A content pushing method and device and a storage medium

Abstract

Embodiments of the present application disclose a content pushing method, a processing device can acquire content behavior data of a target user on a target content in a target time interval, and respectively count a plurality of statistical features related to the target content of the target user based on different statistical periods, the plurality of statistical features constitute multi-period statistical features of the target user on the target content, the multi-period statistical features can reflect the change of intention of the target user on the target content over time, and thus can more accurately depict the preference of the target user. Finally, by using machine learning and other technologies in artificial intelligence technology, the association parameter of the target user and the target content can be determined according to the multi-period statistical features, and whether to push the target content to the target user is determined based on the association parameter, so that the processing device can push the target content to the user with high intention on the target content in the near future, and improve the acceptance probability of the user on the target content and the content recommendation efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, and in particular to a content pushing method and device and a storage medium. BACKGROUND

[0002] A content platform can push content that a user may like to the user through identification and mining of the user, so as to achieve the purpose of expanding the number of users in the platform and increasing the user stickiness of the users in the platform.

[0003] At present, the mining of the user is mainly based on a user portrait of the user. The user portrait of the user identifying the user's preferences and the like can be used as a basis for matching the content that the user may like.

[0004] However, the user portrait is not suitable in some content recommendation scenarios, and the accuracy of the user portrait is difficult to guarantee. When the user portrait is not accurate, incorrect content is determined, which affects the effect of content recommendation. SUMMARY

[0005] To solve the above technical problems, the embodiments of the present application provide a content pushing method. On the one hand, the association between the user and the content is introduced as a basis for determining whether to push or not. On the other hand, the change trend of the user's intention is considered, which is conducive to improving the accuracy of content pushing.

[0006] The embodiments of the present application disclose the following technical solutions:

[0007] In a first aspect, the embodiments of the present application provide a content pushing method, which comprises:

[0008] obtaining content behavior data of a target user corresponding to target content in a target time interval;

[0009] performing data statistical processing on the content behavior data based on different statistical periods to obtain multi-period statistical features of the target user and the target content, the multi-period statistical features comprising first statistical features corresponding to a first statistical period and second statistical features corresponding to a second statistical period, the first statistical period and the second statistical period having different period lengths, and the period lengths of the first statistical period and the second statistical period being less than or equal to a time length identified by the target time interval;

[0010] determining an association parameter of the target user and the target content according to the multi-period statistical features;

[0011] determining whether to push the target content to the target user based on the association parameter.

[0012] In a second aspect, the embodiments of the present application provide a content pushing device, the device comprising an obtaining unit, a first determining unit, a second determining unit and a pushing unit:

[0013] The obtaining unit is configured to obtain content behavior data of a target user on a target content in a target time interval.

[0014] The first determining unit is configured to perform data statistical processing on the content behavior data based on different statistical periods to obtain multi-period statistical features of the target user and the target content, the multi-period statistical features comprising a first statistical feature corresponding to a first statistical period and a second statistical feature corresponding to a second statistical period, the first statistical period and the second statistical period having different period lengths, and the period lengths of the first statistical period and the second statistical period both being less than or equal to a time length identified by the target time interval.

[0015] The second determining unit is configured to determine an association parameter of the target user and the target content according to the multi-period statistical features.

[0016] The pushing unit is configured to determine whether to push the target content to the target user based on the association parameter.

[0017] In a third aspect, the embodiments of the present application provide a computer device, the device comprising a processor and a memory:

[0018] The memory is configured to store program code and transmit the program code to the processor.

[0019] The processor is configured to execute the content pushing method in the first aspect according to instructions in the program code.

[0020] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, the computer readable storage medium being configured to store a computer program, the computer program being configured to execute the content pushing method in the first aspect.

[0021] It can be seen from the technical solution that in the content recommendation scenario, content behavior data of a target user corresponding to target content in a target time interval is obtained, the content behavior data can identify the behavior of the target user related to the target content in the target time interval, and the association between the target user and the target content can be reflected through the content behavior data. According to different statistical periods, a multi-period statistical feature is determined from the content behavior data, and the multi-period statistical feature at least includes a first statistical feature determined based on a first statistical period and a second statistical feature determined based on a second statistical period. Since the first statistical period and the second statistical period have different period lengths, the association between the target user and the target content in the time sequence is reflected through the change of the content behavior data in different statistical periods, that is, the multi-period statistical feature can express the time sequence change trend of the intention of the target user to the target content. The multi-period statistical feature with rich expression ability plays a more accurate role in describing the preferences of the target user, so that when the association parameter determined by the multi-period statistical feature is used to measure whether to push the target content to the target user, if it is determined to push, the possibility of the target user accepting the target content is higher, and the content recommendation efficiency is improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0023] Figure 1 A schematic diagram of a content push method in an actual application scenario provided by an embodiment of the present application;

[0024] Figure 2 A flowchart of a content push method provided by an embodiment of the present application;

[0025] Figure 3 A schematic diagram of a content push method in an actual application scenario provided by an embodiment of the present application;

[0026] Figure 4 A structural block diagram of a content push device provided by an embodiment of the present application;

[0027] Figure 5 A structural diagram of a computer device provided by an embodiment of the present application;

[0028] Figure 6 A structural diagram of a server provided by an embodiment of the present application;

[0029] Figure 7A flowchart of a content pushing method in an actual application scenario provided by an embodiment of the present application. DETAILED DESCRIPTION

[0030] The embodiments of the present application are described below with reference to the accompanying drawings.

[0031] Pushing various related contents to users is one of the common means for product promotion, and whether reasonable content pushing can directly affect the promotion effect of the product. In related technologies, in order to recommend the content with higher acceptance degree to the user, the user portrait corresponding to the user is usually generated based on the information of the user itself, such as height, gender, age, birthplace, etc., and the pushing content that the user may be interested in is determined through the user portrait. However, the judgment basis of this method is too single, and it is difficult to meet the rich and diverse content pushing demand. At the same time, since the user information is relatively fixed, it is difficult to reflect the change of the user's interest and hobby over time, so it is impossible to achieve more accurate content pushing.

[0032] In order to solve the above technical problems, the embodiments of the present application provide a content pushing method, and a processing device can determine the multi-period statistical characteristics of target content corresponding to a target user by counting content behavior data in different time periods. The multi-period statistical characteristics can reflect the change of the association between the target user and the target content over time, so as to introduce the association between the user and the content as the judgment basis for pushing or not, and consider the change trend of the user's intention, which is beneficial to improve the accuracy of content pushing.

