Item recommendation method and device, electronic device and computer-readable storage medium

Abstract

The present disclosure provides an item recommendation method and device, an electronic device, and a computer-readable storage medium, which can be applied to the field of big data technology. The item recommendation method includes: in response to a user request, obtaining a label portrait of a target customer, wherein the label portrait includes at least one label associated with the target customer and a first interest value of the label, wherein the first interest value is used to characterize the target customer's preference for the label; obtaining a target item index list associated with the target label, wherein the target item index list includes at least one target item associated with the target label, and an attention value of the target item, wherein the attention value is used to characterize the degree of attention paid to the target item; determining a list of items to be recommended for the target customer based on the first interest value of the target label and / or the attention value of the target item, so as to return the list of items to be recommended to the user.

Description

Technical Field

[0001] The present disclosure relates to the field of big data technology, and more specifically, to an item recommendation method and device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the development of e-commerce, companies often need to make targeted product recommendations to customers based on their interests when expanding their customer base. Based on this background, intelligent recommendations are becoming more and more widely used.

[0003] In the process of implementing the concepts disclosed herein, the inventors discovered that there are at least the following problems in the related art: existing intelligent recommendation algorithms usually generate tags for customers based on their attributes, and then use the tags to associate with items. They cannot dynamically feedback the customer's interests, resulting in low accuracy of the generated customer interest tags. Summary of the invention

[0004] In view of the above problems, the present disclosure provides an item recommendation method, apparatus, device, medium and program product.

[0005] A first aspect of the present disclosure provides an item recommendation method, comprising:

[0006] In response to a user request, obtaining a tag portrait of a target customer, wherein the tag portrait includes at least one tag associated with the target customer and a first interest value of the tag, wherein the first interest value is used to represent the target customer's preference for the tag;

[0007] Acquire a target item index list associated with the target tag, wherein the target item index list includes at least one target item associated with the target tag and an attention value of the target item, wherein the attention value is used to represent the degree to which the target item is paid attention to;

[0008] According to the first interest value of the target tag and / or the attention value of the target item, a list of items to be recommended for the target customer is determined, so as to return the list of items to the user.

[0009] According to an embodiment of the present disclosure, determining a list of items to be recommended for a target customer according to the first interest value of the target tag and / or the attention value of the target item includes:

[0010] Associating the first interest value of the target tag with the attention value of the target item to obtain a recommendation value of the target item;

[0011] Based on the recommendation value of the target item, determine the list of items to be recommended for the target customers.

[0012] According to an embodiment of the present disclosure, the label portrait of the target customer is constructed by a portrait construction system;

[0013] The label portraits of target customers are constructed through the portrait construction system, including:

[0014] Reading historical behavior log data of a target customer within a preset historical time period from a database, wherein the historical behavior log data is used to characterize the target customer's interested operation behaviors on multiple items;

[0015] Determine at least one tag associated with the target customer based on the historical behavior log data;

[0016] Determining a first interest value of the tag;

[0017] Build a tag profile of the target customer based on the first interest value of the tag.

[0018] According to an embodiment of the present disclosure, determining the first interest value of the tag includes:

[0019] Determine the frequency of the target customer's interested operation behavior within a first time period according to the target customer's historical behavior log data, wherein the first time period is the most recent preset time period within the preset historical time period;

[0020] generating a first reference value of the tag according to the frequency of the operation behavior of interest within a first time period;

[0021] The first interest value of the tag is determined according to the first reference value of the tag and the second interest value of the tag in a second time period, wherein the second time period is a preset time period before the first time period.

[0022] According to an embodiment of the present disclosure, determining the first interest value of the tag according to the first reference value of the tag and the second interest value of the tag within the second time period includes:

[0023] Multiplying a second interest value of the tag in a second time period by a preset attenuation coefficient to obtain a second reference value;

[0024] The first reference value and the second reference value are added to obtain a first interest value.

[0025] According to an embodiment of the present disclosure, determining at least one tag associated with a target customer based on historical behavior log data includes:

[0026] Constructing an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items;

[0027] According to the item tag mapping table and the historical behavior log data, a tag that the target customer is interested in is determined as at least one tag associated with the target customer.

[0028] According to an embodiment of the present disclosure, wherein the target item index list is constructed by an index construction system;

[0029] The target item index list is constructed through the index construction system, including:

[0030] Constructing an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items;

[0031] According to the item tag mapping table, determine the target item associated with the target tag;

[0032] The target items are collected into the same data table to form a target item index list.

[0033] According to an embodiment of the present disclosure, determining a target item associated with a target tag according to the item tag mapping table includes:

[0034] According to the item label mapping table, determine the preliminary items associated with the target label;

[0035] Generate attention values ​​for the preliminary items according to the degree of attention paid to the preliminary items;

[0036] According to the attention value of the preliminary selected items, the target items are determined from the preliminary selected items.

[0037] A second aspect of the present disclosure provides an item recommendation device, including a first acquisition module, a second acquisition module and a determination module.

[0038] A first acquisition module, configured to acquire a tag portrait of a target customer in response to a user request, wherein the tag portrait includes at least one tag associated with the target customer and a first interest value of the tag, wherein the first interest value is used to represent the target customer's preference for the tag;

[0039] A second acquisition module is used to acquire a target item index list associated with the target tag, wherein the target item index list includes at least one target item associated with the target tag and an attention value of the target item, wherein the attention value is used to represent the degree of attention paid to the target item;

[0040] The determination module is used to determine a list of items to be recommended for the target customer according to the first interest value of the target tag and / or the attention value of the target item, so as to return the list of items to be recommended to the user.

[0041] According to an embodiment of the present disclosure, the determination module includes an association unit and a first determination unit.

[0042] The associating unit is used to associate the first interest value of the target tag with the attention value of the target item to obtain a recommendation value of the target item;

[0043] The first determining unit is used to determine a list of items to be recommended for a target customer according to the recommendation value of the target item.

[0044] According to an embodiment of the present disclosure, a label portrait of a target customer is constructed through a portrait construction system; the above-mentioned device also includes a first construction module, which is used to construct a label portrait about the target customer through the portrait construction system, wherein the first construction module includes a reading unit, a second determination unit, a third determination unit and a first construction unit.

[0045] The reading unit is used to read the historical behavior log data of the target customer within a preset historical time period from the database, wherein the historical behavior log data is used to characterize the target customer's interested operation behavior on multiple items;

[0046] A second determination unit, configured to determine at least one tag associated with the target customer based on the historical behavior log data;

[0047] A third determining unit, configured to determine a first interest value of the tag;

[0048] The first construction unit is used to construct a tag portrait of the target customer according to the first interest value of the tag.

