Internet information pushing method and system based on big data service

Abstract

The embodiment of the invention provides an Internet information pushing method and system based on big data service, and the method comprises the steps: obtaining an Internet interaction data sequence of a basic template and a business derivation template from the big data service, and carrying out the parameter learning and optimization of an Internet interest prediction network through the data. And finally, a target Internet interest prediction network is generated, so that the accuracy of interest point prediction is improved, and accurate grasp of personalized demands of the user is realized. By generating and fusing various interactive knowledge graph features, the network parameters are further optimized, so that the pushed information better meets the actual interests and requirements of the user, and the Internet information pushing efficiency and the user satisfaction are remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and more specifically, to an Internet information push method and system based on big data services. Background Art

[0002] With the rapid development of the Internet, the amount of information has exploded. While users enjoy the convenience brought by massive information, they are also facing the problem of information overload. How to filter out the content that users are interested in from the massive amount of Internet information and push it accurately has become a key issue that needs to be solved in the current Internet field.

[0003] Traditional information push methods are often based on user historical behavior data or simple keyword matching. Although this method can meet user needs to a certain extent, it often lacks personalization and accuracy. With the continuous development of big data technology, the method of using big data services to push Internet information has gradually attracted attention.

[0004] In the big data environment, Internet interaction data presents the characteristics of complexity and diversity. How to effectively use this data for information push has become a new challenge. On the one hand, it is necessary to mine users' potential interactive interests in order to more accurately grasp their needs; on the other hand, it is necessary to build an efficient interest prediction network in order to quickly respond to users' interactive behaviors and push relevant information.

[0005] In view of the above problems, although there are some Internet information push methods based on big data in the prior art, these methods often only focus on a single data type or simple data processing method, and cannot fully utilize the rich interactive data sequences provided by big data services. In addition, these methods also have deficiencies in the construction and optimization of interest prediction networks, resulting in limited push effects. Summary of the invention

[0006] In view of this, the purpose of this application is to provide an Internet information push method and system based on big data services.

[0007] In conjunction with the first aspect of the present application, a method for pushing Internet information based on big data services is provided, which is applied to an Internet information push system based on big data services, and the method comprises:

[0008] Obtaining a basic template Internet interaction data sequence and a business-derived template Internet interaction data sequence from a big data service; wherein the basic template Internet interaction data sequence includes basic template Internet interaction data and a priori interaction interest points corresponding to the basic template Internet interaction data, and the business-derived template Internet interaction data sequence includes business-derived template Internet interaction data; the business-derived template Internet interaction data is determined based on the basic template Internet interaction data;

[0009] Performing network parameter learning on an Internet interest prediction network according to the basic template Internet interaction data sequence to generate an Internet interest prediction network that has completed network parameter learning;

[0010] Loading the basic template Internet interaction data in the basic template Internet interaction data sequence into the Internet interest prediction network that has completed network parameter learning, and generating a first interactive knowledge graph feature corresponding to the basic template Internet interaction data;

[0011] Inputting the basic template Internet interaction data in the basic template Internet interaction data sequence and the business-derived template Internet interaction data in the business-derived template Internet interaction data sequence into a neural network model respectively, generating a second interaction knowledge graph feature corresponding to the basic template Internet interaction data and a third interaction knowledge graph feature corresponding to the business-derived template Internet interaction data;

[0012] Based on the first interactive knowledge graph feature, the second interactive knowledge graph feature and the third interactive knowledge graph feature, the network parameter information of the Internet interest prediction network that has completed network parameter learning is optimized to generate a target Internet interest prediction network, and after predicting the points of interest of any input Internet interaction data based on the generated target Internet interest prediction network, Internet information is pushed to the corresponding target user based on the point of interest prediction result.

[0013] In a possible implementation of the first aspect, the business-derived template Internet interaction data is generated by setting and transforming the basic template Internet interaction data; wherein the setting and transformation method includes at least one of the following: data interception, partial masking, and adding data disturbance.

