Article information generation method and device, equipment, medium and program product

Abstract

The embodiment of the invention discloses an article information generation method and device, equipment, a medium and a program product. A specific embodiment of the method comprises the steps of generating article attention information corresponding to a target user and user attention information corresponding to a target article; generating a user representation vector of the target user at the target moment and an article representation vector of the target article at the target moment according to the article attention information and the user attention information; generating a user future representation vector for the future moment according to the user representation vector and the article representation vector; according to the user future representation vector, the user representation vector and the article representation vector, generating an interactive article representation vector for the future moment; recommended item information is generated for the interactive item representation vector. The embodiment is related to artificial intelligence, and the recommended item information for the target user can be accurately and efficiently generated.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technology, and in particular to methods, devices, equipment, media, and program products for generating item information. Background Art

[0002] At present, item recommendation has become the main development direction of many popular fields. For item recommendation, the method usually adopted is: first, obtain the target user's corresponding item interaction behavior information set. Then, generate a user representation vector corresponding to the item interaction behavior information set. Finally, input the user representation vector into the recommended item information generation network model to generate recommended item information. Among them, the recommended item information generation network model can be a long short-term memory (LSTM) network model.

[0003] However, the inventors have found that when the above method is used to generate recommended item information, the following technical problems often occur:

[0004] Since the interaction behaviors between users and items are constantly changing and the interaction time information is usually ignored, the obtained user representation vector is not accurate enough and cannot fully reflect the relevant feature information of the user, making the recommended item information generated subsequently not accurate enough.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art. Summary of the invention

[0006] The content of this disclosure is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of this disclosure is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0007] Some embodiments of the present disclosure propose methods, devices, electronic devices, computer-readable media, and program products for generating item information to solve the technical problems mentioned in the above background technology section.

[0008] In a first aspect, some embodiments of the present disclosure provide a method for generating item information, including: generating item attention information corresponding to a target user and user attention information corresponding to a target item, wherein the target item is an item that has an interactive behavior with the target user at a target moment; generating a user representation vector of the target user at the target moment and an item representation vector of the target item at the target moment based on the item attention information and the user attention information; generating a user future representation vector for a future moment based on the user representation vector and the item representation vector; generating an interactive item representation vector for the future moment based on the user future representation vector, the user representation vector and the item representation vector; and generating recommended item information for the interactive item representation vector.

[0009] Optionally, the above-mentioned generation of item attention information corresponding to the target user and user attention information corresponding to the target item includes: determining the user history representation vector of the above-mentioned target user at the historical moment and the item history representation vector of the above-mentioned target item at the above-mentioned historical moment; determining at least one interactive item information having a historical interactive behavior relationship between the above-mentioned target users, and determining at least one interactive user information having the above-mentioned historical interactive behavior relationship between the above-mentioned target items; determining the first importance information between each interactive item information in the above-mentioned at least one interactive item information and the user history representation vector to obtain at least one first importance information, and determining the second importance information between each interactive user information in the above-mentioned at least one interactive user information and the item history representation vector to obtain at least one second importance information; generating at least one item weight information for the above-mentioned at least one interactive item information and at least one user weight information for the above-mentioned at least one interactive user information based on the above-mentioned at least one first importance information and the above-mentioned at least one second importance information; generating the above-mentioned item attention information and the above-mentioned user attention information based on the above-mentioned at least one item weight information and the above-mentioned at least one user weight information.

[0010] Optionally, the above-mentioned user history representation vector includes: a user history dynamic representation vector; and the above-mentioned determination of the first importance information between each interactive item information in the above-mentioned at least one interactive item information and the user history representation vector, including: determining the interactive item history dynamic representation vector corresponding to the above-mentioned interactive item information; generating vector importance information representing the relationship between the above-mentioned user history dynamic representation vector and the above-mentioned interactive item history dynamic representation vector as the above-mentioned first importance information.

[0011] Optionally, the above-mentioned generating the above-mentioned item attention information and the above-mentioned user attention information based on the above-mentioned at least one item weight information and the above-mentioned at least one user weight information includes: inputting the above-mentioned at least one item weight information and the above-mentioned item history representation vector into the multi-head attention mechanism model to generate the above-mentioned item attention information; inputting the above-mentioned at least one user weight information and the above-mentioned user history representation vector into the above-mentioned multi-head attention mechanism model to generate the above-mentioned user attention information.

[0012] Optionally, the user representation vector includes: a user dynamic representation vector, the item representation vector includes: an item dynamic representation vector; and the method of generating the user representation vector of the target user at the target moment and the item representation vector of the target item at the target moment based on the item attention information and the user attention information includes: obtaining interaction information between the target user and the target item; generating the user dynamic representation vector using a user representation vector generation model based on the interaction information, the user historical dynamic representation vector and the user attention information; generating the item dynamic representation vector using an item representation vector generation model based on the interaction information, the item historical dynamic representation vector and the item attention information.

[0013] Optionally, the generating of the interactive item representation vector for the above-mentioned future moment according to the above-mentioned user future representation vector, the above-mentioned user representation vector and the above-mentioned item representation vector includes: determining adjacent interactive item information having a preset interactive proximity relationship with the above-mentioned target user; determining the adjacent interactive item representation vector of the above-mentioned adjacent interactive item information at the target future moment; generating the above-mentioned interactive item representation vector according to the above-mentioned adjacent interactive item representation vector, the above-mentioned user future representation vector, the above-mentioned user representation vector and the above-mentioned item representation vector.

