Service recommendation processing method and device

Abstract

The embodiment of the invention provides a service recommendation processing method and device.The service recommendation processing method comprises the steps that in the process that service recommendation is conducted on an access user, on the basis of obtained service data of the access user of a service, the service recommendation is conducted on the access user; service data is input into a state extraction module to perform user state feature extraction to obtain user state features, and a corresponding user state feature is selected from the user state features to be input into at least one decision sub-model to perform decision processing of service recommendation elements. And further performing recommendation data generation according to the at least one recommendation element obtained by the decision processing so as to obtain recommendation data for performing service recommendation on the access user.

Description

Technical Field

[0001] This document relates to the field of data processing technology, and in particular to a service recommendation processing method and device. Background Art

[0002] With the continuous development and promotion of the Internet, the application scope of online services provided based on the Internet has become wider and wider, and has gradually covered most users. In this case, various forms of online services have emerged, such as recommending content to users during their access to applications, or providing consultation through a set consultation portal during their access to applications. However, with the widespread application of content recommendation and consultation during user access, users' needs during application access have become more and more personalized, and consultation has become more and more diversified, which has brought certain challenges to the providers of content recommendation and consultation. Summary of the invention

[0003] One or more embodiments of the present specification provide a service recommendation processing method, including: obtaining service data of a user accessing a service. Inputting the service data into each state extraction module to extract user state features, and obtaining each user state feature. Selecting at least one decision sub-model corresponding to the feature input from each user state feature to perform decision processing on a service recommendation element, and obtaining at least one recommendation element. Generating recommendation data according to the at least one recommendation element.

[0004] One or more embodiments of the present specification provide a service recommendation processing device, including: a service data acquisition module, configured to acquire service data of users accessing the service. A state feature extraction module, configured to input the service data into each state extraction module to extract user state features and obtain each user state feature. A decision processing module, configured to select at least one decision sub-model corresponding to the feature input from each user state feature to perform decision processing on a service recommendation element and obtain at least one recommendation element. A recommendation data generation module, configured to generate recommendation data based on the at least one recommendation element.

[0005] One or more embodiments of the present specification provide a service recommendation processing device, including: a processor; and a memory configured to store computer executable instructions, wherein when the computer executable instructions are executed, the processor: obtains service data of a user accessing a service. The service data is input into each state extraction module to extract user state features, and each user state feature is obtained. At least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing on a service recommendation element, and at least one recommendation element is obtained. Recommendation data is generated based on the at least one recommendation element.

[0006] One or more embodiments of the present specification provide a computer-readable storage medium for storing computer-executable instructions, which implement the following process when executed: Obtain service data of users accessing a service. Input the service data into each state extraction module to extract user state features, and obtain each user state feature. Select at least one decision sub-model corresponding to the feature input from each user state feature to perform decision processing on a service recommendation element, and obtain at least one recommendation element. Generate recommendation data based on the at least one recommendation element. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate one or more embodiments of the present specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative labor. Figure 1 A schematic diagram of an implementation environment of a service recommendation processing method provided in one or more embodiments of this specification; Figure 2 A processing flow chart of a service recommendation processing method provided in one or more embodiments of this specification; Figure 3 A processing flow chart of a service recommendation processing method applied to a service recommendation scenario provided by one or more embodiments of this specification; Figure 4 A schematic diagram of an embodiment of a service recommendation processing device provided in one or more embodiments of this specification; Figure 5 A schematic diagram of the structure of a service recommendation processing device provided in one or more embodiments of this specification. DETAILED DESCRIPTION

[0008] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0009] The service recommendation processing method provided in one or more embodiments of this specification can be applied to the implementation environment of client service recommendation. Figure 1 , the implementation environment at least includes: State feature extraction module 101, decision processing module 102 and data assembly module 103; The state feature extraction module 101 is used to extract the user state features from the service data of the accessing user to obtain the state features of each user; The decision processing module 102 is used to select one or more user status features from each user status feature and input them into the corresponding decision sub-model to perform decision processing of service recommendation elements based on the obtained user status features, so as to obtain corresponding recommendation elements; the data assembly module 103 is used to generate recommendation data according to the decision result of the decision processing module 102; In this process, the decision processing module 102 can select text data in the text library 104 when performing decision processing of service recommendation elements, and accordingly, the data assembly module 103 can generate text recommendation data according to the text data selected by the decision processing module 102 in the text library 104; the decision processing module 102 can also select service categories in the service library 105, and accordingly, the data assembly module 103 can generate service recommendation data according to the service categories selected by the decision processing module 102 in the service library 105; In addition, the implementation environment may further include a service status positioning module 106, which is used to locate the service status of the accessing user based on the service data of the accessing user, and obtain corresponding processing that represents the service status of the accessing user.

[0010] In this implementation environment, in the process of recommending services to visiting users, based on the service data of the visiting users, the state feature extraction module 101 first inputs the service data into each state extraction module to extract the user state features, and obtains the state features of each user. On this basis, the decision processing module 102 selects at least one decision sub-model corresponding to the feature input from the obtained state features of each user to perform decision processing on the service recommendation element, and obtains at least one recommendation element. The data assembly module 103 generates recommendation data based on the obtained at least one recommendation element, so as to recommend services to the visiting users based on the generated recommendation data.

[0011] One or more embodiments of a service recommendation processing method provided in this specification are as follows: Reference Figure 2 The service recommendation processing method provided in this embodiment specifically includes steps S202 to S208.

[0012] Step S202: Acquire service data of the user accessing the service.

[0013] The service described in this embodiment refers to the service currently accessed by the accessing user, which may be various services provided within the application, such as the guarantee service provided within the application (sales service for guarantee projects, guarantee application service for guarantee projects); optionally, the service may be an interactive service configured with a virtual digital person, or the service may be configured with a virtual digital person for service interaction, and the accessing user may realize dialogue interaction such as service issues and service recommendations by interacting with the virtual digital person during the service access process.

[0014] The service data of the accessing user may be identity-related data of the accessing user himself, as well as data related to the accessing user during the service access process.