[0014] In a possible implementation of the first aspect, the performing network parameter learning on the Internet interest prediction network according to the basic template Internet interaction data sequence to generate the Internet interest prediction network after completing the network parameter learning includes:

[0015] Loading the basic template Internet interaction data in the basic template Internet interaction data sequence into the Internet interest prediction network to generate a first interactive knowledge graph feature corresponding to the basic template Internet interaction data;

[0016] Inputting the first interactive knowledge graph feature corresponding to the basic template Internet interactive data into a set fully connected mapping layer to generate a mapping interest point corresponding to the basic template Internet interactive data;

[0017] Determining the interest matching degree corresponding to the basic template Internet interaction data based on the mapped interest points and the prior interaction interest points corresponding to the basic template Internet interaction data;

[0018] Based on the basic template Internet interaction data, the prior interaction interest points corresponding to the basic template Internet interaction data, the mapped interest points corresponding to the basic template Internet interaction data and the interest matching degree corresponding to the basic template Internet interaction data, the network parameter information of the Internet interest prediction network is optimized to generate an Internet interest prediction network that has completed network parameter learning.

[0019] In a possible implementation of the first aspect, the network parameter information of the Internet interest prediction network that has completed network parameter learning is optimized based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature to generate a target Internet interest prediction network, and after predicting points of interest for any input Internet interaction data based on the generated target Internet interest prediction network, Internet information is pushed to the corresponding target user based on the point of interest prediction result, including:

[0020] According to the first interactive knowledge graph feature, the second interactive knowledge graph feature and the third interactive knowledge graph feature, respectively optimizing the network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model to generate an optimized Internet interest prediction network and an optimized neural network model;

[0021] If the network parameter information of the optimized Internet interest prediction network meets the set requirements, the optimized Internet interest prediction network is used as the target Internet interest prediction network.

[0022] In a possible implementation of the first aspect, the method of optimizing the network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature to generate an optimized Internet interest prediction network and an optimized neural network model includes:

[0023] Determining a feature error parameter based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature;

[0024] The network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model are optimized respectively according to the characteristic error parameters to generate an optimized Internet interest prediction network and an optimized neural network model.

[0025] In a possible implementation of the first aspect, the method further includes:

[0026] If the network parameter information of the optimized Internet interest prediction network does not meet the set requirements, determining target network parameter information based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model;

[0027] Based on the target network parameter information, the network parameter information of the optimized neural network model is optimized to generate an optimized neural network model; and the step of performing network parameter learning on the Internet interest prediction network according to the basic template Internet interaction data sequence to generate an Internet interest prediction network that has completed network parameter learning is continued until the network parameter information of the Internet interest prediction network meets the set requirements.

[0028] In a possible implementation of the first aspect, determining the target network parameter information based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model includes:

[0029] The network parameter information of the optimized neural network model and the weighted result of the set first influencing factor are used as the first network parameter data;

[0030] The weighted result of the network parameter information of the optimized Internet interest prediction network and the preset second influencing factor is used as the second network parameter data; wherein the second influencing factor is determined based on the first influencing factor;

[0031] The fusion data of the first network parameter data and the second network parameter data is used as the target network parameter information.

[0032] In a possible implementation of the first aspect, the model structures of the Internet interest prediction network and the neural network model are the same, and the basic network parameter information of the Internet interest prediction network and the neural network model are the same.

[0033] In combination with the second aspect of the present application, a big data service-based Internet information push system is provided, wherein the big data service-based Internet information push system comprises a machine-readable storage medium and a processor, wherein the machine-readable storage medium stores machine-executable instructions, and when the processor executes the machine-executable instructions, the big data service-based Internet information push system implements the aforementioned big data service-based Internet information push method.

[0034] In conjunction with the third aspect of the present application, a computer-readable storage medium is provided, in which computer-executable instructions are stored. When the computer-executable instructions are executed, the aforementioned Internet information push method based on big data services is implemented.

[0035] In combination with any of the above aspects, the embodiment of the present application obtains the Internet interaction data sequence of the basic template and the business derivative template from the big data service, and uses these data to learn and optimize the parameters of the Internet interest prediction network, and finally generates the target Internet interest prediction network, which not only improves the accuracy of interest point prediction, but also realizes the accurate grasp of the personalized needs of users. By generating and integrating multiple interactive knowledge graph features, the network parameters are further optimized, so that the pushed information is more in line with the actual interests and needs of users, thereby significantly improving the efficiency of Internet information push and user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained by combining these drawings without paying creative work.