[0014] Optionally, the generating of the recommended item information for the interactive item representation vector includes: inputting the interactive item representation vector into a pre-trained local sensitivity hashing algorithm model to generate the recommended item information.

[0015] In a second aspect, some embodiments of the present disclosure provide an item information generating device, comprising: a first generating unit, configured to generate item attention information corresponding to a target user and user attention information corresponding to a target item, wherein the target item is an item that has an interactive behavior with the target user at a target moment; a second generating unit, configured to generate a user representation vector of the target user at the target moment and an item representation vector of the target item at the target moment based on the item attention information and the user attention information; a third generating unit, configured to generate a future user representation vector for a future moment based on the user representation vector and the item representation vector; a fourth generating unit, configured to generate an interactive item representation vector for the future moment based on the user future representation vector, the user representation vector and the item representation vector; and a fifth generating unit, configured to generate recommended item information for the interactive item representation vector.

[0016] Optionally, the first generating unit may be configured to: determine the user history representation vector of the target user at the historical moment and the item history representation vector of the target item at the historical moment; determine at least one interactive item information having a historical interactive behavior relationship between the target users, and determine at least one interactive user information having the historical interactive behavior relationship between the target items; determine the first importance information between each interactive item information in the at least one interactive item information and the user history representation vector to obtain at least one first importance information, and determine the second importance information between each interactive user information in the at least one interactive user information and the item history representation vector to obtain at least one second importance information; generate at least one item weight information for the at least one interactive item information and at least one user weight information for the at least one interactive user information based on the at least one first importance information and the at least one second importance information; generate the item attention information and the user attention information based on the at least one item weight information and the at least one user weight information.

[0017] Optionally, the above-mentioned user history representation vector includes: a user history dynamic representation vector; and the first generation unit can be configured to: determine the interactive item history dynamic representation vector corresponding to the above-mentioned interactive item information; generate vector importance information representing the relationship between the above-mentioned user history dynamic representation vector and the above-mentioned interactive item history dynamic representation vector as the above-mentioned first importance information.

[0018] Optionally, the first generation unit can be configured to: input the at least one item weight information and the item history representation vector into the multi-head attention mechanism model to generate the item attention information; input the at least one user weight information and the user history representation vector into the multi-head attention mechanism model to generate the user attention information.

[0019] Optionally, the user representation vector includes: a user dynamic representation vector, the item representation vector includes: an item dynamic representation vector; and the second generation unit can be configured to: obtain interaction information between the target user and the target item; generate the user dynamic representation vector using a user representation vector generation model based on the interaction information, the user historical dynamic representation vector and the user attention information; generate the item dynamic representation vector using an item representation vector generation model based on the interaction information, the item historical dynamic representation vector and the item attention information.

[0020] Optionally, the fourth generation unit can be configured to: determine the nearby interactive item information that has a preset interactive proximity relationship with the above-mentioned target user; determine the nearby interactive item representation vector of the above-mentioned nearby interactive item information at the target future moment; generate the above-mentioned interactive item representation vector based on the above-mentioned nearby interactive item representation vector, the above-mentioned user future representation vector, the above-mentioned user representation vector and the above-mentioned item representation vector.

[0021] Optionally, the fifth generating unit may be configured to: input the above-mentioned interactive item representation vector into a pre-trained local sensitivity hashing algorithm model to generate the above-mentioned recommended item information.

[0022] In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0023] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner in the first aspect is implemented.

[0024] In a fifth aspect, some embodiments of the present disclosure provide a computer program product, including a computer program, which implements the method described in any implementation manner in the above-mentioned first aspect when executed by a processor.

[0025] The above-mentioned embodiments of the present disclosure have the following beneficial effects: the item information generation method of some embodiments of the present disclosure can accurately and efficiently generate recommended item information for the target user. Specifically, the reason why the relevant recommended item information is not accurate is that the interaction behavior corresponding to the user and the item is constantly changing, and the interaction time information is usually ignored, resulting in the obtained user representation vector being not accurate enough, and the relevant feature information of the user cannot be fully reflected, so that the recommended item information generated subsequently is not accurate enough. Based on this, the item information generation method of some embodiments of the present disclosure first generates the item attention information corresponding to the target user and the user attention information corresponding to the target item. Among them, the above-mentioned target item is an item that has an interaction behavior with the above-mentioned target user at the target moment. Here, through multi-faceted feature importance analysis, the item attention information and user attention information are generated for the subsequent generation of more accurate user representation vectors and item representation vectors. In addition, the item attention information and user attention information are generated at the target moment in a targeted manner, and the interaction time information is considered, so that at the first time of the interaction, the user's corresponding representation vector and the item representation vector of the corresponding item are fully generated subsequently. Then, according to the above-mentioned item attention information and the above-mentioned user attention information, the user representation vector of the above-mentioned target user at the above-mentioned target time and the item representation vector of the above-mentioned target item at the above-mentioned target time can be generated more accurately. Here, through the item attention information and the above-mentioned user attention information, the generated user representation vector can reflect more user demand characteristics and user interaction characteristics, and the generated item representation vector can reflect more item demand characteristics and item interaction characteristics, so that the recommended item information can be generated more accurately in the future. Then, according to the above-mentioned user representation vector and the above-mentioned item representation vector, the user future representation vector for the future moment can be accurately generated. Here, by predicting the user future representation vector, the transformation of the target user corresponding to the user representation vector can be effectively grasped, so that the item information (i.e., the interactive item representation vector) that the user may interact with can be generated more accurately in the future. Therefore, by considering the user representation vector at the future moment, more accurate information of items of interest for the target user can be learned from the perspective of interaction time. Furthermore, according to the above-mentioned user future representation vector, the above-mentioned user representation vector and the above-mentioned item representation vector, the interactive item representation vector for the above-mentioned future moment can be accurately generated. Finally, the recommended item information for the above-mentioned interactive item representation vector can be accurately generated. In summary, from the perspective of interaction time, we not only fully consider the user representation vector and the corresponding item interaction vector at the target moment and the future moment, but also fully consider that the interaction behaviors corresponding to the user and the item are constantly changing, so that we can generate recommended item information more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0027] Figure 1 is a schematic diagram of an application scenario of the method for generating item information according to some embodiments of the present disclosure;