[0049] According to an embodiment of the present disclosure, the third determining unit includes a first determining subunit, a first generating subunit and a second generating subunit.

[0050] The first determination subunit is used to determine the frequency of the target customer's interested operation behavior within a first time period according to the target customer's historical behavior log data, wherein the first time period is the most recent preset time period within the preset historical time period;

[0051] A first generating subunit, configured to generate a first reference value of the tag according to a frequency of an operation behavior of interest within a first time period;

[0052] The second generating subunit is used to determine the first interest value of the tag according to the first reference value of the tag and the second interest value of the tag in a second time period, wherein the second time period is a preset time period before the first time period.

[0053] According to an embodiment of the present disclosure, in the second generating subunit, generating the interest value of the tag in the first time period according to the first value of the tag and the interest value of the tag in the second time period includes:

[0054] Multiplying a second interest value of the tag in a second time period by a preset attenuation coefficient to obtain a second reference value;

[0055] The first reference value and the second reference value are added to obtain a first interest value.

[0056] According to an embodiment of the present disclosure, the second determining unit includes a constructing subunit and a second determining subunit.

[0057] The construction subunit is used to construct an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items;

[0058] The second determining subunit is used to determine a tag that the target customer is interested in according to the item tag mapping table and the historical behavior log data as at least one tag associated with the target customer.

[0059] According to an embodiment of the present disclosure, the target item index list is constructed by an index construction system; the above-mentioned device also includes a second construction module, which is used to construct the target item index list through the index construction system, wherein the second construction module includes a second construction unit, a fourth determination unit and a collection unit.

[0060] The second construction unit is used to construct an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items;

[0061] A fourth determining unit, configured to determine a target item associated with the target tag according to the item tag mapping table;

[0062] The collection unit is used to collect the target items into the same data table to form a target item index list.

[0063] According to an embodiment of the present disclosure, the fourth determining unit includes a third determining subunit, a third generating subunit and a fourth determining subunit.

[0064] The third determination subunit is used to determine the preliminary selected items associated with the target tag according to the item tag mapping table;

[0065] The third generating subunit is used to generate the attention value of the preliminary selected items according to the degree of attention paid to the preliminary selected items;

[0066] The fourth determining subunit is used to determine the target object from the preliminary selected objects according to the attention value of the preliminary selected objects.

[0067] The third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned item recommendation method.

[0068] The fourth aspect of the present disclosure also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above-mentioned item recommendation method.

[0069] The fifth aspect of the present disclosure also provides a computer program product, including a computer program, which implements the above-mentioned item recommendation method when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0071] Figure 1 A system architecture diagram schematically shows an item recommendation method, apparatus, device, medium, and program product according to an embodiment of the present disclosure;

[0072] Figure 2 A flowchart of an item recommendation method according to an embodiment of the present disclosure is schematically shown;

[0073] Figure 3 A system principle diagram of an item recommendation method according to an embodiment of the present disclosure is schematically shown;

[0074] Figure 4 A structural block diagram of an item recommendation device according to an embodiment of the present disclosure is schematically shown; and

[0075] Figure 5 A block diagram of an electronic device suitable for implementing an item recommendation method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0076] Hereinafter, embodiments of the present disclosure will 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 present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0077] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0078] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0079] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (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, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0080] It should be noted that in the technical solution of the present disclosure, the customer's authorization or consent is obtained before obtaining or collecting the customer's personal information.

[0081] In the process of expanding customer base, enterprises often need to make targeted product recommendations to customers based on their interests. For example, the intelligent recommendation method that can be used in related technologies can be: when both label data and item data already exist, first generate labels for customers based on customer attributes (such as age, gender, occupation, etc.) based on a predetermined rule, and then use the labels to associate with items. However, this method cannot dynamically feedback customer interests, making the generated customer interest labels inaccurate, and there is no way to generate interest labels for customers who lack attributes. In addition, because the method for matching customer interest labels and items in related technologies is the direct data association method, the time complexity is N squared, the time complexity is high, and the amount of calculation is large.

[0082] In view of this, an embodiment of the present disclosure provides an item recommendation method, comprising:

[0083] In response to a customer request for obtaining items to be recommended for a target customer, obtaining a tag portrait of the target customer, wherein the tag portrait includes at least one tag associated with the target customer and an interest value of each tag, wherein the interest value is used to represent the target customer's preference for the tag;

[0084] Acquire a target item index list associated with the target tag, wherein the target item index list includes at least one target item associated with the target tag and an attention value of each target item, wherein the attention value is used to represent the degree of attention paid to the target item;

[0085] According to the interest value of the target tag and / or the attention value of each target item, the proposed recommended items for the target customer are determined so as to return the proposed recommended items for the target customer to the customer.

[0086] Figure 1The system architecture diagram of the item recommendation method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.

[0087] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a terminal device 101, a server 102, and a database 103. The terminal device 101, the server 102, and the database 103 may communicate with each other via a network, and the network may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0088] The client can use the terminal device 101 to interact with the server 102 through the network to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0089] The terminal device 101 may be any electronic device having a display screen and supporting web browsing, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, and the like.

[0090] The server 102 may be a server that provides various services, such as a background management server (only an example) that provides support for websites browsed by clients using the terminal device 101. The background management server may analyze and process the received data such as client requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to client requests) to the terminal device.

[0091] In the application scenario of the embodiment of the present disclosure, the customer can use the terminal device 101 to initiate a request to the server 102 for obtaining the proposed recommended items for the target customer. In response to the customer request, the server 102 can be used to execute the item recommendation method of the embodiment of the present disclosure, obtain the label portrait of the target customer and the item index list representing the mapping relationship between the label and the item from the database 103, associate the label portrait of the target customer with the item index list, obtain the proposed recommended item list for the target customer, and return it to the customer through the terminal device 101.

[0092] According to the embodiments of the present disclosure, the target customer's tag portrait and item index list may be pre-built or built online in real time according to the customer's request. The target customer's tag portrait and item index list may be built by the server 102 and stored in the database 103, or may be built by other servers or service clusters and stored in the database 103.

[0093] It should be noted that the item recommendation method provided in the embodiment of the present disclosure can generally be executed by the server 102. Accordingly, the item recommendation device provided in the embodiment of the present disclosure can generally be set in the server 102. The item recommendation method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 102 and can communicate with the terminal device 101 and / or the server 102. Accordingly, the item recommendation device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 102 and can communicate with the terminal device 101 and / or the server 102.

