Product recommendation method and device, electronic equipment, storage medium and program product

Abstract

The present disclosure provides a product recommendation method and device, electronic equipment and storage medium, which can be applied in the field of finance or other fields. The method comprises: extracting at least one first content feature of a new product; assigning an initial first weight to each of the first content features, and updating the first weight of the first content feature according to user feedback, which is obtained based on the first content feature of the new product in a first time period; obtaining second content features of at least one old product; determining the weight Euclidean distance between each two products in a product set according to the first content features, the updated first weight and the second content features; determining a neighbor set of the new product according to the weight Euclidean distance, wherein the neighbor set contains at least one old product; determining a predicted score of the new product according to the weight Euclidean distance between the old product in the neighbor set and the new product and the historical score of the old product in the neighbor set; and determining at least one new product recommended to the user according to the predicted score.

Description

Technical Field

[0001] This disclosure relates to the field of big data technology, and more specifically to a product recommendation method, apparatus, electronic device, storage medium, and program product. Background Technology

[0002] Traditional recommendation methods employ collaborative filtering. However, because collaborative filtering relies on historical rating data for users and products, which is lacking for new users or new products, it suffers from a cold start problem, leading to situations where recommendations fail or are poorly implemented. Existing technologies address this by using Euclidean distance between two products as the clustering criterion. However, since users have varying preferences for different content features, the Euclidean distance calculated using this method, treating all content features equally, sometimes fails to accurately reflect the similarity between two products in practical applications of new product recommendations. This negatively impacts subsequent hierarchical clustering and the final result of collaborative filtering, ultimately reducing the quality of new item (product) recommendations. Summary of the Invention

[0003] In view of the above problems, this disclosure provides a product recommendation method, apparatus, electronic device, readable storage medium, and computer program product, which can not only get rid of the predicament of traditional collaborative filtering methods relying on historical product ratings, but also effectively solve the problem of low product recommendation quality in the prior art, achieve the effect of matching product recommendations with user preferences, and achieve the goal of accurate recommendation.

[0004] According to a first aspect of this disclosure, a product recommendation method is provided, the method comprising: extracting at least one first content feature of a new product; assigning an initial first weight to each of the at least one first content feature, and updating the first weight of the first content feature based on user feedback, the user feedback being obtained based on feedback of the first content feature of the new product within a first time period; obtaining a second content feature of at least one old product; determining a weighted Euclidean distance between every two products in a product set consisting of the new product and the old product based on the first content feature, the updated first weight, and the second content feature; determining a neighbor set of the new product based on the weighted Euclidean distance, the neighbor set containing at least one old product; determining a predicted rating of the new product based on the weighted Euclidean distance between the old product and the new product in the neighbor set and the historical rating of the old product in the neighbor set; and determining at least one new product to recommend to a user based on the predicted rating.

[0005] In some exemplary embodiments of this disclosure, after extracting at least one first content feature of the new product, the first content feature is subjected to initialization preprocessing, wherein the initialization preprocessing includes: for first content features containing numbers, extracting numbers and removing non-numeric content; for first content features containing a range of numbers with upper and lower limits, extracting the median value of the range of numbers; and for first content features not containing numbers, leaving them unchanged.

[0006] In some exemplary embodiments of this disclosure, when the new product is first recommended, the initial first weight of the first content feature included in each of the new products is 1.

[0007] In some exemplary embodiments of this disclosure, updating the first weight of the first content feature based on user feedback includes: updating the first weight of each of the first content features of the new product in a second time period based on the user feedback associated with each of the first content features obtained in the first time period.

[0008] In some exemplary embodiments of this disclosure, determining the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product based on the first content feature, the updated first weight, and the second content feature includes: determining a first value based on the first content feature of the new product and the second content feature of the old product; and determining the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product based on the first weight and the first value.

[0009] In some exemplary embodiments of this disclosure, when the first content feature and the second content feature are numbers, the first value is determined based on the numbers contained in the first content feature and the second content feature; when the first content feature and the second content feature are non-numbers, text matching is performed, and if the texts are the same, the first value is determined to be 0, and if the texts are different, the first value is determined to be 1.

[0010] In some exemplary embodiments of this disclosure, determining the neighbor set of the new product based on the weighted Euclidean distance includes: step 1, each product in the product set composed of the new product and old products is respectively regarded as a cluster; step 2, the two closest clusters in the cluster are merged; step 3, step 2 is repeated until the number of clusters is equal to a preset threshold, and the merging of clusters ends, and the other old products in the cluster where the new product is located are determined as the neighbor set of the new product.

[0011] In some exemplary embodiments of this disclosure, determining a predicted score for the new product based on the weighted Euclidean distance between the old products in the neighborhood set and the new product, as well as the historical scores of the old products in the neighborhood set, includes: determining a second value based on the historical scores of the old products in the neighborhood set and the weighted Euclidean distance of the new product associated with the old products in the neighborhood set; and determining a predicted score for the new product based on the second value and the sum of the weighted Euclidean distances between the old products in the neighborhood set and the new product.