[0028] Figure 2 is a flow chart of some embodiments of the method for generating item information according to the present disclosure;

[0029] Figure 3 is a schematic diagram of a graph structure corresponding to a graph in some embodiments of the method for generating item information according to the present disclosure;

[0030] Figure 4 is a flow chart of other embodiments of the method for generating item information according to the present disclosure;

[0031] Figure 5 is a schematic diagram of the structure of some embodiments of the device for generating item information according to the present disclosure;

[0032] Figure 6 It is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0033] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0034] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other.

[0035] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0036] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0037] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0038] With regard to the collection, storage, and use of information (such as recommended item information) involved in this disclosure, before performing corresponding operations, the relevant organizations or individuals shall fulfill obligations such as conducting information security impact assessments, fulfilling the obligation to inform the information subject, and obtaining the authorization and consent of the information subject in advance.

[0039] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0040] Figure 1 It is a schematic diagram of an application scenario of the method for generating item information according to some embodiments of the present disclosure.

[0041] exist Figure 1 In the application scenario, first, the electronic device 101 can generate the item attention information 103 corresponding to the target user 102 and the user attention information 105 corresponding to the target item 104. Among them, the above-mentioned target item 104 is an item that has an interactive behavior with the above-mentioned target user 102 at the target moment. In this application scenario, the target user 102 can be "A user". The target item 104 can be "B item". Then, the electronic device 101 can generate the user representation vector 106 of the above-mentioned target user 102 at the above-mentioned target moment and the item representation vector 107 of the above-mentioned target item 104 at the above-mentioned target moment according to the above-mentioned item attention information 103 and the above-mentioned user attention information 105. Next, the electronic device 101 can generate the user future representation vector 108 for the future moment according to the above-mentioned user representation vector 106 and the above-mentioned item representation vector 107. Furthermore, the electronic device 101 can generate the interactive item representation vector 109 for the above-mentioned future moment according to the above-mentioned user future representation vector 108, the above-mentioned user representation vector 106 and the above-mentioned item representation vector 107. Finally, the electronic device 101 may generate recommended item information 110 for the above-mentioned interactive item representation vector 109. In this application scenario, the recommended item information 110 may be "B1 item".

[0042] It should be noted that the electronic device 101 can be hardware or software. When the electronic device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or it can be implemented as a single server or a single terminal device. When the electronic device is embodied as software, it can be installed in the hardware devices listed above. It can be implemented as multiple software or software modules for providing distributed services, for example, or it can be implemented as a single software or software module. No specific limitation is made here.

[0043] It should be understood that Figure 1 The number of electronic devices in the embodiment is only for illustration. Any number of electronic devices may be provided according to implementation requirements.

[0044] Continue to refer Figure 2 , shows a process 200 of some embodiments of the method for generating item information according to the present disclosure. The method for generating item information comprises the following steps:

[0045] Step 201, generating object attention information corresponding to a target user and user attention information corresponding to a target object.

[0046] In some embodiments, the execution subject of the above-mentioned item information generation method (for example Figure 1 The electronic device 101 shown) can generate item attention information corresponding to the target user and user attention information corresponding to the target item. Among them, the target user can be the item to be recommended. The item attention information can characterize the target user's preference for each item. That is, the importance of the item relative to the target user. In practice, the item attention information can be information in the form of a vector. The user attention information corresponding to the target item can characterize the user population information that likes the target item. The user attention information can be information in the form of a vector. The above-mentioned target item is an item that has an interactive behavior with the above-mentioned target user at the target moment. The target moment can be a predetermined moment. For example, the target moment can be the current moment. In practice, for e-commerce scenarios, the target user can be a user browsing an e-commerce application. The target item can be a commodity that has an interactive behavior with the target user. For example, the interactive behavior can be but is not limited to at least one of the following: transaction behavior, collection behavior, and adding to shopping cart behavior.

[0047] It should be noted that the user information corresponding to the target user and the item information corresponding to the target item can be nodes in the graph. The graph includes multiple nodes and multiple edges. The node can be one of the following: user information, interactive item information. The edge can include: interactive information and interactive time that characterize the interactive behavior between the user and the item.

[0048] In practice, such as Figure 3 As shown, a schematic diagram of the graph structure corresponding to the spectrum is shown.

[0049] Among them, user A has interactive behavior relationships with item A and item C. User B has interactive behavior relationships with item A, item C, and item B. User C has interactive behavior relationships with item A and item B.