[0094] It should be understood that Figure 1 The number of terminal devices, servers, and databases in the embodiment is only for illustration. Any number of terminal devices, servers, and databases may be provided according to implementation requirements.

[0095] The following will be based on Figure 1 The scene described by Figure 2 to Figure 5 The item recommendation method of the disclosed embodiment is described in detail.

[0096] Figure 2 The flowchart of the item recommendation method according to an embodiment of the present disclosure is schematically shown.

[0097] like Figure 2 As shown, the item recommendation method of this embodiment includes operations S201 to S203.

[0098] In operation S201, in response to a user request, a tag portrait of a target customer is obtained, wherein the tag portrait includes at least one tag associated with the target customer and a first interest value of the tag, wherein the first interest value is used to characterize the target customer's preference for the tag.

[0099] In operation S202, a target item index list associated with a target tag is obtained, wherein the target item index list includes at least one target item associated with the target tag and an attention value of the target item, wherein the attention value is used to represent the degree of attention paid to the target item.

[0100] In operation S203, a list of items to be recommended for the target customer is determined according to the first interest value of the target tag and / or the attention value of the target item, so as to return the list of items to the user.

[0101] According to an embodiment of the present disclosure, the item recommendation method is based on the tag data of the item, where the item can be a general recommendation system material, such as articles, e-commerce products, news, videos, advertisements, etc. The tag can be any mark that can represent the item, such as the subject word of the article, the brand or category of the product, the category of the product, the classification of the news, the tag of the video, the style of clothing, etc.

[0102] According to the embodiments of the present disclosure, the target customer's label portrait can be pre-built through an offline computing system, or can be built online in real time according to customer requests. The target customer's label portrait can be pre-built and stored in a database, and the database can store label portraits of multiple customers. The target customer is the current processing customer, and can be any one of the multiple customers.

[0103] According to an embodiment of the present disclosure, a tag portrait characterizes the mapping relationship between customers and tags, representing which tags the customer is interested in, and the different preferences for multiple tags. Each customer can be associated with multiple different tags. The target tag is the currently processed tag, which can be any one of the multiple tags associated with the target customer. For example, the tag portrait currently being processed can be created for customer A, and the tag portrait includes the tags that the customer is interested in and the first interest value of each tag: mobile phone - 85; pants - 70; cat food - 59..., indicating that the current customer is most interested in mobile phones, followed by pants. The interest value characterizes the target customer's preference for tags. The customer's interest value for tags will change over time. For example, in different time periods, the customer's interest value for tags is different. The first interest value can be the interest value in the most recent time period, which can indicate the tags of items that the customer is interested in in the recent period of time.

[0104] According to an embodiment of the present disclosure, the target item index list may be pre-built or constructed online in real time according to customer requests. The target item index list may be pre-built and stored in a database. The database may store multiple item index lists, each item index list corresponds to a tag, and each item index list contains one or more items associated with the tag and the attention value of each item. The attention value may indicate the degree to which each item is paid attention to by all customers (e.g., the degree to which it is clicked by all customers, the degree to which it is purchased by all customers, etc.), and the target item index list corresponds to the target tag. For example, if the target tag is "mobile phone", the target item index list may be: Brand A Model Mobile Phone-96, Brand B Model Mobile Phone-84, Brand C Model Mobile Phone-70, etc. Among the items related to the tag "mobile phone", Brand A Model Mobile Phone has the highest degree of attention from customers.

[0105] Pre-building the label portraits and corresponding item index lists of each customer can save online server computing resources, but when the number of customers is large (large e-commerce platforms or short video recommendation platforms have hundreds of millions of customers), it will cause storage overflow problems. Online real-time construction of label portraits and item index lists can avoid storage problems caused by too many customers, but when the number of requests is too large, it will affect the speed of request response.

[0106] According to an embodiment of the present disclosure, the proposed recommended items for the target customer are determined based on the first interest value of the target tag and / or the attention value of each target item. The proposed recommended items for the target customer can be determined based only on the size of the first interest value of the target tag or the size of the attention value of each target item. For example, based on the size of the first interest value of each tag in the tag portrait of the target customer, it is determined that the target customer is currently most interested in the product as a mobile phone (first interest value 85). Based on this information alone, products related to mobile phones can be randomly recommended to the customer, such as recommending a certain brand of mobile phones, recommending a certain brand of mobile phone supplies, and so on. For example, after determining that the target customer is currently most interested in the product as a mobile phone based on the size of the first interest value of each tag in the tag portrait of the target customer, the target item index list associated with the mobile phone tag is found, and according to the size of the attention value of each item in the index list, several products with higher attention values ​​are selected to recommend to the customer.

[0107] According to an embodiment of the present disclosure, based on the first interest value of the target tag and / or the attention value of the target item, the recommended items for the target customer are determined. For example, the first interest value of the target tag can be associated with the attention value of each target item to obtain the recommendation value of each target item, and then the preferred items are recommended to the target customer in a targeted manner according to the size of the recommendation value. For example, based on the size of the first interest value of each tag in the tag portrait of the target customer, after determining that the target customer is currently most interested in the product is a mobile phone (first interest value 85), the target item index list associated with the mobile phone tag is found, and the first interest value 85 of the target tag is respectively associated with the attention value of each target item (for example, 96, 86, 73...), and the recommendation value of each target item (83, 67, 54...) is obtained. According to the size of the recommendation value, the product with a higher recommendation value is recommended to the customer.

[0108] According to the embodiments of the present disclosure, the above-mentioned item recommendation method is a personalized recommendation algorithm based on item tags, which is different from the method of generating tags for customers in related technologies and then using tags to associate with items. The items referred to can be general recommendation system materials, and the tags referred to can be any mark that can represent items, which is more versatile. Because the tag portrait is associated with the item tags that customers are interested in and the first interest value of each tag, the customer's interests can be dynamically fed back, making the generated tags more accurate; because the tags are item tags, the problem of not being able to generate customer interest tags due to lack of customer attributes in related technologies is avoided.

[0109] According to an embodiment of the present disclosure, since the tag portrait is associated with the item tags of the customer and the first interest value of each tag, and the target item index list is associated with the items associated with the target tag and the attention value of each item, by determining the proposed recommended items for the target customer based on the first interest value of the target tag and / or the attention value of each target item, the customer's interest identification accuracy is higher, the recommended items are more targeted, and are more easily adopted by the customer.