[0015] Optionally, the service data includes at least one of the following: user identity data, user portrait data, user behavior records, service interaction records, real-time interaction data, historical interaction data, and user access nodes; wherein the user portrait data may be data or data sets describing user characteristics in one or more dimensions; the user behavior record may be a behavior record of the accessing user in the application, or a behavior record of the accessing user for the service in the application; the service interaction record may be an interaction record obtained by the accessing user during the process of accessing one or more services in the application, or an interaction record obtained by the accessing user during the process of accessing the service in the application; Real-time interaction data refers to the interaction data for an application or services within an application obtained during the current access process; historical interaction data refers to the interaction data for an application or services within an application obtained and stored during previous access processes; user access nodes refer to the access nodes of users during the process of accessing an application or service, such as access nodes for security services, including nodes for product awareness, product understanding, purchase decisions, usage experience, and after-sales service.

[0016] In specific implementation, the service data of the accessing user of the service can be obtained from the client, and the service data can also be obtained from the server. Alternatively, the client can obtain the service data from the server, such as the client calling a data acquisition interface to obtain the service data of the accessing user from the server. Alternatively, the service data can also be composed of the service data of the accessing user obtained from the client and the service data obtained from the server.

[0017] It should be pointed out that, considering that the service data or any one or more data contained in the service data (such as user identity data, user portrait data) belong to the privacy of the accessing user to a certain extent, if you want to obtain the service data of the accessing user or any one or more data contained in the service data, you can obtain the accessing user's authorization before obtaining the service data or any one or more data contained in the service data. The method of authorizing the accessing user can be to send a data authorization reminder to the accessing user during the process of obtaining the service data or any one or more data contained in the service data. After the accessing user confirms the data authorization reminder through an instruction, the service data or any one or more data contained in the service data can be obtained. In addition, the data authorization reminder can also be sent to the accessing user through other methods, and the authorization to obtain the service data or any one or more data contained in the service data can be obtained after the user confirms the instruction.

[0018] In actual applications, in the process of recommending services to visiting users, the service status data of the visiting users can also be obtained based on the service data of the visiting users, so as to locate the service status of the visiting users in the service recommendation scenario through the service status data, so as to perceive the status of the visiting users more comprehensively and deeply in the service recommendation process, and realize the refined perception of the service status of the visiting users.

[0019] Specifically, after obtaining the service data of the accessing user, the service status data of the accessing user may be determined based on the service data, that is, the service status of the accessing user may be located based on the service data. In an optional implementation manner provided by this embodiment, determining the service status data of the accessing user based on the service data includes at least one of the following operations: Label the user portrait data contained in the service data to obtain user portrait labels; Performing grouping of service status according to user identity data and user behavior records contained in the service data to obtain service status grouping of visiting users; Determine the user access node based on the user behavior record contained in the service data and determine the access path based on the user access node; The service interaction records of accessing users are extracted from the service data and stored.

[0020] Among them, in the process of labeling the user portrait data contained in the service data, the corresponding user portrait labels can be obtained by processing the service data and labeling the user portrait data based on the data processing. Furthermore, the service data can be input into the portrait generation model to label the user portrait data, and the user portrait labels output by the portrait generation model can be obtained. In the process of grouping the service status of visiting users, based on multiple pre-set service status groups (such as consultation status group, repeat purchase status group, purchase status group, etc.), starting from the user identity data and user behavior records contained in the service data, the matching degree between the user identity data (such as occupation, date of birth, geographic location, income level, purchase history, behavior pattern, etc.) and user behavior records and each service status group can be calculated, and the service status group with a greater matching degree than other service status groups is selected as the service status group of the visiting user; In the process of determining the user access nodes and access paths of the accessing user, starting from the user behavior records contained in the service data, the access nodes of the accessing user in the service (such as product awareness, product understanding, purchase decision, user experience and after-sales service nodes) are determined, and based on the determined access nodes, the access path of the accessing user can be determined according to the access nodes of the accessing user; In the process of extracting the service interaction records of the visiting users from the service data, the interaction records related to the service interaction with the visiting users can be extracted from the service data, such as extracting the visiting users' customer service records, complaint records, security application records and evaluation records from the service data.

[0021] It should be noted that, among the four methods for determining the service status data of accessing users provided above, any one of them can be selected for implementation according to actual needs during the specific implementation process, or, at least two of the four methods can be combined to form a new implementation method according to actual needs.

[0022] Step S204: input the service data into each state extraction module to extract user state features, and obtain each user state feature.

[0023] In specific implementation, based on obtaining the service data of the accessing user, the service data is input into one or more state extraction modules to extract user state features, thereby obtaining the state features of each user; wherein, the state extraction module refers to a feature extraction module used to extract state features of the service data of the accessing user, and the state extraction module may use a corresponding machine learning network for feature extraction to extract state features according to the different extracted state features, such as extracting state features by using a state extraction module of a neural network, or extracting state features by using a machine learning model or configuring a model to call an interface of the state extraction module, such as extracting state features by calling a large language model.

[0024] This embodiment provides four state extraction modules for extracting user state features: an intention extraction module for extracting interaction intention features, a behavior extraction module for extracting behavior features, a path extraction module for extracting access path features, and an emotion extraction module for extracting service emotion features. The following specifically describes the four processes of extracting user state features.

[0025] (1) Intent Extraction Module In an optional implementation provided by this embodiment, the service data is input into each state extraction module respectively to extract user state features to obtain each user state feature, including: inputting the service data into the intention extraction module, extracting interaction intention features based on the real-time interaction data and historical interaction data contained in the service data, and obtaining user intention features.

[0026] In order to improve the feature extraction efficiency of interaction intent feature extraction, and also to improve the accuracy and effectiveness of the extracted interaction intent features, the intent extraction module may also adopt a corresponding machine learning network or machine learning model to extract interaction intent features. Optionally, the intent extraction module includes an intent feature extraction model using a neural network (such as a recurrent neural network (RNN) or a long short-term memory network (LSTM)), or a large language model for intent feature extraction.

[0027] (2) Behavior Extraction Module In an optional implementation provided by this embodiment, the service data is input into each state extraction module to extract user state features to obtain each user state feature, including: inputting the service data into the behavior extraction module to extract behavior features to obtain user behavior features.