[0033] It can be understood that the method can be applied to a processing device, and the processing device is a processing device with a content pushing function, for example, it can be a terminal device or a server with a content pushing function. The method is independently executed by the terminal device or the server, and can also be applied to a network scenario in which the terminal device and the server communicate, and the terminal device and the server cooperate to run. The terminal device can be a mobile phone, a desktop computer, a personal digital assistant (PDA), a tablet computer, etc. The server can be understood as an application server, or a Web server. In actual deployment, the server can be a standalone physical server, or a server cluster or distributed system composed of multiple physical servers. The terminal and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the present application.

[0034] In addition, the present application also relates to artificial intelligence technology (Artificial Intelligence, AI). Artificial intelligence is a theory, method, technology and application system for using a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive an environment, acquire knowledge and use the knowledge to obtain optimal results. In other words, artificial intelligence is a comprehensive technology of computer science, which attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that the machines have the functions of perception, reasoning and decision-making.

[0035] Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, both hardware and software technologies. Artificial intelligence basic technologies generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc. Among them, the technical solutions of the present application mainly involve machine learning technology.

[0036] Machine learning (Machine Learning, ML) is a multi-field interdisciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, etc. It is a subject that studies how a computer simulates or implements human learning behavior to acquire new knowledge or skills, and reorganizes existing knowledge structure to continuously improve its own performance. Machine learning is the core of artificial intelligence and the fundamental approach to making computers intelligent, and its applications are widespread in various fields of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and example-based learning technologies. For example, the embodiments of the present application can use machine learning technology to train a classifier model, so that the classifier model can accurately determine the correlation parameters.

[0037] In order to facilitate understanding of the technical solutions of the present application, the content pushing method provided by the embodiments of the present application will be introduced in combination with an actual application scenario.

[0038] Referring to Figure 1 , Figure 1A schematic diagram of a content pushing method provided by an embodiment of the present application in an actual application scenario is shown in FIG. 1. In the actual application scenario, the processing device is a server 101 having a content pushing function. The server 101 stores content behavior data between a user and content. The content behavior data refers to data capable of reflecting the association between the user and the content. For example, when the content is an advertisement of a product, the content behavior data can be the number of times of pushing the advertisement to the user, and the like. These data can reflect the intention of the user to the advertisement.

[0039] Before pushing a target content to a target user, the server 101 can first acquire content behavior data of the target user corresponding to the target content in a target time interval, for example, the number of times of pushing the target content to the target user in one year, and the like. The server 101 can perform data statistical processing on the content behavior data based on different statistical periods to obtain multi-period statistical features of the target user and the target content. In order to reflect the change of the association between the target user and the target content over time, the server 101 can perform statistics on the content behavior data in multiple different statistical periods, for example, a first statistical period and a second statistical period. The first statistical period and the second statistical period have different period lengths. In the actual application scenario, when the target time interval is one year, the first statistical period can be one month, and the second statistical period can be three months, as long as the period length is less than or equal to the target time.

[0040] Thus, through the first statistical feature and the second statistical feature in the multi-period statistical features, the server 101 can know the change of the association between the target user and the target content in the last three months to the last one month, and further, through the multi-period statistical features, the server 101 can more accurately depict the preference of the target user. The server 101 can determine an association parameter of the target user and the target content according to the multi-period statistical features. The association parameter can reflect the acceptance of the target user to the target content at the current time to a certain extent. For example, if the content behavior data is the number of times of pushing, the proportion of the number of times of pushing in the first statistical feature in the number of times of pushing in the second statistical feature is high. It is indicated that the number of times of pushing in the last two months is less than the number of times of pushing in the last one month, that is, the number of times of pushing is increasing, and thus, to a certain extent, it can be indicated that the acceptance of the target user to the target content at the current time can be high, and the determined association parameter can be high.

[0041] The server 101 can determine whether to push the target content to the target user based on the correlation parameter, for example, if the correlation parameter is higher than a certain threshold, the target content is pushed to the target user, and if it is not higher than the threshold, the pushing is not performed, so that the server 101 can push the target content to the user with higher acceptance based on the perspective of the target content, and improve the pushing efficiency and effectiveness; based on the perspective of the target user, the server 101 can push the content with higher acceptance to the target user, improve the experience of the target user being pushed, and finally improve the rationality of the entire content pushing process.

[0042] Next, a content pushing method provided by the embodiment of the application will be introduced in combination with the drawings.

[0043] Referring to Figure 2 , Figure 2 A flowchart of a content pushing method provided by the embodiment of the application is shown in FIG. 2, and the method comprises the following steps.

[0044] S201: Obtain content behavior data of a target user corresponding to target content in a target time interval.

[0045] The content behavior data refers to data capable of reflecting the behavior of the target user related to the target content in the target time interval, and the correlation between the target user and the target content can be reflected through the content behavior data. The target time interval can be a time interval of any length, the target user can be any user, and the target content can be content related to any product. For example, the content behavior data can be the number of times of pushing an advertisement of a product to a user in a target time interval.

[0046] S202: Perform data statistical processing based on different statistical periods on the content behavior data to obtain multi-period statistical features of the target user related to the target content.

[0047] In order to perform more accurate and effective content pushing, the processing device first needs to accurately judge the preferences of the user, so as to push the content interested and intended by the user to the user. It can be understood that the intention degree of the user to the content is not fixed, but changes continuously along with the time sequence. For example, a user may need to purchase product A three months ago, so the intention to the product A related advertisement is higher, and the number of accepting such advertisement pushing is more; in the past three months, the user has purchased product A, so the intention to the product A related advertisement is reduced, and if the user is pushed with the advertisement of the product again, the acceptance of the user to the advertisement may be lower.

[0048] Meanwhile, content behavior data corresponding to the content of the product usually has a highly sparse characteristic, and the interaction between the target user and the target content can be separated by months, so if the content is directly pushed according to the content behavior data in the target time interval, the processing device can not accurately determine the intention degree of the target user to the target content.

[0049] Based on this, after obtaining the content behavior data of the target content corresponding to the target time interval, in order to more accurately depict the preferences of the target user to the target content, the processing device can analyze the association relationship between the target user and the target content based on the time sequence, so as to determine the change of the intention of the target user to the target content through the association relationship changing with the time sequence. In the embodiments of the present application, the processing device can first set a first statistical period and a second statistical period, the first statistical period and the second statistical period have different period lengths, and in order to improve the effectiveness of the statistics, the period lengths of the first statistical period and the second statistical period are less than or equal to the length of time identified by the target time interval.

[0050] The processing device can combine the above-mentioned multiple statistical periods to perform data statistical processing on the content behavior data to obtain a multi-period statistical feature related to the target user and the target content, wherein the data statistical processing refers to statistical processing of data according to a certain statistical period, and the multi-period statistical feature is used to identify the characteristics of the content behavior data in multiple statistical periods. In the embodiments of the present application, the multi-period statistical feature includes a first statistical feature corresponding to the first statistical period and a second statistical feature corresponding to the second statistical period. Since the first statistical period and the second statistical period have different period lengths, the association relationship between the target user and the target content in different time periods can be reflected through the first statistical feature and the second statistical feature, so that the change of the association relationship over time can be reflected through the multi-period statistical feature, and the intention change of the target user to the target content can be reflected. Since the intention of the target user to the target content changes over time, the processing device can more vividly and accurately depict the user preferences through the multi-period statistical feature, which is beneficial to subsequent content pushing to the target user.

