Information recommendation method, device, medium and equipment

Abstract

The present disclosure relates to the field of data processing technology, and provides an information recommendation method and device, as well as a computer storage medium and an electronic device. The method includes: obtaining the browsing behavior data of the target user for each historical recommendation information and obtaining the identification data of each historical recommendation information; performing hash processing on the identification data of each historical recommendation information to convert the identification data of each historical recommendation information into a hash value respectively; determining the interest weight of the target user for each historical recommendation information according to the browsing behavior data of the target user for each historical recommendation information; determining the interest vector of the target user based on the hash value and interest weight corresponding to each historical recommendation information, so as to determine the recommended information for the target user based on the interest vector. This technical solution can improve the calculation efficiency of the recommended information, and is conducive to improving the real-time performance of determining the recommended information.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular, to an information recommendation method and device, as well as a computer storage medium and an electronic device for implementing the information recommendation method. Background Art

[0002] Feeds refers to information dynamically displayed in social media, for example, sports news, entertainment news, etc. dynamically displayed in social media.

[0003] In order to meet the personalized needs of users, the information flow determines the user's interests based on the user's historical browsing and viewing behavior, thereby realizing personalized recommendation display information (Item) according to the user's interests. In related technologies, the recommended display information for the user is generally determined based on the user's browsing behavior for different display information.

[0004] However, the related technology has the problem of low computational efficiency of recommending and displaying information, which is not conducive to determining the recommended information for users in real time.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure. Summary of the invention

[0006] The purpose of the present disclosure is to provide an information recommendation method and device, as well as a computer storage medium and electronic device for implementing the above-mentioned information recommendation method, so as to improve the calculation efficiency of the recommended information at least to a certain extent while ensuring the accuracy of the recommended information, which is conducive to real-time determination of the recommended information for the user.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to one aspect of the present disclosure, there is provided an information recommendation method, comprising:

[0009] Obtaining browsing behavior data of target users for each historical recommendation information and obtaining identification data of each historical recommendation information;

[0010] Performing hash processing on the identification data of each piece of the historical recommendation information to convert the identification data of each piece of the historical recommendation information into a hash value respectively;

[0011] Determining the interest weight of the target user for each of the historical recommendation information according to the browsing behavior data of the target user for each of the historical recommendation information; and

[0012] Based on the hash value and the interest weight corresponding to each of the historical recommendation information, the interest vector of the target user is determined, so as to determine the recommendation information for the target user based on the interest vector.

[0013] According to one aspect of the present disclosure, there is provided an information recommendation device, comprising: a data acquisition module, a hash processing module, an interest weight determination module, and an information recommendation module.

[0014] The data acquisition module is configured to: acquire the browsing behavior data of the target user for each historical recommendation information and acquire the identification data of each historical recommendation information;

[0015] The hash processing module is configured to: perform hash processing on the identification data of each piece of the historical recommendation information, so as to convert the identification data of each piece of the historical recommendation information into a hash value respectively;

[0016] The interest weight determination module is configured to: determine the interest weight of the target user for each of the historical recommendation information according to the browsing behavior data of the target user for each of the historical recommendation information; and

[0017] The information recommendation module is configured to determine the interest vector of the target user based on the hash value and interest weight corresponding to each of the historical recommendation information, so as to determine the recommended information for the target user based on the interest vector.

[0018] In some embodiments of the present disclosure, based on the above scheme, the hash processing module is specifically configured as follows:

[0019] Hash processing is performed on the identification number of each of the historical recommendation information, and one or more of the content tag, display type and listing time.

[0020] In some embodiments of the present disclosure, based on the above scheme, the hash processing module is further specifically configured as follows:

[0021] The character string, integer or floating point identification data of each piece of the historical recommendation information is processed by a hash function to obtain a hash value corresponding to the identification data of each piece of the historical recommendation information.

[0022] In some embodiments of the present disclosure, based on the above solution, the above information recommendation module includes: a weighted processing unit and a superposition unit. Among them:

[0023] The weighted processing unit is configured to: perform weighted processing on the hash value corresponding to each of the historical recommendation information according to the interest weight of each of the historical recommendation information to obtain the interest sub-vector of each of the historical recommendation information; and

[0024] The superposition unit is configured to: superimpose the interest sub-vectors of the historical recommendation information to obtain the interest vector of the target user.

[0025] In some embodiments of the present disclosure, based on the above scheme, the hash value includes a first data code and a second data code; wherein:

[0026] The weighted processing unit is specifically configured to: when the data bit of the hash value is the first data code, replace the data bit from the first data code with the interest weight of each of the historical recommendation information; and

[0027] When the data bit of the hash value is the second data code, the data bit is replaced by the opposite number of the interest weight of each piece of the historical recommendation information from the second data code.

[0028] In some embodiments of the present disclosure, based on the aforementioned scheme, the above-mentioned superposition unit includes: a statistical period determination subunit, a candidate interest subvector determination subunit and an interest vector determination subunit.

[0029] in:

[0030] The statistical period determination subunit is configured to: determine the statistical period of each of the historical recommendation information;

[0031] The candidate interest sub-vector determining sub-unit is configured to: process the interest sub-vectors of each of the historical recommendation information according to the time decay coefficient of the statistical period to obtain the candidate interest sub-vectors of each of the historical recommendation information; and

[0032] The interest vector determination subunit is configured to: superimpose the candidate interest subvectors of each of the historical recommendation information to obtain the interest vector of the target user.

[0033] In some embodiments of the present disclosure, based on the above-mentioned solution, the statistical period determination subunit is specifically configured as follows:

[0034] Determine the statistical period of each of the historical recommendation information according to the timestamp of the browsing behavior data of the target user for each of the historical recommendation information; or,

[0035] The statistical period to which each piece of the historical recommendation information belongs is determined according to the time when each piece of the historical recommendation information is put on the shelves.

[0036] In some embodiments of the present disclosure, based on the aforementioned solution, the above-mentioned information recommendation module includes: a recommended information determination unit.

[0037] The above-mentioned recommendation information determination unit includes: a splicing subunit and a recommendation subunit. Specifically:

[0038] The data acquisition module is further configured to: acquire basic attribute data of the target user, acquire attribute data of each of the historical recommendation information, and acquire scene data of the display page;

[0039] The above-mentioned splicing subunit is configured to: splice the interest vector of the target user, the basic attribute data of the target user, the attribute data of each of the historical recommendation information and the scene data of the display page to obtain a splicing feature; and,

[0040] The above-mentioned splicing subunit is configured to: predict the recommendation information for the target user according to the splicing features.

