Information recommendation method and device, electronic equipment and medium

Abstract

The embodiment of the invention provides an information recommendation method and device, electronic equipment and a medium, and relates to the technical field of computers. The method comprises the following steps: acquiring user feature information of a target user; screening out a plurality of recommendation models from a pre-constructed model pool, respectively inputting the user feature information as input data into the plurality of recommendation models, and obtaining candidate service packages output by each recommendation model; and comprehensively analyzing the candidate service packages output by the plurality of recommendation models, and determining a service package corresponding to the target user. According to the method, the service package is recommended through the multiple recommendation models, the advantages of all the recommendation models are fully utilized, the defects of a single model are overcome, and therefore the recommendation accuracy and effectiveness are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an information recommendation method, device, electronic equipment and medium. Background Art

[0002] In order to meet the needs of different users, telecom operators bundle different products or product combinations to form different service packages. The current methods for operators to recommend packages are mainly manual recommendations in business halls and recommendations using artificial intelligence algorithms. The first recommendation method focuses on popular packages, which are incomplete and subjective. Only experienced marketers can recommend suitable and personalized service packages for users. The second recommendation method mainly calculates the similarity between users based on the attributes and behavioral habits of each user, and recommends service packages to users with high similarity. This method has a large user base, high computational complexity, and low prediction accuracy. It is difficult to accurately predict the real needs of users, resulting in unsatisfactory recommendation results and low user satisfaction. Therefore, a recommendation model that can more accurately predict the packages that users may use is needed to improve the accuracy of recommendations and user satisfaction. Summary of the invention

[0003] In order to solve the above technical problems or at least partially solve the above technical problems, embodiments of the present invention provide an information recommendation method, device, electronic device and medium.

[0004] In a first aspect, an embodiment of the present invention provides an information recommendation method, comprising:

[0005] Obtain user characteristic information of target users;

[0006] Filtering out multiple recommendation models from a pre-built model pool, inputting the user feature information as input data into the multiple recommendation models respectively, and obtaining candidate service packages output by each of the recommendation models;

[0007] Comprehensively analyze the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user.

[0008] In an optional embodiment, the user characteristic information of the target user includes one or more of the following: historical package usage data, consumption habit data and basic information data of the target user.

[0009] In an optional embodiment, before inputting the user characteristic information as input data into the plurality of recommendation models respectively, the method further includes:

[0010] Performing field item detection on the user characteristic information to determine whether there are missing field items in the user characteristic information;

[0011] In the case where it is determined that there are missing field items in the user characteristic information, determining a filling method corresponding to the missing field item according to the attributes of the missing field item;

[0012] According to the filling method, the missing field items of the user characteristic information are filled.

[0013] In an optional embodiment, the comprehensive analysis of the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user includes: determining the weight corresponding to each of the recommendation models; calculating the score of each candidate service package according to the weight corresponding to each of the recommendation models; and determining the service package corresponding to the target user according to the score of each candidate service package.

[0014] In an optional embodiment, the comprehensive analysis of the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user includes: determining the recommendation value corresponding to the candidate service packages output by each recommendation model to determine the service package corresponding to the target user.

[0015] In an optional embodiment, determining the recommendation value corresponding to the candidate service packages output by each recommendation model includes: according to a user tag corresponding to the target user; determining the recommendation value of each candidate service package under the user tag.

[0016] In an optional embodiment, determining the recommendation value of each of the candidate business packages under the user tag includes: counting the historical recommendation times and historical selection times of each of the candidate business packages under the user tag; and determining the recommendation value of each of the candidate business packages under the user tag based on the ratio of the historical recommendation times to the historical selection times.

[0017] In a second aspect, an embodiment of the present invention provides an information recommendation device, including:

[0018] An acquisition module is used to acquire user characteristic information of a target user;

[0019] A processing module, used to select multiple recommendation models from a pre-built model pool, input the user feature information as input data into the multiple recommendation models respectively, and obtain candidate service packages output by each of the recommendation models;

[0020] The screening module is used to comprehensively analyze the candidate service packages output by the multiple recommendation models and determine the service package corresponding to the target user.