[0051] For example, in a possible implementation, the target time interval can be 1 year, the first statistical period can be 3 months, the second statistical period can be 2 weeks, and the content behavior data can be the number of times of pushing the target content to the target user. The first statistical feature can be 20 times in 3 months, and the second statistical feature can be 0 times in 2 weeks, so that the intention degree of the target user to the target content in the recent period can be determined to be lower and lower through the multi-period statistical feature.

[0052] S203: Determine the correlation parameters between target users and target content based on multi-period statistical characteristics.

[0053] In order to determine which users to push content to more reasonably, the processing device can determine the correlation parameters corresponding to multiple users based on specific judgment criteria. These correlation parameters can reflect the user's acceptance of the content, thereby enabling the processing device to select users with higher acceptance for content push.

[0054] As mentioned above, through this multi-period statistical feature, the processing device can determine the changing trend of the target user's intention for the target content over time. The target user's intention for the target content can, to a certain extent, reflect the target user's acceptance of the target content. Therefore, in this embodiment, the processing device can reasonably determine the correlation parameters between the target user and the target content based on the multi-period statistical feature. These correlation parameters can reflect the target user's acceptance of the target content. For example, continuing the above example, if the number of times the target content is pushed to the target user within two weeks is much less than the number of times it is pushed within three months, it indicates that the target user's recent intention for the target content has decreased, and the determined correlation parameter may be lower. If the number of times the target content is pushed to the target user within two weeks is close to the number of times it is pushed within three months, it indicates that the push of the target content to the target user mainly targets the last two weeks. In this case, the target user's intention for the target content is higher, and the determined correlation parameter may be higher.

[0055] S204: Determine whether to push target content to the target user based on the associated parameters.

[0056] By using correlation parameters, the processing device can determine the target user's acceptance level of the target content. Based on these correlation parameters, the processing device can push content to the target user only when the target user's acceptance level is high. From the user's perspective, the processing device can push more content that the user may be interested in, improving the user's content push experience. From the content's perspective, the processing device can push content to users with high acceptance levels, improving the efficiency and effectiveness of content push, achieving a win-win situation for both content and users.

[0057] It can be seen from the technical solution that, in the content recommendation scenario, due to the different cycle lengths of the first statistical period and the second statistical period, the changes of the content behavior data in different statistical periods reflect the changes in the time sequence of the association between the target user and the target content, that is, the multi-period statistical feature can express the time sequence change trend of the target user's intention to the target content. The multi-period statistical feature with rich expression ability plays a more accurate role in describing the target user's preference, so that when the association parameter determined by the multi-period statistical feature is used to measure whether to push the target content to the target user, if it is determined to push, the target user is more likely to accept the target content, thereby improving the content recommendation efficiency.

[0058] In order to further improve the accuracy of the association parameter, the processing device can also perform more detailed analysis on the information used to determine the association parameter. Next, more detailed schemes will be described from the perspectives of target content and content behavior data.

[0059] (1) Target content angle

[0060] It can be understood that in real life, the content pushed to the user usually includes multiple push types, which can be divided according to the push purpose of the content pushed to the user. Among them, when the content is pushed to the user with different push purposes, the specific content information and the content push stage of the user may be different, which can reflect the interaction degree between the user and the product corresponding to the content.

[0061] For example, when the product is an insurance product, the content can be an advertisement related to the insurance product, wherein the push type can include marketing, verification and member interaction, the push purpose of the marketing advertisement is to let more users know about the insurance product, the verification advertisement can be pushed when the user needs to verify the insurance product, and the push purpose can be to introduce more information related to the insurance product to the user when the user is already interested in the insurance product; the member interaction advertisement refers to the advertisement pushed to the user when the user interacts with the insurance product, and the push purpose can be to introduce updated content related to the insurance product to the user. As can be seen, in the content push stages corresponding to the marketing advertisement, the verification advertisement and the member interaction advertisement, the interaction degree between the user and the insurance product is continuously improved.

[0062] Since the degree of interaction between the user and the product can be different, if different types of target content can be classified and analyzed, the processing device can obtain the association between the target user in different content pushing stages and the target content to some extent, thereby enriching the information content included in the multi-period statistical data and further improving the determination accuracy of the association parameter. In a possible implementation, the processing device can divide the target content into types based on the pushing type, and obtain multiple pushing type dimensions of the target content. The pushing type dimensions can be used for classified statistics of the content behavior data, for example, the multiple pushing type dimensions can include marketing type dimensions, verification type dimensions, and member interaction type dimensions, and the like. When determining the multi-period statistical features, the processing device can perform data statistical processing on the content behavior data based on different statistical periods, and obtain the multi-period statistical features of the target user and the target content respectively associated under the multiple pushing type dimensions, so as to reflect the association between different types of target content and the target user for the same product, and further reflect the intention degree of the target user to the target content when in different content pushing stages.

[0063] It can be understood that, in addition to dividing the pushing type according to the pushing purpose, the processing device can also determine the pushing type based on other manners, so as to obtain other information related to the intention degree of the target user by determining the association between different types of target content and the target user, which is not limited here. Based on the association between the target content of different pushing types and the target user, the processing device can more accurately depict the preferences of the target user, and thus can more accurately push the content.

[0064] (2) Content behavior data perspective

[0065] Since the association mode between the user and the content can include multiple modes, the content included in the content behavior data also has certain diversity. For example, in a possible implementation, the content behavior data can include first data of the target user and / or second data of the target user interacting with the pushed target content. The first data can reflect the association between the target user and the target content from the perspective of content pushing, and the second data can reflect the association between the target user and the target content from the perspective of user and content interaction, for example, the second data can be the click rate, conversion rate, and the like of the target user to the target content. Thus, by the first data and the second data, the processing device can determine the association between the target user and the target content from different perspectives, further enriching the content that can be expressed by the content behavior data, so that the multi-period statistical features determined based on the content behavior data can reflect the intention change of the target user to the target content from multiple perspectives, and ultimately more accurate association parameters can be determined.

[0066] Besides increasing the granularity of data analysis, expanding the amount of data used for analysis is also a way to improve the accuracy of the final results. It's understandable that if a user has a high level of interest in content related to a particular product, they are likely to also have a high level of interest in content related to related products. For example, if a user has a high level of interest in content related to insurance products, they are likely to be interested in financial products that are highly related to insurance. Based on this, processing equipment can combine the content of related products to the target content to further analyze the target user's level of interest in the target content, thereby improving the richness of data representation.