[0041] In some embodiments of the present disclosure, based on the above-mentioned solution, the above-mentioned splicing subunit is specifically configured as follows:

[0042] The splicing features are input into a prediction model trained offline, so that the prediction model determines the output of the model based on the splicing features to obtain a recommendation value for each of the historical recommendation information; and the recommendation information for the target user is determined according to the recommendation score.

[0043] In some embodiments of the present disclosure, based on the above-mentioned solution, the above-mentioned splicing subunit is specifically configured as follows:

[0044] Load the offline trained prediction model into the engine of the information recommendation system;

[0045] In response to a request for access to the display page, triggering a recommendation request to the information recommendation system;

[0046] Acquire the splicing features of the target user in real time through the engine; and,

[0047] The splicing features are input into the prediction model to perform real-time prediction through the prediction model to obtain recommendation information for the target user.

[0048] In some embodiments of the present disclosure, based on the above solution, the above data acquisition module is specifically configured as follows:

[0049] Obtain one or more of the viewing time, click count, collection count, and recommendation count of each historical recommendation information by the target user.

[0050] In some embodiments of the present disclosure, based on the above scheme, the interest weight determination module is specifically configured as follows:

[0051] The interest weight of the target user for each of the historical recommendation information is determined according to one or more of the viewing time, click times, collection times, and recommendation times of each of the historical recommendation information by the target user.

[0052] According to one aspect of the present disclosure, a computer storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the information recommendation method described in the first aspect is implemented.

[0053] According to one aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the information recommendation method described in the first aspect above by executing the executable instructions.

[0054] It can be seen from the above technical solutions that the information recommendation method and device in the exemplary embodiments of the present disclosure, as well as the computer storage medium and electronic device for implementing the above information recommendation method, have at least the following advantages and positive effects:

[0055] In the technical solutions provided by some embodiments of the present disclosure, the target user's interest weight for each historical recommendation information is determined based on the identification data of each historical recommendation information of the target user being hashed, and the target user's browsing behavior data for each historical recommendation information. Then, based on the hash value and interest weight corresponding to each historical recommendation information, the target user's interest vector is determined, so as to determine the recommended information for the target user according to the interest vector.

[0056] On the one hand, this technical solution obtains the interest vector of the target user based on the identification information of each historical recommendation information of the target user and the interest weight; it can be seen that the interest characteristics determined by this technical solution can more comprehensively reflect the preferences of the target user. Furthermore, the recommendation information is determined based on the above interest vector characteristics, which is conducive to ensuring the accuracy of the recommended information for the target user. On the other hand, in this technical solution, the identification information of each historical recommendation information is hashed, and the interest vector is determined based on the hash value obtained by the processing. It is possible to effectively reflect the interests of the target user with a smaller amount of data, thereby improving the calculation efficiency of the recommendation information, which is conducive to improving the real-time nature of determining the recommendation information.

[0057] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present disclosure, and together with the specification, are used to explain the principles of the present disclosure. Obviously, the drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0059] Figure 1 A schematic diagram showing a system architecture of an exemplary application environment in which an information recommendation method and device according to an embodiment of the present disclosure can be applied;

[0060] Figure 2 A flowchart of an information recommendation method according to an embodiment of the present disclosure is schematically shown;

[0061] Figure 3 A flowchart of a method for determining a user's interest vector according to an embodiment of the present disclosure is schematically shown;

[0062] Figure 4 A flowchart schematically shows a method for determining an interest sub-vector of historical recommendation information according to an embodiment of the present disclosure;

[0063] Figure 5 A flowchart of a method for determining a user's interest vector according to another embodiment of the present disclosure is schematically shown;

[0064] Figure 6 The flowchart schematically shows a method for determining a user's interest vector according to yet another embodiment of the present disclosure;

[0065] Figure 7 A flowchart schematically illustrates a method for determining a user's interest vector according to another embodiment of the present disclosure;

[0066] Figure 8 A flowchart of an information recommendation method according to another embodiment of the present disclosure is schematically shown;

[0067] Fig. 9 The flowchart of the information recommendation method according to another embodiment of the present disclosure is schematically shown;

[0068] Fig.10 The flowchart of the information recommendation method according to another embodiment of the present disclosure is schematically shown;

[0069] Fig.11 A schematic diagram showing a structure of an information recommendation device in an exemplary embodiment of the present disclosure; and,

[0070] Fig.12 A schematic structural diagram of an electronic device in an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0071] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be more comprehensive and complete and will fully convey the concept of the example embodiments to those skilled in the art.

[0072] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the present disclosure.

[0073] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0074] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0075] Figure 1 A schematic diagram of a system architecture of an exemplary application environment in which an information recommendation method and device according to an embodiment of the present disclosure can be applied is shown.

[0076] like Figure 1 As shown, the system architecture 100 may include one or more of terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc. The terminal devices 101, 102, 103 may be various electronic devices with display screens, including but not limited to desktop computers, portable computers, smart phones, tablet computers, etc. It should be understood that Figure 1The number of terminal devices, networks and servers in the example is only illustrative. Any number of terminal devices, networks and servers may be provided as required. For example, the server 105 may be a server cluster composed of multiple servers.

[0077] The information recommendation method provided in the embodiment of the present disclosure is generally executed by the server 105, and accordingly, the information recommendation device is generally set in the server 105. However, it is easy for those skilled in the art to understand that the information recommendation method provided in the embodiment of the present disclosure can also be executed by the terminal devices 101, 102, and 103, and accordingly, the information recommendation device can also be set in the terminal devices 101, 102, and 103, which is not particularly limited in this exemplary embodiment.

[0078] For example, in an exemplary embodiment, the terminal devices 101, 102, and 103 may obtain the browsing behavior data of the target user for each historical recommendation information and the identification data of each historical recommendation information and send them to the server 105. Thus, the server 105 obtains the browsing behavior data of the target user for each historical recommendation information and obtains the identification data of each historical recommendation information; then, the server 105 performs hash processing on the identification data of each historical recommendation information to convert the identification data of each historical recommendation information into a hash value respectively. Further, the server 105 determines the interest weight of the target user for each historical recommendation information based on the browsing behavior data of the target user for each historical recommendation information; finally, the server 105 determines the interest vector of the target user based on the hash value and interest weight corresponding to each historical recommendation information, so as to determine the recommendation information for the target user based on the interest vector.