[0028] Specifically, during the process of extracting behavior features, the behavior extraction module may adopt the following optional implementation method to extract behavior features: determine the user behavior sequence according to the user behavior records included in the service data, and extract behavior features based on the user behavior sequence.

[0029] Furthermore, in the process of extracting behavior features by the behavior extraction module, in order to improve the feature extraction efficiency of the behavior feature extraction, and also to improve the accuracy and effectiveness of the extracted user behavior features, the behavior extraction module may also use a corresponding machine learning network to extract behavior features. Optionally, the behavior extraction module includes an intent feature extraction model using a neural network based on an attention mechanism, such as an intent feature extraction model using a recurrent neural network (CNN, Convolutional Neural Networks) using an attention mechanism.

[0030] (3) Path extraction module In an optional implementation provided by this embodiment, the service data is respectively input into each state extraction module to extract user state features to obtain each user state feature, including: inputting the service data into the path extraction module to extract access path features to obtain access path features; Specifically, during the process of extracting access path features, the path extraction module may adopt the following optional implementation method to extract access path features: determine the access path based on the user access nodes contained in the service data, and extract features of the key access nodes determined based on the access path to obtain access path features.

[0031] Furthermore, in the process of access path feature extraction performed by the path extraction module, in order to improve the feature extraction efficiency of the access path feature extraction, and also to improve the accuracy and effectiveness of the extracted access path features, the path extraction module may also adopt a corresponding machine learning network to extract access path features. Optionally, the path extraction module includes a path extraction model that uses a graph neural network and a path prediction algorithm, such as a path extraction model that uses a graph neural network (GNN, Graph NeuralNetwork) and a path prediction algorithm.

[0032] (4) Emotion extraction module In an optional implementation provided by this embodiment, the service data is input into each state extraction module to extract user state features to obtain each user state feature, including: inputting the service data into the emotion extraction module to extract service emotion features to obtain user emotion features.

[0033] The machine learning network or machine learning model used by the four state extraction modules provided above can be obtained after model training. In order to improve the efficiency and effectiveness of model training, distributed learning can be used for model training. In addition, other training methods can be used for model training according to actual needs, such as federated learning for model training. In the specific implementation process, the four state extraction modules provided above can be deployed on the server side, and can also be deployed on the client side, or at least one of the four state extraction modules is deployed on the server side and at least one of the four state extraction modules is deployed on the client side; Among them, model training can be performed on the server side, and after the training is completed, the state extraction module using the trained machine learning network or machine learning model can be deployed on the server side, and the state extraction module using the trained machine learning network or machine learning model can be deployed on the client side; in addition, model training can also be performed on the client side, and after the training is completed, the state extraction module using the trained machine learning network or machine learning model can be deployed on the server side, and the state extraction module using the trained machine learning network or machine learning model can be deployed on the client side.

[0034] Furthermore, when at least one state extraction module using the machine learning network or machine learning model obtained after training is deployed on the client, in order to reduce the operating pressure of the terminal device that carries or runs the client, the machine learning network or machine learning model obtained after the above model training can be compressed, so that the state extraction module using the compressed machine learning network or machine learning model can be deployed on the client, such as by pruning (removing unimportant weights in the network or model), quantization (reducing the accuracy of weights) and / or knowledge distillation (using a smaller model to imitate a large model).

[0035] It should be noted that the four state extraction modules provided above and the corresponding four implementation methods of respectively inputting the service data into the state extraction modules for user state feature extraction can be implemented according to actual needs in the specific implementation process. Any one or more state extraction modules and one or more implementation methods of corresponding state extraction modules for user state feature extraction can be selected for implementation. For example, the service data is respectively input into each state extraction module for user state feature extraction to obtain each user state feature, including: inputting the service data into the intention extraction module, and performing interaction intention feature extraction based on the real-time interaction data and historical interaction data contained in the service data to obtain the user intention feature; inputting the service data into the emotion extraction module to perform service emotion feature extraction to obtain the user emotion feature; or, inputting the service data into each state extraction module for user state feature extraction to obtain each user state feature, including: inputting the service data into the intention extraction module, and performing interaction intention feature extraction based on the real-time interaction data and historical interaction data contained in the service data to obtain the user intention feature; inputting the service data into the path extraction module for access path feature extraction to obtain access path feature; inputting the service data into the emotion extraction module for service emotion feature extraction to obtain the user emotion feature; It should also be noted that the above four implementation methods can also be combined as a whole of any two, three or four according to actual needs, and corresponding technical means can be selected from the four implementation methods according to actual needs to combine to form a new implementation method. For example, the service data can be input into each state extraction module for user state feature extraction to obtain each user state feature, including: inputting the service data into the state extraction model for interaction intention feature extraction and service emotion feature extraction to obtain user intention features and user emotion features; for another example, inputting the service data into the state extraction module for interaction intention feature extraction, behavior feature extraction, access path feature extraction and / or service emotion feature extraction to obtain user intention features, user behavior features, access path features and / or user emotion features.

[0036] Step S206 , selecting at least one decision sub-model corresponding to the feature input from among the user status features to perform decision processing on a service recommendation element, and obtaining at least one recommendation element.

[0037] In specific implementation, based on extracting user state features to obtain each user state feature, features are selected from each user state feature, and the selected features are input into at least one decision sub-model for decision processing of service recommendation elements to obtain at least one corresponding recommendation element.

[0038] Among them, the recommendation factor refers to the factor used to generate the recommendation data recommended to the visiting user in the process of recommending services to the visiting user. Optionally, the recommendation factor includes at least one of the following: the recommended copy, the service category, the recommendation time, the user intention, and the user interest. The decision sub-model refers to the decision model used to make service factor decisions based on the input user status characteristics; in the process of the decision sub-model making service factor decisions, according to the different input user status characteristics or the different service factors for decision making, the corresponding feature extraction machine learning network can be used to make the corresponding service factor decision.

[0039] The present embodiment provides four decision sub-models for making decisions on service recommendation elements: a text decision sub-model for making decisions on text to be recommended, a service decision sub-model for making decisions on service categories, a recommendation time decision sub-model for making decisions on recommendation time, and a status decision sub-model for making decisions on service status. The following specifically describes the process of making decisions on service recommendation elements using these four decision sub-models.