[0067] In one possible implementation, the processing device can first determine a first product domain corresponding to the target content, and then obtain first behavioral data of the target user corresponding to the target content within a target time interval from the first product domain. This first behavioral data is used to reflect the association between the target user and the target content within the target time interval. Subsequently, the processing device can determine associated products that belong to a second product domain. The processing device can then obtain second behavioral data of the target user corresponding to the associated content within the target time interval from the second product domain. This second behavioral data is used to reflect the association between the target user and the associated content within the target time interval, and the associated content is content related to the associated product. The processing device can determine content behavioral data based on the first and second behavioral data. Thus, this content behavioral data can not only reflect the association between the target user and the target content, but also further reflect the association between the target user and related content in related domains. Therefore, based on the association with related content, it can further reflect the target user's intention towards the target content, enriching the expression and content of the data.

[0068] Understandably, to achieve the aforementioned technical content, associated products only need to be related to the product corresponding to the target content; there is no specific limitation on the product field to which the product belongs. The second product field can be the same as the first product field, or it can be a different product field. Here, associated products refer to multiple products that a user might be simultaneously interested in, without limiting the specific form of the products. For example, to protect the returns of financial products, many users will choose to purchase insurance for those financial products. Therefore, financial products and insurance products are products that users might be simultaneously interested in, thus constituting associated products.

[0069] As mentioned above, the content data of the related product can expand the data amount, enrich the data expression, and make the processing device more accurate in judging the user's intention. In some possible cases, the first behavior data obtained by the processing device from the first product field already has sufficient data amount, and the processing device can determine the multi-period statistical feature based on the first behavior data, without obtaining the second behavior data, so as to control the data amount processed by the processing device within a reasonable range and improve the processing efficiency. In other possible cases, the first behavior data obtained by the processing device from the first product field is low, and the processing device can combine the second behavior data obtained from the second product field to accurately judge the intention of the target user based on sufficient data amount.

[0070] Therefore, in order to analyze the intention of the target user with a reasonable data amount, the processing device can determine whether to obtain the second behavior data according to the data amount of the first behavior data. In a possible implementation, the processing device can determine a preset data amount, which is the data amount required to accurately analyze the intention degree of the target user. The processing device can determine whether the data amount of the first behavior data meets the preset data amount. If yes, the first behavior data is determined as the content behavior data, and the second behavior data does not need to be obtained. If no, the step of determining the associated product associated with the product is performed, so that the second behavior data is obtained from the second product field to expand the data amount.

[0071] Meanwhile, since the application can use the behavior data of the associated product and the associated content to determine the associated parameter, for some content that the user has not associated, even if the processing device cannot obtain the corresponding content behavior data between the user and the content, the processing device can also obtain other related behavior data to more accurately determine the intention of the user, thereby properly solving the problem of "cold start" and avoiding the situation that the user cannot be accurately pushed due to lack of data.

[0072] In addition, in order to more accurately determine the associated parameter between the target user and the target content, in addition to the multi-period statistical feature, the processing device can integrate more features to enrich the data dimension. For example, in a possible implementation, the processing device can not only reflect the association between the user and the product through the multi-period statistical feature, but also introduce user attribute features and product attribute features. The user attribute features can reflect the attribute characteristics of the user himself, and the product attribute features can reflect the attribute characteristics of the product itself. Therefore, based on the above information, the processing device can more accurately judge the intention degree of the user from the three aspects of the user, the product, and the association between the user and the content.

[0073] Firstly, the processing device can determine the product attribute feature of the product corresponding to the target content. The product attribute can be divided into numerical attribute and non-numerical attribute. The numerical attribute can include price, weight, quantity, etc. The non-numerical attribute can include, for example, product category, place of origin, etc. The processing device can convert the non-numerical attribute into a numerical feature that can be calculated, so as to determine the product attribute feature. For example, the product category can be represented by a specific number. When there are 10 categories of products, the processing device can number the 10 categories as 1-10, and convert them into a one-hot vector for feature calculation. If a product belongs to the 3rd category of the 10 categories, the product attribute feature corresponding to the product can be [0, 0, 1, 0, 0, 0, 0, 0, 0, 0].

[0074] Then, the processing device can determine the user attribute feature of the target user according to the user information of the target user. The user data feature can include gender, age, region, etc. The processing device can represent the gender of the user by 0 and 1, and represent the age of the user by age range, for example, the age can be divided into 7 age ranges, each age is marked by an integer 1-7, and the division method is 0-18, 18-25, 25-30, 30-40, 50-60 and above 60. For the regional information, the processing device can number all the regions and use a one-hot sparse vector to represent the regional information.

[0075] The processing device can determine the classification fusion feature according to the product attribute feature, the user attribute feature and the multi-period statistical feature. The classification fusion feature can reflect the intention degree of the target user to the target product from the above three aspects. The processing device can determine the association parameter of the target user and the target content according to the classification fusion feature, so that the association parameter can more accurately depict the preferences of the target user through more rich data dimensions. Of course, in addition to the above three features, the processing device can also fuse more dimensional features, which are not limited here.

[0076] As mentioned above, through the multi-period statistical feature related to the target user and the target content, the change of the intention degree of the target user to the target content over time sequence can be reflected. The user characteristics of a user can often be analyzed from the things he is interested in. For example, if a user is interested in the content related to fitness equipment and outdoor sports equipment, this user is probably a sports-loving user. The processing device can subsequently push more content related to sports products to this user. As can be seen, through the multi-period statistical feature, in addition to being able to judge whether to push content to the user, the user's own preference characteristics can also be depicted.

[0077] In a possible implementation, in order to obtain a more rich and stereoscopic user portrait, the processing device can obtain total content behavior data of the target user corresponding to different contents in a target time interval, the different contents including the target content, the total content behavior data including content behavior data corresponding to each content. The processing device can perform data statistical processing based on different statistical periods on the total content behavior data to obtain multi-period statistical features of the target user related to different contents, determine a user interest portrait of the target user according to the multi-period statistical features related to different contents, and the user interest portrait is used to reflect the interest orientation of the user to the content. Since the multi-period statistical features can reflect the change of the user intention degree along the time sequence, the user interest portrait determined based on the multi-period statistical features can not only reflect the interest of the user in multiple dimensions, but also can dynamically analyze the user interest, and has a more flexible and accurate information reflecting effect compared with some fixed user attributes such as age, gender and the like. Therefore, based on the user interest portrait, the processing device can mine users with higher interest degree in the related content from the user group when the user needs to be analyzed.

[0078] In the above process, the processing device can adopt various ways to determine the association parameter. In a possible implementation, the processing device can determine the association parameter of the target user and the target content according to the multi-period statistical features through a classifier model, for example, an XGBoost model. The classifier model can be trained according to the following manner.