[0012] A second aspect of this disclosure provides a product recommendation apparatus, the apparatus comprising: an extraction module configured to extract at least one first content feature of a new product; an update module configured to assign an initial first weight to each of the at least one first content feature and update the first weight of the first content feature based on user feedback, wherein the user feedback is based on feedback of the first content feature of the new product within a first time period; an acquisition module configured to acquire a second content feature of at least one old product; a first determination module configured to determine a weighted Euclidean distance between every two products in a product set consisting of the new product and the old product based on the first content feature, the updated first weight, and the second content feature; a second determination module configured to determine a neighbor set of the new product based on the weighted Euclidean distance, wherein the neighbor set contains at least one old product; a third determination module configured to determine a predicted rating of the new product based on the weighted Euclidean distance between the old product and the new product in the neighbor set and the historical rating of the old product in the neighbor set; and a fourth determination module configured to determine at least one new product to recommend to a user based on the predicted rating.

[0013] In some exemplary embodiments of this disclosure, the product recommendation device further includes a preprocessing module, which is configured to: after extracting at least one first content feature of a new product, perform initialization preprocessing on the first content feature, wherein the initialization preprocessing includes: for the first content feature containing numbers, extracting numbers and removing non-numeric content; for the first content feature containing a range of numbers with upper and lower limits, extracting the median value of the range of numbers; and for the first content feature not containing numbers, leaving it unchanged.

[0014] In some exemplary embodiments of this disclosure, the product recommendation device further includes an initial weight determination module, which is configured such that when a new product is recommended for the first time, the initial first weight of each first content feature included in the new product is 1.

[0015] In some exemplary embodiments of this disclosure, the update module further includes an update submodule configured to: update the first weight of each of the first content features of the new product in a second time period based on the user feedback obtained during the first time period and associated with each of the first content features.

[0016] In some exemplary embodiments of this disclosure, the first determining module includes a first determining submodule, which is configured to: determine the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product based on the first content feature, the updated first weight, and the second content feature, including: determining a first value based on the first content feature of the new product and the second content feature of the old product; and determining the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product based on the first weight and the first value.

[0017] In some exemplary embodiments of this disclosure, the first determining submodule further includes a first determining subunit, which is configured to: when the first content feature and the second content feature are numbers, determine the first value based on the numbers contained in the first content feature and the second content feature; when the first content feature and the second content feature are non-numeric, perform text matching, and if the text is the same, determine the first value as 0, and if the text is different, determine the first value as 1.

[0018] In some exemplary embodiments of this disclosure, the second determining module includes a second determining submodule, which is configured to: Step 1, each product in the product set composed of the new product and the old products is respectively regarded as a cluster; Step 2, the two closest clusters in the cluster are merged; Step 3, Step 2 is repeated until the number of clusters is equal to a preset threshold, and the merging of clusters ends, and the other old products in the cluster where the new product is located are determined as the neighbor set of the new product.

[0019] In some exemplary embodiments of this disclosure, the third determining module includes a third determining submodule, which is configured to: determine a second value based on the historical ratings of old products in the neighbor set and the weighted Euclidean distance of the new product associated with the old products in the neighbor set; and determine a predicted rating of the new product based on the second value and the sum of the weighted Euclidean distances between the old products in the neighbor set and the new product.

[0020] A third aspect of this disclosure provides an electronic device comprising: one or more processors; and a storage device for storing executable instructions that, when executed by the processor, implement the method described above.

[0021] A fourth aspect of this disclosure provides a computer-readable storage medium having executable instructions stored thereon that, when executed by a processor, implement the method described above.

[0022] A fifth aspect of this disclosure provides a computer program product including a computer program that, when executed by a processor, implements the method described above.

[0023] According to embodiments of this disclosure, by extracting multiple first content features of a new product and assigning an initial first weight to each first content feature, and dynamically updating the first weight of each first content feature based on user feedback, it is possible to more accurately obtain the degree of user preference for different content features of the new product. Based on the first content features of the new product, the updated first weight, and the second content features of the old product, the weighted Euclidean distance between each pair of products is calculated. Based on this weighted Euclidean distance, the old products in the neighbor set of the new product are determined. These old products in the neighbor set have a high degree of similarity to the new product. Then, based on the historical ratings of the old products in the neighbor set and the weighted Euclidean distance between the old products and the new product, the rating of the new product is predicted. Recommending the new product based on this predicted rating can overcome the limitations of traditional collaborative filtering that relies on historical product ratings, align with user preferences, achieve precise marketing, and thus improve user satisfaction. Attached Figure Description

[0024] The foregoing contents, as well as other objects, features, and advantages of this disclosure, will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0025] Figure 1 The diagram illustrates a system architecture to which the product recommendation method of the present disclosure can be applied;

[0026] Figure 2 A flowchart illustrating a product recommendation method according to an embodiment of the present disclosure is shown schematically;

[0027] Figure 3 This illustration schematically shows the flow of the product recommendation method in the initialization preprocessing of an embodiment of the present disclosure;

[0028] Figure 4 This illustration schematically shows the process of the product recommendation method of this disclosure when making an initial recommendation of a new product;

[0029] Figure 5 This illustration schematically shows the flow of the product recommendation method in operation S220 according to an embodiment of the present disclosure;

[0030] Figure 6 The flowchart illustrating the product recommendation method of this disclosure in operation S240 is shown in the schematic diagram.

[0031] Figure 7 The flowchart illustrating the product recommendation method of this disclosure in operation S250 is shown in the schematic diagram.

[0032] Figure 8 The flowchart illustrating the product recommendation method of this disclosure in operation S260 is shown in the schematic diagram.

[0033] Figure 9 A schematic block diagram of a product recommendation device according to an embodiment of the present disclosure is shown.