[0021] In an optional embodiment, the device further includes a preprocessing module, which is used to:

[0022] Performing field item detection on the user characteristic information to determine whether there are missing field items in the user characteristic information;

[0023] In the case where it is determined that there are missing field items in the user characteristic information, determining a filling method corresponding to the missing field item according to the attributes of the missing field item;

[0024] According to the filling method, the missing field items of the user characteristic information are filled.

[0025] In an optional embodiment, the screening module is used to: determine the weights corresponding to each of the recommendation models; calculate the scores of each candidate service package based on the weights corresponding to each of the recommendation models; and determine the service package corresponding to the target user based on the scores of each candidate service package.

[0026] In an optional embodiment, the screening module is used to: determine the recommendation value corresponding to the candidate service packages output by each recommendation model, and determine the service package corresponding to the target user.

[0027] In an optional embodiment, the screening module is used to: determine the recommended value of each of the candidate service packages under the user tag according to the user tag corresponding to the target user.

[0028] In an optional embodiment, the screening module is used to: count the historical recommendation times and historical selection times of each of the candidate service packages under the user tag; and determine the recommendation value of each of the candidate service packages under the user tag based on the ratio of the historical recommendation times to the historical selection times.

[0029] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to implement the information recommendation method provided by any embodiment of the present invention when executing the program stored in the memory.

[0030] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the information recommendation method provided by any embodiment of the present invention.

[0031] The technical solution provided by the present method embodiment brings at least the following beneficial effects:

[0032] The information recommendation method provided by the embodiment of the present invention uses multiple recommendation models to analyze the user characteristic information of the target user, obtains multiple candidate service packages, and comprehensively analyzes the multiple candidate service packages to determine the target service package that matches the target user. The method introduces multiple recommendation models, fully utilizes the advantages of each recommendation model, and makes up for the defects of a single model, thereby improving the accuracy of the recommendation as a whole, and then comprehensively analyzes multiple candidate service packages to determine the target service package that matches the target user, providing users with more accurate package recommendation services. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0034] Figure 1 A schematic diagram showing a flow chart of an information recommendation method provided by an embodiment of the present invention;

[0035] Figure 2 The system architecture of the information recommendation method applicable to the embodiment of the present invention is shown;

[0036] Figure 3 A schematic diagram showing a flow chart of an information recommendation method provided by another embodiment of the present invention;

[0037] Figure 4 A schematic diagram showing a flow chart of an information recommendation method provided by another embodiment of the present invention;

[0038] Figure 5 A schematic diagram showing the structure of an information recommendation device according to an embodiment of the present invention is shown;

[0039] Figure 6 A schematic structural diagram of an electronic device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0040] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0041] Figure 1 FIG. 1 is a flow chart showing the information recommendation method provided by an embodiment of the present invention. Figure 1 As shown, the method includes:

[0042] Step S101: Obtain user characteristic information of the target user. The user characteristic information may include the target user's personal identity information (such as age, name, job, place of residence, etc.), historical package usage data (such as package usage in the past three months), and the target user's account status (such as the account balance corresponding to the mobile phone number).

[0043] In an optional embodiment, as shown in Table 1 below, the user characteristic information of the target user includes one or more of the following: historical package usage data, consumption habit data, and basic information data of the target user. In Table 1, "VARCHAR2" indicates a character type, and "NUMBER" indicates a numeric type.

[0044] Table 1:

[0045]

[0046]

[0047] In an optional embodiment, after obtaining the user characteristic information, the user characteristic information is normalized according to the above Table 1, and converted into a data form that can be recognized by the recommendation model.

[0048] In an optional embodiment, after obtaining the user characteristic information, it is preprocessed, such as data cleaning, filling missing values, etc. Optionally, the process of preprocessing the user characteristic information in the embodiment of the present invention includes:

[0049] Step a: Perform field item detection on the user characteristic information to determine whether there are missing field items in the user characteristic information.

[0050] Step b: When it is determined that there are missing fields in the user characteristic information, a filling method corresponding to the missing fields is determined according to the attributes of the missing fields.

[0051] Step c: fill in the missing field items of the user characteristic information according to the filling method.