[0079] Firstly, the processing device can determine a training sample according to the target content, the training sample including historical content behavior data of a sample user and a sample label used to identify whether the historical user has an association with the target content, the historical content behavior data being content behavior data that has been used to determine the association between the target content in a historical time. The processing device can train an initial classifier according to the training sample to obtain the classifier model. In this training process, the processing device can determine multi-period statistical features corresponding to the target content based on the historical content behavior data through various statistical periods, and then learn the relationship between the multi-period statistical features and the association parameter, so as to learn how to accurately determine the corresponding association parameter through the multi-period statistical features.

[0080] In addition to the above classifier model, the processing device can also use a logistic regression model, a decision tree model, a neural network model, a naive Bayes model and the like to determine the association parameter, which is not limited here.

[0081] In order to obtain more accurate association parameters through the classifier model, in a possible implementation, the number of the classifier models can be multiple. When determining the association parameters through the classifier model, the processing device can determine the pending association parameters of the target user and the target content through multiple classifier models according to the multi-period statistical characteristics, and then determine the association parameters of the target user and the target content according to the pending association parameters determined respectively, so that the processing device can comprehensively judge the association parameters by combining the output results of multiple classifier models, and avoid the problem that the output result is inaccurate due to small probability anomaly of a small number of classifier models. For example, the processing device can take the average of multiple pending association parameters as the finally determined association parameters.

[0082] Taking the XGBoost model as an example, the XGBoost model can include multiple base classifiers, and each base classifier corresponds to a sub-model. When training, the processing device can fuse the results of multiple base classifiers in the XGBoost, and consider the complexity of each sub-model, and optimize the model complexity and the output result at the same time, and the objective function is:

[0083]

[0084] wherein, is the loss function between the expected output and the predicted output, Ω(f k ) is the model complexity of the kth sub-classifier, and w is the sub-classifier parameter to be learned. The processing device can perform multiple rounds of training optimization on the objective function, and the loss function expression of the tth time is:

[0085]

[0086] The processing device can perform second-order Taylor expansion on the above formula, g is the first-order derivative, and h is the second-order derivative:

[0087]

[0088] In use, the processing device can determine i pending association parameters through i sub-models in the XGBoost, and then determine the association parameters based on the i pending association parameters through the XGBoost.

[0089] In addition to the above, the multiple classifier models can also be different models, for example, the XGBoost classifier model, the Naive Bayes classifier and the neural network classifier can be used for training at the same time. In use, the processing device can obtain three pending association parameters through the three different classifiers respectively, and then combine the three pending association parameters, for example, through weighted fusion or the like, to obtain the final association parameters.

[0090] In order to more clearly understand the technical solutions of the present application, next, a content pushing method provided by the embodiments of the present application will be introduced in combination with an actual application scenario.

[0091] Referring to Figure 3 , Figure 3 A flowchart of a content pushing method in an actual application scenario provided by the embodiments of the present application, in which the processing device can be a content recommendation server, and the content behavior data can be advertisement pushing data pushed by financial and insurance advertisements to users. Figure 7 , Figure 7 A flowchart of a content pushing method in an actual application scenario provided by the embodiments of the present application, in which the processing device can be a content recommendation server, and the content behavior data can be advertisement pushing data pushed by financial and insurance advertisements to users.

[0092] S701: Obtain total content behavior data of a target user corresponding to different contents in a target time interval.

[0093] The content pushing server can first obtain the advertisement pushing data, wherein the advertisement pushing data can include not only the advertisement pushing data related to the target product to be pushed this time, but also the advertisement pushing data corresponding to the associated products, for example, the target content can be the advertisement pushing data corresponding to A goods, B goods is the associated product of A goods, the total content behavior data includes the advertisement pushing data of A goods and the advertisement pushing data of B goods, and the target time interval is 12 months.

[0094] S702: Determine the user attribute features of the target user according to the user attributes of the target user.

[0095] S703: Data preprocessing.

[0096] The content pushing server can perform data preprocessing operation on the advertisement pushing data and the user attribute features. In the data preprocessing process, the content pushing server can eliminate dirty data and invalid data, and select the product-related advertisement data from the advertisement pushing data according to the product information to which the advertisement pushing data is directed. In addition, in order to make the data more reasonable, when obtaining the advertisement pushing data of the associated products, if the number of the advertisement pushing data corresponding to a certain associated product is less than N, the advertisement pushing data corresponding to the associated product can be eliminated, the number of N can be determined based on historical data processing experience, so as to avoid the problem of too large error caused by too small data quantity, and further improve the accuracy of content pushing.

[0097] S704: Divide different contents into types based on the pushing types, to obtain multiple pushing type dimensions of different contents.

[0098] After preprocessing the ad push data, the content push server can classify the push data according to the push type corresponding to the ad push data, and divide the statistical period according to the classified ad push data.

[0099] S705: Perform data statistical processing on the total content behavior data based on different statistical periods to obtain multi-period statistical characteristics of target users related to different content under multiple push type dimensions.

[0100] The statistical period can be 1 week, 2 weeks, 1 month, 3 months, 6 months, or 12 months. The processed ad push data is shown in the table below:

[0101] Table 1. Ad push data for target users and product A

[0102]

[0103]

[0104] Table 2. Ad push data for target users and product B.

[0105]

[0106] The content push server can determine multi-period statistical features based on the classified and statistically analyzed advertising push data. If the actual application scenario has 6 statistical periods, 10 products, and 3 push categories, then a feature vector of 6*10*3=180-dimensional multi-period statistical features can be generated.

[0107] S706: Determine the user interest profile of the target user based on multi-period statistical characteristics related to different content.

[0108] In this embodiment of the application, the advertising push data may include advertising push data for product A corresponding to the target user and product B corresponding to the target user. The user interest profile corresponding to the target user can be constructed by multi-period statistical features determined based on these data.

[0109] S707: Determine the product attribute characteristics of the product corresponding to the target content.

[0110] In this practical application scenario, the target content can correspond to product A, and the content push server can determine the product attribute characteristics corresponding to product A.

[0111] S708: Determine the classification fusion features based on product attribute characteristics, user attribute characteristics, and multi-period statistical characteristics.

[0112] Content push servers can fuse multi-period statistical features, product attribute features, and user attribute features to obtain categorized fusion features.

[0113] S709: determining an association parameter of the target user and the target content according to the classified fusion feature.

[0114] The content pushing server can determine the association parameter corresponding to the target user through the XGBoost classifier according to the classified fusion feature, and the association parameter can reflect the intention degree of the target user to the A commodity related advertisement.

[0115] S710: determining whether to push the target content to the target user based on the association parameter.