[0050] As an example, first, an interactive item information set corresponding to the target user and an interactive user information set corresponding to the target item are obtained. Then, the item category corresponding to each interactive item information in the above interactive item information set and the user category corresponding to each interactive user information in the interactive user information set are determined. Then, according to the item category corresponding to each interactive item information, the interactive item information set is divided into multiple interactive item information subsets, and according to the user category corresponding to each interactive user information, the interactive user information set is divided into multiple interactive user information subsets. According to the number ratio of the interactive item information included in each interactive item information subset, each first weight information corresponding to each interactive item information subset is generated. And according to the number ratio of the interactive user information included in each interactive user information subset, each second weight information corresponding to each interactive user information subset is generated. Then, an interactive item vector corresponding to each interactive item information in the multiple interactive item information subsets is generated to obtain multiple interactive item vector subsets, and an interactive user vector corresponding to each interactive user information in the multiple interactive user information subsets is generated to obtain multiple interactive user vector subsets. Finally, multiple interactive item vector subsets and each first weight information are correspondingly multiplied to obtain a first multiplication result as the item attention information, and multiple interactive user vector subsets and each second weight information are correspondingly multiplied to obtain a second multiplication result as the user attention information.

[0051] In some optional implementations of some embodiments, the above-mentioned generating the object attention information corresponding to the target user and the user attention information corresponding to the target object may include the following steps:

[0052] In the first step, the execution entity may determine the user history representation vector of the target user at the historical moment and the item history representation vector of the target item at the historical moment. The historical moment may be a moment before the target moment. The user history representation vector may be a user representation vector before the target moment with the closest corresponding time. Similarly, the item history representation vector may be an item representation vector before the target moment with the closest corresponding time. For example, for a target moment of 11 o'clock, the user representation vector sequence of the target user before 11 o'clock includes: a first user representation vector, a second user representation vector, and a third user representation vector. The corresponding user history representation vector may be the third user representation vector. The item representation vector sequence of the target item before 11 o'clock includes: a first item representation vector, a second item representation vector, and a third item representation vector. The corresponding item history representation vector may be the third item representation vector.

[0053] The second step is to determine at least one interactive item information having a historical interactive behavior relationship between the target user and at least one interactive user information having the historical interactive behavior relationship between the target item. The historical interactive behavior relationship may represent the interactive behavior that has occurred. The at least one interactive item information includes item information corresponding to the target item. The at least one interactive user information includes user information corresponding to the target user.

[0054] As an example, the above-mentioned execution entity can determine the target number of interactive item information with the closest time and historical interactive behavior relationship between the above-mentioned target users as at least one interactive item information, and determine the target number of interactive user information with the closest time and historical interactive behavior relationship between the above-mentioned target items as at least one interactive user information.

[0055] As another example, the above-mentioned execution entity may determine an interactive item information set within a historical target time period in which a historical interactive behavior relationship exists between the above-mentioned target users as at least one interactive item information, and determine an interactive user information set within a historical target time period in which a historical interactive behavior relationship exists between the above-mentioned target items as at least one interactive user information.

[0056] The third step is to determine the first importance information between each interactive item information in the at least one interactive item information and the user history representation vector to obtain at least one first importance information, and determine the second importance information between each interactive user information in the at least one interactive user information and the item history representation vector to obtain at least one second importance information. The first importance information represents the importance of the interactive item information relative to the target user. The second importance information represents the importance of the interactive user information relative to the target item.

[0057] The fourth step is to generate at least one item weight information for the at least one interactive item information and at least one user weight information for the at least one interactive user information based on the at least one first importance information and the at least one second importance information.

[0058] As an example, the execution entity may normalize at least one first importance information and at least one second importance information to obtain at least one item weight information and at least one user weight information.

[0059] As another example, the execution entity may input the at least one first importance information and the at least one second importance information into a Softmax function to generate at least one item weight information and at least one user weight information.

[0060] The fifth step is to generate the item attention information and the user attention information based on the at least one item weight information and the at least one user weight information.

[0061] As an example, first, the execution entity may perform corresponding multiplication of at least one item vector and at least one item weight information to obtain at least one first multiplication result. Then, the execution entity may perform corresponding multiplication of at least one user vector and at least one user weight information to obtain at least one second multiplication result. Finally, each first multiplication result in the at least one first multiplication result is concatenated to obtain a first concatenated result as item attention information, and each second multiplication result in the at least one second multiplication result is concatenated to obtain a second concatenated result as user attention information.

[0062] In some optional implementations of some embodiments, the user history representation vector includes: a user history dynamic representation vector, and the item history representation vector includes: an item history dynamic representation vector. The user history dynamic representation vector may be a user dynamic representation vector with the shortest corresponding time before the target moment. The item history dynamic representation vector is an item dynamic representation vector with the shortest corresponding time before the target moment. Among them, the user dynamic representation vector may represent a representation vector of user attributes for the user that may change continuously with the change of the interactive behavior. That is, the user dynamic representation vector corresponds to at least one user attribute as a dynamically changing attribute. In practice, the user dynamic representation vector corresponds to at least one user attribute including: user transaction volume, average number of user orders, and number of user orders. The item dynamic representation vector may be a representation vector of item attributes for the item that may change continuously with the change of the interactive behavior. That is, the item dynamic representation vector corresponds to at least one item attribute as a dynamically changing attribute. In practice, the item dynamic representation vector corresponds to at least one item attribute including: item price, item transaction volume, item order quantity, and item inventory quantity.

[0063] Optionally, the determining of the first importance information between each interactive item information in the at least one interactive item information and the user history representation vector may include the following steps:

[0064] In the first step, the execution subject may determine the interactive item historical dynamic representation vector corresponding to the interactive item information.

[0065] As an example, the execution subject may determine the interactive item historical dynamic representation vector corresponding to the interactive item information by means of vector query.

[0066] The second step is to generate vector importance information representing the historical dynamic representation vector of the user and the historical dynamic representation vector of the interactive item as the first importance information.