[0040] (1) Decision-making process for recommended documents Four state extraction modules for extracting user state features are provided above. Accordingly, four user state features are obtained by extracting user state features through the four state extraction modules: user intention feature, user behavior feature, access path feature, and user emotion feature. Based on the four user state features obtained above, any one or more user state features can be selected from the four user state features and input into the corresponding decision sub-model for decision processing of service recommendation elements. Here, in the process of making a decision on the recommended copy, a feature can be selected from the above-mentioned user state features to be input into the copy decision sub-model for copy decision, that is, one or more of the four user state features, namely, user intention feature, user behavior feature, access path feature, and user emotion feature, can be selected to be input into the copy decision sub-model for copy decision; Specifically, in an optional implementation manner provided by this embodiment, at least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing on a service recommendation element, and at least one recommendation element is obtained, including: Selecting user intention features, user behavior features, access path features, and user emotion features from among the user status features; The selected features are input into the copy decision sub-model to make copy decisions and obtain the copy to be recommended.

[0041] Optionally, the copy decision includes: selecting from the copy library the copy to be recommended that matches the visiting user or the service status group to which the visiting user belongs, display parameters and / or illustrations of the copy to be recommended; in addition, the copy decision may also be generating the copy to be recommended that matches the service status group to which the visiting user or the user being asked belongs, determining the display parameters of the copy to be recommended and generating illustrations of the copy to be recommended.

[0042] Among them, the display parameters of the copy to be recommended include copy style, style, color and / or size and other parameters related to copy display; the picture of the copy to be recommended can be the background picture of the copy to be recommended, or it can be an illustration in the copy content, and the picture can be generated by an image generation model.

[0043] In actual applications, in the process of recommending services to visiting users, there is also a need to recommend rights and interests to visiting users, so as to improve the recommendation conversion rate by combining rights and interests recommendations on the basis of copy recommendations. In view of this, in the process of recommending copy to visiting users on the basis of copy decisions, rights and interests can also be recommended to visiting users, such as recommending discount coupons, membership privileges or gift information to visiting users. Accordingly, here, rights and interests recommendation decisions can also be made in the process of making copy decisions. Optionally, copy decisions include: determining the rights and interests to be recommended to visiting users or service status groups; updating the rights and interests description information in the copy to be recommended based on the rights and interests, or writing the rights and interests description information into the copy to be recommended.

[0044] (2) Decision-making process for service categories In the process of service category decision processing, features can be selected from the above-obtained user status features and input into the service decision sub-model to make service recommendation decisions, that is, one or more of the four user status features, namely, user intention features, user behavior features, access path features, and user emotion features, can be selected and input into the service decision sub-model to make service recommendation decisions; Specifically, in an optional implementation manner provided by this embodiment, at least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing on a service recommendation element, and at least one recommendation element is obtained, including: Selecting user intention features, user behavior features, access path features and / or user emotion features from among the user status features; The selected features are input into the service decision sub-model to make service recommendation decisions and obtain the service category.

[0045] The service category refers to the category of the service recommended in the service recommendation process. Optionally, the service category includes at least one of the following: product recommendation category (service category for product recommendation to visiting users), service interaction category (service category for service interaction recommendation to users), message push category (service category for message push to visiting users).

[0046] In an optional implementation manner provided by this embodiment, the service recommendation decision includes: determining user preferences based on user behavior characteristics, access path characteristics and / or user emotional characteristics; selecting a service category in the service library that matches the user intention characteristics and user preferences; or selecting a service category in the service library that matches the user behavior characteristics, access path characteristics and / or user emotional characteristics.

[0047] (3) Decision-making process for recommendation time In the process of making a decision on the recommendation time, a feature can be selected from the above-obtained user status features and input into the recommendation time decision sub-model to make a recommendation time decision, that is, one or more of the four user status features, namely, user intention feature, user behavior feature, access path feature, and user emotion feature, can be selected and input into the recommendation time decision sub-model to make a recommendation time decision; Specifically, in an optional implementation provided by this embodiment, at least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing on the service recommendation element to obtain at least one recommendation element, including: inputting user intention features, user behavior features and / or access path features into the recommendation time decision sub-model to make a recommendation time decision to obtain the recommendation time.

[0048] (4) Decision-making process for service status In the process of service status decision processing, a feature input status decision sub-model can be selected from the above-obtained user status features to make a service status decision, that is, one or more input status decision sub-models can be selected from the four user status features of user intention features, user behavior features, access path features, and user emotion features to make a service recommendation decision; Specifically, in an optional implementation provided by this embodiment, at least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing on the service recommendation element to obtain at least one recommendation element, including: inputting user intention features and user behavior features into the state decision sub-model to perform user intention recognition and user interest prediction to obtain user intention and user interest.

[0049] The four decision sub-models provided above can be obtained after model training. In order to improve the efficiency and effectiveness of model training, distributed learning can be used for model training. In addition, other training methods can be used for model training according to actual needs, such as federated learning for model training. In the specific implementation process, the four decision sub-models provided above can be deployed on the server side or on the client side, or at least one of the four decision sub-models is deployed on the server side and at least one of the four decision sub-models is deployed on the client side. Among them, model training can be performed on the server side, and after the training is completed, the machine learning network decision sub-model obtained by training will be deployed on the server side, and the decision sub-model obtained by training will be deployed to the client side; in addition, model training can also be performed on the client side, and after the training is completed, the decision sub-model obtained by training will be deployed to the server side, and the decision sub-model obtained by training will be deployed to the client side.

[0050] Furthermore, when the decision sub-model obtained after training is deployed on the client, in order to reduce the operating pressure of the terminal device that carries or runs the client, the decision sub-model obtained after the above model training can be compressed, so that the compressed decision sub-model can be deployed on the client, such as by pruning (removing unimportant weights in the network or model), quantization (reducing the accuracy of weights) and / or knowledge distillation (using a smaller model to imitate a large model).