[0110] According to the embodiments of the present disclosure, the above-mentioned recommendation method adopts an online response to customer requests in real time, which can avoid the problem of storage overflow when the number of customers is large (large e-commerce platforms or short video recommendation platforms, with hundreds of millions of customers), compared with pre-calculating the recommendation results for all customers offline.

[0111] According to an embodiment of the present disclosure, determining a list of items to be recommended for a target customer according to the first interest value of the target tag and / or the attention value of the target item includes:

[0112] The first interest value of the target tag is associated with the attention value of the target item to obtain a recommendation value of the target item; and a list of items to be recommended for the target customer is determined according to the recommendation value of the target item.

[0113] According to an embodiment of the present disclosure, the first interest value of the target tag is associated with the attention value of the target item to obtain a recommendation value of the target item. For example, the interest value of the target tag may be multiplied by the attention value of each target item to obtain a recommendation value, or the interest value of the target tag may be added to the attention value of each target item to obtain a recommendation value. Other association methods may also be used to associate the interest value of the target tag with the attention value of each target item to obtain a recommendation value, and then the preferred items may be recommended to the target customers in a targeted manner according to the size of the recommendation value.

[0114] According to an embodiment of the present disclosure, a list of items to be recommended to target customers is determined based on the recommendation value of each target item. For example, several items with higher recommendation values ​​may be selected for recommendation to customers for each target item index list (i.e., for the items under each label). Alternatively, the recommendation values ​​of the items in the target item index list corresponding to all labels may be sorted, and several items with higher recommendation values ​​may be selected to form a list of items to be recommended.

[0115] According to an embodiment of the present disclosure, the above method is, for example: based on the interest value of each label in the label portrait of the target customer, after determining that the product that the target customer is currently most interested in is a mobile phone (interest value 8), find the target item index list associated with the mobile phone label, and multiply the interest value 8 of the target label by the attention value of each target item (for example, 2, 1, 0.5...) to obtain the recommended value of each target item (Brand A mobile phone-16, Brand B mobile phone-8, Brand C mobile phone holder-4...), and recommend products with higher recommended values ​​to customers based on the size of the recommended values.

[0116] According to an embodiment of the present disclosure, the above method can also be, for example: the interest value of each label in the target customer's label portrait is determined, and the top 10 product labels that the target customer is currently interested in are determined as follows: mobile phone (interest value 8), pants (interest value 7), shoes (interest value 6), cat food (interest value 5), ... Then, the attention value of each product in the item index list associated with each label is respectively associated with the attention value of each product in the item index list associated with each label, and the recommended value of a series of products under each label is obtained and sorted as follows: Brand A mobile phone-16, Brand B cat food-15, Brand C mobile phone holder-14..., and according to the size of the recommended value, products with higher recommended values ​​are recommended to customers.

[0117] According to an embodiment of the present disclosure, by associating the first interest value of the target tag with the attention value of the target item to obtain the recommendation value of the target item, and then recommending related items to customers based on the recommendation value, it is achieved to recommend items to customers based on the degree of customer interest and the degree of attention paid to the items, and fully consider the various factors that may affect the customer's purchasing behavior. The recommended items are more targeted and easier to be adopted by customers, thereby achieving the purpose of accurately recommending items of interest to customers.

[0118] According to an embodiment of the present disclosure, the target customer's label portrait is constructed by a portrait construction system. According to an embodiment of the present disclosure, the target customer's label portrait can be pre-constructed offline by the portrait construction system, or can be constructed online in real time by the portrait construction system according to the customer's request.

[0119] The above-mentioned label portraits of target customers constructed by the portrait construction system include:

[0120] The historical behavior log data of the target customer within a preset historical time period is read from the database, where the historical behavior log data is used to characterize the target customer's interest operation behavior on multiple items, such as click logs, add-to-cart logs, purchase logs, etc. Each log includes at least customer information, product information, and time information.

[0121] At least one tag associated with the target customer is determined based on historical behavior log data; for example, products with high click rates can be found based on the click frequency of each product, and these products can be classified to obtain products under multiple categories, and labels are extracted for each category of products based on common attributes of the products in that category as labels that the customer is interested in.

[0122] The first interest value of the tag may be determined, for example, based on the click frequency of each commodity associated with each tag.

[0123] According to the first interest value of the tag, a tag portrait of the target customer is constructed. For example, a tag portrait may be constructed for each customer, and each customer's tag portrait includes at least one tag associated with the customer and the first interest value of each tag.

[0124] According to the embodiments of the present disclosure, the method of the embodiments of the present disclosure automatically constructs an interest portrait based on the customer's implicit feedback (log behavior data), and does not need to construct tags based on customer attributes, thereby avoiding the problem of not being able to generate interest tags for customers who lack attributes. At the same time, compared with the method of generating tags based on customer attributes (highly subjective), this method is an objective analysis method, which can more objectively and accurately identify the customer's tag interests through behavior logs.

[0125] According to an embodiment of the present disclosure, in the above method, determining the first interest value of the tag includes:

[0126] According to the historical behavior log data of the target customer, the frequency of the target customer's interested operation behavior within the first time period is determined, wherein the first time period is the most recent preset time period within the preset historical time period; the frequency of the interested operation behavior may be, for example, click frequency, add-to-cart frequency, purchase frequency, etc. The preset historical time period may be, for example, the most recent week, two weeks, one month, one year, etc. The first time period is the most recent preset time period within the preset historical time period, for example, the most recent day within the most recent week, or the most recent week within the most recent month, etc.

[0127] The first reference value of the tag is generated according to the frequency of the operation of interest in the first time period; for example, the first reference value of each tag can be determined according to the click frequency of each product associated with each tag on the most recent day of the most recent week.

[0128] The first interest value of the tag is determined according to the first reference value of the tag and the second interest value of the tag in a second time period, wherein the second time period is a preset time period before the first time period. The second time period can be, for example, the day before the most recent day in the most recent week, or the week before the most recent week in the most recent month, etc.

[0129] According to the embodiments of the present disclosure, the interest value represents the target customer's preference for the tag. The customer's interest value for the tag will change over time. In different time periods, the customer's interest value for the tag is different. The first interest value is the interest value in the most recent time period, which can indicate the item tags that the customer is interested in in the most recent period. The second interest value refers to the interest value in the second time period (the time period before the most recent time period), which refers to the historical interest value, indicating the item tags that the customer is interested in in the previous time period in the past.