[0034] Figure 10 A block diagram schematically illustrates an electronic device suitable for implementing a product recommendation method according to an embodiment of the present disclosure. Detailed Implementation

[0035] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0036] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0037] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0038] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "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 A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0039] In the technical solution disclosed herein, the acquisition, storage, and application of user personal information comply with the provisions of relevant laws and regulations, necessary confidentiality measures have been taken, and there is no violation of public order and good morals.

[0040] In the technical solution disclosed herein, all operations involving the acquisition, storage, and application of user personal information have been authorized by the user.

[0041] In this disclosure, the term "content features" may refer to all the key elements of a product that are visible to the user. For example, the content features of a financial product may include features that the user can see when purchasing a wealth management product, such as self-operated / agent-operated, minimum purchase amount, risk rating, minimum holding period, performance benchmark, currency, etc.

[0042] The term "collaborative filtering" can refer to recommending information that a user might be interested in by leveraging the preferences of a group of like-minded individuals with shared experiences. Individuals respond to information to a considerable extent (such as rating) through a collaborative mechanism and record their responses to achieve the purpose of filtering, thereby helping others to select information. The responses are not necessarily limited to those who are particularly interested; recording information that is particularly uninteresting is also quite important.

[0043] In response to the "information overload" problem, various product recommendation technologies have emerged in financial systems, including content-based recommendation, collaborative filtering-based recommendation, and hybrid recommendation technologies. Collaborative filtering is one of the most widely used recommendation technologies. However, collaborative filtering suffers from a cold start problem because it relies on historical rating data for users and products. New users or new products lack this data, leading to situations where recommendations are either impossible or ineffective, thus reducing user satisfaction and impacting sales performance.

[0044] For example, existing literature discloses a solution to the cold-start problem in collaborative filtering algorithms. This solution utilizes the content information features of items (products) combined with an agglomerative hierarchical clustering algorithm to obtain the classification results of items (products), find the neighboring items (products) of the target item (product), and thus complete the recommendation. It primarily addresses the cold-start problem of new items (products) in traditional collaborative filtering recommendation techniques, completing the recommendation of new items (products) and improving the algorithm's recommendation accuracy and quality.

[0045] In the aforementioned existing technologies, to address the cold start problem of new projects (products), the Euclidean distance between two products is calculated based on the content features of the project (product) as the basis for subsequent hierarchical clustering. However, in the actual business scenario of financial product recommendation, users have different preferences for different content features. Therefore, the existing technology's method of calculating the Euclidean distance for all content features "equally" and performing hierarchical clustering based on this Euclidean distance sometimes fails to accurately reflect the similarity between two products in practical applications of financial product recommendation. This can affect the final results of subsequent hierarchical clustering and collaborative filtering recommendations, thus reducing the recommendation quality of new projects (products).

[0046] To address the aforementioned issues, this disclosure provides a product recommendation method, apparatus, electronic device, readable storage medium, and computer program product. This method can acquire a user's preference for multiple content features of a new product, and perform hierarchical clustering based on the weighted Euclidean distance between the multiple content features of new and old products. It then obtains a neighbor set containing old products highly similar to the new product, predicts a rating, and makes dynamic recommendations. The recommended new products align with user preferences, thereby improving recommendation quality and user satisfaction. The product recommendation method includes, but is not limited to: extracting at least one first content feature of a new product; assigning an initial first weight to each of the at least one first content feature, and updating the first weight of the first content feature based on user feedback, where user feedback is based on the first content features of the new product within a first time period; acquiring at least one second content feature of an old product; determining the weighted Euclidean distance between every two products in the product set consisting of the new and old products based on the first content feature, the updated first weight, and the second content feature; determining the neighbor set of the new product based on the weighted Euclidean distance, where the neighbor set contains at least one old product; determining the predicted rating of the new product based on the weighted Euclidean distance between the old and new products in the neighbor set and the historical ratings of the old products in the neighbor set; and determining at least one new product to recommend to the user based on the predicted rating.

[0047] According to embodiments of this disclosure, by extracting multiple first content features of a new product and assigning an initial first weight to each first content feature, and dynamically updating the first weight of each first content feature based on user feedback, it is possible to more accurately obtain the degree of user preference for different content features of the new product. Based on the first content features of the new product, the updated first weight, and the second content features of the old product, the weighted Euclidean distance between each pair of products is calculated. Based on this weighted Euclidean distance, the old products in the neighbor set of the new product are determined. These old products in the neighbor set have a high degree of similarity to the new product. Then, based on the historical ratings of the old products in the neighbor set and the weighted Euclidean distance between the old products and the new product, the rating of the new product is predicted. Recommending the new product based on this predicted rating can overcome the limitations of traditional collaborative filtering that relies on historical product ratings, align with user preferences, achieve precise marketing, and thus improve user satisfaction.

[0048] Figure 1 The diagram illustrates a system architecture to which the product recommendation method of embodiments of this disclosure can be applied. It should be noted that... Figure 1The examples shown are merely examples of system architectures applicable to the embodiments of this disclosure, intended to help those skilled in the art understand the technical content of this disclosure. However, they do not imply that the embodiments of this disclosure cannot be used in other devices, systems, environments, or scenarios. It should be noted that the product recommendation method provided in the embodiments of this disclosure can be used in the fields of big data technology and finance in relation to product recommendation, and can also be used in any field other than finance. The product recommendation method and apparatus provided in the embodiments of this disclosure do not limit the application field.