[0116] The content pushing server can select the intended population based on the association parameter, so as to determine the user with high intention degree to the A commodity related advertisement for content pushing, improve the success rate and efficiency of the A commodity related advertisement pushing, and thus facilitate the promotion of the A commodity.

[0117] Based on the content pushing method provided in the above embodiment, the embodiment of the present application further provides a content pushing device. Referring to Figure 4 , Figure 4 The structure block diagram of the content pushing device 400 provided in the embodiment of the present application is shown in the figure, and the device 400 includes an acquisition unit 401, a first determination unit 402, a second determination unit 403 and a pushing unit 404:

[0118] The acquisition unit 401 is configured to acquire content behavior data of a target user corresponding to a target content in a target time interval.

[0119] The first determination unit 402 is configured to determine a multi-period statistical feature of the target user related to the target content through data periodicity statistics according to the content behavior data, the multi-period statistical feature including a first statistical feature corresponding to a first statistical period and a second statistical feature corresponding to a second statistical period, the first statistical period and the second statistical period having different period lengths, and the period lengths of the first statistical period and the second statistical period being less than or equal to the time length identified by the target time interval.

[0120] The second determination unit 403 is configured to determine an association parameter of the target user and the target content according to the multi-period statistical feature.

[0121] The pushing unit 404 is configured to determine whether to push the target content to the target user based on the association parameter.

[0122] In a possible implementation manner, the device 400 further includes a classification unit:

[0123] The classification unit is configured to divide the target content into multiple pushing type dimensions based on a pushing type.

[0124] The first determining unit 402 is specifically configured to:

[0125] According to the content behavior data, a multi-period statistical feature of the target user and the target content in the plurality of push type dimensions is determined by periodic data statistics.

[0126] In a possible implementation, the content behavior data includes first data of the target user being pushed the target content and / or second data of the target user interacting with the pushed target content.

[0127] In a possible implementation, the obtaining unit 401 is specifically configured to:

[0128] determine a first product field of a product corresponding to the target content;

[0129] obtain first behavior data of the target user corresponding to the target content in the target time interval from the first product field;

[0130] determine an associated product having an association with the product, the associated product belonging to a second product field;

[0131] obtain second behavior data of the target user corresponding to an associated content in the target time interval from the second product field, the associated content being content related to the associated product;

[0132] determine the content behavior data according to the first behavior data and the second behavior data.

[0133] In a possible implementation, the apparatus 400 further includes a third determining unit:

[0134] The third determining unit is configured to determine whether a data amount of the first behavior data meets a preset data amount.

[0135] If yes, the first behavior data is determined as the content behavior data.

[0136] If no, the step of determining the associated product having the association with the product is performed.

[0137] In a possible implementation, the apparatus 400 further includes a fourth determining unit and a fifth determining unit:

[0138] The fourth determining unit is configured to determine a product attribute feature of the product corresponding to the target content.

[0139] The fifth determining unit is configured to determine a user attribute feature of the target user according to user information of the target user.

[0140] The second determining unit 403 is specifically used for

[0141] The classification fusion features are determined based on the product attribute features, the user attribute features, and the multi-period statistical features;

[0142] The association parameters between the target user and the target content are determined based on the classification fusion features.

[0143] In one possible implementation, the acquisition unit 401 is specifically used for:

[0144] Obtain the total content behavior data of the target user corresponding to different content within the target time interval, wherein the different content includes the target content;

[0145] The first determining unit 401 is specifically used for

[0146] Based on the total content behavior data, multi-period statistical characteristics of the target user related to different content are determined through periodic data statistics.

[0147] The user interest profile of the target user is determined based on the multi-period statistical features related to the different content.

[0148] In one possible implementation, the second determining unit 403 is specifically used for:

[0149] Based on the multi-period statistical characteristics, the association parameters between the target user and the target content are determined by a classifier model;

[0150] The classifier model is trained in the following manner:

[0151] Training samples are determined based on the target content. The training samples include historical content behavior data of sample users and sample tags used to identify whether the historical users are associated with the target content.

[0152] The initial classifier is trained based on the training samples to obtain the classifier model.

[0153] In one possible implementation, the number of classifier models is multiple, and the second determining unit 403 is specifically used for:

[0154] Based on the multi-period statistical characteristics, the undetermined association parameters between the target user and the target content are determined by multiple classifier models.

[0155] Based on the determined undetermined association parameters, the association parameters between the target user and the target content are determined.

[0156] The embodiments of the present application further provide a computer device, which is described below in combination with the drawings. Please refer to Figure 5 As shown in the drawings, the embodiments of the present application provide a device, which can also be a terminal device. The terminal device can be any intelligent terminal, including a mobile phone, a tablet computer, a personal digital assistant (PDA), a point of sales (POS), a vehicle-mounted computer, etc. Take the mobile phone as an example of the terminal device:

[0157] Figure 5 As shown in the drawings, the embodiments of the present application provide a device, which can also be a terminal device. The terminal device can be any intelligent terminal, including a mobile phone, a tablet computer, a personal digital assistant (PDA), a point of sales (POS), a vehicle-mounted computer, etc. Take the mobile phone as an example of the terminal device: Figure 5 The mobile phone includes a radio frequency (RF) circuit 510, a memory 520, an input unit 530, a display unit 540, a sensor 550, an audio circuit 560, a wireless fidelity (WiFi) module 570, a processor 580, and a power supply 590, etc. Those skilled in the art can understand that the mobile phone structure shown in the drawings does not constitute a limitation on the mobile phone, which can include more or fewer components than those shown in the drawings, or combine some components, or arrange different components. Figure 5 The mobile phone structure shown in the drawings does not constitute a limitation on the mobile phone, which can include more or fewer components than those shown in the drawings, or combine some components, or arrange different components.

[0158] The embodiments of the present application further provide a computer device, which is described below in combination with the drawings. Please refer to Figure 5 The various components of the mobile phone are described in detail as follows:

[0159] The RF circuit 510 can be used for receiving and sending signals in the process of information or communication, in particular, receiving the downlink information from the base station and processing by the processor 580; in addition, sending the uplink data to the base station. Generally, the RF circuit 510 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc. In addition, the RF circuit 510 can also communicate with the network and other devices through wireless communication. The above wireless communication can use any communication standard or protocol, including but not limited to global system for mobile communication (GSM), general packet radio service (GPRS), code division multiple access (CDMA), wideband code division multiple access (WCDMA), long term evolution (LTE), email, short message service (SMS), etc.

[0160] The memory 520 can be used to store software programs and modules, and the processor 580 executes various functions of the mobile phone and data processing by running the software programs and modules stored in the memory 520. The memory 520 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 520 can include a high-speed random access memory, and can also include a non-volatile memory, for example, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state memory device.