[0052] As shown in Table 1, the user characteristic information includes multiple fields, such as gender, traffic usage, and online time. The user characteristic information is detected according to the fields shown in Table 1 to determine whether the user characteristic information includes all the fields. In the case where it is determined that there are missing fields in the user characteristic information, the corresponding filling method is determined according to the attributes of the missing field item, such as the data type, such as filling zero, mean value, etc. under the missing field item. Then, the missing field item is filled according to the filling method corresponding to the missing field item.

[0053] Step S102: Filter out multiple recommendation models from the pre-built model pool, input the user feature information as input data into the multiple recommendation models respectively, and obtain candidate service packages output by each of the recommendation models.

[0054] The recommendation models in the model pool are used to mine user feature information, fully explore the relationship between user identity information, consumption behavior and package selection, and infer the package type that best matches the user. The recommendation models in the model pool independently learn the user feature information of the training samples.

[0055] In an optional embodiment, the recommended models in the model pool include but are not limited to decision trees, recurrent neural networks, multi-layer perceptrons, reinforcement learning models, etc. Among them, a decision tree is a machine learning model with a tree structure, which is used for classification and regression tasks. Each node represents a test for a certain attribute, each branch represents a test result, and each leaf node represents a category or regression value. The decision tree is modeled by recursively dividing the data into smaller subsets until the data points in each subset have similar outputs. A recurrent neural network (RNN) is a neural network model with a cyclic connection, which can use the output of the previous step as the input of the current step, so that the network can remember the previous input information, and is suitable for sequence data. A multi-layer perceptron (MLP) is a feedforward neural network consisting of an input layer, one or more hidden layers, and an output layer. Each layer is composed of a number of neurons, and the layers are fully connected. The reinforcement learning model describes how an agent forms the intrinsic value of an action based on feedback in the environment, and guides the action based on the value.

[0056] In this embodiment, each recommendation model independently processes the user feature information of the target user and independently predicts the candidate recommendation model corresponding to the user feature information. The training process of the decision tree, recurrent neural network and multi-layer perceptron can be expressed as:

[0057] M tree =DecisionTreeClassifier(X train )

[0058] M nn =NeuralNetworkClassifier(X train )

[0059] M mlp =MultiLayerPerceptronClassifier(X train )

[0060] Among them, M tree Used to represent the output data of the decision tree, M nn Represents the output data of the recurrent neural network, M mlp Represents the output data of the multilayer perceptron, X train Represents training samples, DecisionTreeClassifier(X train ) represents the learning of the decision tree for the training sample, NeuralNetworkClassifier(X train ) represents the learning of the recurrent neural network on the training samples, MultiLayerPerceptronClassifier(X train ) represents the learning of the multilayer perceptron on the training samples. Suppose the training samples are where x i Represents user feature information, y i Represents the service package selected by the user. For each recommendation model C j , by minimizing the loss function L(C j (x i ),y i ) to learn user feature information.

[0061] Optionally, you can set the number of recommended models screened from the model pool each time and specify the recommended models. For example, the model pool includes 4 recommended models, namely Model A, Model B, Model C and Model D. Model A, Model B and Model C are designated as the recommended models for analyzing user characteristic information. Still optionally, you can specify the recommended models according to the number of service packages. For example, there are a total of m1 service packages in the first interval of package rates. Model A and Model B are designated as the recommended models for analyzing user characteristic information. There are a total of m2 service packages in the first interval of package rates (m2>m1). Model A, Model B and Model C are designated as the recommended models for analyzing user characteristic information.

[0062] Step S103: comprehensively analyzing the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user.

[0063] In an optional embodiment, the candidate service packages output by each recommendation model can be pushed to the target user for screening by the target user. In other optional embodiments, each candidate service package can be comprehensively analyzed to determine the service package corresponding to the target user. For example, if step S102 selects three recommendation models (model A, model B, and model C) to analyze user feature information, a total of three candidate service packages are output. If the same candidate service packages exist in the three candidate service packages, the same candidate service packages are determined as the service packages corresponding to the target user. If the three candidate service packages are different, the service package corresponding to the target user is determined according to the effective level of the recommendation model. For example, if the effective level of model A is higher than that of model B and model C, the candidate service package output by model A is used as the service package corresponding to the target user.