[0079] Exemplarily, the server 105 may also send the recommendation information to the terminal devices 101 , 102 , 103 , so that the target user may conveniently browse the recommendation information through the terminal devices 101 , 102 , 103 .

[0080] In particular, according to an embodiment of the present disclosure, the process described below with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes 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.

[0081] In such an embodiment, the computer program can be downloaded and installed from the Internet. When the computer program is executed by the central processing unit (CPU), various functions defined in the method and apparatus of the present application are executed. In some embodiments, the server 105 may also include an artificial intelligence (AI) processor, which is used to process computing operations related to machine learning.

[0082] Among them, Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0083] Machine Learning (ML) is a multi-disciplinary subject involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0084] The solution provided by the embodiments of the present disclosure involves artificial intelligence machine learning technology, which is specifically described by the following embodiments:

[0085] The usual approach to recommending information to users is to maintain a tag system and determine the corresponding tags in the tag system based on the display content of the display items. The tags corresponding to the display items are obtained based on the user's actions such as clicking on the display items, so as to determine the user's interests based on the tags. Specifically, when implementing personalized recommendations for thousands of people, in order to improve the accuracy of the recommendation of the display items, it is necessary to obtain a larger number of user tags to more comprehensively reflect the user's interest vector. However, a large number of tags will affect the calculation efficiency of the recommended information, which is not conducive to the real-time determination of the recommended information for the user.

[0086] Based on the above problems, the inventors have provided a solution that extracts some tags from the many tags corresponding to the user as the user's portrait features, so as to reduce the number of tags, thereby improving the calculation efficiency of recommended information to a certain extent, and further improving the real-time nature of determining the recommended information for the user.

[0087] However, the inventors found that in the above solution, the user's portrait features simplify the number of tags, which will damage the user's interest features and cannot fully cover the user's interests, which will lead to a decrease in the accuracy of the recommended information.

[0088] Based on the problem of decreased accuracy of recommended information, the inventors have provided another solution, which is to build the tag system into a hierarchical system in order to reduce the dimension of user interest. Only the top tags of the hierarchical system are selected as the user's interest features. However, for a tag system where the relationship between tags is not a simple inclusion relationship, the relationship between tags is relatively complex and cannot be built into a hierarchical system.

[0089] For example, for sports information recommendations, the hierarchical relationship of tags in sports information is not clear at the beginning of construction due to the particularity of the product, such as basketball, international football and domestic football are at the same level. At the same time, the parent-child relationship of tags is not a simple one-to-many relationship, such as the Barcelona team has multiple parent nodes such as "La Liga" and "Champions League".

[0090] However, the inventors found that in the above-mentioned alternative solution, the versatility of the technical solution is low.

[0091] In response to one or more technical problems existing in the related technologies, the present technical solution provides an information recommendation method and device, as well as a computer storage medium and electronic device for implementing the above-mentioned information recommendation method, which can improve the calculation efficiency of the recommended information at least to a certain extent while ensuring the accuracy of the recommended information, and is conducive to real-time determination of the recommended information for the user.

[0092] It should be particularly emphasized that the information recommendation method provided in this application involves the user's browsing behavior data for each historical recommended information. When applying the various embodiments in this application to specific products or technologies, data collection can only be carried out after obtaining the permission or consent of the relevant users, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0093] The following is a detailed description of an embodiment of the information recommendation method provided by the present disclosure:

[0094] Figure 2 The flowchart of the information recommendation method according to an embodiment of the present disclosure is schematically shown. Figure 2, the embodiment shown in the figure includes:

[0095] Step S210, obtaining browsing behavior data of the target user for each historical recommendation information and obtaining identification data of each historical recommendation information;

[0096] Step S220, performing hash processing on the identification data of each piece of the historical recommendation information to convert the identification data of each piece of the historical recommendation information into a hash value respectively;

[0097] Step S230, determining the interest weight of the target user for each of the historical recommendation information according to the browsing behavior data of the target user for each of the historical recommendation information; and

[0098] Step S240: determining the interest vector of the target user based on the hash value and interest weight corresponding to each of the historical recommendation information, so as to determine the recommendation information for the target user based on the interest vector.

[0099] exist Figure 2 In the technical solution provided by the illustrated embodiment: on the one hand, the technical solution obtains the interest vector of the target user based on the identification information and interest weight of each historical recommendation information of the target user; it can be seen that the interest features determined by the technical solution can more comprehensively reflect the preferences of the target user. Furthermore, the recommendation information is determined based on the above interest vector features, which is conducive to ensuring the accuracy of the recommended information for the target user.

[0100] On the other hand, in this technical solution, the identification information of each historical recommendation information is hashed, and the interest vector is determined based on the hash value obtained by the processing. It is possible to effectively reflect the interests of the target user with a small amount of data, thereby improving the calculation efficiency of the recommendation information and facilitating the real-time determination of the recommendation information.

[0101] On the other hand, the technical solution determines the user's interest vector based on the identification data of the historical recommendation information and the user's browsing behavior data, without determining the label relationship of each historical recommendation information or recommending information based on the label relationship. Therefore, the information recommendation method provided by the technical solution has high versatility.

[0102] This technical solution is explained by taking the recommendation information of any user (referred to as "target user") as an example. Figure 2 The specific implementation methods of each step of the embodiment shown are described in detail:

[0103] In an exemplary embodiment, the historical recommendation information for the target user in step S210 refers to multimedia presentation information such as sports news and entertainment news that have been recommended to the target user in the past, which can be recorded as Item.

[0104] In an exemplary embodiment, the above-mentioned historical recommendation information of the target user can be display information recommended to the user according to the user's interests, or can be display information recommended to the user randomly. Among them, one implementation method of recommending display information to the user according to the user's interests is:

[0105] Maintain a tag system; then select the corresponding tag in the tag system according to the content of each display information Item. For example: Item1:tag1; Item2:tag1,tag2; Item3:tag2, and so on. Then, according to the behavior of the target user viewing the Item, determine the target's tag according to the tags of the viewed Items, and record it as the target user's interest tag. For example: user user1 has watched Item1 and Item3, then user user1's interest tags are tag1 and tag2. Further, determine the above recommended display information according to the target user's interest tags.