[0067] As an example, the above execution entity can generate importance information through the following formula:

[0068]

[0069] in, It can be the importance information when the target time is t, the corresponding target item is the i-th item, and the corresponding target user is the u-th user. - The target time can be t - The item history dynamic representation vector corresponding to the i-th item at the moment. nodeIt can be a node-level attention model. σ is the activation function. a can be a node-level attention vector. u(t - ) can be the target time t - The user historical dynamic representation vector corresponding to the i-th item at the moment.

[0070] In some optional implementations of some embodiments, generating the item attention information and the user attention information according to the at least one item weight information and the at least one user weight information may include the following steps:

[0071] In the first step, the execution subject may input the at least one item weight information and the item history representation vector into a multi-head attention mechanism model to generate the item attention information. In practice, the multi-head attention mechanism model may be an attention mechanism model in a Transformer model.

[0072] In the second step, the above-mentioned execution entity can input the above-mentioned at least one user weight information and the above-mentioned user history representation vector into the above-mentioned multi-head attention mechanism model to generate the above-mentioned user attention information.

[0073] Step 202: Generate a user representation vector of the target user at the target time and an item representation vector of the target item at the target time based on the item attention information and the user attention information.

[0074] In some embodiments, the execution entity may generate a user representation vector of the target user at the target moment and an item representation vector of the target item at the target moment based on the item attention information and the user attention information. The user representation vector may represent a vector of user feature information corresponding to the user. Specifically, the user feature information may be feature information of features related to interaction with items, or user attribute feature information. User attribute feature information may be feature information related to user attributes. For example, user attributes may include, but are not limited to, at least one of the following: user name, user credit, user contact information. The item representation vector may represent a vector of item feature information corresponding to the item. Specifically, the item feature information may be feature information of features related to interaction with the user, or item attribute feature information. The item attribute feature information may be feature information related to user attributes. For example, item attributes may include, but are not limited to, at least one of the following: item identification, item value, and item origin.

[0075] As an example, first, the execution subject can obtain the user attribute information set and the item attribute information set of the target user at the target time. Then, a user attribute vector set corresponding to the user attribute information set and an item attribute vector set corresponding to the item attribute information set are generated. Next, an item attribute matrix for the user attribute vector set and an item attribute matrix for the pair of item attribute vector sets are generated. Finally, the item attribute matrix and the item attention information are input into the multi-head attention mechanism model to generate an item representation vector, and the user attribute matrix and the user attention information are input into the multi-head attention mechanism model to generate a user representation vector.

[0076] In some optional implementations of some embodiments, the user representation vector includes: a user dynamic representation vector, and the item representation vector includes: an item dynamic representation vector.

[0077] Optionally, the generating, according to the object attention information and the user attention information, a user representation vector of the target user at the target time and an item representation vector of the target item at the target time may include the following steps:

[0078] The first step is to obtain the interaction information between the target user and the target item. The interaction information may be a representation of the interaction behavior between the target user and the target item. Specifically, the interaction behavior may be an order placement behavior. The interaction information may be in the form of a vector of the interaction behavior.

[0079] As an example, the above execution subject may perform information word embedding processing on the interaction behavior information to generate interaction information.

[0080] In the second step, based on the interaction information, the user historical dynamic representation vector and the user attention information, a user representation vector generation model is used to generate the user dynamic representation vector. The user representation vector generation model may be a model for generating user representation vectors. For example, the user representation vector generation model may be a fully connected model.

[0081] As an example, first, the time difference between the future moment and the target moment is obtained. Then, the time difference, the interaction information, the user history dynamic representation vector, the item history dynamic representation vector, and the user attention information are input into the user representation vector generation model to generate the user dynamic representation vector.

[0082] In practice, the user representation vector generation model can be the following formula:

[0083]

[0084] Among them, u(t` can be the user representation vector at time t. It can be the first parameter matrix for the user. u(t - It can be the user's historical dynamic representation vector at time t. It can be the second parameter matrix for the user. i(t - It can be a dynamic representation vector of item history. attn (i can be the user's attention information at time t. It can be the third parameter matrix for the user. f can be the interaction information. It can be the fourth parameter matrix for the user. ι It could be a time difference. It can be a fifth parameter matrix for the user.

[0085] In the third step, based on the interaction information, the item historical dynamic representation vector and the item attention information, the item representation vector generation model is used to generate the item dynamic representation vector. The item representation vector generation model may be a model for generating item representation vectors. For example, the item representation vector generation model may be a fully connected model.

[0086] As an example, first, the time difference between the future moment and the target moment is obtained. Then, the time difference, the interaction information, the user history dynamic representation vector, the item history dynamic representation vector, and the item attention information are input into the item representation vector generation model to generate the item dynamic representation vector.

[0087] It should be noted that the model structure corresponding to the item representation vector generation model may be the same as the model structure corresponding to the user representation vector generation model, and details will not be repeated here.

[0088] Step 203: Generate a user future representation vector for a future moment according to the user representation vector and the item representation vector.

[0089] In some embodiments, the execution subject may generate a user future representation vector for a future moment based on the user representation vector and the item representation vector. The future moment may be a moment set relative to a target moment. That is, the future moment is a moment after the target moment. The time length between the target moment and the future moment may be preset. The user future representation vector may be a vector after the user representation of the target user at the future moment is transformed. That is, the user future representation vector may represent the corresponding behavioral feature information of the user at the future moment.

[0090] As an example, the above execution entity can directly input the user representation vector and the item representation vector into a pre-trained fully connected neural network model to output the user's future representation vector for a future moment.