[0037] Figure 1 A flowchart of an Internet information push method based on big data services provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.

[0040] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0041] Figure 1 The flowchart of the Internet information push method based on big data service provided by the embodiment of the present application is shown. It should be understood that in other embodiments, the order of some steps of the Internet information push method based on big data service of this embodiment can be shared with each other based on actual needs, or some steps can be omitted or maintained. The details of the Internet information push method based on big data service include:

[0042] Step S110, obtaining a basic template Internet interaction data sequence and a business derived template Internet interaction data sequence from a big data service. The basic template Internet interaction data sequence includes basic template Internet interaction data and a priori interaction interest points corresponding to the basic template Internet interaction data, and the business derived template Internet interaction data sequence includes business derived template Internet interaction data. The business derived template Internet interaction data is determined based on the basic template Internet interaction data.

[0043] In this embodiment, a server of a large Internet service provider is used, and the server is responsible for providing personalized Internet information push services to many users. The server is connected to a huge big data service platform, which stores a large amount of user Internet interaction data.

[0044] In terms of the basic template Internet interaction data sequence, let's take an online shopping platform as an example. The basic template Internet interaction data may include the user's browsing history on the shopping platform, such as which product pages the user viewed, how long they stayed, whether they added the product to the shopping cart, and other operation records. The prior interaction interest points are predetermined based on the user's previous purchase behavior or clear preference settings. For example, if the user often bought sports equipment in the past, then "sports equipment" is a prior interaction interest point. For this shopping platform user, the basic template Internet interaction data sequence may be as follows: User A viewed the pages of sports shoes and sportswear in the past month, stayed for about 5-10 minutes each time, and added a pair of sports shoes to the shopping cart. The corresponding prior interaction interest point is "sports equipment".

[0045] The business-derived template Internet interaction data sequence is determined based on the basic template Internet interaction data. For example, for the basic template Internet interaction data of user A on the above-mentioned shopping platform, the business-derived template Internet interaction data is generated by data interception. Assume that the original basic template Internet interaction data contains the daily browsing records of user A in the past month, and now the browsing records of the most recent week are intercepted as the business-derived template Internet interaction data. Or, a partial masking method is adopted to mask part of the product details of the sports shoes page viewed by the user, and only basic information such as the approximate category of the product is retained as the business-derived template Internet interaction data. Or, data perturbation is added, and a small random perturbation is added to the dwell time data of the user viewing the product page, such as the original stay time of 5 minutes, which becomes 5.5 minutes after the perturbation. These processed data constitute the business-derived template Internet interaction data sequence.

[0046] Step S120, performing network parameter learning on the Internet interest prediction network according to the basic template Internet interaction data sequence, and generating an Internet interest prediction network after completing network parameter learning.

[0047] In this embodiment, the server starts to learn network parameters for the Internet interest prediction network, which is a neural network structure specifically used to predict the Internet interests of users.

[0048] First, the basic template Internet interaction data in the basic template Internet interaction data sequence is loaded into the Internet interest prediction network. For example, taking user A of the shopping platform as an example, user A's browsing history (operation records such as viewing sports shoes and sportswear pages) is loaded into the network. The neurons in the network calculate based on these input data to generate the first interactive knowledge graph features corresponding to the basic template Internet interaction data.

[0049] Next, this first interactive knowledge graph feature is input into the set fully connected mapping layer. The fully connected mapping layer maps the input features according to predefined weights and biases. For example, assume that the weight matrix of this mapping layer is set based on a large amount of historical data and prior knowledge. For the first interactive knowledge graph feature of user A, after calculation by the fully connected mapping layer, the mapping interest point corresponding to the basic template Internet interaction data is generated. It is possible that after the first interactive knowledge graph feature corresponding to the browsing history of user A is mapped, the mapping interest point obtained is "high-end sports equipment".