[0161] The input unit 530 can be used to receive input digital or character information, and to generate key signal inputs used for user settings and function controls of the mobile phone. Specifically, the input unit 530 can include a touch panel 531 and other input devices 532. The touch panel 531, also called a touch screen, can collect touch operations (such as operations of a user using a finger, a stylus, or any suitable object or accessory on or near the touch panel 531) of the user on or near it, and drive the corresponding connection device according to the pre-set program. Optionally, the touch panel 531 can include two parts, a touch detection device and a touch controller. The touch detection device detects the touch position of the user and detects the signal generated by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into touch coordinates, and sends it to the processor 580, and can also receive commands from the processor 580 and execute them. In addition, the touch panel 531 can be implemented in various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel 531, the input unit 530 can also include other input devices 532. Specifically, the other input devices 532 can include one or more of a physical keyboard, function keys (such as volume control keys, on / off keys, etc.), trackballs, mice, joysticks, etc.

[0162] The display unit 540 can be used to display information input by the user or information provided to the user, as well as various menus of the mobile phone. The display unit 540 can include a display panel 541, which can be configured in the form of a liquid crystal display (LCD) or an organic light-emitting diode (OLED), etc. Further, the touch panel 531 can cover the display panel 541, and when the touch panel 531 detects a touch operation on or near it, it transmits to the processor 580 to determine the type of touch event, and then the processor 580 provides corresponding visual output on the display panel 541 according to the type of touch event. Although in the above description, the touch panel 531 and the display panel 541 are implemented as two independent components to realize the input and output functions of the mobile phone, in some embodiments, the touch panel 531 and the display panel 541 can be integrated to realize the input and output functions of the mobile phone. Figure 5

[0163] ​The phone can also include at least one sensor 550, such as an optical sensor, a motion sensor, and other sensors. Specifically, the optical sensor can include an ambient light sensor to adjust the brightness of the display panel 541 according to the brightness of ambient light, and a proximity sensor to turn off the display panel 541 and / or the backlight when the phone is moved to the ear. As one of the motion sensors, the accelerometer sensor can detect the magnitude of the acceleration in each direction (generally three axes), and when at rest, the magnitude and direction of gravity, which can be used for applications that identify the phone posture (such as switching between landscape and portrait screens, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometers, taps), and the like. As for other sensors that the phone can also be configured, such as a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, and the like, will not be described here.

[0164] The audio circuit 560, the speaker 561, and the microphone 562 can provide an audio interface between the user and the phone. The audio circuit 560 can convert the received audio data into an electrical signal, transmit it to the speaker 561, and convert it into a sound signal output by the speaker 561; on the other hand, the microphone 562 converts the collected sound signal into an electrical signal, which is received by the audio circuit 560 and converted into audio data, which is then processed by the processor 580 and transmitted to another phone via the RF circuit 510, or output to the memory 520 for further processing.

[0165] WiFi is a short-range wireless transmission technology. The WiFi module 570 can help users send and receive emails, browse web pages, and access streaming media, and it provides users with wireless broadband Internet access. Although Figure 5 Although the WiFi module 570 is shown, it is understood that it does not belong to the necessary components of the phone, and can be omitted as needed without changing the essence of the invention.

[0166] The processor 580 is the control center of the phone, which connects all parts of the phone through various interfaces and lines, and performs various functions of the phone and processes data by running or executing software programs and / or modules stored in the memory 520, and calling data stored in the memory 520. Optionally, the processor 580 can include one or more processing units; preferably, the processor 580 can integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communication. It is understood that the above-mentioned modem processor can also not be integrated into the processor 580.

[0167] The mobile phone also includes a power supply 590 (such as a battery) that supplies power to various components. Preferably, the power supply can be logically connected to the processor 580 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0168] Although not shown, mobile phones may also include a camera, Bluetooth module, etc., which will not be described in detail here.

[0169] In this embodiment, the processor 580 included in the terminal device also has the following functions:

[0170] Acquire content behavior data of target users within a target time interval for the target content;

[0171] Based on the content behavior data, multi-period statistical features related to the target user and the target content are determined through periodic data statistics. The multi-period statistical features include a first statistical feature corresponding to a first statistical period and a second statistical feature corresponding to a second statistical period. The first statistical period and the second statistical period have different period lengths, and the period lengths of the first statistical period and the second statistical period are both less than or equal to the duration identified by the target time interval.

[0172] The association parameters between the target user and the target content are determined based on the multi-period statistical characteristics.

[0173] Based on the associated parameters, it is determined whether to push the target content to the target user.

[0174] This application also provides a server; please refer to [link / reference]. Figure 6 As shown, Figure 6 This is a structural diagram of a server 600 provided in an embodiment of this application. The server 600 can vary significantly due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 622 (e.g., one or more processors) and a memory 632, and one or more storage media 630 (e.g., one or more mass storage devices) for storing application programs 642 or data 644. The memory 632 and storage media 630 can be temporary or persistent storage. The program stored in the storage media 630 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the server. Furthermore, the CPU 622 may be configured to communicate with the storage media 630 and execute the series of instruction operations stored in the storage media 630 on the server 600.

[0175] The server 600 can also include one or more power supplies 626, one or more wired or wireless network interfaces 650, one or more input / output interfaces 658, and / or one or more operating systems 641, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0176] The steps performed by the server in the above embodiments can be based on Figure 6 the server structure shown.

[0177] The embodiments of the present application also provide a computer readable storage medium for storing a computer program for performing any of the content pushing methods described in the above embodiments.

[0178] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program performs the steps of the above-mentioned method embodiments when executed; and the foregoing storage medium can be at least one of the following: read-only memory (English: read-only memory, abbreviation: ROM), RAM, magnetic disk or optical disk and various media that can store program codes.

[0179] It should be noted that each of the embodiments in the present specification is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, they are described more simply, and the relevant parts can be referred to the part of the description of the method embodiments. The above-described device and system embodiments are only illustrative, and the units described as separate components can be or can not be physically separated, and the components shown as units can be or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiments according to the actual needs. Those skilled in the art can understand and implement without creative labor.