[0049] like Figure 1 As shown, an exemplary system architecture 100 to which the product recommendation method can be applied may include terminal devices 101, 102, and 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the terminal devices 101, 102, and 103 and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0050] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as email client applications, file processing applications, shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0051] Terminal devices 101, 102, and 103 can be various electronic devices with displays that support functions such as data input, file sending, data analysis, data processing, and web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0052] Server 105 can be a server providing various services, such as a backend management server (for example only) that supports data obtained by users using terminal devices 101, 102, and 103 or websites they browse. The backend management server can analyze and process received user requests and other data, and feed back the processing results (such as web pages, information, or data obtained or generated based on user requests) to the terminal devices. It can also analyze or process files sent by users and control the terminal devices based on the processing results, such as restricting access to the terminal devices.

[0053] It should be noted that the product recommendation method provided in this disclosure embodiment can generally be executed by server 105. Correspondingly, the product recommendation device provided in this disclosure embodiment can generally be located in server 105. The product recommendation method provided in this disclosure embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105. Correspondingly, the product recommendation device provided in this disclosure embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105.

[0054] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0055] The following will be through Figures 2 to 8 The product recommendation method of the disclosed embodiments is described in detail.

[0056] Figure 2 A flowchart illustrating a product recommendation method according to an embodiment of this disclosure is shown schematically.

[0057] like Figure 2 As shown, the process 200 of the product recommendation method disclosed herein includes operations S210 to S270.

[0058] In operation S210, at least one first content feature of the new product is extracted.

[0059] For example, a new product may be a financial product or project in a business scenario of financial product recommendation, and the first content feature may be all the key elements of the financial product that are visible to the user.

[0060] For example, when users purchase financial products, the features they can see, such as whether the product is self-operated or distributed, the minimum purchase amount, the risk rating, the minimum holding period, the performance benchmark, and the currency, are the primary content features of the financial products they are using.

[0061] Figure 3 The illustration schematically shows the flow of the product recommendation method in the initialization preprocessing of an embodiment of the present disclosure.

[0062] like Figure 3 As shown, the process 300 for initialization preprocessing after extracting at least one first content feature of the new product includes operations S310 to S330.

[0063] In operation S310, for the first content feature containing numbers, the numbers are extracted and the non-digit content is removed.

[0064] In operation S320, for the first content feature containing an upper and lower limit of the number range, the median value of the number range is extracted.

[0065] In operation S330, the first content feature that does not contain numbers remains unchanged.

[0066] According to embodiments of this disclosure, initialization preprocessing facilitates subsequent process operations.

[0067] In operation S220, each of the at least one first content feature is assigned an initial first weight, and the first weight of the first content feature is updated based on user feedback, which is obtained based on the feedback of the first content feature of the new product within a first time period.

[0068] Figure 4 The illustration schematically shows the operation S400 of the product recommendation method of this disclosure when making an initial recommendation of a new product.

[0069] In operation S400, when a new product is initially recommended, the initial first weight of the first content feature included in each new product is assigned a value of 1. According to operation S400, by setting the initial first weight of each first content feature to 1 during the initial recommendation, it is possible to prevent the distortion of subsequent user feedback due to random recommendations, thereby making subsequent product recommendations more accurate.

[0070] Figure 5 The specific flow of the product recommendation method in operation S220 of this disclosure embodiment is illustrated schematically.

[0071] For example, operation S220 further includes operation S221, in which the first weight of each first content feature of the new product in the second time period is updated based on user feedback associated with each first content feature obtained in the first time period.

[0072] In some exemplary embodiments of this disclosure, user feedback is a preference feedback count aggregated from day one to day N for each first content feature of the new product, and an updated first weight for each first content feature on day N+1 is calculated based on the user feedback data collected from day one to day N. The specific formula is as follows:

[0073] (1)

[0074] in, It is the first weight of the k-th primary content feature among the n primary content features of the new product after the update on day N+1. It is the total number of user preference feedbacks for the k-th primary content feature from day 1 to day N, where n is the total number of primary content features of the new product.

[0075] For example, if the new product is a bank wealth management product, and five first content features are extracted from Table 1 below, and the user preference feedback counts for each first content feature of the new product in the first two days are shown in Table 1 below, then the updated first weight of each first content feature of the new product on the third day is calculated based on the user feedback in the first two days, as shown in Table 2 below.

[0076] Table 1

[0077]

[0078] Table 2

[0079]

[0080] In some exemplary embodiments of this disclosure, within a specified time period, when a user purchases a new product, or when a user chooses to refuse a new product recommended to them, the system collects the user's preferences for content features and records feedback data.

[0081] For example, when a new product is recommended to a user but the user chooses to decline, feedback is collected on which features of the new product the user is dissatisfied with; after the user purchases the new product, feedback is collected on which features of the new product the user is satisfied with. The specific recording method is as follows:

[0082] When a new product is launched, the initial count of each first content feature of the new product is 0. Then, regardless of whether the collected feedback is positive or negative from users regarding a certain first content feature, the preference feedback count of that first content feature is incremented by 1. User preference feedback is collected and counted daily within a specified time period. The preference feedback count accumulated from the first day to the Nth day is used to calculate the first weight updated on the N+1th day using formula (1).

[0083] For example, user feedback can be obtained through the following steps: When recommending users to log in to the product purchase system, a new product advertisement pops up to recommend the new product to the user. The advertisement describes several primary content features of the new product before preprocessing. When the user rejects the advertisement, Table 3 below is provided for the user to provide feedback on which primary content features of the new product they are dissatisfied with; similarly, after the user purchases the new product, Table 4 below is provided for the user to provide feedback.