[0050] Then, based on the mapped interest points and prior interaction interest points corresponding to the basic template Internet interaction data, the interest matching degree corresponding to the basic template Internet interaction data is determined. For user A, the prior interaction interest point is "sports equipment" and the mapped interest point is "high-end sports equipment". The server calculates the interest matching degree by comparing the semantic similarity of the two interest points or predefined matching rules. For example, if the semantic similarity is set to be above 80% as a high match, and it is found through calculation that the semantic similarity between "sports equipment" and "high-end sports equipment" is 85%, then the interest matching degree is 85%.

[0051] Finally, the network parameter information of the Internet interest prediction network is optimized based on the basic template Internet interaction data, the prior interaction interest points corresponding to the basic template Internet interaction data, the mapped interest points corresponding to the basic template Internet interaction data, and the interest matching degree corresponding to the basic template Internet interaction data. The server adjusts parameters such as weights and biases in the network based on these data. For example, if the interest matching degree is high, the network parameters may be fine-tuned to further improve the accuracy of the prediction; if the interest matching degree is low, the weights and other parameters may be adjusted to a greater extent so that the network can more accurately map to the correct interest points when processing similar data. After multiple such adjustment processes, when the network parameters converge to a stable state, an Internet interest prediction network that completes network parameter learning is generated.

[0052] Step S130, loading the basic template Internet interaction data in the basic template Internet interaction data sequence into the Internet interest prediction network that has completed network parameter learning, and generating a first interactive knowledge graph feature corresponding to the basic template Internet interaction data.

[0053] In this embodiment, the server has obtained an Internet interest prediction network that has completed network parameter learning. Now the basic template Internet interaction data (taking the browsing history of user A on the shopping platform as an example) is loaded into this network again.

[0054] Each neuron in the network processes the input data according to the pre-learned parameters. For example, for user A's browsing of sports shoes and sportswear pages, as well as the length of stay and adding to the shopping cart, the first layer of neurons in the network may extract features from these data and convert them into an intermediate representation that is more suitable for subsequent network processing. Then, after calculations and information transmission by multiple layers of neurons, the first interactive knowledge graph feature corresponding to the basic template Internet interactive data is finally generated. This first interactive knowledge graph feature may be a high-dimensional vector representation that contains information such as the relationship between user A's browsing behavior and various predefined interest categories in the network. For example, some dimensions in this vector may represent user A's brand preference for sports equipment, and some dimensions may represent sensitivity to price ranges, etc.

[0055] Step S140, input the basic template Internet interaction data in the basic template Internet interaction data sequence and the business-derived template Internet interaction data in the business-derived template Internet interaction data sequence into the neural network model respectively, and generate a second interaction knowledge graph feature corresponding to the basic template Internet interaction data and a third interaction knowledge graph feature corresponding to the business-derived template Internet interaction data.

[0056] The server has a pre-built neural network model whose structure is similar to the Internet interest prediction network.

[0057] First, the basic template Internet interaction data (still taking the shopping platform browsing history of user A as an example) is input into the neural network model. The neurons in the neural network model begin to process these data. For example, the input layer of the model receives data such as user A viewing sports shoes and sportswear pages, and then performs complex calculations through the neurons in the hidden layer. The neurons in the hidden layer may use different activation functions, such as the ReLU function, to perform nonlinear transformations on the input data to mine more complex feature relationships in the data. After calculations in multiple hidden layers, the second interactive knowledge graph feature corresponding to the basic template Internet interaction data is finally generated. This second interactive knowledge graph feature may differ from the first interactive knowledge graph feature generated in step S130 in some dimensions, because although the two networks have similar structures, the parameters may be different. It is also a high-dimensional vector representation that contains feature information from different angles about user A's browsing behavior.

[0058] Next, the business-derived template Internet interaction data (such as the browsing history of user A in the past week after data interception) is input into the neural network model. Similarly, the neural network model processes these data. For example, since the data is intercepted, the model may pay more attention to recent browsing behavior characteristics when processing. After a process similar to that of processing the basic template Internet interaction data, the third interactive knowledge graph feature corresponding to the business-derived template Internet interaction data is finally generated. This feature is also represented in the form of a high-dimensional vector, which reflects the unique information in the business-derived template Internet interaction data, such as recent changes in user A's attention to specific types of sports equipment.