[0091] As another example, the above execution subject may generate the user future representation vector by the following formula:

[0092]

[0093] in, It can be the user's future representation vector for the future moment. n Can be the parameter vector corresponding to the linear layer.

[0094] Step 204: Generate an interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector.

[0095] In some embodiments, the execution subject may generate an interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector, wherein the interactive item representation vector may be an item representation vector of an interactive item with which the target user may have an interactive behavior at the future moment.

[0096] As an example, the execution entity may input the user future representation vector, the user representation vector and the item representation vector into a fully connected neural network model to generate an interactive item representation vector for the future moment.

[0097] In some optional implementations of some embodiments, generating the interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector may include the following steps:

[0098] The first step is to determine the adjacent interactive item information that has a preset interactive proximity relationship with the target user, wherein the preset interactive proximity relationship may be a pre-set interactive relationship that characterizes the interactive behavior of the nearest neighbor.

[0099] The second step is to determine the adjacent interactive item representation vector of the adjacent interactive item information at the target future time. The specific implementation method can refer to the generation of the item representation vector.

[0100] The third step is to generate the above-mentioned interactive item representation vector according to the above-mentioned adjacent interactive item representation vector, the above-mentioned user future representation vector, the above-mentioned user representation vector and the above-mentioned item representation vector.

[0101] As an example, the above execution entity can generate the interactive item representation vector through the following formula:

[0102]

[0103] in, It can be the interactive item representation vector. W1, W2, W3, W4 and B are the parameter matrices in the fully connected linear layer. i(t+Δ - It can be a vector representing adjacent interactive items. It can be a static representation vector of the item corresponding to the target user.

[0104] As another example, the execution entity may input the neighboring interactive item representation vector, the user future representation vector, the user representation vector, and the item representation vector into the attention mechanism model to generate the interactive item representation vector.

[0105] Step 205: Generate recommended item information for the above interactive item representation vector.

[0106] In some embodiments, the execution entity may generate recommended item information for the interactive item representation vector. The recommended item information may be item information of the recommended item. The recommended item may be an item to be recommended to the target user.

[0107] As an example, first, the execution subject may obtain multiple category vectors corresponding to multiple item categories. Then, the category vector with the smallest corresponding cosine distance with the interactive item representation vector among the multiple category vectors is determined as the target category vector. Next, multiple item information corresponding to the target category vector is determined. Finally, the item information with the smallest corresponding cosine distance between the corresponding item representation vector and the interactive item representation vector is determined from the multiple item information as the recommended item information.

[0108] The above-mentioned embodiments of the present disclosure have the following beneficial effects: the item information generation method of some embodiments of the present disclosure can accurately and efficiently generate recommended item information for the target user. Specifically, the reason why the relevant recommended item information is not accurate is that the interaction behavior corresponding to the user and the item is constantly changing, and the interaction time information is usually ignored, resulting in the obtained user representation vector being not accurate enough, and the relevant feature information of the user cannot be fully reflected, so that the recommended item information generated subsequently is not accurate enough. Based on this, the item information generation method of some embodiments of the present disclosure first generates the item attention information corresponding to the target user and the user attention information corresponding to the target item. Among them, the above-mentioned target item is an item that has an interaction behavior with the above-mentioned target user at the target moment. Here, through multi-faceted feature importance analysis, the item attention information and user attention information are generated for the subsequent generation of more accurate user representation vectors and item representation vectors. In addition, the item attention information and user attention information are generated at the target moment in a targeted manner, and the interaction time information is considered, so that at the first time of the interaction, the user's corresponding representation vector and the item representation vector of the corresponding item are fully generated subsequently. Then, according to the above-mentioned item attention information and the above-mentioned user attention information, the user representation vector of the above-mentioned target user at the above-mentioned target time and the item representation vector of the above-mentioned target item at the above-mentioned target time can be generated more accurately. Here, through the item attention information and the above-mentioned user attention information, the generated user representation vector can reflect more user demand characteristics and user interaction characteristics, and the generated item representation vector can reflect more item demand characteristics and item interaction characteristics, so that the recommended item information can be generated more accurately in the future. Then, according to the above-mentioned user representation vector and the above-mentioned item representation vector, the user future representation vector for the future moment can be accurately generated. Here, by predicting the user future representation vector, the transformation of the target user corresponding to the user representation vector can be effectively grasped, so that the item information (i.e., the interactive item representation vector) that the user may interact with can be generated more accurately in the future. Therefore, by considering the user representation vector at the future moment, more accurate information of items of interest for the target user can be learned from the perspective of interaction time. Furthermore, according to the above-mentioned user future representation vector, the above-mentioned user representation vector and the above-mentioned item representation vector, the interactive item representation vector for the above-mentioned future moment can be accurately generated. Finally, the recommended item information for the above-mentioned interactive item representation vector can be accurately generated. In summary, from the perspective of interaction time, we not only fully consider the user representation vector and the corresponding item interaction vector at the target moment and the future moment, but also fully consider that the interaction behaviors corresponding to the user and the item are constantly changing, so that we can generate recommended item information more accurately.

[0109] Further references Figure 4 , shows a process 400 of another embodiment of the method for generating item information according to the present disclosure. The method for generating item information includes the following steps:

[0110] Step 401, generating object attention information corresponding to a target user and user attention information corresponding to a target object.

[0111] Step 402: Generate a user representation vector of the target user at the target time and an item representation vector of the target item at the target time based on the item attention information and the user attention information.