[0084] Table 3

[0085]

[0086] Table 4

[0087]

[0088] For example, if a user is dissatisfied with a new product's risk rating being too high, holding period being too long, or performance benchmark (return rate) being too low, they can check the boxes for risk rating, minimum holding period, and performance benchmark (return rate) in Table 3 above, and the preference feedback count for these three features will increment by 1. Similarly, if a user is satisfied with the new product's low minimum purchase amount and high performance benchmark (return rate), they can check the boxes for minimum purchase amount and performance benchmark (return rate) in Table 4 above, and the preference feedback count for these two features will increment by 1.

[0089] According to embodiments of this disclosure, by dynamically adjusting the weights corresponding to each first content feature of a new product based on user feedback (daily) in the short term after the product launch, it is possible to more accurately obtain the degree of user preference for different first content features of the new product, providing a basis for subsequently determining product recommendations that fit user preferences.

[0090] In operation S230, at least one second content feature of an old product is acquired.

[0091] In some exemplary embodiments of this disclosure, after obtaining the second content feature of the old product, the obtained second content feature is subjected to initialization preprocessing, which is the same as the initialization preprocessing performed on the first content feature of the new product in operation S210.

[0092] For example, old products A and B, and new product C are bank wealth management products. Each of products A, B, and C has five content features extracted from Table 5. Each content feature of products A, B, and C undergoes initialization preprocessing: the first content feature does not contain numbers and therefore remains unchanged; the second, third, and fourth content features contain numbers and therefore directly take the numbers; the fifth content feature contains a range of numbers and therefore takes the midpoint of the range. This yields the content features of products A, B, and C in Table 5 after initialization preprocessing, as detailed in Table 6.

[0093] Table 5

[0094]

[0095] Table 6

[0096]

[0097] According to embodiments of this disclosure, initialization preprocessing of the first content features of the new product and the second content features of the old product facilitates subsequent calculations based on the content features.

[0098] In operation S240, the weighted Euclidean distance between every two products in the product set consisting of new products and old products is determined based on the first content feature, the updated first weight, and the second content feature.

[0099] Figure 6 The flowchart illustrating the product recommendation method of an embodiment of the present disclosure during operation S240 is shown in the illustration.

[0100] like Figure 6 As shown, operation S240 includes operations S241 to S242.

[0101] In operation S241, a first value is determined based on the first content feature of the new product and the second content feature of the old product.

[0102] In operation S242, the weighted Euclidean distance between every two products in the product set consisting of new products and old products is determined based on the first weight and the first value.

[0103] For example, the weighted Euclidean distance between every two products in the product set consisting of all new and old products is calculated according to the following formula (2):

[0104] (2)

[0105] In formula (2), It is the weighted Euclidean distance between product i and product j in the product set. It is the k-th content feature of product i. It is the k-th content feature of product j. - It is a first value determined based on the first content characteristics of the new product and the second content characteristics of the old product; It is the first weight after updating the kth content feature of the current time period, calculated according to the aforementioned formula (1).

[0106] In embodiments of this disclosure, determining the first value based on the first content feature of the new product and the second content feature of the old product includes: when the first content feature and the second content feature are numbers, determining the first value based on the numbers contained in the first content feature and the second content feature; when the first content feature and the second content feature are non-numbers, performing text matching, and if the text is the same, determining the first value as 0, and if the text is different, determining the first value as 1.

[0107] For example, on the third day, the first weight updated on the third day is calculated according to the aforementioned formula (1), and then the weight Euclidean distance between every two products in the entire product set is calculated according to the aforementioned formula (2), as shown in Table 7.

[0108] Table 7

[0109]

[0110] In operation S250, the neighborhood set of the new product is determined based on the weighted Euclidean distance. The neighborhood set contains at least one old product.

[0111] In some exemplary embodiments of this disclosure, the neighborhood set of new products is determined by agglomerative hierarchical clustering based on the weighted Euclidean distance between products obtained by operation S240.

[0112] Figure 7 The flowchart of the product recommendation method according to an embodiment of the present disclosure in operation S250 is illustrated schematically.

[0113] The following is combined with Figure 7 The procedure for operating S250 is explained in detail, such as... Figure 7 As shown, determining the neighbor set of the new product based on the weighted Euclidean distance includes operations S251 to S254.

[0114] In operation S251, each product in the product set consisting of new and old products is treated as a separate cluster.

[0115] In operation S252, the two closest clusters in the cluster are merged.

[0116] In some exemplary embodiments of this disclosure, two clusters and The distance between them is calculated using the shortest distance method, which is the weighted Euclidean distance between the two closest points (product i and product j) in two clusters. This is the distance between two clusters. The specific calculation formula is as follows:

[0117] (4)

[0118] In formula (4), It is a cluster and The distance between them.

[0119] In operation S253, it is determined whether the number of clusters is equal to a preset threshold. If the number of clusters is equal to the preset threshold, operation S254 is executed. If the number of clusters is greater than the preset threshold, operation S252 is executed.

[0120] In operation S254, the cluster merging is completed, and the other old products in the cluster where the new product is located are identified as the neighbor set of the new product.