[0051] It should be noted that the four decision sub-models and the corresponding four implementation methods for decision processing of service recommendation elements provided above can be implemented by selecting any one or more decision sub-models and the corresponding implementation methods for decision processing of service recommendation elements according to actual needs. For example, at least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing of the service recommendation element, and at least one recommendation element is obtained, including: Select user intention features, user behavior features, access path features and user emotion features from each user status feature; input the selected features into the copy decision sub-model to make a copy decision, and obtain the copy to be recommended; input the user intention features, user behavior features and access path features into the recommendation time decision sub-model to make a recommendation time decision, and obtain the recommendation time; Alternatively, user intention features, user behavior features, access path features and user emotion features are selected from each user status feature; the selected features are input into the service decision sub-model to make a service recommendation decision and obtain a service category; the user intention features, user behavior features and access path features are input into the recommendation time decision sub-model to make a recommendation time decision and obtain a recommendation time.

[0052] For another example, selecting at least one decision sub-model corresponding to the feature input from each user state feature to perform decision processing on a service recommendation element to obtain at least one recommendation element includes: Select user intention features, user behavior features, access path features and user emotion features from each user status feature; input the selected features into the copy decision sub-model to make a copy decision, and obtain the copy to be recommended; input the user intention features and user behavior features into the state decision sub-model to identify user intention and predict user interest, and obtain user intention and user interest; Alternatively, user intention features, user behavior features, access path features and user emotion features are selected from each user state feature; the selected features are input into the service decision sub-model to make service recommendation decisions and obtain service categories; user intention features and user behavior features are input into the state decision sub-model to identify user intentions and predict user interests and obtain user intentions and user interests.

[0053] For another example, selecting at least one decision sub-model corresponding to the feature input from each user status feature to perform decision processing on a service recommendation element to obtain at least one recommendation element includes: Selecting user intention features, user behavior features, access path features, and user emotion features from among the user status features; Input the selected features (user intention features, user behavior features, access path features, and user emotion features) into the copy decision sub-model to make copy decisions and obtain the copy to be recommended; Input the selected features (user intention features, user behavior features, access path features, and user emotion features) into the service decision sub-model to make service recommendation decisions and obtain service categories; Input the user intention features, user behavior features and access path features into the recommendation time decision sub-model to make a recommendation time decision and obtain the recommendation time; or, Selecting user intention features, user behavior features, access path features, and user emotion features from among the user status features; Input the selected features (user intention features, user behavior features, access path features, and user emotion features) into the copy decision sub-model to make copy decisions and obtain the copy to be recommended; Input the user intention features, user behavior features and access path features into the recommendation time decision sub-model to make a recommendation time decision and obtain the recommendation time; The user intention features and user behavior features are input into the state decision sub-model to perform user intention recognition and user interest prediction to obtain user intention and user interest.

[0054] It should also be noted that the above four implementation methods can also be combined as a whole of any two, three or four according to actual needs, and corresponding technical means can be selected from the four implementation methods according to actual needs to form a new implementation method. For example, in an optional implementation provided by this embodiment, at least one decision sub-model corresponding to the feature input is selected from each user state feature to perform decision processing on the service recommendation element to obtain at least one recommendation element, including: selecting a feature from each user state feature (such as selecting a user intention feature, a user behavior feature and an access path feature), and inputting the selected feature into a copy decision sub-model to perform copy decision to obtain a copy to be recommended; inputting the selected feature into a service decision sub-model to perform service recommendation decision to obtain a service category; Alternatively, features are selected from each user status feature (such as user intention features, user behavior features, and access path features); the selected features are input into the copy decision sub-model to make a copy decision to obtain the copy to be recommended, and the selected features are input into the recommendation time decision sub-model to make a recommendation time decision to obtain the recommendation time.

[0055] In addition, in another optional implementation provided by this embodiment, at least one decision sub-model corresponding to the feature input is selected from each user status feature to perform decision processing on the service recommendation element to obtain at least one recommendation element, including: selecting a feature from each user status feature, and using the selected feature service decision sub-model to perform service recommendation decision to obtain a service category; and inputting the selected feature into a recommendation time decision sub-model to perform recommendation time decision to obtain a recommendation time.

[0056] Step S208: Generate recommendation data according to the at least one recommendation element.

[0057] In this embodiment, based on the above-mentioned decision-making processing of the service recommendation elements to obtain at least one recommendation element, recommendation data is generated according to the at least one recommendation element obtained, so as to obtain recommendation data recommended to the visiting user; further, the recommendation data can also be recommended to the visiting user. Specifically, in the process of recommending the recommendation data to the visiting user, the recommendation data can also be recommended to the visiting user through the virtual digital person configured by the service. For example, the recommendation data can be recommended to the visiting user in the form of a dialogue as the dialogue content of the virtual digital person, so as to enhance the interest of the service recommendation and help improve the user conversion rate of the service recommendation.

[0058] During the specific execution process, in the above-mentioned process of selecting features from each user status feature and inputting the selected features into the copy decision sub-model for copy decision, on the basis of obtaining the copy to be recommended, the display parameters of the copy to be recommended, and the rights and interests description information in the accompanying pictures and / or the recommended copy, in the process of generating recommendation data according to at least one recommendation element, the recommendation data can be generated according to at least one of the copy to be recommended, the display parameters of the copy to be recommended, and the accompanying pictures, and the rights and interests description information in the recommended copy. Specifically, in an optional implementation manner provided by this embodiment, the recommendation data is generated according to at least one recommendation element, including: generating copy recommendation data based on the copy to be recommended and at least one of the display parameters, the accompanying pictures and the rights and interests description information.

[0059] Based on the above-mentioned selecting a feature from each user state feature and inputting the selected feature into the service decision sub-model to make a service recommendation decision to obtain a service category, in the process of generating recommendation data according to at least one recommendation element, the recommendation data may be generated according to a data generation method corresponding to the service category. In an optional implementation manner provided by this embodiment, generating recommendation data according to at least one recommendation element includes: generating recommendation data according to a data generation method corresponding to the service category to obtain service recommendation data; Among them, for product recommendation categories, recommendation data is generated according to the data generation method corresponding to the service category, and the following optional implementation methods are adopted: service products are selected according to a preset selection method or a selection method of the visiting user, and a recommendation page or recommendation message for the selected target service product is generated; for example, the current selection method is determined among the three pre-set selection methods of popular selection, hot selection, and interactive selection, and service products are selected according to the determined selection method, and a recommendation page or recommendation message for the selected target service product is generated.