[0130] The second interest value of each tag in the second time period is a result value determined by the method of the embodiment of the present disclosure. In the embodiment of the present disclosure, the first interest value of each current time period is obtained by iteratively calculating the second interest value of the previous time period and the first reference value of the current time period. The second interest value in the initial second time period (that is, the initial second time period is the first preset time period in the preset historical time period) adopts the first reference value in the initial second time period, that is, in the initial second time period, the first reference value of the tag is determined according to the click frequency of each commodity associated with the tag as the second interest value of the tag.

[0131] According to an embodiment of the present disclosure, in the above method, determining the first interest value of the tag based on the first reference value of the tag and the second interest value of the tag within a second time period includes: multiplying the second interest value of the tag within the second time period by a preset attenuation coefficient to obtain a second reference value, and then adding the first reference value and the second reference value to obtain the first interest value.

[0132] According to an embodiment of the present disclosure, the first interest value of a tag generated within a first time period may be, for example: the application scenario is that the calculation period is the most recent week, and the current calculation tag is "mobile phone", then the first day's tag interest value is first calculated based on the click frequency of the goods associated with the mobile phone within the first day; when calculating the second day's tag interest value, first calculate the second day's first reference value based on the click frequency of the goods associated with the mobile phone within the second day, and then associate the first reference value of the second day with the tag interest value (second interest value) of the first day to obtain the first interest value of the second day (it may be the interest value of the first day multiplied by the attenuation coefficient, and then added to the first reference value of the second day to obtain the first interest value of the second day), and so on, to obtain the interest value of the "mobile phone" tag for each day in the most recent week, and use the interest value of the most recent day as the interest value of the "mobile phone" tag, that is, store the interest value corresponding to the latest time period as the interest value of the tag in the database.

[0133] According to an embodiment of the present disclosure, through the above method, a customer's interest portrait is constructed according to the behavior log. In the process of determining the interest value of each tag associated with the target customer, the first interest value of the tag is determined according to the first reference value of the tag and the second interest value of the tag in the second time period, thereby realizing the use of the interest value corresponding to the latest time period as the interest value of the tag. At the same time, combined with the preset attenuation coefficient, the factor of customer interest changing over time is taken into account, the customer's latest interests can be tracked in time, and the customer's latest interest changes can be accurately captured according to the customer's behavioral changes, thereby improving the reference value of the tag portrait.

[0134] According to an embodiment of the present disclosure, in the above method, determining at least one tag associated with the target customer based on historical behavior log data includes:

[0135] Construct an item label mapping table, where the item label mapping table records multiple items and labels associated with the items. For example, the key attributes of each item can be extracted to obtain the label of the item, and the item label mapping table can be constructed by establishing a mapping relationship between the item and the label. For example, if the item is a brand A mobile phone holder, its label can be extracted as: mobile phone or holder. If the item is a brand A shoe, its label can be extracted as: shoe or sports.

[0136] According to the item tag mapping table and the historical behavior log data, the tags that the target customer is interested in are determined as at least one tag associated with the target customer. For example, the items with high click-through rates can be found as the items that the customer is interested in based on the frequency of each item being clicked by the customer, and then the items and tags that are mapped to the tags that the customer is interested in are mapped as at least one tag associated with the target customer based on the mapping relationship between the items and tags in the item tag mapping table.

[0137] According to an embodiment of the present disclosure, the target item index list is constructed by an index construction system. According to an embodiment of the present disclosure, the target item index list may be pre-constructed by an index construction system, or may be constructed online in real time by an index construction system according to a customer request. The target item index list may be pre-constructed and stored in a database, and the database may store multiple item index lists, each item index list corresponds to a tag, and the target item index list corresponds to a target tag.

[0138] Among them, building a target item index list through the index building system includes:

[0139] First, an item label mapping table is constructed, wherein the item label mapping table records multiple items and labels respectively associated with the items (the construction method refers to the construction method of the item label mapping table in the above embodiment, which will not be repeated here); then, according to the item label mapping table, the target item associated with the target label is determined; the target items are collected into the same data table to form a target item index list. For example, the items associated with the target label can be found according to the mapping relationship between the goods and labels in the item label mapping table, and these items are collected into the same data table to form a target item index list.

[0140] According to the embodiments of the present disclosure, the target item index list established by using the item label mapping relationship includes multiple items associated with the target label, which is an inverted index data structure. When the label in the label portrait is associated with the relevant items, the inverted index query method can ensure the fastest query time. Compared with the direct association method (traversing and searching for the goods corresponding to each label), the complexity of time query can be reduced (the query time complexity of using the inverted index is N, and the time complexity of traversing and searching for the goods corresponding to each label is N squared), and the response speed is improved.

[0141] According to an embodiment of the present disclosure, in the above method, determining the target item associated with the target tag according to the item tag mapping table includes:

[0142] According to the item label mapping table, determine the preliminary items associated with the target label;

[0143] Generate attention values ​​for the preliminary items according to the degree of attention paid to the preliminary items;

[0144] According to the attention value of the preliminary selected items, the target items are determined from the preliminary selected items.

[0145] According to an embodiment of the present disclosure, the above method can, for example, find items associated with the target tag based on the mapping relationship between goods and tags in the item tag mapping table, collect these items into the same data table to form a list of preliminary items associated with the target tag, and then generate an attention value for each preliminary item according to the degree of attention received by each preliminary item (for example, according to the frequency of being clicked or purchased), sort the preliminary items according to the attention value, and select multiple preliminary items with the highest ranking as target items.

[0146] According to an embodiment of the present disclosure, the target item index list constructed by the above method includes the attention value of each item, and the target item is an item with a higher degree of attention. The recommended items determined based on this are more targeted and easier to be adopted by customers.

[0147] Figure 3The following schematically shows the system principle diagram of the item recommendation method according to the embodiment of the present disclosure. Figure 3 , an exemplary description is given of the item recommendation method of the embodiment of the present disclosure.

[0148] like Figure 3 As shown, the system of the item recommendation method according to the embodiment of the present disclosure may include a label extraction module, a portrait construction module, an inverted index module, a portrait service module, an index service module, etc.

[0149] According to an embodiment of the present disclosure, the label extraction module is used to construct an item label mapping table.

[0150] According to an embodiment of the present disclosure, a portrait construction module (a portrait construction system) is used to construct a label portrait of each customer based on the customer's historical behavior log data.

[0151] According to an embodiment of the present disclosure, an inverted index module (index building system) is used to build an index list of items under each tag.

[0152] According to the embodiment of the present disclosure, the portrait service module is used to store the pre-built customer tag portrait in the online Redis storage and develop a real-time query interface. The service supports the query interface externally, and can query the interest tags corresponding to each customer and the interest value corresponding to each tag according to the customer ID.