[0112] Step 403: Generate a user future representation vector for a future moment according to the user representation vector and the item representation vector.

[0113] Step 404: Generate an interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector.

[0114] In some embodiments, the specific implementation of steps 401-404 and the technical effects thereof can be referred to in Figure 2 The steps 201-204 in the corresponding embodiment are not described in detail here.

[0115] Step 405: input the above-mentioned interactive item representation vector into a pre-trained local sensitivity hashing algorithm model to generate the above-mentioned recommended item information.

[0116] In some embodiments, an execution entity (e.g. Figure 1 The electronic device 101 shown can input the above interactive item representation vector into a pre-trained locality-sensitive hashing algorithm model to generate the above recommended item information. The locality-sensitive hashing algorithm model can be a network model based on a locality-sensitive hashing (LSH) algorithm.

[0117] from Figure 4 It can be seen that Figure 2 Compared with the description of some corresponding embodiments, Figure 4 The process 400 of the item information generation method in some corresponding embodiments uses a local sensitivity hash algorithm model to generate recommended item information for an interactive item representation vector, which can accurately generate item information with the closest interactive item representation vector in a nearly constant time, greatly reducing the amount of calculation.

[0118] Further references Figure 5As an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of an item information generating device. These device embodiments are similar to Figure 2 Corresponding to the method embodiments shown, the device for generating item information can be specifically applied to various electronic devices.

[0119] like Figure 5 As shown, an item information generating device 500 includes: a first generating unit 501, a second generating unit 502, a third generating unit 503, a fourth generating unit 504 and a fifth generating unit 505. The first generating unit 501 is configured to generate item attention information corresponding to a target user and user attention information corresponding to a target item, wherein the target item is an item that has an interactive behavior with the target user at a target moment; the second generating unit 502 is configured to generate a user representation vector of the target user at the target moment and an item representation vector of the target item at the target moment according to the item attention information and the user attention information; the third generating unit 503 is configured to generate a user future representation vector for a future moment according to the user representation vector and the item representation vector; the fourth generating unit 504 is configured to generate an interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector; the fifth generating unit 505 is configured to generate recommended item information for the interactive item representation vector.

[0120] In some optional implementations of some embodiments, the first generation unit 501 may be further configured to: determine the user history representation vector of the target user at the historical moment and the item history representation vector of the target item at the historical moment; determine at least one interactive item information having a historical interactive behavior relationship between the target users, and determine at least one interactive user information having the historical interactive behavior relationship between the target items; determine the first importance information between each interactive item information in the at least one interactive item information and the user history representation vector, and obtain at least one first importance information, and determine the second importance information between each interactive user information in the at least one interactive user information and the item history representation vector, and obtain at least one second importance information; generate at least one item weight information for the at least one interactive item information and at least one user weight information for the at least one interactive user information based on the at least one first importance information and the at least one second importance information; generate the item attention information and the user attention information based on the at least one item weight information and the at least one user weight information.

[0121] In some optional implementations of some embodiments, the user history representation vector includes: a user history dynamic representation vector; and the first generating unit 501 may be further configured to:

[0122] The above-mentioned determination of the first importance information between each interactive item information in the above-mentioned at least one interactive item information and the user history representation vector includes: determining the interactive item historical dynamic representation vector corresponding to the above-mentioned interactive item information; generating vector importance information representing the relationship between the above-mentioned user history dynamic representation vector and the above-mentioned interactive item history dynamic representation vector as the above-mentioned first importance information.

[0123] In some optional implementations of some embodiments, the first generation unit 501 may be further configured to: input the at least one item weight information and the item history representation vector into a multi-head attention mechanism model to generate the item attention information; input the at least one user weight information and the user history representation vector into the multi-head attention mechanism model to generate the user attention information.

[0124] In some optional implementations of some embodiments, the user representation vector includes: a user dynamic representation vector, the item representation vector includes: an item dynamic representation vector; and the second generation unit 502 may be further configured to: obtain interaction information between the target user and the target item; generate the user dynamic representation vector using a user representation vector generation model based on the interaction information, the user historical dynamic representation vector and the user attention information; generate the item dynamic representation vector using an item representation vector generation model based on the interaction information, the item historical dynamic representation vector and the item attention information.

[0125] In some optional implementations of some embodiments, the fourth generation unit 504 may be further configured to: determine the adjacent interactive item information that has a preset interactive proximity relationship with the target user; determine the adjacent interactive item representation vector of the adjacent interactive item information at the target future moment; and generate the interactive item representation vector based on the adjacent interactive item representation vector, the user future representation vector, the user representation vector and the item representation vector.

[0126] In some optional implementations of some embodiments, the fifth generation unit 505 may be further configured to: input the interactive item representation vector into a pre-trained local sensitivity hashing algorithm model to generate the recommended item information.

[0127] It is understandable that the units recorded in the article information generating device 500 are similar to those in the reference Figure 2Therefore, the operations, features and beneficial effects described above for the method are also applicable to the item information generating device 500 and the units included therein, and will not be described in detail here.

[0128] Reference below Figure 6 , which shows an electronic device (eg, Figure 1 Schematic diagram of the structure of the electronic device 101)600. Figure 6 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0129] like Figure 6 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory 602 or a program loaded from a storage device 608 into a random access memory 603. Various programs and data required for the operation of the electronic device 600 are also stored in the random access memory 603. The processing device 601, the read-only memory 602, and the random access memory 603 are connected to each other via a bus 604. An input / output interface 605 is also connected to the bus 604.