[0060] For service interaction categories, recommendation data is generated in accordance with the data generation method corresponding to the service category, and the following optional implementation method is used to achieve it: service interaction data of service products visited by visiting users is generated; for example, the service interaction data of service products may be product interpretation, product Q&A and / or user retention-related interaction data. In the process of generating service interaction data, service interaction data may be generated through natural language generation and speech synthesis technology. Product descriptions or usage guides, product function demonstrations, operation tutorials or maintenance suggestions may be generated as service interaction data recommended to visiting users.

[0061] For the message push category, recommendation data is generated according to the data generation method corresponding to the service category, and the following optional implementation method is adopted: generate push messages for service products associated with the visiting user; for example, generate renewal reminders for purchased products that the visiting user has purchased, or generate payment reminders for products that the visiting user is to pay for, etc., wherein the renewal reminders and payment reminders may include renewal discount information and service change notifications, etc. The push of renewal reminders and payment reminders can be pushed through one or more channels such as SMS, email and / or in-application message push.

[0062] In addition, in the process of generating recommendation data based on at least one recommendation element, a data relationship between the generated recommendation data can be established, such as an access trigger relationship between the recommendation data. A data relationship between the recommendation element and the generated recommendation data can also be established in the process of generating recommendation data, such as a recommendation trigger relationship between the recommendation time and the recommendation data, that is, the recommendation data is recommended when the recommendation time arrives.

[0063] Specifically, in an optional implementation manner provided by this embodiment, generating recommendation data according to at least one recommendation factor includes: Generate copy recommendation data based on the copy to be recommended and at least one of the display parameters, the accompanying pictures and the rights description information; Generate service recommendation data corresponding to the service category, and establish an access trigger relationship between the service recommendation data and the copy recommendation data.

[0064] Optionally, the copy recommendation data is displayed through a window message or a page message, and the service recommendation data is displayed after the window message or the page message is triggered.

[0065] In another optional implementation manner provided by this embodiment, after the step of generating recommendation data according to at least one recommendation element is executed, it also includes: recommending text recommendation data or service recommendation data to the visiting user according to the recommendation time.

[0066] In summary, the service recommendation processing method provided in this embodiment, based on obtaining the service data of the visiting user, inputs the service data into one or more state extraction modules to extract the user state features, obtains the state features of each user, further selects features from the state features of each user, and inputs the selected features into at least one decision sub-model to perform decision processing of service recommendation elements, obtains at least one corresponding recommendation element, and generates recommendation data based on the obtained at least one recommendation element, thereby obtaining recommendation data recommended to the visiting user, thereby realizing multi-dimensional state feature extraction through multiple state extraction modules, and performing decision processing of service recommendation elements and recommendation data generation through multiple decision sub-models based on the extracted multi-dimensional state features, thereby improving the efficiency and comprehensiveness of recommendation data generation; The state extraction module can also use the corresponding machine learning network or machine learning model to extract state features, which improves the feature extraction efficiency of interaction intention feature extraction, and also improves the accuracy and effectiveness of the extracted interaction intention features; After generating and obtaining the recommendation data, in the process of recommending the recommendation data to the visiting user, the virtual digital person configured by the service can also recommend the recommendation data to the visiting user. For example, the recommendation data can be used as the conversation content of the virtual digital person to recommend the recommendation data to the visiting user in the form of a conversation, so as to enhance the fun of the service recommendation and help improve the user conversion rate of the service recommendation; In addition, in the process of recommending services to visiting users, the service status data of the visiting users can also be obtained based on the service data of the visiting users, so as to locate the service status of the visiting users in the service recommendation scenario through the service status data, so as to perceive the status of the visiting users more comprehensively and deeply in the service recommendation process, and realize the refined perception of the service status of the visiting users.

[0067] The following takes the application of a service recommendation processing method provided in this embodiment in a service recommendation scenario as an example. Figure 3 , the service recommendation processing method provided in this embodiment is further described, see Figure 3 , a service recommendation processing method applied to a service recommendation scenario specifically includes the following steps.

[0068] Step S302: Obtain service data of the user accessing the service.

[0069] Step S304: input the service data into a plurality of state extraction modules respectively to extract user state features, and obtain a plurality of user state features.

[0070] Optionally, the state extraction module includes an intention extraction module, a behavior extraction module, a path extraction module, and an emotion extraction module; the service data is input into the intention extraction module to perform interaction intention feature extraction to obtain user intention features, the service data is input into the behavior extraction module to perform behavior feature extraction to obtain user behavior features, the service data is input into the path extraction module to perform access path feature extraction to obtain access path features, and the service data is input into the emotion extraction module to perform service emotion feature extraction to obtain user emotion features.

[0071] Step S306, selecting a feature from a plurality of user status features, and inputting the selected feature into a copy decision sub-model to perform copy decision to obtain a copy to be recommended.

[0072] Step S308: input the selected features into the service decision sub-model to make a service recommendation decision to obtain a service category.

[0073] Step S310: Input the selected features into the recommendation time decision sub-model to make a recommendation time decision to obtain the recommendation time.

[0074] Step S312, generating text recommendation data based on the text to be recommended.

[0075] Step S314: Generate service recommendation data corresponding to the service category, and establish an access trigger relationship between the service recommendation data and the copy recommendation data.

[0076] Step S316, recommending copy recommendation data to the visiting user according to the recommendation time.

[0077] Optionally, the copy recommendation data is displayed through a window message or a page message, and the service recommendation data is displayed after the window message or the page message is triggered.

[0078] It should be noted that any one of steps S302 to S316 or any combination of multiple steps can be combined with any one of steps S202 to S208 to form a new implementation method according to the needs of implementation deployment; in addition, according to the needs of actual deployment, any one or multiple technical features can be selected from steps S302 to S316 and combined with any one or multiple technical features provided by steps S202 to S208 to form a new implementation method; or, any one or multiple technical features in steps S302 to S316 can be replaced with any one or multiple technical features provided by steps S202 to S208 to form a new implementation method according to the needs of actual deployment, which will not be repeated here.