[0106] Exemplarily, for each interest tag corresponding to the target user, the Item set corresponding to the tag is obtained. Then, the above historical recommendation information is determined according to the Item set corresponding to the interest tag of the target user.

[0107] In an exemplary embodiment, in step S210, the browsing behavior data of the target user for each historical recommendation information is obtained. Exemplarily, the browsing behavior data of the target user for each historical recommendation information can be obtained by obtaining the user's log information.

[0108] Specifically, the viewing time of the target user for the historical recommendation information A can be obtained as the browsing behavior data of the target user for the historical recommendation information A, the number of clicks of the target user for the historical recommendation information B can be obtained as the browsing behavior data of the target user for the historical recommendation information B, the number of favorites of the target user for the historical recommendation information C can be obtained as the browsing behavior data click count of the target user for the historical recommendation information C, and the number of favorites of the target user for the historical recommendation information D can also be obtained as the browsing behavior data click count of the target user for the historical recommendation information D. Of course, the browsing behavior data of a certain historical recommendation information can also be determined according to a combination of the above-mentioned implementation modes.

[0109] The above browsing behavior data can be used to measure the target user's preference for each historical recommendation information to further determine the target user's interest vector. For example:

[0110] In step S230, the interest weight of the target user for each of the historical recommendation information is determined according to the browsing behavior data of the target user for each of the historical recommendation information.

[0111] In an exemplary embodiment, if the target user's viewing time for historical recommendation information is longer / the degree of completion of viewing historical recommendation information is higher, it means that the target user has a higher degree of liking for the historical recommendation information, and thus a higher interest weight value can be determined for the historical recommendation information. Exemplarily, if user S has completed viewing of historical recommendation information E, then the interest weight of user S for historical recommendation information E is determined to be 1. Conversely, the shorter the viewing time for historical recommendation information by the target user / the lower the degree of completion of viewing historical recommendation information, the lower the degree of liking for the historical recommendation information by the target user, and thus a lower interest weight value can be determined for the historical recommendation information. Exemplarily, if user S clicks on historical recommendation information F (a 5-minute video), but only watches the entire video for more than ten seconds, then the interest weight of user S for historical recommendation information F is determined to be a value close to 0.

[0112] Exemplarily, similar to the “watching time / watching completion level” in the above embodiment, the interest weight of the target user for each of the historical recommendation information may also be determined by one or more of the number of clicks, the number of favorites, and the number of recommendations.

[0113] In an exemplary embodiment, in step S210, identification data of each piece of historical recommendation information is also obtained, and in step S220: hash processing is performed on the identification data of each piece of historical recommendation information to convert the identification data of each piece of historical recommendation information into a hash value respectively.

[0114] For example, in order to accurately locate different historical recommendation information, the identification data of the historical recommendation information includes its identification number (such as ItemID in the reference figure), so that the hash value corresponding to each historical recommendation information can be obtained by hashing the identification number of each historical recommendation information. Figure 3 , for the N historical recommendation information of the target user 300, refer to part 310, which can be represented as [Item1, Item2, Item3, ... ItemN] according to their respective identification numbers.

[0115] Furthermore, after performing the above hash processing on ItemID, a portion 320 is obtained. For example, for the historical recommendation information with identification number Item1, its hash value is "1100...11"; for the historical recommendation information with identification number Item2, its hash value is "1000...11", etc. Since the hash value lengths corresponding to each historical recommendation information are the same, it is helpful to accurately determine the user's interest vector by recommending historical information.

[0116] For example, on the basis of ensuring accurate positioning of each historical recommendation information by identification number, the attribute characteristics of each historical recommendation information are added to finally reflect the interests and hobbies of the user. Then, the identification number of the historical recommendation information and one or more of the content tag, display type and listing time of the historical recommendation information can be used as the identification data of the historical recommendation information, and hash processing can be performed to obtain the hash value of the length corresponding to each historical recommendation information.

[0117] The above-mentioned content tag refers to at least one tag marked according to the content of the historical recommendation information, and the above-mentioned display type may include: video type, text type, image type, etc.

[0118] Compared with the related art: determining the user's interest tags by obtaining the information that the user is interested in, and then directly predicting the recommended information for the user through as many interest tags as possible. However, a large number of tags will affect the calculation efficiency of the recommended information. It can be seen that the related art is not conducive to determining the recommended display items for the user in real time.

[0119] In the technical solution, the specific implementation method of converting the identification data of each historical recommendation information into a hash value includes: processing the string type, integer type or floating point type identification data of each historical recommendation information through a hash function to obtain the hash value corresponding to the identification data of each historical recommendation information. Thus, on the basis of preserving the original information amount to the maximum extent, the data amount is effectively reduced, thereby achieving the technical effect of improving the calculation efficiency of the recommendation information, which is conducive to real-time determination of the recommendation information for the target user.

[0120] For example, compared with the related technical solutions, the click-through rate (CTR) of the recommended information can be increased by 2.04%. The specific data is shown in Table 1 below:

[0121] Table 1

[0122] Indicator Adopt relevant technical solutions Adopt this technical solution Lift CTR 10.28％ 10.49％ 2.04％

[0123] It should be noted that the specific implementation methods of the above-mentioned step S220 and step S230 are not in order, and step S220 can be executed first and then step S230, step S230 can be executed first and then step S220, or step S220 and step S230 can be executed simultaneously.

[0124] Specifically, the interest weight of each historical recommendation information is determined by the specific implementation method corresponding to step S230, such as Figure 3In part 310, the interest weights corresponding to each of the N historical recommendation information for the target user 300 can be expressed as [w1, w2, w3, ... wN]; the hash value corresponding to each of the historical recommendation information is determined by the specific embodiment corresponding to step S220, such as Figure 3 The interest weights corresponding to the N historical recommendation information in part 320 can be expressed as [1100…11,1000…11,0100…11,…,0010…11].

[0125] Further, in step S240: based on the hash value and interest weight corresponding to each of the historical recommendation information, the interest vector of the target user is determined. Figure 4 The flowchart of a method for determining a user's interest vector according to an embodiment of the present disclosure is schematically shown.

[0126] refer to Figure 4 , the method shown in the embodiment of the figure includes:

[0127] Step S410, weighting the hash values ​​corresponding to each of the historical recommendation information according to the interest weight of each of the historical recommendation information to obtain the interest sub-vector of each of the historical recommendation information; and step S420, superimposing the interest sub-vectors of each of the historical recommendation information to obtain the interest vector of the target user.