[0130] Typically, the following devices may be connected to the input / output interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 6 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 6 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0131] In particular, according to some embodiments of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from a network through a communication device 609, or installed from a storage device 608, or installed from a read-only memory 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of some embodiments of the present disclosure are executed.

[0132] It should be noted that the computer-readable medium in some embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, 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), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In some embodiments of the present disclosure, the computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0133] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0134] The computer-readable medium may be included in the electronic device; or it may exist independently without being installed in the electronic device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: generates item attention information corresponding to the target user and user attention information corresponding to the target item, wherein the target item is an item that has an interactive behavior with the target user at the target moment; generates a user representation vector of the target user at the target moment and an item representation vector of the target item at the target moment based on the item attention information and the user attention information; generates a user future representation vector for a future moment based on the user representation vector and the item representation vector; generates an interactive item representation vector for the future moment based on the user future representation vector, the user representation vector and the item representation vector; and generates recommended item information for the interactive item representation vector.

[0135] Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0136] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square 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 square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / 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.

[0137] The units described in some embodiments of the present disclosure may be implemented by software or by hardware. The described units may also be provided in a processor, for example, may be described as: a processor including a first generation unit, a second generation unit, a third generation unit, a fourth generation unit, and a fifth generation unit. The names of these units do not, in some cases, constitute limitations on the units themselves, for example, the first generation unit may also be described as a "unit for generating object attention information corresponding to a target user and user attention information corresponding to a target object".

[0138] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip (SOCs), complex programmable logic devices (CPLDs), and the like.

[0139] Some embodiments of the present disclosure further provide a computer program product, including a computer program, which implements any of the above-mentioned methods for generating item information when executed by a processor.

[0140] The above descriptions are only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) and the technical solutions formed.

Claims

1. A method for generating item information, comprising: Generate item attention information corresponding to a target user and user attention information corresponding to a target item, wherein the target item is an item that has an interactive behavior with the target user at a target time; Generate a user representation vector of the target user at the target time and an item representation vector of the target item at the target time according to the item attention information and the user attention information; Generate a user future representation vector for a future moment according to the user representation vector and the item representation vector; Generate an interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector; Generate recommended item information for the interactive item representation vector.

2. The method according to claim 1, wherein: The generating of the object attention information corresponding to the target user and the user attention information corresponding to the target object includes: Determine a user history representation vector of the target user at a historical moment and an item history representation vector of the target item at the historical moment; Determine at least one interactive item information having a historical interactive behavior relationship between the target users, and determine at least one interactive user information having the historical interactive behavior relationship between the target items; Determine first importance information between each interactive item information in the at least one interactive item information and the user history representation vector to obtain at least one first importance information, and determine second importance information between each interactive user information in the at least one interactive user information and the item history representation vector to obtain at least one second importance information; generating, according to the at least one first importance information and the at least one second importance information, at least one item weight information for the at least one interactive item information and at least one user weight information for the at least one interactive user information; The item attention information and the user attention information are generated according to the at least one item weight information and the at least one user weight information.

3. The method according to claim 2, wherein: The user history representation vector includes: a user history dynamic representation vector; and The determining of first importance information between each interactive item information in the at least one interactive item information and the user history representation vector comprises: Determine an interactive item historical dynamic representation vector corresponding to the interactive item information; Vector importance information representing the historical dynamic representation vector of the user and the historical dynamic representation vector of the interactive item is generated as the first importance information.

4. The method according to claim 2, wherein: The generating the item attention information and the user attention information according to the at least one item weight information and the at least one user weight information comprises: Inputting the at least one item weight information and the item history representation vector into a multi-head attention mechanism model to generate the item attention information; The at least one user weight information and the user history representation vector are input into the multi-head attention mechanism model to generate the user attention information.

5. The method according to claim 3, wherein: The user representation vector includes: a user dynamic representation vector, and the item representation vector includes: an item dynamic representation vector; and The step of generating a user representation vector of the target user at the target time and an item representation vector of the target item at the target time according to the item attention information and the user attention information includes: Acquiring interaction information between the target user and the target object; generating the user dynamic representation vector using a user representation vector generation model according to the interaction information, the user historical dynamic representation vector and the user attention information; According to the interaction information, the item historical dynamic representation vector and the item attention information, an item representation vector generation model is used to generate the item dynamic representation vector.

6. The method according to claim 1, wherein: The step of generating the interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector comprises: Determine information of adjacent interactive items that have a preset interactive proximity relationship with the target user; Determine a neighboring interactive item representation vector of the neighboring interactive item information at a target future time; The interactive item representation vector is generated according to the adjacent interactive item representation vector, the user future representation vector, the user representation vector and the item representation vector.

7. The method according to claim 1, wherein: The generating the recommended item information for the interactive item representation vector includes: The interactive item representation vector is input into a pre-trained local sensitivity hashing algorithm model to generate the recommended item information.

8. An item information generating device, comprising: A first generating unit is configured to generate item attention information corresponding to a target user and user attention information corresponding to a target item, wherein the target item is an item that has an interactive behavior with the target user at a target moment; A second generating unit is configured to generate a user representation vector of the target user at the target time and an item representation vector of the target item at the target time according to the item attention information and the user attention information; A third generating unit is configured to generate a user future representation vector for a future moment according to the user representation vector and the item representation vector; a fourth generating unit, configured to generate an interactive item representation vector for the future moment according to the user future representation vector, the user representation vector and the item representation vector; A fifth generating unit is configured to generate recommended item information for the interactive item representation vector.

9. An electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

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

Images