[0059] Step S150, based on the first interactive knowledge graph feature, the second interactive knowledge graph feature and the third interactive knowledge graph feature, the network parameter information of the Internet interest prediction network that has completed network parameter learning is optimized to generate a target Internet interest prediction network, and after predicting the points of interest of any input Internet interaction data based on the generated target Internet interest prediction network, Internet information is pushed to the corresponding target user based on the point of interest prediction result.

[0060] The server first optimizes the network parameter information of the Internet interest prediction network that has completed network parameter learning based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature.

[0061] The feature error parameter is determined based on these three interactive knowledge graph features. For example, the server calculates the feature error parameter by comparing the difference between the first interactive knowledge graph feature and the second interactive knowledge graph feature, and the difference between the first interactive knowledge graph feature and the third interactive knowledge graph feature. Assume that the first interactive knowledge graph feature represents the comprehensive interest feature of user A in sports equipment, the second interactive knowledge graph feature emphasizes user A's interest in a specific brand of sports equipment in some dimensions, and the third interactive knowledge graph feature is generated based on the business-derived template Internet interaction data (such as recent browsing history), and may pay more attention to user A's recent interest in lower-priced sports equipment. By calculating the differences between these features, the feature error parameter is obtained.

[0062] Then, the network parameter information of the Internet interest prediction network that has completed network parameter learning is optimized based on this feature error parameter. If the feature error parameter indicates that the first interactive knowledge graph feature and the second interactive knowledge graph feature have a large difference in a certain interest dimension, such as the brand preference dimension, then the server will adjust the weights of the neuron connections related to brand preference in the Internet interest prediction network so that the network can more accurately reflect brand preference information when processing similar data. Similarly, similar network parameter adjustments are made to the differences related to the third interactive knowledge graph feature. After such an optimization process, if the network parameter information of the optimized Internet interest prediction network meets the set requirements (for example, the prediction accuracy reaches more than 90%), this optimized network will be used as the target Internet interest prediction network.

[0063] Next, the server uses the target Internet interest prediction network to predict points of interest for any input Internet interaction data. For example, when a new user B's Internet interaction data (such as browsing news, social media likes and comments, etc.) is input into the target Internet interest prediction network, the network processes this data according to the pre-learned parameters and predicts points of interest that user B may be interested in, such as "technology news", "travel information", etc.

[0064] Finally, Internet information is pushed to the corresponding target user based on the predicted points of interest. For user B, the server selects content related to "technology news" and "travel information" from the massive amount of Internet information based on the predicted points of interest, and then pushes these contents to user B. For example, the server can push the latest technology news articles, travel destination recommendations and other information to user B's device, such as the news client or notification bar of a mobile phone or computer, thereby realizing personalized Internet information push service.

[0065] If the network parameter information of the optimized Internet interest prediction network does not meet the set requirements during the optimization process, the server will determine the target network parameter information based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model. For example, assume that the prediction accuracy of the optimized Internet interest prediction network is only 80%, which does not meet the set 90% requirement. The server uses the weighted result of the network parameter information of the optimized neural network model and the set first impact factor (such as 0.3) as the first network parameter data. Assuming that the optimized neural network model performs better when processing certain types of interactive data, this weighted result will comprehensively consider the advantages of the neural network model. At the same time, the weighted result of the network parameter information of the optimized Internet interest prediction network and the preset second impact factor (since the second impact factor is determined based on the first impact factor, it is assumed to be 0.7) is used as the second network parameter data. Then, the fusion data of the first network parameter data and the second network parameter data is used as the target network parameter information.

[0066] Based on this target network parameter information, the network parameter information of the optimized neural network model is optimized to generate an optimized neural network model. Then, the server continues to perform the steps of learning the network parameters of the Internet interest prediction network based on the basic template Internet interaction data sequence, repeating the whole process until the network parameter information of the Internet interest prediction network meets the set requirements, and finally obtains the target Internet interest prediction network that can accurately predict points of interest, and realizes efficient Internet information push service.

[0067] Based on the above steps, the embodiment of the present application obtains the Internet interaction data sequence of the basic template and the business derivative template from the big data service, and uses these data to learn and optimize the parameters of the Internet interest prediction network, and finally generates the target Internet interest prediction network, which not only improves the accuracy of interest point prediction, but also realizes the accurate grasp of user personalized needs. By generating and integrating multiple interactive knowledge graph features, the network parameters are further optimized, so that the pushed information is more in line with the actual interests and needs of users, thereby significantly improving the efficiency of Internet information push and user satisfaction.