[0128] In an exemplary embodiment, since the interest weight of each historical recommendation information can reflect the target user's preference for each historical recommendation information, and the above hash value can uniquely identify each historical recommendation information, the target user's preference for each historical recommendation information can be accurately located by weighting the interest weight with the hash value of the corresponding historical recommendation information.

[0129] For example, Figure 5 The flowchart of the method for determining the interest subvector of the historical recommendation information according to an embodiment of the present disclosure is schematically shown, which can be used as a specific implementation of step S410. Specifically, it is used to illustrate how to implement weighted processing when the hash value corresponding to each historical recommendation information includes the first data code and the second data code.

[0130] refer to Figure 5 , the method shown in the embodiment of the figure includes:

[0131] Step S510, when the data bit of the hash value is the first data code, the data bit is replaced by the first data code with the interest weight of each historical recommendation information; and, step S520, when the data bit of the hash value is the second data code, the data bit is replaced by the second data code with the opposite number of the interest weight of each historical recommendation information.

[0132] The following combination Figure 3 The specific implementation methods of step S510 and step S520 are explained. For the historical recommendation information with identification number Item1, its corresponding hash value is binary "1100...11", and at the same time, the interest weight of the target user 300 for the historical recommendation information is w1. The "1" in the hash value is used as the above-mentioned first data code, and the "0" therein is used as the above-mentioned second data code. That is to say, when the data bit of the hash value is "1", the data bit is replaced by "1" with the interest weight "w1"; when the data bit of the hash value is "0", the data bit is replaced by "0" with "-w1". Thereby, the interest sub-vector "w1,w1,-w1,-w1...,w1" corresponding to the historical recommendation information is obtained (reference Figure 3 330 of the .

[0133] Similarly, for the historical recommendation information with identification number Item2, after weighting the corresponding hash value with interest weight w2, the interest sub-vector "w2,-w2,-w2,-w2…,w2" corresponding to the historical recommendation information is obtained.

[0134] Continue to refer Figure 4 In step S420, the interest sub-vectors of the historical recommendation information are superimposed to obtain the interest vector of the target user. Figure 3 , after superimposing the interest sub-vectors of part 330, the interest vector of the target user 300 represented by part 340 is obtained [w1+w2-w3…-wN,w1-w2+w3…-wN,-w1-w2-w3…+Wn,…w1+w2+w3…+wN]. Exemplarily, the interest vector of the target user is [12,-23,19,-32,29,…30].

[0135] In an exemplary embodiment, Figure 6 The flowchart of the method for determining the user's interest vector according to another embodiment of the present disclosure is schematically shown. Specifically, another specific implementation of step S420 is provided in consideration of the time decay parameter.

[0136] refer to Figure 6 The method shown in the embodiment of the figure includes steps S610 to S630.

[0137] In step S610, the statistical period of each piece of historical recommendation information is determined.

[0138] In an exemplary embodiment, the closer the date of the browsing behavior data generated by the target user is to the date of the statistical analysis of the above interest vector, the more accurately the current interests and hobbies of the target user can be reflected; on the contrary, the older the date of the browsing behavior data is, the more likely the user's interests and hobbies have changed over time, and the current interests and hobbies of the target user can no longer be reflected accurately. It can be seen that the current interests and hobbies of the target user are affected by the timestamp of the browsing behavior data.

[0139] Therefore, a specific implementation of step S610 may be: determining the statistical period of each historical recommendation information according to the timestamp of the target user's browsing behavior data for each historical recommendation information.

[0140] In an exemplary embodiment, each historical recommended information is generally information popular during the corresponding time period of its listing. For example, during the XX basketball league, the information listed / popular is mostly basketball-related information, while during the XX football league, the information listed / popular is mostly football-related information. Therefore, the listing time of the historical recommended information also affects the interests and hobbies of the target user to a certain extent. It can be seen that the current interests and hobbies of the target user are affected by the listing time of the historical recommended information.

[0141] Therefore, a specific implementation of step S610 may also be: determining the statistical period of each piece of historical recommendation information according to the time when each piece of historical recommendation information was put on the shelves.

[0142] In step S620, the interest sub-vectors of each of the historical recommendation information are processed according to the time decay coefficient of the statistical period to obtain candidate interest sub-vectors of each of the historical recommendation information.

[0143] In an exemplary embodiment, reference Figure 7 , still Figure 3 The historical recommendation information of target user 300 is used as an example for explanation. In section 710, for each N historical recommendation information, suppose: the interest sub-vector corresponding to Item 1 is determined in the first statistical period, and its corresponding time decay coefficient is s (where s is a positive number less than 1); the interest sub-vector corresponding to Item 2 is determined in the second statistical period, and its corresponding time decay coefficient is s 2 Item3's interest sub-vector is determined in the third statistical period, and its corresponding time decay coefficient is s 3 ; ..., the interest word vector corresponding to ItemN is determined in the Nth (N is an integer greater than 3) statistical period, and its corresponding time decay coefficient is s N Among them, the first statistical period is the statistical period closest to the current moment, and the Nth statistical period is the statistical period farthest from the current moment.

[0144] In an exemplary embodiment, the interest sub-vectors of each historical recommendation information are processed according to the time decay coefficient of the statistical period to obtain the candidate interest sub-vectors of each historical recommendation information. Exemplarily, the time decay coefficient corresponding to Item1 is s, and the candidate interest word vector corresponding to Item1 is determined by multiplying s with the interest sub-vector corresponding to Item1 "s*w1,s*w1,-s*w1,-s*w1…,s*w1". Similarly, the candidate interest word vector corresponding to Item2 is "s 2 *w2,-s 2 *w2,-s 2 *w2,-s 2 *w2…,s 2 *w2”.

[0145] In step S630, the candidate interest sub-vectors of the historical recommendation information are superimposed to obtain the interest vector of the target user.

[0146] refer to Figure 7 The 720 part of the target user is superimposed with the candidate interest sub-vectors of N historical recommendation information to obtain the interest vector of the target user 300:

[0147] [s*w1+s 2 *w2-s 3 *w3…-s N *wN,s*w1-s 2 *w2+s 3 *w3…-s N *wN,-s*w1-s 2 *w2-s 3 *w3…+s N *wN,……s*w1+s 2 *w2+s 3 *w3…-s N *wN].