[0079] An embodiment of a service recommendation processing device provided in this specification is as follows: In the above-mentioned embodiment, a service recommendation processing method is provided, and correspondingly, a service recommendation processing device is also provided, which is described below with reference to the accompanying drawings.

[0080] Reference Figure 4 , which shows a schematic diagram of an embodiment of a service recommendation processing device provided in this embodiment.

[0081] Since the device embodiment corresponds to the method embodiment, the description is relatively simple, and the relevant parts can refer to the corresponding description of the method embodiment provided above. The device embodiment described below is only illustrative.

[0082] This embodiment provides a service recommendation processing device, the device comprising: A service data acquisition module 402 is configured to acquire service data of a user accessing a service; The state feature extraction module 404 is configured to input the service data into each state extraction module to extract the user state feature and obtain the state feature of each user; A decision processing module 406 is configured to select at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtain at least one recommendation element; The recommendation data generating module 408 is configured to generate recommendation data according to the at least one recommendation element.

[0083] An embodiment of a service recommendation processing device provided in this specification is as follows: Corresponding to the service recommendation processing method described above, based on the same technical concept, one or more embodiments of this specification also provide a service recommendation processing device, which is used to execute the service recommendation processing method provided above. Figure 5 A schematic diagram of the structure of a service recommendation processing device provided in one or more embodiments of this specification.

[0084] This embodiment provides a service recommendation processing device, including: like Figure 5 As shown, the service recommendation processing device may have relatively large differences due to different configurations or performances, and may include one or more processors 501 and a memory 502, and the memory 502 may store one or more storage applications or data. Among them, the memory 502 may be a short-term storage or a persistent storage. The application stored in the memory 502 may include one or more modules (not shown in the figure), and each module may include a series of computer executable instructions in the service recommendation processing device. Furthermore, the processor 501 may be configured to communicate with the memory 502 and execute a series of computer executable instructions in the memory 502 on the service recommendation processing device. The service recommendation processing device may also include one or more power supplies 503, one or more wired or wireless network interfaces 504, one or more input / output interfaces 505, one or more keyboards 506, etc.

[0085] In a specific embodiment, the service recommendation processing device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer executable instructions for the service recommendation processing device, and the one or more programs are configured to be executed by one or more processors, including computer executable instructions for performing the following: Obtain service data of users accessing the service; Input the service data into each state extraction module to extract user state features, and obtain each user state feature; Selecting at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtaining at least one recommendation element; Recommendation data is generated according to the at least one recommendation element.

[0086] An embodiment of a computer-readable storage medium provided in this specification is as follows: Corresponding to the service recommendation processing method described above, based on the same technical concept, one or more embodiments of this specification also provide a computer-readable storage medium.

[0087] The computer-readable storage medium provided in this embodiment is used to store computer-executable instructions, and the computer-executable instructions implement the following process when executed: Obtain service data of users accessing the service; Input the service data into each state extraction module to extract user state features, and obtain each user state feature; Selecting at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtaining at least one recommendation element; Recommendation data is generated according to the at least one recommendation element.

[0088] It should be noted that an embodiment of a computer-readable storage medium in this specification and an embodiment of a service recommendation processing method in this specification are based on the same inventive concept, so the specific implementation of this embodiment can refer to the implementation of the aforementioned corresponding method, and the repeated parts will not be repeated.

[0089] An embodiment of a computer program product provided in this specification is as follows: Corresponding to the service recommendation processing method described above, based on the same technical concept, one or more embodiments of this specification also provide a computer program product.

[0090] A computer program product, comprising a computer program / instruction, which, when executed by a processor, implements the following steps: Obtain service data of users accessing the service; Input the service data into each state extraction module to extract user state features, and obtain each user state feature; Selecting at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtaining at least one recommendation element; Recommendation data is generated according to the at least one recommendation element.

[0091] It should be noted that an embodiment of a computer program product in this specification and an embodiment of a service recommendation processing method in this specification are based on the same inventive concept, so the specific implementation of this embodiment can refer to the implementation of the aforementioned corresponding method, and the repeated parts will not be repeated.

[0092] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referenced to each other. Each embodiment focuses on the differences from other embodiments. For example, the device embodiment, equipment embodiment, and computer-readable storage medium embodiment are similar to the method embodiment, so the description is relatively simple. To read the relevant content in the device embodiment, equipment embodiment, and computer-readable storage medium embodiment, please refer to the partial description of the method embodiment.

[0093] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0094] In the 1930s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to circuit structures such as diodes, transistors, switches, etc.) or software improvements (improvements to method flows). However, with the development of technology, many of today's method flow improvements can be regarded as direct improvements to hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using hardware entity modules. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask chip manufacturers to design and produce dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0095] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, and the memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.

[0096] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0097] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0098] It should be understood by those skilled in the art that one or more embodiments of the present specification may be provided as a method, system or computer program product. Therefore, one or more embodiments of the present specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present specification may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0099] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0100] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0101] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0102] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0103] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0104] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0105] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of further restrictions, the elements defined by the sentence "includes at least one ..." do not exclude the presence of other identical elements in the process, method, commodity or device including the elements.

[0106] One or more embodiments of the present specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0107] The above description is only an embodiment of this document and is not intended to limit this document. For those skilled in the art, this document may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this document should be included in the scope of the claims of this document.

Claims

1. A service recommendation processing method, comprising: Obtain service data of users accessing the service; Input the service data into each state extraction module to extract user state features, and obtain each user state feature; Selecting at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtaining at least one recommendation element; Recommendation data is generated according to the at least one recommendation element.

2. The service recommendation processing method according to claim 1, wherein the service data is input into each state extraction module to extract user state features to obtain each user state feature, including: Inputting the service data into an intention extraction module, extracting interaction intention features based on real-time interaction data and historical interaction data contained in the service data, and obtaining user intention features; Wherein, the intention extraction module includes an intention feature extraction model using a neural network, or a large language model for intention feature extraction.

3. The service recommendation processing method according to claim 1, wherein the service data is input into each state extraction module to extract user state features to obtain each user state feature, including: Inputting the service data into a behavior extraction module to extract behavior features and obtain user behavior features; The behavior feature extraction includes: determining a user behavior sequence according to the user behavior records included in the service data, and performing behavior feature extraction based on the user behavior sequence.