[0068] In a possible implementation, the business-derived template Internet interaction data is generated by setting and transforming the basic template Internet interaction data, wherein the setting and transformation method includes at least one of the following: data interception, partial masking, and adding data disturbance.

[0069] In a possible implementation, the performing network parameter learning on the Internet interest prediction network according to the basic template Internet interaction data sequence to generate the Internet interest prediction network after completing the network parameter learning includes:

[0070] The basic template Internet interaction data in the basic template Internet interaction data sequence is loaded into the Internet interest prediction network to generate a first interactive knowledge graph feature corresponding to the basic template Internet interaction data.

[0071] The first interactive knowledge graph feature corresponding to the basic template Internet interactive data is input into the set fully connected mapping layer to generate mapping interest points corresponding to the basic template Internet interactive data.

[0072] Based on the mapped interest points and the prior interaction interest points corresponding to the basic template Internet interaction data, the interest matching degree corresponding to the basic template Internet interaction data is determined.

[0073] Based on the basic template Internet interaction data, the prior interaction interest points corresponding to the basic template Internet interaction data, the mapped interest points corresponding to the basic template Internet interaction data and the interest matching degree corresponding to the basic template Internet interaction data, the network parameter information of the Internet interest prediction network is optimized to generate an Internet interest prediction network that has completed network parameter learning.

[0074] In a possible implementation, the optimizing the network parameter information of the Internet interest prediction network that has completed network parameter learning based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature to generate a target Internet interest prediction network includes:

[0075] Based on the first interactive knowledge graph feature, the second interactive knowledge graph feature and the third interactive knowledge graph feature, the network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model are optimized respectively to generate an optimized Internet interest prediction network and an optimized neural network model.

[0076] If the network parameter information of the optimized Internet interest prediction network meets the set requirements, the optimized Internet interest prediction network is used as the target Internet interest prediction network.

[0077] In a possible implementation, the network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model are optimized respectively based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature to generate an optimized Internet interest prediction network and an optimized neural network model, including:

[0078] Determine a feature error parameter based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature.

[0079] The network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model are optimized respectively according to the characteristic error parameters to generate an optimized Internet interest prediction network and an optimized neural network model.

[0080] In a possible implementation, the method further includes:

[0081] If the network parameter information of the optimized Internet interest prediction network does not meet the set requirements, the target network parameter information is determined based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model.

[0082] Based on the target network parameter information, the network parameter information of the optimized neural network model is optimized to generate an optimized neural network model. And, the steps of learning the network parameters of the Internet interest prediction network based on the basic template Internet interaction data sequence to generate an Internet interest prediction network that has completed the network parameter learning are continued until the network parameter information of the Internet interest prediction network meets the set requirements.

[0083] In a possible implementation, the determining target network parameter information based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model includes:

[0084] The network parameter information of the optimized neural network model and the weighted result of the set first influencing factor are used as the first network parameter data.

[0085] The weighted result of the network parameter information of the optimized Internet interest prediction network and the preset second influencing factor is used as the second network parameter data, wherein the second influencing factor is determined based on the first influencing factor.

[0086] The fusion data of the first network parameter data and the second network parameter data is used as the target network parameter information.

[0087] In a possible implementation, the model structures of the Internet interest prediction network and the neural network model are the same, and the basic network parameter information of the Internet interest prediction network and the neural network model are the same.

[0088] In the above embodiments, the Internet information push system based on big data services for executing the above method embodiments has at least one processor, a control module (chip set) coupled to at least one of the (at least one) processors, a memory coupled to the control module, a non-volatile memory (NVM) / storage device coupled to the control module, at least one load / output device coupled to the control module, and a network interface coupled to the control module.

[0089] The processor may include at least one single-core or multi-core processor, and the processor may include any combination of general-purpose processors or dedicated processors (such as graphics processors, application processors, baseband processors, etc.). For some alternative implementations, the Internet information push system based on big data services can be used as the gateway and other electronic devices described in the embodiments of the present application.