[0148] In an exemplary embodiment, after determining the interest vector of the target user, in step S240: determining recommendation information for the target user based on the interest vector. Figure 8 The following schematically shows a flow chart of an information recommendation method according to another embodiment of the present disclosure.

[0149] refer to Figure 8 The method shown in the embodiment of the figure includes steps S810 to S830.

[0150] In step S810, basic attribute data of the target user is obtained, attribute data of each of the historical recommendation information is obtained, and scene data of the display page is obtained.

[0151] This embodiment uses a machine learning model to predict the recommended information of the target user. In order to improve the prediction accuracy of the recommended information, in addition to the interest vector of the target user determined in the above embodiment, the basic attribute data of the target user (such as gender, age, place of residence, education level, etc.), the attribute data of each historical recommended information (such as content tags, display type and listing time, etc.), and the scene data of the current display page (such as contextual information of the recommended information, etc.) should also be obtained.

[0152] refer to Fig. 9 , a gradient boosting decision tree (GBDT) 920 is used to predict the recommended information. Specifically, the model input information 910 includes: the interest vector 912 of the target user, which can be: [0.9, 0.3, 1.2, 12.8…]; the basic attribute data 911 of the target user can be: male, 15-35 years old, Beijing (residence), master (education), etc.; the attribute data 913 of a certain historical recommendation information of the target user, including: James (content label), Lakers (content label), news (display type); the scene data 914 of the current display page, including: Feeds flow, first place, etc. In an exemplary embodiment, the model input information can also include other features 915, such as statistics, etc.

[0153] In step S820, the interest vector of the target user, the basic attribute data of the target user, the attribute data of each of the historical recommendation information, and the scene data of the display page are spliced ​​to obtain a splicing feature.

[0154] In an exemplary embodiment, the plurality of feature data may be concatenated by the gradient boosting tree model, or the plurality of feature data may be concatenated to obtain concatenated features before inputting the model, so that in step S830, the recommended information for the target user is predicted based on the concatenated features.

[0155] In a specific implementation of step S830: the splicing features are input into a prediction model trained offline, so that the prediction model determines the output of the model based on the splicing features to obtain a recommendation value for each of the historical recommendation information; and the recommendation information for the target user is determined based on the recommendation score.

[0156] In this embodiment, an offline training prediction model is adopted and real-time prediction is achieved through the trained prediction model. Specifically, the spliced ​​features obtained by splicing the above-mentioned multiple feature data are input into the offline trained prediction model, and the prediction model determines the output of the model based on the spliced ​​features to obtain the recommendation value of each historical recommendation information. In other words, a new recommendation value score is obtained for each historical recommendation. Further, these scores are sorted, and the topN' (where N' is less than the number N of the above-mentioned recommended historical information) is taken as the recommended information, and displayed to the above-mentioned target user.

[0157] It should be noted that, in the process of offline training the model, the method of determining the interest vector of each user in the training data is the same as the method of determining the interest vector of the target user described in the above embodiment. Therefore, the format of the interest vector of the target user in the real-time prediction process is the same as the format of the interest vector of each user in the offline training process, which is conducive to ensuring the prediction accuracy of the recommendation information.

[0158] In an exemplary embodiment, Fig.10 A specific implementation of step S830 is shown, referring to Fig.10 :

[0159] First, the prediction model trained offline is loaded into the engine 1020 of the information recommendation system; in response to the access request for the display page issued to the user terminal 1010 (step S101), a recommendation request to the above-mentioned information recommendation system is triggered. Further, the splicing features of the target user are obtained in real time through the engine 1020 of the above-mentioned recommendation system (step S102); then, the splicing features are input into the loaded prediction model to perform real-time prediction through the prediction model to obtain the recommended information for the target user (step S103). Then, in step S104: the recommended information is sent to the user terminal 1010, so that the above-mentioned target user can view the recommended information determined according to his interests and hobbies.

[0160] This technical solution adopts offline training prediction model and determines the recommended information by realizing real-time prediction through the trained prediction model, which is conducive to further improving the calculation efficiency of the recommended information and the timeliness of determining the recommended information for the target user. Furthermore, by obtaining the browsing behavior of the target user on the above recommended information, the target user's more accurate interests and hobbies can be captured, which is conducive to improving the recommendation accuracy of the next round of recommended information.

[0161] Those skilled in the art will appreciate that all or part of the steps to implement the above-mentioned embodiments are implemented as a computer program executed by a processor (including a CPU and a GPU). For example, the model training of the above-mentioned prediction model is implemented by a GPU; based on the trained prediction model, the CPU is used to implement information recommendation for the target user; or, based on the trained prediction model, the GPU is used to implement information recommendation for the target user, etc. When the computer program is executed by the CPU or GPU, the above-mentioned functions defined by the above-mentioned method provided in the present disclosure are executed. The program can be stored in a computer-readable storage medium, which can be a read-only memory, a disk or an optical disk, etc.

[0162] In addition, it should be noted that the above figures are only schematic illustrations of the processes included in the method according to the exemplary embodiment of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0163] The following introduces an embodiment of an information recommendation device of the present disclosure, which can be used to execute the above-mentioned information recommendation method of the present disclosure.

[0164] Fig.11 The structure diagram of the information recommendation device in the exemplary embodiment of the present disclosure is schematically shown. Fig.11 As shown, the above-mentioned information recommendation device 1100 includes: a data acquisition module 1101, a hash processing module 1102, an interest weight determination module 1103, and an information recommendation module 1104. Among them:

[0165] The data acquisition module 1101 is configured to: acquire the browsing behavior data of the target user for each historical recommendation information and acquire the identification data of each historical recommendation information;

[0166] The hash processing module 1102 is configured to: perform hash processing on the identification data of each piece of the historical recommendation information, so as to convert the identification data of each piece of the historical recommendation information into a hash value respectively;

[0167] The interest weight determination module 1103 is configured to: determine the interest weight of the target user for each of the historical recommendation information according to the browsing behavior data of the target user for each of the historical recommendation information;

[0168] The information recommendation module 1104 is configured to determine the interest vector of the target user based on the hash value and interest weight corresponding to each of the historical recommendation information, so as to determine the recommended information for the target user based on the interest vector.