[0153] According to an embodiment of the present disclosure, the index service module is used to load the pre-built item index list into the memory of each online server and develop a real-time index query interface. The service may have two interfaces, one for building indexes and the other for querying indexes. The building index interface is used to read the inverted index file and load it into the online server memory; the query index interface supports querying the corresponding item list according to the tag ID and returns the result to the caller in real time.

[0154] According to the embodiments of the present disclosure, the item label mapping table, the customer's label portrait, and the item index list can be pre-built offline through the corresponding modules, or they can be built online in real time through the corresponding modules according to the customer's request. The services provided by the portrait service module and the index service module are used for online real-time service processes. Below, exemplary descriptions are given respectively. Here, the item label mapping table, the customer's label portrait, and the item index list are pre-built offline through the corresponding modules as an example.

[0155] According to the embodiment of the present disclosure, the method for the label extraction module to construct the item label mapping table offline is: extracting the key attributes of each item raw material (original commodity) to obtain the label of the commodity, and establishing a mapping relationship between the item and the label to construct the item label mapping table. The construction method thereof refers to the construction method of the item label mapping table in the aforementioned embodiment, and will not be repeated here.

[0156] According to the embodiments of the present disclosure, the portrait construction module mainly constructs the customer's interest preference for different tags based on the customer's multivariate behavior log, using data statistical methods and time decay models. The weight of its interest preference will decay with the decay of time. The newer the behavior, the higher the weight of the contribution, and the older the behavior, the lower the weight of the contribution. The specific method of constructing each customer's label portrait based on the customer's historical behavior log data is:

[0157] The input of this module is the customer behavior log, which includes click log, add-to-cart log, purchase log, etc. Each log includes at least three fields: (customer ID, product ID, time and date). The following is a detailed explanation using the click log as an example. Other log methods are similar and will not be repeated here.

[0158] Step 1: Read the click behavior log of the first day <customer ID, product ID, time and date>, read the constructed item label mapping table <product ID, label ID>, and associate the mapping to obtain the mapping relationship between customers and labels: <customer ID, label ID, time and date>; for example, the item clicked by the customer can be determined as the item that the customer is interested in based on the click log, and then mapped to the label that the customer is interested in based on the mapping relationship between the item and the label in the item label mapping table as at least one label associated with the customer.

[0159] Step 2: Count the frequency of customers clicking on the product on that day and get the triplet of <customer ID, tag ID, frequency cnt>;

[0160] Step 3: Normalize the frequency cnt to generate a score between (0 and 1), <customer ID, tag ID, score>, where the normalization formula is:

[0161]

[0162] Step 4: Read the data of the second day and map it to get the triple of <customer ID, tag ID, frequency cnt>;

[0163] Step 5: Calculate the score of the next day to get <customer ID, tag ID, score next_day >, where the calculation formula for the score on the second day is:

[0164]

[0165] Step 6: Add up the scores of these two days to get the latest <customer ID, tag ID, score new >Triple, where the scores of the two days are not directly added together, but the score of the previous day is multiplied by the decay factor Decay Factor and then added to the score of the next day. The reason for this is that the newer the click log, the higher the weight. The calculation formula is as follows:

[0166] score new =score*decayFactor+score next _ day Formula (III)

[0167] Step 7: Sort by scores in descending order and retain the top K tags for each customer, for example, K = 200;

[0168] Step 8: Read the behavior log of the new day every day and repeat steps 4 to 7 to continuously update the customer's preference profile for tags.

[0169] According to the embodiment of the present disclosure, in the above formula (III), score new That is, the interest value of the tag in the first time period, score is the interest value of the tag in the second time period, and score is the interest value of the tag in the second time period. next_day The first value of the tag in the first time period. The latest interest value of the tag is calculated by multiplying the interest value of the tag in the second time period by a preset attenuation coefficient to obtain a second value, and then calculating the sum of the first value and the second value to generate the interest value of the tag in the first time period as the latest interest value of the tag.

[0170] According to the embodiment of the present disclosure, the inverted index module (index construction system) is used to construct an index list of items under each tag. The module determines the mapping relationship between items and tags based on the pre-constructed item tag mapping table, and then determines the target item associated with the target tag based on the mapping relationship between items and tags, that is, constructs a key-value pair data structure with the tag ID as the key and the item ID list as the value, and obtains the corresponding inverted index list of items for each tag.

[0171] According to the embodiment of the present disclosure, the input of this module is a pre-built item label mapping table and item information, and after the inverted index is constructed, an inverted index file is generated. The specific operations are:

[0172] Step 1: Read in two parts of data: (1) the mapping relationship between items and tags, a binary tuple <item ID, tag ID>; (2) the item’s positive information, such as item ctr, item price, item sales, etc.

[0173] Step 2: Aggregate data based on tag ID to obtain a list of item IDs for each tag <tag ID, item ID list>; that is, find the items associated with the target tag based on the mapping relationship between products and tags in the item tag mapping table, and aggregate these items into the same data table to form a preliminary list of items associated with the target tag.

[0174] Step 3: Sort the item ID list obtained in step 2 according to the business goal. For example, in descending order of ctr, the sorted <label ID, sorted item ID list> is obtained. The item ID list contains the item ID and the item's sorting score. item (Attention value), for example, this score can be set to be the single item ctr of the item. The above method is: according to the degree of attention of each preliminary item (for example, according to the frequency of being clicked or purchased), the attention value of each preliminary item is generated, and the preliminary items are sorted according to the attention value.

[0175] Step 4: In order to save storage, the item ID list obtained in step 3 is truncated in sorted order, and only the topK items are retained, for example, K = 500, to obtain the final inverted index file, that is, multiple preliminary items with the highest ranking are selected from the preliminary items as target items, and the item index list under the target label is obtained.

[0176] According to an embodiment of the present disclosure, the above example illustrates a method in which an item label mapping table, a customer label portrait, and an item index list are pre-built offline through corresponding modules. The following exemplifies the process of online processing:

[0177] Step 1: In response to a customer request for obtaining recommended items for a target customer, obtain a tag profile of the target customer, where the tag profile includes at least one tag associated with the target customer and an interest value for each tag. Take out the topK tag profile data of the customer, and select the tag scores from high to low. The tag score (interest value) is the interest score of the tag ID associated with the customer. user，tag , which is the score generated by the aforementioned portrait construction module.

[0178] Step 2: Get the target item index list associated with the target tag, where the target item index list includes at least one target item associated with the target tag and the attention value of each target item. According to the customer's topK tag ID, request the query interface of the inverted index to obtain the item ID list corresponding to each tag.