[0180] The above describes only one specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any skilled person in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A content push method characterized by, The method comprises: obtaining content behavior data of a target user corresponding to target content in a target time interval, the content behavior data comprising first data of the target user being pushed the target content and second data of the target user interacting with the pushed target content, the first data reflecting the association between the target user and the target content from the aspect of content pushing, and the second data reflecting the association between the target user and the target content from the aspect of user-content interaction, the first data being the number of times of pushing the target content to the target user, the target content being classified into multiple push type dimensions based on push types, the push type dimensions being used for classified statistics of the content behavior data, the push types being classified according to the push purpose of content being pushed to users, and the first data and the second data being data of different product fields; performing data statistical processing on the content behavior data based on different statistical periods to obtain multiple-period statistical features of the target user and the target content being respectively associated under multiple push type dimensions, the multiple-period statistical features of different types of the same product reflecting the time sequence change trend of the target user's intention degree to the target content when being in different content pushing stages, the multiple-period statistical features comprising first statistical features corresponding to a first statistical period and second statistical features corresponding to a second statistical period, the first statistical period and the second statistical period having different period lengths, and the period lengths of the first statistical period and the second statistical period both being less than or equal to the time length identified by the target time interval; determining product attribute features of a product corresponding to the target content; determining user attribute features of the target user according to user information of the target user; determining an association parameter of the target user and the target content according to the multiple-period statistical features, wherein the association parameter is determined according to the product attribute features, the user attribute features and the multiple-period statistical features, and the association parameter reflects the acceptance of the target user to the target content; determining whether to push the target content to the target user based on the association parameter.

2. The method of claim 1, wherein, The method comprises: determining a first product field of a product corresponding to the target content; obtaining first behavior data of the target user corresponding to the target content in the target time interval from the first product field; determining an associated product having an association with the product, the associated product belonging to a second product field; obtaining second behavior data of the target user corresponding to an associated content in the target time interval from the second product field, the associated content being content related to the associated product; determining the content behavior data according to the first behavior data and the second behavior data.

3. The method of claim 2, wherein, The method further comprises: determining whether the data amount of the first behavior data meets a preset data amount. If yes, the first behavior data is determined as the content behavior data. If no, the step of determining an associated product associated with the product is performed.

4. The method according to any one of claims 1 to 3, characterized in that, The content behavior data of the target user corresponding to the target content in a target time interval is obtained, including: The total content behavior data of the target user corresponding to different contents in the target time interval is obtained, the different contents including the target content; The content behavior data is subjected to data statistical processing based on different statistical periods to obtain multi-period statistical features of the target user related to the target content in multiple push type dimensions, including: The total content behavior data is subjected to data statistical processing based on different statistical periods to obtain multi-period statistical features of the target user related to the different contents; The user interest portrait of the target user is determined according to the multi-period statistical features related to the different contents.

5. The method according to any one of claims 1 to 3, characterized in that, The association parameter of the target user and the target content is determined according to the multi-period statistical features, including: The association parameter of the target user and the target content is determined by a classifier model according to the multi-period statistical features; The classifier model is trained according to the following manner: Training samples are determined according to the target content, the training samples including historical content behavior data of sample users and sample labels for identifying whether historical users are associated with the target content; An initial classifier is trained according to the training samples to obtain the classifier model.

6. The method of claim 5, wherein, The number of the classifier models is multiple, and the association parameter of the target user and the target content is determined by a classifier model according to the multi-period statistical features, including: The association parameter of the target user and the target content is determined by multiple classifier models according to the multi-period statistical features. The association parameter of the target user and the target content is determined according to the determined association parameters.

7. A content pusher device, characterized by The device includes an acquisition unit, a first determination unit, a second determination unit, a fourth determination unit, a fifth determination unit, and a push unit: The acquisition unit is configured to obtain content behavior data of a target user corresponding to a target content in a target time interval, the content behavior data including first data of a target user pushed target content and second data of a target user interacting with the pushed target content, the first data reflecting the association between the target user and the target content from the aspect of content push, and the second data reflecting the association between the target user and the target content from the aspect of user and content interaction, the first data being the number of times of pushing the target content to the target user, the target content being classified into multiple push type dimensions based on push type, the push type dimensions being used for classified statistics of content behavior data, the push type being classified according to the push purpose of content pushed to the user, and the first data and the second data being data of different product fields. The first determining unit is configured to perform data statistical processing on the content behavior data based on different statistical periods to obtain multi-period statistical features of the target user and the target content respectively in multiple push type dimensions, the multi-period statistical features of different types of the same product reflecting a time sequence change trend of the target user in the intention degree to the target content when being in different content push stages, the multi-period statistical features including a first statistical feature corresponding to a first statistical period and a second statistical feature corresponding to a second statistical period, the first statistical period and the second statistical period having different period lengths, and the period lengths of the first statistical period and the second statistical period being less than or equal to a time length identified by the target time interval; The fourth determining unit is configured to determine a product attribute feature of a product corresponding to the target content; The fifth determining unit is configured to determine a user attribute feature of the target user according to user information of the target user; The second determining unit is configured to determine an association parameter of the target user and the target content according to the multi-period statistical features, including: determining a classification fusion feature according to the product attribute feature, the user attribute feature and the multi-period statistical features; and determining the association parameter of the target user and the target content according to the classification fusion feature, the association parameter reflecting an acceptance degree of the target user to the target content; The pushing unit is configured to determine whether to push the target content to the target user based on the association parameter.

8. The apparatus of claim 7, wherein, The obtaining unit is specifically configured to: determine a first product field of a product corresponding to the target content; obtain first behavior data of the target user corresponding to the target content in the target time interval from the first product field; determine an associated product having an association with the product, the associated product belonging to a second product field; obtain second behavior data of the target user corresponding to an associated content in the target time interval from the second product field, the associated content being content related to the associated product; determine the content behavior data according to the first behavior data and the second behavior data.

9. The apparatus of claim 8, wherein, The device further includes a third determining unit; The third determining unit is configured to determine whether a data amount of the first behavior data meets a preset data amount, and if yes, determine the first behavior data as the content behavior data; if no, perform the step of determining the associated product having the association with the product.

10. The apparatus of any of claims 7-9, wherein, The obtaining unit is specifically configured to: obtain total content behavior data of the target user corresponding to different contents in the target time interval, the different contents including the target content; The first determining unit is specifically configured to: perform data statistical processing on the total content behavior data based on different statistical periods to obtain multi-period statistical features of the target user and the different contents respectively; determine a user interest portrait of the target user according to the multi-period statistical features related to the different contents.

11. The apparatus of any of claims 7-9, wherein, The second determining unit is specifically configured to: According to the multi-period statistical feature, determine, by a classifier model, an association parameter of the target user and the target content; The classifier model is trained according to the following manner: Determine a training sample according to the target content, the training sample including historical content behavior data of a sample user and a sample label used to identify whether a historical user has an association with the target content; Train an initial classifier according to the training sample to obtain the classifier model.

12. The apparatus of claim 11, wherein, The number of the classifier models is multiple, and the second determination unit is specifically configured to: According to the multi-period statistical feature, determine, by multiple classifier models, a pending association parameter of the target user and the target content respectively; According to the pending association parameters determined respectively, determine the association parameter of the target user and the target content.

13. A computer device, comprising: The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the content pushing method according to the instructions in the program code.

14. A computer-readable storage medium, characterized in that, The computer readable storage medium is used to store a computer program, and the computer program is used to execute the content pushing method.

Images