[0169] In some embodiments of the present disclosure, based on the aforementioned scheme, the hash processing module 1102 is specifically configured to: perform hash processing on the identification number of each of the historical recommendation information, and one or more of the content label, display type and listing time.

[0170] In some embodiments of the present disclosure, based on the aforementioned scheme, the above-mentioned hash processing module 1102 is also specifically configured to: process the string type, integer type or floating point type identification data of each of the historical recommendation information through a hash function to obtain a hash value corresponding to the identification data of each of the historical recommendation information.

[0171] In some embodiments of the present disclosure, based on the above solution, the information recommendation module 1104 includes: a weighted processing unit and a superposition unit.

[0172] The weighted processing unit 11041 is configured to: perform weighted processing on the hash value corresponding to each of the historical recommendation information according to the interest weight of each of the historical recommendation information to obtain the interest sub-vector of each of the historical recommendation information; and

[0173] The superposition unit 11042 is configured to: superimpose the interest sub-vectors of the historical recommendation information to obtain the interest vector of the target user.

[0174] In some embodiments of the present disclosure, based on the above scheme, the hash value includes a first data code and a second data code; wherein:

[0175] The above-mentioned weighted processing unit 11041 is specifically configured as: when the data bit of the hash value is the first data code, the data bit is replaced by the first data code with the interest weight of each historical recommendation information; and, when the data bit of the hash value is the second data code, the data bit is replaced by the second data code with the opposite number of the interest weight of each historical recommendation information.

[0176] In some embodiments of the present disclosure, based on the above-mentioned solution, the superposition unit 11042 includes: a statistical period determination subunit, a candidate interest subvector determination subunit and an interest vector determination subunit. Among them:

[0177] The statistical period determination subunit is configured to: determine the statistical period of each of the historical recommendation information;

[0178] The candidate interest sub-vector determining sub-unit is configured to: process the interest sub-vectors of each of the historical recommendation information according to the time decay coefficient of the statistical period to obtain the candidate interest sub-vectors of each of the historical recommendation information; and

[0179] The interest vector determination subunit is configured to: superimpose the candidate interest subvectors of each of the historical recommendation information to obtain the interest vector of the target user.

[0180] In some embodiments of the present disclosure, based on the above-mentioned solution, the statistical period determination subunit is specifically configured as follows:

[0181] The statistical period of each historical recommendation information is determined according to the timestamp of the target user's browsing behavior data for each historical recommendation information; or, the statistical period of each historical recommendation information is determined according to the listing time of each historical recommendation information.

[0182] In some embodiments of the present disclosure, based on the above scheme, the information recommendation module 1104 includes: a recommended information determination unit 11043. The recommended information determination unit 11043 includes: a splicing subunit and a recommendation subunit.

[0183] The data acquisition module 1101 is further configured to: acquire basic attribute data of the target user, acquire attribute data of each of the historical recommendation information, and acquire scene data of the display page;

[0184] The above-mentioned splicing subunit is configured to: splice the interest vector of the target user, the basic attribute data of the target user, the attribute data of each of the historical recommendation information and the scene data of the display page to obtain a splicing feature; and,

[0185] The above-mentioned splicing subunit is configured to: predict the recommendation information for the target user according to the splicing features.

[0186] In some embodiments of the present disclosure, based on the above-mentioned solution, the above-mentioned splicing subunit is specifically configured as follows:

[0187] The splicing features are input into a prediction model trained offline, so that the prediction model determines the output of the model based on the splicing features to obtain a recommendation value for each of the historical recommendation information; and the recommendation information for the target user is determined according to the recommendation score.

[0188] In some embodiments of the present disclosure, based on the above-mentioned solution, the above-mentioned splicing subunit is specifically configured as follows:

[0189] The prediction model trained offline is loaded into the engine of the information recommendation system; in response to the access request to the display page, a recommendation request to the information recommendation system is triggered; the splicing features of the target user are obtained in real time through the engine; and the splicing features are input into the prediction model to perform real-time prediction through the prediction model to obtain recommendation information for the target user.

[0190] In some embodiments of the present disclosure, based on the aforementioned scheme, the data acquisition module 1101 is specifically configured to: obtain one or more of the viewing time, number of clicks, number of favorites, and number of recommendations of the target user for each historical recommendation information.

[0191] In some embodiments of the present disclosure, based on the aforementioned scheme, the above-mentioned interest weight determination module 1103 is specifically configured to: determine the interest weight of the target user for each of the historical recommendation information according to one or more of the viewing time, number of clicks, number of collections and number of recommendations of the target user for each of the historical recommendation information.

[0192] The specific details of each unit in the above information recommendation device are already in the appendix of the manual. Figures 2 to 10 The corresponding information recommendation method is described in detail, so it will not be repeated here.

[0193] Fig.12 A schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present disclosure is shown.

[0194] It should be noted that Fig.12 The computer system 1200 of 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.

[0195] like Fig.12 As shown, the computer system 1200 includes a processor 1201 (including a graphics processing unit (GPU), a central processing unit (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1202 or the program loaded from the storage part 1208 to the random access memory (RAM) 1203. Various programs and data required for system operation are also stored in RAM 1203. The processor (CPU or GPU) 1201, ROM 1202 and RAM 1203 are connected to each other through a bus 1204. An input / output (I / O) interface 1205 is also connected to the bus 1004.

[0196] The following components are connected to the I / O interface 1205: an input section 1206 including a keyboard, a mouse, etc.; an output section 1207 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a local area network (LAN) card, a modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to the I / O interface 1205 as needed. A removable medium 1211, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1210 as needed so that a computer program read therefrom is installed into the storage section 1208 as needed.

[0197] In particular, according to an embodiment of the present disclosure, the process described below with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes 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 such an embodiment, the computer program can be downloaded and installed from a network through the communication part 1209, and / or installed from a removable medium 1211. When the computer program is executed by the processor (CPU or GPU) 1201, various functions defined in the system of the present application are executed.

[0198] It should be noted that the computer-readable medium shown in the embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above 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.

[0199] More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), flash memory, 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 foregoing.

[0200] In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.

[0201] A 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 appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0202] 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 implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings.

[0203] For example, two boxes shown in succession may actually be executed substantially in parallel, and they may sometimes be executed in the reverse 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 boxes in the block diagram or flow chart, can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0204] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, and the units described may also be arranged in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0205] As another aspect, the present application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiment.