[0090] For some alternative embodiments, the Internet information push system based on big data services may include at least one computer-readable medium (e.g., a memory or NVM / storage device) having instructions and at least one processor integrated with the at least one computer-readable medium and configured to execute instructions to implement a module to perform the actions described in the present disclosure.

[0091] For one embodiment, the control module may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) and / or any suitable device or component in communication with the control module.

[0092] The control module may include a memory controller module to provide an interface to the memory. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0093] The memory may be used, for example, to load and store data and / or instructions for an Internet information push system based on big data services. For one embodiment, the memory may include any suitable volatile memory, for example, a suitable DRAM.

[0094] For one embodiment, the control module may include at least one load-in / load-out controller to provide an interface to the NVM / storage device and the (at least one) load-in / load-out device.

[0095] For example, the NVM / storage device may be used to store data and / or instructions. The NVM / storage device may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (at least one) non-volatile storage device (e.g., at least one hard disk drive (HDD), at least one compact disk (CD) drive, and / or at least one digital versatile disk (DVD) drive).

[0096] The NVM / storage device may include storage resources that are physically part of the device on which the Internet information push system based on big data services is installed, or it may be accessible to the device without being part of the device. For example, the NVM / storage device may be accessed via (at least one) load / output device based on a network.

[0097] The (at least one) loading / output device may provide an interface for the Internet information push system based on big data services to communicate with any other appropriate device, and the loading / output device may include a communication component, a pinyin component, a sensor component, etc. The network interface may provide an interface for the Internet information push system based on big data services to communicate based on at least one network, and the Internet information push system based on big data services may wirelessly communicate with at least one component of a wireless network based on any of at least one wireless network priors and / or protocols, such as accessing a wireless network based on a communication prior.

[0098] For one embodiment, at least one of the (at least one) processor may be loaded together with the logic of at least one controller of the control module (e.g., a memory controller module). For one embodiment, at least one of the (at least one) processor may be loaded together with the logic of at least one controller of the control module to form a system level load. For one embodiment, at least one of the (at least one) processor may be fused on the same die with the logic of at least one controller of the control module. For one embodiment, at least one of the (at least one) processor may be fused on the same die with the logic of at least one controller of the control module to form a system on chip (SoC).

[0099] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, based on the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0100] An embodiment of the present invention discloses a computer-readable storage medium that stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the Internet information push method based on big data services described in the aforementioned embodiment.

[0101] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps in the Internet information push method based on big data services described in the aforementioned embodiment.

[0102] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed over multiple modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. Those of ordinary skill in the art may understand and implement the present invention without creative effort.

[0103] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, a magnetic disk storage, a magnetic tape storage, or any other medium that can be used to be computer-readable with or store data.

[0104] Finally, it should be noted that what is disclosed above is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for pushing Internet information based on big data services, characterized in that: The method comprises: Obtaining a basic template Internet interaction data sequence and a business-derived template Internet interaction data sequence from a big data service; wherein the basic template Internet interaction data sequence includes basic template Internet interaction data and a priori interaction interest points corresponding to the basic template Internet interaction data, and the business-derived template Internet interaction data sequence includes business-derived template Internet interaction data; the business-derived template Internet interaction data is determined based on the basic template Internet interaction data; Performing network parameter learning on an Internet interest prediction network according to the basic template Internet interaction data sequence to generate an Internet interest prediction network that has completed network parameter learning; Loading the basic template Internet interaction data in the basic template Internet interaction data sequence into the Internet interest prediction network that has completed network parameter learning, and generating a first interactive knowledge graph feature corresponding to the basic template Internet interaction data; Inputting the basic template Internet interaction data in the basic template Internet interaction data sequence and the business-derived template Internet interaction data in the business-derived template Internet interaction data sequence into a neural network model respectively, generating a second interaction knowledge graph feature corresponding to the basic template Internet interaction data and a third interaction knowledge graph feature corresponding to the business-derived template Internet interaction data; Based on the first interactive knowledge graph feature, the second interactive knowledge graph feature and the third interactive knowledge graph feature, the network parameter information of the Internet interest prediction network that has completed network parameter learning is optimized to generate a target Internet interest prediction network, and after predicting the points of interest of any input Internet interaction data based on the generated target Internet interest prediction network, Internet information is pushed to the corresponding target user based on the point of interest prediction result.