4. The service recommendation processing method according to claim 1, wherein the service data is input into each state extraction module to extract user state features to obtain each user state feature, including: Inputting the service data into a path extraction module to extract access path features and obtain access path features; The access path feature extraction includes: determining the access path according to the user access nodes included in the service data, and performing feature extraction on the key access nodes determined based on the access path to obtain the access path feature.

5. The service recommendation processing method according to claim 1, wherein the service data is input into each state extraction module to extract user state features to obtain each user state feature, including: The service data is input into the emotion extraction module to extract service emotion features to obtain user emotion features.

6. The service recommendation processing method according to claim 1, wherein the step of selecting at least one decision sub-model corresponding to the feature input in each of the user status features to perform decision processing on the service recommendation element to obtain at least one recommendation element comprises: Selecting user intention features, user behavior features, access path features and user emotion features from among the user status features; The selected features are input into the copy decision sub-model to make copy decisions and obtain the copy to be recommended.

7. The service recommendation processing method according to claim 6, wherein the copywriting decision comprises: Select from the copy library a copy to be recommended that matches the visiting user or the service status group to which the visiting user belongs, and display parameters and accompanying pictures of the copy to be recommended.

8. The service recommendation processing method according to claim 7, wherein the copywriting decision further comprises: Determining the rights and interests recommended to the accessing user or the service status group; The interest description information in the document to be recommended is updated based on the interest, or the interest description information is written into the document to be recommended.

9. The service recommendation processing method according to claim 8, wherein generating recommendation data according to the at least one recommendation element comprises: Generate copy recommendation data based on the copy to be recommended and at least one of the display parameters, the illustrations, and the rights and interests description information.

10. The service recommendation processing method according to claim 1, wherein selecting at least one decision sub-model corresponding to the feature input in each of the user state features to perform decision processing on the service recommendation element to obtain at least one recommendation element comprises: Selecting user intention features, user behavior features, access path features and user emotion features from among the user status features; The selected features are input into the service decision sub-model to make service recommendation decisions and obtain the service category.

11. The service recommendation processing method according to claim 10, wherein the service recommendation decision comprises: Determine user preference according to the user behavior characteristics, the access path characteristics and the user emotion characteristics; A service category matching the user intention feature and the user preference is selected in a service library.

12. The service recommendation processing method according to claim 11, wherein generating recommendation data according to the at least one recommendation element comprises: Generate recommendation data according to the data generation method corresponding to the service category to obtain service recommendation data; The generating of the recommended data according to the data generating method corresponding to the service category includes: If the service category is a product recommendation category, the service product is selected according to the product selection method of the visiting user, and a recommendation page for the selected target service product is generated; If the service category is a service interaction category, generating service interaction data of the service product accessed by the accessing user; If the service category is a message push category, a push message of the service product associated with the accessing user is generated.

13. The service recommendation processing method according to claim 6 or 10, wherein the selecting at least one decision sub-model corresponding to the feature input in each of the user state features to perform decision processing on the service recommendation element to obtain at least one recommendation element further comprises: The user intention feature, the user behavior feature and the access path feature are input into a recommendation time decision sub-model to make a recommendation time decision and obtain a recommendation time.

14. The service recommendation processing method according to claim 13, after the step of generating recommendation data according to the at least one recommendation element is performed, further comprising: Recommending the copy recommendation data or the service recommendation data to the visiting user according to the recommendation time.

15. The service recommendation processing method according to claim 6 or 10, wherein the selecting at least one decision sub-model corresponding to the feature input in each of the user state features to perform decision processing on the service recommendation element to obtain at least one recommendation element further comprises: The user intention feature and the user behavior feature are input into the state decision sub-model to perform user intention recognition and user interest prediction to obtain user intention and user interest.

16. The service recommendation processing method according to claim 1, wherein selecting at least one decision sub-model corresponding to the feature input in each of the user status features to perform decision processing on the service recommendation element to obtain at least one recommendation element comprises: Selecting features from the user status features, and inputting the selected features into the copy decision sub-model to make a copy decision, thereby obtaining a copy to be recommended; The selected features are input into the service decision sub-model to make a service recommendation decision and obtain a service category.

17. The service recommendation processing method according to claim 16, wherein generating recommendation data according to the at least one recommendation element comprises: Generate copy recommendation data based on the copy to be recommended and at least one of the display parameters, the accompanying pictures and the rights and interests description information; Generate service recommendation data corresponding to the service category, and establish an access trigger relationship between the service recommendation data and the text recommendation data.

18. According to the service recommendation processing method of claim 17, the copy recommendation data is displayed through a window message or a page message, and the service recommendation data is displayed after the window message or the page message is triggered.

19. The service recommendation processing method according to claim 1, wherein each state extraction module and / or the at least one decision sub-model is deployed on a client; The at least one decision sub-model is obtained by performing model training and model compression after the training is completed.

20. The service recommendation processing method according to claim 1, after the step of obtaining the service data of the accessing user of the service is performed, at least one of the following operations is performed: Labeling the user portrait data contained in the service data to obtain a user portrait label; Performing grouping of service status according to the user identity data and user behavior records contained in the service data to obtain service status groups of the accessing users; Determine a user access node according to the user behavior record contained in the service data and determine an access path according to the user access node; The service interaction record of the accessing user is extracted from the service data and stored.

21. A service recommendation processing device, comprising: A service data acquisition module, configured to acquire service data of a user accessing the service; A state feature extraction module is configured to input the service data into each state extraction module to extract user state features and obtain each user state feature; A decision processing module is configured to select at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtain at least one recommendation element; The recommendation data generation module is configured to generate recommendation data according to the at least one recommendation element.

22. A service recommendation processing device, comprising: processor; and a memory configured to store computer executable instructions that, when executed, cause the processor to: Obtain service data of users accessing the service; Input the service data into each state extraction module to extract user state features, and obtain each user state feature; Selecting at least one decision sub-model corresponding to the feature input from each of the user status features to perform decision processing on a service recommendation element, and obtaining at least one recommendation element; Recommendation data is generated according to the at least one recommendation element.

23. A computer-readable storage medium for storing computer-executable instructions, wherein the computer-executable instructions implement the steps of the method of claim 1 when executed.

Images