[0206] For example, the electronic device can implement Figure 2 The information recommendation method shown in , specifically: step S210, obtaining the browsing behavior data of the target user for each historical recommendation information and obtaining the identification data of each historical recommendation information; step S220, hashing the identification data of each historical recommendation information to convert the identification data of each historical recommendation information into a hash value respectively; step S230, determining the interest weight of the target user for each historical recommendation information according to the browsing behavior data of the target user for each historical recommendation information; and step S240, determining the interest vector of the target user based on the hash value and interest weight corresponding to each historical recommendation information, so as to determine the recommended information for the target user based on the interest vector.

[0207] Exemplarily, the electronic device may also implement the following Figure 3 As Fig.10 The information recommendation method shown in any of the figures.

[0208] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0209] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0210] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0211] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. An information recommendation method, characterized in that: The method comprises: Obtaining browsing behavior data of target users for each historical recommendation information and obtaining identification data of each historical recommendation information; Performing hash processing on the identification data of each piece of the historical recommendation information to convert the identification data of each piece of the historical recommendation information into a hash value respectively; Determining the interest weight of the target user for each of the historical recommendation information according to the browsing behavior data of the target user for each of the historical recommendation information; Performing weighted processing on the hash value corresponding to each piece of the historical recommendation information according to the interest weight of each piece of the historical recommendation information to obtain an interest sub-vector of each piece of the historical recommendation information; Determine the statistical period of each of the historical recommendation information according to the timestamp of the browsing behavior data of the target user for each of the historical recommendation information, or according to the time when each of the historical recommendation information was put on the shelves; Processing the interest sub-vectors of each of the historical recommendation information according to the time attenuation coefficient of the statistical period to obtain candidate interest sub-vectors of each of the historical recommendation information; Superimposing the candidate interest sub-vectors of the historical recommendation information to obtain the interest vector of the target user; Obtaining basic attribute data of the target user, obtaining attribute data of each of the historical recommendation information, and obtaining scene data of the display page; Constructing a gradient boosting decision tree based on the interest vector of the target user, the basic attribute data of the target user, the attribute data of each of the historical recommendation information, and the scene data of the display page; The recommendation information for the target user is predicted according to the gradient boosting decision tree.

2. The information recommendation method according to claim 1, characterized in that: The performing hash processing on the identification data of each of the historical recommendation information includes: Hash processing is performed on the identification number of each of the historical recommendation information, and one or more of the content tag, display type and listing time.

3. The information recommendation method according to claim 1, characterized in that: The performing hash processing on the identification data of each piece of the historical recommendation information to convert the identification data of each piece of the historical recommendation information into a hash value respectively includes: The character string, integer or floating point identification data of each piece of the historical recommendation information is processed by a hash function to obtain a hash value corresponding to the identification data of each piece of the historical recommendation information.

4. The information recommendation method according to claim 1, characterized in that: The hash value includes a first data code and a second data code; wherein: The weighting process of the hash value corresponding to each piece of the historical recommendation information according to the interest weight of each piece of the historical recommendation information includes: When the data bit of the hash value is the first data code, the data bit is replaced by the first data code with the interest weight of each piece of the historical recommendation information; When the data bit of the hash value is the second data code, the data bit is replaced by the opposite number of the interest weight of each piece of the historical recommendation information from the second data code.

5. The information recommendation method according to claim 1, characterized in that: The predicting the recommendation information for the target user according to the gradient boosting decision tree includes: Inputting the gradient boosting decision tree into the offline trained prediction model, so that the prediction model determines the output of the model based on the gradient boosting decision tree to obtain the recommendation value of each of the historical recommendation information; Determine the recommended information for the target user according to the recommendation score.

6. The information recommendation method according to claim 1, characterized in that: The predicting the recommendation information for the target user according to the gradient boosting decision tree includes: Load the offline trained prediction model into the engine of the information recommendation system; In response to a request for access to the display page, triggering a recommendation request to the information recommendation system; Acquire the gradient boosting decision tree in real time through the engine; The gradient boosting decision tree is input into the prediction model to perform real-time prediction through the prediction model to obtain recommendation information for the target user.

7. The information recommendation method according to any one of claims 1 to 3, characterized in that: The acquiring of the target user's browsing behavior data for each historical recommendation information includes: Obtain one or more of the viewing time, click count, collection count, and recommendation count of each historical recommendation information by the target user.

8. The information recommendation method according to any one of claims 1 to 3, characterized in that: The determining, based on the browsing behavior data of the target user for each of the historical recommendation information, the interest weight of the target user for each of the historical recommendation information includes: The interest weight of the target user for each of the historical recommendation information is determined according to one or more of the viewing time, click times, collection times, and recommendation times of each of the historical recommendation information by the target user.

9. An information recommendation device, characterized in that: The device comprises: The data acquisition module is configured to: acquire the browsing behavior data of the target user for each historical recommendation information and acquire the identification data of each historical recommendation information; A hash processing module is configured to: perform hash processing on the identification data of each piece of the historical recommendation information to convert the identification data of each piece of the historical recommendation information into a hash value respectively; The interest weight determination module is configured to: determine the interest weight of the target user for each of the historical recommendation information according to the browsing behavior data of the target user for each of the historical recommendation information; The information recommendation module is configured to: perform weighted processing on the hash value corresponding to each of the historical recommendation information according to the interest weight of each of the historical recommendation information to obtain the interest sub-vector of each of the historical recommendation information; determine the statistical period in which each of the historical recommendation information belongs according to the timestamp of the browsing behavior data of the target user for each of the historical recommendation information, or according to the shelf time of each of the historical recommendation information; process the interest sub-vector of each of the historical recommendation information according to the time attenuation coefficient of the statistical period to obtain the candidate interest sub-vector of each of the historical recommendation information; superimpose the candidate interest sub-vectors of each of the historical recommendation information to obtain the interest vector of the target user; obtain the basic attribute data of the target user, obtain the attribute data of each of the historical recommendation information, and obtain the scene data of the display page; construct a gradient boosting decision tree based on the interest vector of the target user, the basic attribute data of the target user, the attribute data of each of the historical recommendation information, and the scene data of the display page; predict the recommended information for the target user according to the gradient boosting decision tree.

10. A computer storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the information recommendation method according to any one of claims 1 to 8 is implemented.

11. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device, used to store one or more programs, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the information recommendation method as described in any one of claims 1 to 8.

Images