[0121] For example, a product set consisting of new and old products includes {old product A, old product B, new product C}, and a preset threshold for the cluster is set to 1. The following steps are performed to determine the neighbor set of the new product:

[0122] Initially, each product is a cluster, resulting in three clusters: {Product A}, {Product B}, and {Product C}. The distances between each cluster are calculated using formula (4). The distance between clusters {Product A} and {Product B} is 372.03, the distance between clusters {Product A} and {Product C} is 223.10, and the distance between clusters {Product B} and {Product C} is 159.69. Clusters {Product B} and {Product C} are the closest and are merged to obtain two new clusters: {Product B, Product C} and {Product A}. Cluster merging continues until one cluster, {Product B, Product C, Product A}, is obtained, at which point the merging process terminates. Therefore, [Product B, Product A] is the neighbor set of the new product C.

[0123] Similarly, when the preset threshold for clusters is 2, we eventually get two clusters: {Product B, Product C} and {Product A}. Then, [Product B] is the neighbor set of the new product C.

[0124] In some exemplary embodiments of this disclosure, the preset threshold for clusters can be preset to a fixed value, such as 2, based on experience, or it can be set to a certain proportion of the initial number of clusters, such as 10%. In other optional embodiments of this disclosure, the preset threshold for clusters can be adjusted according to actual needs to meet the requirements of different scenarios.

[0125] According to embodiments of this disclosure, the similarity between new and old products is measured by Euclidean distance determined based on multiple content features of the product, and then the neighbor set is found by upward clustering. This avoids the dilemma of traditional collaborative filtering algorithms being unable to calculate similarity and find neighbors for new products due to the lack of historical ratings, and to a certain extent solves the cold start problem of using traditional collaborative filtering algorithms when recommending new financial products.

[0126] In operation S260, the predicted score of the new product is determined based on the weighted Euclidean distance between the old and new products in the neighborhood set and the historical scores of the old products in the neighborhood set.

[0127] Figure 8 The flowchart illustrating the product recommendation method of an embodiment of the present disclosure during operation S260 is shown in the illustration.

[0128] like Figure 8 As shown, operation S260 includes operations S261 to S262.

[0129] In operation S261, the second value is determined based on the historical scores of the old products in the neighbor set and the weighted Euclidean distance of the new products associated with the old products in the neighbor set.

[0130] In operation S262, the predicted score of the new product is determined based on the second value and the sum of the weighted Euclidean distances between the old and new products in the neighbor set.

[0131] For example, the predicted score for the new product is calculated according to the following formula (3):

[0132] (3)

[0133] In formula (3), It is the predicted score for new product j. It is the set of old products i whose neighbors are the new product j. It is the historical rating of old product i by user u for the neighborhood set of new product j. It is the weighted Euclidean distance between product i and product j calculated according to the aforementioned formula (2). and The product is a second value determined based on the historical scores of the old products in the neighborhood set and the weighted Euclidean distance between the old product and the new product associated with it.

[0134] For example, according to operation S250, the neighbor set of new product C is determined to be [old product B, old product A]. The user's historical ratings for old products B and A are 80 and 90 respectively. On the third day, according to operation S240, the weighted Euclidean distances of BC and AC are 159.69 and 223.10 (see Table 7). Then, according to the aforementioned formula (3), the predicted rating of new product C on the third day is calculated as ((159.69×80+223.10×90)) / (159.69+223.10)=85.82826092635649. The calculation result is rounded to two decimal places, and the predicted rating of new product C is 85.83.

[0135] In operation S270, at least one new product is determined to be recommended to the user based on the predicted score.

[0136] In some exemplary embodiments of this disclosure, the predicted score of each new product is determined according to operations S210 to S250, and the new products are sorted from high to low according to the predicted scores of all new products. The top M new products in the sorted list are selected for recommendation, where M is a positive integer, that is, the recommended new products can be one or more.

[0137] For example, within a specified short period of time (not enough time to generate sufficient user ratings for the new product, but enough user preference feedback on content features can be obtained), taking ten days as an example, on the first day, the predicted rating of the new product is determined and the product is recommended based on an initial first weight of 1 for each first content feature of the new product; on the second day, the first weight is updated based on the user preferences for different first content features from the first day, and the predicted rating of the new product is determined and the product is recommended based on the updated weight; on the third day, the first weight is updated based on the user preferences for different first content features from the first and second days, and the predicted rating of the new product is determined and the product is recommended based on the updated weight, and so on, until the tenth day. During this period, the first weight is dynamically adjusted every day, thereby realizing the dynamic updating of weight Euclidean distance, neighbor set, predicted rating, and user recommendation list, and the final recommendation result tends to be more accurate.

[0138] According to embodiments of this disclosure, by acquiring multiple content features of a new product and multiple second content features of an old product, and dynamically updating the first weight of each first content feature based on user feedback, it is possible to more accurately obtain the user's preference for different first content features of the new product. Then, based on each content feature and the updated first weight, the weighted Euclidean distance between products can be determined, enabling the identification of products with high similarity. Furthermore, by predicting the rating of the new product based on the historical ratings of the old products in the highly similar neighbor set and the weighted Euclidean distance between the old and new products, recommending the new product based on this predicted rating can overcome the limitations of traditional collaborative filtering that relies on historical product ratings. This addresses the cold start problem when recommending products using traditional collaborative filtering algorithms to some extent, achieving the goal of aligning with user preferences and precise marketing, thereby improving user satisfaction. Especially in the financial sector, such as banking, by dynamically adjusting the weights corresponding to each first content feature of the new product daily based on user feedback within a short period after its launch, dynamic updates of the weighted Euclidean distance, neighbor set, predicted rating, and user recommendation list can be achieved, better meeting the application scenario requirements of financial products and improving recommendation quality.