2. The Internet information push method based on big data service according to claim 1, characterized in that: The business-derived template Internet interaction data is generated by setting and transforming the basic template Internet interaction data; wherein, the setting and transformation method includes at least one of the following: data interception, partial masking, and adding data disturbance.

3. The Internet information push method based on big data service according to claim 1 is characterized in that: The method of performing network parameter learning on the Internet interest prediction network according to the basic template Internet interaction data sequence to generate the Internet interest prediction network after completing the network parameter learning comprises: Loading the basic template Internet interaction data in the basic template Internet interaction data sequence into the Internet interest prediction network to generate a first interactive knowledge graph feature corresponding to the basic template Internet interaction data; Inputting the first interactive knowledge graph feature corresponding to the basic template Internet interactive data into a set fully connected mapping layer to generate a mapping interest point corresponding to the basic template Internet interactive data; Determining the interest matching degree corresponding to the basic template Internet interaction data based on the mapped interest points and the prior interaction interest points corresponding to the basic template Internet interaction data; Based on the basic template Internet interaction data, the prior interaction interest points corresponding to the basic template Internet interaction data, the mapped interest points corresponding to the basic template Internet interaction data and the interest matching degree corresponding to the basic template Internet interaction data, the network parameter information of the Internet interest prediction network is optimized to generate an Internet interest prediction network that has completed network parameter learning.

4. The Internet information push method based on big data service according to claim 1 is characterized in that: The step of optimizing the network parameter information of the Internet interest prediction network that has completed network parameter learning based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature to generate a target Internet interest prediction network includes: According to the first interactive knowledge graph feature, the second interactive knowledge graph feature and the third interactive knowledge graph feature, respectively optimizing the network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model to generate an optimized Internet interest prediction network and an optimized neural network model; If the network parameter information of the optimized Internet interest prediction network meets the set requirements, the optimized Internet interest prediction network is used as the target Internet interest prediction network.

5. The Internet information push method based on big data service according to claim 4 is characterized in that: The method of optimizing the network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature to generate an optimized Internet interest prediction network and an optimized neural network model includes: Determining a feature error parameter based on the first interactive knowledge graph feature, the second interactive knowledge graph feature, and the third interactive knowledge graph feature; The network parameter information of the Internet interest prediction network that has completed network parameter learning and the network parameter information of the neural network model are optimized respectively according to the characteristic error parameters to generate an optimized Internet interest prediction network and an optimized neural network model.

6. The Internet information push method based on big data service according to claim 4 is characterized in that: The method further comprises: If the network parameter information of the optimized Internet interest prediction network does not meet the set requirements, determining target network parameter information based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model; Based on the target network parameter information, the network parameter information of the optimized neural network model is optimized to generate an optimized neural network model; and the step of performing network parameter learning on the Internet interest prediction network according to the basic template Internet interaction data sequence to generate an Internet interest prediction network that has completed network parameter learning is continued until the network parameter information of the Internet interest prediction network meets the set requirements.

7. The Internet information push method based on big data service according to claim 6 is characterized in that: The determining target network parameter information based on the network parameter information of the optimized Internet interest prediction network and the network parameter information of the optimized neural network model includes: The network parameter information of the optimized neural network model and the weighted result of the set first influencing factor are used as the first network parameter data; The weighted result of the network parameter information of the optimized Internet interest prediction network and the preset second influencing factor is used as the second network parameter data; wherein the second influencing factor is determined based on the first influencing factor; The fusion data of the first network parameter data and the second network parameter data is used as the target network parameter information.

8. The Internet information push method based on big data service according to any one of claims 1 to 7, characterized in that: The model structures of the Internet interest prediction network and the neural network model are the same, and the basic network parameter information of the Internet interest prediction network and the neural network model are the same.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the Internet information push method based on big data services described in any one of claims 1 to 8 is implemented.

10. An Internet information push system based on big data service, characterized in that: It includes a processor and a computer-readable storage medium, wherein the computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the Internet information push method based on big data service described in any one of claims 1 to 8 is implemented.

Images