time series model to obtain the predicted correlation between the preferred product and the intermediate product. If the predicted correlation is less than a preset value, the target product is determined.

4. The method according to claim 3, characterized in that, The step of using a time series model to process the correlation degree and the correlation impact value to obtain the predicted correlation between the preferred product and the intermediate product includes: The total correlation value is obtained by multiplying the correlation degree between the preferred product and the intermediate product by the correlation impact value. The total correlation value is processed using a time series model to obtain the predicted correlation between the preferred product and the intermediate product.

5. The method according to claim 1, characterized in that, Obtaining the preferred products of the target object further includes: Receive a list of answers returned by the target object for a preset list of questions, the preset list of questions including at least one question for mining product preference information; Semantic parsing is performed on the answer list to extract preference feature vectors related to product attributes; The preference feature vector is matched with the feature vectors of each product in the candidate product feature library to obtain the matching score of each candidate product. Based on the matching score, the target object's preferred product is determined from the candidate products.

6. The method according to claim 1, characterized in that, The method further includes: From the candidate products, a benchmark product whose click-through rate (CTR) meets a preset CTR threshold is determined. The benchmark product's CTR attributes are the same as the product attributes of at least one of the target product and the preferred product. Based on the benchmark product resource data of the benchmark product, calculate the product similarity between the benchmark product and the target product and the preferred product, respectively. If the product similarity is greater than the preset product similarity, the target product and the preferred product are recommended to the target object.

7. A product recommendation device, characterized in that, The device includes: The acquisition module is used to acquire product resource data of the target object's preferred products and candidate resource data of candidate products. The candidate text data in the candidate resource data is used to describe the product operation activities of the candidate products in a preset historical period. The preset historical period is determined based on the interactive operations for the preferred products. The correlation determination module is used to process the product resource data and the candidate resource data based on a correlation algorithm to obtain the degree of correlation between the preferred product and the candidate product. An intermediate product determination module is used to determine intermediate products that do not meet a preset relevance threshold from the candidate products based on the relevance degree. The semantic understanding module is used to perform semantic understanding on the candidate text data using a large language model to obtain the correlation influence relationship between the preferred product and the intermediate product. The correlation influence relationship represents the click tendency state of the candidate text data towards the intermediate product. The recommendation module is used to determine a target product from the intermediate products when the correlation influence relationship meets a preset negative correlation condition, and to push the target product and the preferred product to the target object.

8. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 6.

Images