[0179] Step 3: Intelligent sorting, that is, determining the recommended items for the target customers based on the interest value of the target tag and / or the attention value of each target item. This step requires the scores passed down from the previous two steps: (1) The interest value taken from the customer portrait: interest score user，tag ; (2) Query the target item index list to get the attention value: item score item , after associating the two scores, the recommendation value of each target item is obtained. According to the embodiment of the present disclosure, there are multiple methods for associating the interest value of the target tag with the attention value of each target item. An example of one of the association methods is given below:

[0180] score sort =score user，tag *score item Formula (IV)

[0181] Among them, score sort Refers to the recommended value of the target item, score user，tag Refers to the interest value of the target tag, score item Refers to the attention value of the target item.

[0182] Step 4: Return the data and obtain personalized recommendation results based on item tags.

[0183] Based on the above-mentioned item recommendation method, the present disclosure also provides an item recommendation device. Figure 4 The device is described in detail.

[0184] Figure 4 The structural block diagram of the item recommendation device according to an embodiment of the present disclosure is schematically shown.

[0185] like Figure 4 As shown, the item recommendation device of this embodiment includes a first acquisition module 401, a second acquisition module 402 and a determination module 403.

[0186] The first acquisition module 401 is used to obtain a tag portrait of a target customer in response to a user request, wherein the tag portrait includes at least one tag associated with the target customer and a first interest value of the tag, wherein the first interest value is used to represent the target customer's preference for the tag. ;

[0187] The second acquisition module 402 is used to acquire a target item index list associated with the target tag, wherein the target item index list includes at least one target item associated with the target tag and an attention value of the target item, wherein the attention value is used to represent the degree of attention paid to the target item. ;

[0188] The determination module 403 is used to determine a list of items to be recommended for the target customer according to the first interest value of the target tag and / or the attention value of the target item, so as to return the list of items to be recommended to the user.

[0189] According to the embodiments of the present disclosure, the item recommendation algorithm implemented by the above-mentioned item recommendation device is a personalized recommendation algorithm based on item tags, which is different from the method of generating tags for customers in the related art and then using the tags to associate with items. The items referred to can be general recommendation system materials, and the tags referred to can be any mark that can represent the items, which is more universal. In the above-mentioned device, the tag portrait obtained by the first acquisition module 401 is associated with the item tags of interest to the customer and the first interest value of the tag, which can dynamically feedback the customer's interests, so that the generated tags are more accurate; because the tags are item tags, the problem of not being able to generate customer interest tags due to lack of customer attributes in the related art is avoided.

[0190] According to an embodiment of the present disclosure, in the above-mentioned device, the label portrait obtained by the first acquisition module 401 is associated with the item label that the customer is interested in and the first interest value of the label, and the target item index list obtained by the second acquisition module 402 is associated with the items associated with the target label and the attention value of the item. The determination module 403 determines the recommended items for the target customer based on the interest value of the target label and / or the attention value of the target item. The customer's interest identification accuracy is higher, the recommended items are more targeted, and are more likely to be adopted by the customer.

[0191] According to the embodiments of the present disclosure, the recommendation method implemented by the above-mentioned device adopts an online response to customer requests in real time, which can avoid the problem of storage overflow when the number of customers is large (large e-commerce platforms or short video recommendation platforms, with hundreds of millions of customers), compared with pre-calculating the recommendation results for all customers offline.

[0192] According to an embodiment of the present disclosure,

[0193] According to an embodiment of the present disclosure, the determination module includes an association unit and a first determination unit.

[0194] The associating unit is used to associate the first interest value of the target tag with the attention value of the target item to obtain a recommendation value of the target item;

[0195] The first determining unit is used to determine a list of items to be recommended for a target customer according to the recommendation value of the target item.

[0196] According to an embodiment of the present disclosure, a label portrait of a target customer is constructed through a portrait construction system; the above-mentioned device also includes a first construction module, which is used to construct a label portrait about the target customer through the portrait construction system, wherein the first construction module includes a reading unit, a second determination unit, a third determination unit and a first construction unit.

[0197] The reading unit is used to read the historical behavior log data of the target customer within a preset historical time period from the database, wherein the historical behavior log data is used to characterize the target customer's interested operation behavior on multiple items;

[0198] A second determination unit, configured to determine at least one tag associated with the target customer based on the historical behavior log data;

[0199] A third determining unit, configured to determine a first interest value of the tag;

[0200] The first construction unit is used to construct a tag portrait of the target customer according to the first interest value of the tag.

[0201] According to an embodiment of the present disclosure, the third determining unit includes a first determining subunit, a first generating subunit and a second generating subunit.

[0202] The first determination subunit is used to determine the frequency of the target customer's interested operation behavior within a first time period according to the target customer's historical behavior log data, wherein the first time period is the most recent preset time period within the preset historical time period;

[0203] A first generating subunit, configured to generate a first reference value of the tag according to a frequency of an operation behavior of interest within a first time period;

[0204] The second generating subunit is used to determine the first interest value of the tag according to the first reference value of the tag and the second interest value of the tag in a second time period, wherein the second time period is a preset time period before the first time period.

[0205] According to an embodiment of the present disclosure, in the second generating subunit, generating the interest value of the tag in the first time period according to the first value of the tag and the interest value of the tag in the second time period includes:

[0206] Multiplying a second interest value of the tag in a second time period by a preset attenuation coefficient to obtain a second reference value;

[0207] The first reference value and the second reference value are added to obtain a first interest value.

[0208] According to an embodiment of the present disclosure, the second determining unit includes a constructing subunit and a second determining subunit.

[0209] The construction subunit is used to construct an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items;

[0210] The second determining subunit is used to determine a tag that the target customer is interested in according to the item tag mapping table and the historical behavior log data as at least one tag associated with the target customer.

[0211] According to an embodiment of the present disclosure, the target item index list is constructed by an index construction system; the above-mentioned device also includes a second construction module, which is used to construct the target item index list through the index construction system, wherein the second construction module includes a second construction unit, a fourth determination unit and a collection unit.

[0212] The second construction unit is used to construct an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items;

[0213] A fourth determining unit, configured to determine a target item associated with the target tag according to the item tag mapping table;

[0214] The collection unit is used to collect the target items into the same data table to form a target item index list.

[0215] According to an embodiment of the present disclosure, the fourth determining unit includes a third determining subunit, a third generating subunit and a fourth determining subunit.