[0139] Figure 9 A schematic block diagram of a product recommendation device according to an embodiment of the present disclosure is shown.

[0140] like Figure 9 As shown, the product recommendation device 500 of this embodiment includes an extraction module 501, an update module 502, an acquisition module 503, a first determination module 504, a second determination module 505, a third determination module 506, and a fourth determination module 507.

[0141] Extraction module 501 is configured to extract at least one first content feature of the new product. In one embodiment, extraction module 501 may be used to perform operation S210 described above, which will not be repeated here.

[0142] The update module 502 is configured to assign an initial first weight to each of the at least one first content feature, and update the first weight of the first content feature based on user feedback, wherein the user feedback is obtained based on the first content features of new products within a first time period. In one embodiment, the update module 502 may be used to perform the operation S220 described above, which will not be repeated here.

[0143] The acquisition module 503 is configured to acquire at least one second content feature of an older product. In one embodiment, the acquisition module 503 may be used to perform the operation S230 described above, which will not be repeated here.

[0144] The first determining module 504 is configured to determine the weighted Euclidean distance between every two products in the product set composed of the new product and the old product based on the first content feature, the updated first weight, and the second content feature. In one embodiment, the first determining module 504 can be used to perform the operation S240 described above, which will not be repeated here.

[0145] The second determining module 505 is configured to determine the neighbor set of the new product based on the weighted Euclidean distance, wherein the neighbor set contains at least one old product. In one embodiment, the second determining module 505 may be used to perform the operation S250 described above, which will not be repeated here.

[0146] The third determining module 506 is configured to determine the predicted score of the new product based on the weighted Euclidean distance between the old products in the neighbor set and the new product, as well as the historical scores of the old products in the neighbor set. In one embodiment, the third determining module 506 may be used to perform the operation S260 described above, which will not be repeated here.

[0147] The fourth determining module 507 is configured to determine at least one new product to recommend to the user based on the predicted score. In one embodiment, the fourth determining module 507 may be used to perform the operation S270 described above, which will not be repeated here.

[0148] In some exemplary embodiments of this disclosure, the product recommendation device further includes a preprocessing module, which is configured to: after extracting at least one first content feature of a new product, perform initialization preprocessing on the first content feature, wherein the initialization preprocessing includes: for the first content feature containing numbers, extracting numbers and removing non-numeric content; for the first content feature containing a range of numbers with upper and lower limits, extracting the median value of the range of numbers; and for the first content feature not containing numbers, leaving it unchanged.

[0149] In some exemplary embodiments of this disclosure, the product recommendation device further includes an initial weight determination module, which is configured such that when a new product is recommended for the first time, the initial first weight of each first content feature included in the new product is 1.

[0150] In some exemplary embodiments of this disclosure, the update module further includes an update submodule configured to: update the first weight of each of the first content features of the new product in a second time period based on the user feedback obtained during the first time period and associated with each of the first content features.

[0151] In some exemplary embodiments of this disclosure, the first determining module includes a first determining submodule, which is configured to: determine the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product based on the first content feature, the updated first weight, and the second content feature, including: determining a first value based on the first content feature of the new product and the second content feature of the old product; and determining the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product based on the first weight and the first value.

[0152] In some exemplary embodiments of this disclosure, the first determining submodule further includes a first determining subunit, which is configured to: when the first content feature and the second content feature are numbers, determine the first value based on the numbers contained in the first content feature and the second content feature; when the first content feature and the second content feature are non-numeric, perform text matching, and if the text is the same, determine the first value as 0, and if the text is different, determine the first value as 1.

[0153] In some exemplary embodiments of this disclosure, the second determining module includes a second determining submodule, which is configured to: Step 1, each product in the product set composed of the new product and the old products is respectively regarded as a cluster; Step 2, the two closest clusters in the cluster are merged; Step 3, Step 2 is repeated until the number of clusters is equal to a preset threshold, and the merging of clusters ends, and the other old products in the cluster where the new product is located are determined as the neighbor set of the new product.

[0154] In some exemplary embodiments of this disclosure, the third determining module includes a third determining submodule, which is configured to: determine a second value based on the historical ratings of old products in the neighbor set and the weighted Euclidean distance of the new product associated with the old products in the neighbor set; and determine a predicted rating of the new product based on the second value and the sum of the weighted Euclidean distances between the old products in the neighbor set and the new product.

[0155] According to embodiments of this disclosure, any multiple modules among the extraction module 501, update module 502, acquisition module 503, first determination module 504, second determination module 505, third determination module 506, and fourth determination module 507 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this disclosure, at least one of the extraction module 501, update module 502, acquisition module 503, first determination module 504, second determination module 505, third determination module 506, and fourth determination module 507 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in any one of software, hardware, and firmware methods, or in a suitable combination of any of these methods. Alternatively, at least one of the extraction module 501, update module 502, acquisition module 503, first determination module 504, second determination module 505, third determination module 506, and fourth determination module 507 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0156] Figure 10 A block diagram schematically illustrates an electronic device suitable for implementing a product recommendation method according to an embodiment of the present disclosure. Figure 10 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0157] like Figure 10 As shown, an electronic device 600 according to an embodiment of this disclosure includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage portion 608 into a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this disclosure.

[0158] RAM 603 stores various programs and data required for the operation of electronic device 600. Processor 601, ROM 602, and RAM 603 are interconnected via bus 604. Processor 601 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 602 and / or RAM 603. It should be noted that the programs may also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in said one or more memories.