[0216] The third determination subunit is used to determine the preliminary selected items associated with the target tag according to the item tag mapping table;

[0217] The third generating subunit is used to generate the attention value of the preliminary selected items according to the degree of attention paid to the preliminary selected items;

[0218] The fourth determining subunit is used to determine the target object from the preliminary selected objects according to the attention value of the preliminary selected objects.

[0219] According to an embodiment of the present disclosure, any multiple modules in the first acquisition module 401, the second acquisition module 402 and the determination module 403 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the first acquisition module 401, the second acquisition module 402 and the determination module 403 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 a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in any appropriate combination of any of them. Alternatively, at least one of the first acquisition module 401, the second acquisition module 402 and the determination module 403 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0220] Figure 5 A block diagram of an electronic device suitable for implementing an item recommendation method according to an embodiment of the present disclosure is schematically shown.

[0221] like Figure 5 As shown, the electronic device 500 according to an embodiment of the present disclosure includes a processor 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage part 508 into a random access memory (RAM) 503. The processor 501 may 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 may also include an onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0222] In RAM 503, various programs and data required for the operation of electronic device 500 are stored. Processor 501, ROM 502 and RAM 503 are connected to each other via bus 504. Processor 501 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 502 and / or RAM 503. It should be noted that the program can also be stored in one or more memories other than ROM 502 and RAM 503. Processor 501 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0223] According to an embodiment of the present disclosure, the electronic device 500 may further include an input / output (I / O) interface 505, which is also connected to the bus 504. The electronic device 500 may further include one or more of the following components connected to the I / O interface 505: an input portion 506 including a keyboard, a mouse, etc.; an output portion 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 508 including a hard disk, etc.; and a communication portion 509 including a network interface card such as a LAN card, a modem, etc. The communication portion 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage portion 508 as needed.

[0224] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0225] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 502 and / or RAM 503 described above and / or one or more memories other than ROM 502 and RAM 503.

[0226] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the item recommendation method provided by the embodiment of the present disclosure.

[0227] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 501. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0228] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 509, and / or installed from the removable medium 511. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0229] 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 removable medium 511. When the computer program is executed by the processor 501, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0230] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the client computing device, partially on the client device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the client computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0231] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0232] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0233] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. An item recommendation method, comprising: In response to a user request, obtaining a tag portrait of a target customer, wherein the tag portrait includes at least one tag associated with the target customer and a first interest value of the tag, wherein the first interest value is used to represent the target customer's preference for the tag; Acquire a target item index list associated with a target tag, wherein the target item index list includes at least one target item associated with the target tag and an attention value of the target item, wherein the attention value is used to represent the degree to which the target item is paid attention to; Determine a list of items to be recommended for the target customer according to the first interest value of the target tag and / or the attention value of the target item, so as to return the list of items to the user; The first interest value of the tag is determined by the following method: Determining the frequency of the target customer's interested operation behavior within a first time period according to the target customer's historical behavior log data, wherein the first time period is a most recent preset time period within a preset historical time period; generating a first reference value of the tag according to the frequency of the operation behavior of interest within the first time period; multiplying a second interest value of the tag in a second time period by a preset attenuation coefficient to obtain a second reference value; adding the first reference value and the second reference value to obtain the first interest value, wherein the second time period is a preset time period before the first time period; The calculation formula of the first interest value is as follows: in, is the first interest value of the tag in the first time period, is the second interest value of the tag in the second time period, is the first reference value of the tag in the first time period, wherein, is the preset attenuation coefficient.

2. The method according to claim 1, wherein: Determining the list of items to be recommended for the target customer according to the first interest value of the target tag and / or the attention value of the target item includes: Associating the first interest value of the target tag with the attention value of the target item to obtain a recommendation value of the target item; According to the recommendation value of the target item, a list of items to be recommended for the target customer is determined.

3. The method according to claim 1, wherein: The label portrait of the target customer is constructed by a portrait construction system; Constructing a label portrait of the target customer by the portrait construction system includes: Reading historical behavior log data of the target customer within a preset historical time period from a database, wherein the historical behavior log data is used to characterize the target customer's interested operation behavior on multiple items; Determining at least one tag associated with the target customer based on the historical behavior log data; determining the first interest value of the tag; A tag profile of the target customer is constructed according to the first interest value of the tag.

4. The method according to claim 3, wherein: The determining, based on the historical behavior log data, at least one tag associated with the target customer includes: Constructing an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items; According to the item tag mapping table and the historical behavior log data, a tag that the target customer is interested in is determined as at least one tag associated with the target customer.

5. The method according to claim 1, wherein: The target item index list is constructed by an index construction system; Constructing the target item index list by the index construction system includes: Constructing an item label mapping table, wherein the item label mapping table records a plurality of items and labels respectively associated with the items; Determining a target item associated with a target tag according to the item tag mapping table; The target items are collected into the same data table to form the target item index list.

6. The method according to claim 5, wherein: The determining, according to the item tag mapping table, a target item associated with a target tag comprises: Determining preliminary selected items associated with the target tag according to the item tag mapping table; generating the attention value of the preliminary selected item according to the degree of attention the preliminary selected item receives; According to the attention value of the preliminary selected items, the target item is determined from the preliminary selected items.

7. An item recommendation device, comprising: A first acquisition module, configured to acquire a tag portrait of a target customer in response to a user request, wherein the tag portrait includes at least one tag associated with the target customer and a first interest value of the tag, wherein the first interest value is used to represent the target customer's preference for the tag; A second acquisition module is used to acquire a target item index list associated with the target tag, wherein the target item index list includes at least one target item associated with the target tag and an attention value of the target item, wherein the attention value is used to represent the degree of attention paid to the target item; a determination module, configured to determine a list of items to be recommended for the target customer according to the first interest value of the target tag and / or the attention value of the target item, so as to return the list of items to be recommended to the user; The first building module includes a third determining unit, configured to determine a first interest value of the tag; The third determining unit includes: A first determination subunit is used to determine the frequency of the target customer's interested operation behavior within a first time period according to the target customer's historical behavior log data, wherein the first time period is the most recent preset time period within the preset historical time period; A first generating subunit, configured to generate a first reference value of the tag according to a frequency of an operation behavior of interest within a first time period; A second generating subunit is configured to multiply a second interest value of the tag in a second time period by a preset attenuation coefficient to obtain a second reference value; and add the first reference value and the second reference value to obtain the first interest value, wherein the second time period is a preset time period before the first time period; The calculation formula of the first interest value is as follows: in, is the first interest value of the tag in the first time period, is the second interest value of the tag in the second time period, is the first reference value of the tag in the first time period, wherein, is the preset attenuation coefficient.

8. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images