[0159] According to embodiments of this disclosure, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to a bus 604. The electronic device 600 may also include one or more of the following components connected to the I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 610 as needed so that computer programs read from it can be installed into the storage section 608 as needed.

[0160] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0161] According to embodiments of this disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this disclosure, the computer-readable storage medium may include ROM 602 and / or RAM 603 and / or one or more memories other than ROM 602 and RAM 603 described above.

[0162] Embodiments of this disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement the methods provided in the embodiments of this disclosure.

[0163] When the computer program is executed by the processor 601, it performs the functions defined in the system / apparatus of this disclosure embodiments. According to embodiments of this disclosure, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0164] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 609, and / or installed from the removable medium 611. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0165] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, it performs the functions defined in the system of this disclosure embodiment. According to embodiments of this disclosure, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0166] According to embodiments of this disclosure, program code for executing the computer programs provided in embodiments of this disclosure can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can execute entirely on a user's computing device, partially on a user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0167] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0168] Those skilled in the art will understand that the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. In particular, the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0169] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this disclosure is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A product recommendation method, wherein, include: Extract at least one primary content feature of the new product; Each of the at least one first content feature is assigned an initial first weight, and the first weight of the first content feature is updated based on user feedback, wherein the user feedback is obtained based on the feedback of the first content features of the new product within a first time period; Obtain at least one second content feature of an old product; Based on the first content feature, the updated first weight, and the second content feature, determine the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product; The neighborhood set of the new product is determined based on the weighted Euclidean distance, and the neighborhood set contains at least one old product. The predicted score of the new product is determined based on the weighted Euclidean distance between the old and new products in the neighborhood set and the historical scores of the old products in the neighborhood set. Based on the predicted score, at least one new product is determined to be recommended to the user; Based on the first content feature, the updated first weight, and the second content feature, determine the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product, including: The first value is determined based on the first content feature of the new product and the second content feature of the old product; Based on the first weight and the first value, determine the weighted Euclidean distance between every two products in the product set consisting of the new product and the old product; The weighted Euclidean distance between every two products in the product set consisting of all new and old products is calculated using the following formula: In the formula, It is the weighted Euclidean distance between product i and product j in the product set. It is the k-th content feature of product i. It is the k-th content feature of product j. - It is a first value determined based on the first content characteristics of the new product and the second content characteristics of the old product; It is the first weight after the update of the k-th content feature in the current time period.

2. The method according to claim 1, wherein, After extracting at least one first content feature of the new product, the first content feature is subjected to initialization preprocessing. The initialization preprocessing includes: For the first content feature containing numbers, extract the numbers and remove the non-numeric content; For a first content feature containing an upper and lower limit for a number range, extract the median value of the number range; and For the first content feature that does not contain numbers, it remains unchanged.

3. The method according to claim 1, wherein, When a new product is first recommended, the initial first weight of the first content feature included in each new product is 1.

4. The method according to claim 1, wherein, The first weight of the first content feature is updated based on user feedback, including: Based on the user feedback obtained during the first time period and associated with each of the first content features, the first weight of each of the first content features of the new product during the second time period is updated.

5. The method according to claim 1, wherein, When the first content feature and the second content feature are numbers, the first value is determined based on the numbers contained in the first content feature and the second content feature. When the first content feature and the second content feature are not numbers, text matching is performed. If the texts are the same, the first value is determined to be 0; if the texts are different, the first value is determined to be 1.

6. The method according to claim 1, wherein, Determining the neighbor set of the new product based on the weighted Euclidean distance includes: Step 1: Each product in the product set consisting of the new product and the old product is treated as a cluster. Step 2: Merge the two closest clusters in the cluster; Step 3: Repeat step 2 until the number of clusters equals a preset threshold, then end the cluster merging process and determine the other old products in the cluster where the new product is located as the neighbor set of the new product.

7. The method according to claim 1, wherein, The predicted score of the new product is determined based on the weighted Euclidean distance between the old and new products in the neighborhood set and the historical scores of the old products in the neighborhood set, including: The second value is determined based on the historical scores of the old products in the neighborhood set and the weighted Euclidean distance of the new products associated with the old products in the neighborhood set; The predicted score of the new product is determined based on the second value and the sum of the weighted Euclidean distances between the old products and the new product in the neighbor set.

8. A product recommendation device for performing the method according to any one of claims 1 to 7, wherein, The device includes: The extraction module is configured to extract at least one first content feature of the new product; The update module is configured to assign an initial first weight to each of the at least one first content feature, and update the first weight of the first content feature based on user feedback, wherein the user feedback is obtained based on the feedback of the first content feature of the new product within a first time period; The acquisition module is configured to acquire at least one second content feature of an old product; The first determining module is configured to determine the weighted Euclidean distance between every two products in the product set composed of the new product and the old product based on the first content feature, the updated first weight, and the second content feature. The second determining module is configured to determine the neighbor set of the new product based on the weighted Euclidean distance, wherein the neighbor set contains at least one old product. The third determining module is configured to determine the predicted score of the new product based on the weighted Euclidean distance between the old products and the new product in the neighbor set and the historical scores of the old products in the neighbor set. The fourth determining module is configured to determine at least one new product to recommend to the user based on the predicted score.

9. An electronic device, comprising: One or more processors; A storage device for storing executable instructions, which, when executed by the processor, implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, implement the method according to any one of claims 1 to 7.

11. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images