[0064] Optionally, after determining the service package corresponding to the target user, the service package and the candidate service packages output by each recommendation model are recorded in a database (eg, as shown in Table 2 below) for subsequent calls.

[0065] Table 2:

[0066] Fields Chinese name Data Types SERV_ID User ID NUMBER DC_Tree_ID Output of Recommendation Model 1 NUMBER PyTorchNN_ID Output of Recommendation Model 2 NUMBER PytorchMLP_ID Output of Recommendation Model 3 NUMBER voting_ID Package filter results NUMBER

[0067] The information recommendation method provided by the embodiment of the present invention uses multiple recommendation models to analyze the user characteristic information of the target user, obtains multiple candidate service packages, and comprehensively analyzes the multiple candidate service packages to determine the target service package that matches the target user. The method introduces multiple recommendation models, fully utilizes the advantages of each recommendation model, and makes up for the defects of a single model, thereby improving the accuracy of the recommendation as a whole, and then comprehensively analyzes multiple candidate service packages to determine the target service package that matches the target user, providing users with more accurate package recommendation services.

[0068] The information recommendation method provided by the embodiment of the present invention can be applied not only to service packages in the telecommunications field, but also to content recommendation in other scenarios. For example, the information used for recommendation can include but is not limited to: book information, film and television drama information, short video information, news information, etc. An exemplary application scenario can be: the historical behavior data of the target user (such as historical browsing records, purchase records, shopping cart records, and collection records) are respectively input as input data into multiple recommendation models to obtain multiple candidate books, and the books to be recommended that best match the target user are screened out from the multiple candidate books. Alternatively, a recommendation is made for a certain movie, a certain TV series, or a specific plot / fragment in a certain movie or a certain TV series in the film and television drama information. For example, the historical behavior data of the target user (such as historical browsing records, browsing time, comment data, and collection records) are respectively input as input data into multiple recommendation models to obtain multiple candidate film and television dramas, and the film and television drama to be recommended that best matches the target user is screened out from the multiple candidate film and television dramas.

[0069] Figure 2 The system architecture of the information recommendation method applicable to the embodiment of the present invention is shown.

[0070] like Figure 2 As shown, the system architecture 200 applicable to the method for generating recommendation information and the method for displaying recommendation information according to the embodiment of the present invention includes: terminal devices 201, 202, 203, a network 204 and a server 205. The network 204 is a medium for providing a communication link between the terminal devices 201, 202, 203 and the server 205. The network 204 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0071] The user can use the terminal devices 201, 202, 203 to interact with the server 205 through the network 204 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 201, 202, 203, such as: e-book reading software / applications (APP), shopping applications, short video applications, web browser applications, search applications, news client applications, instant messaging tools, food delivery applications, social platform software, etc. (only as examples).

[0072] The terminal devices 201, 202, and 203 may be electronic devices with a display screen and supporting at least one of text, picture, and video playback, or the terminal devices 201, 202, and 203 may also be electronic devices with a display screen and supporting web browsing. For example, the electronic devices may include, but are not limited to: smart phones, tablet computers, laptop computers, desktop computers, smart watches, etc.

[0073] The server 205 may be a server that provides various services, such as a background management server (only as an example) that provides service support for the application interface or web page browsed by the user using the terminal devices 201, 202, and 203. The background management server may analyze and process the received recommendation request, and feed back the result to be recommended (such as the target information containing the information to be recommended generated according to the user's recommendation request) obtained after data processing to the terminal device.

[0074] It should be noted that the information recommendation method provided in the embodiment of the present invention can generally be executed by the server 205. The information recommendation method provided in the embodiment of the present invention can also be executed by a server or server cluster different from the server 205 and capable of communicating with the terminal devices 201, 202, 203 and / or the server 205.

[0075] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0076] In an optional implementation scenario, the user queries the current service package details, usage, and various service packages provided by the telecom operator through the software provided by the telecom operator on the terminal device. The telecom operator provides data support. If a user expects to change the service package, he logs in to the software provided by the telecom operator to query the various optional service packages. The software is provided with a recommendation control, and the user clicks the recommendation control to initiate a package recommendation request. The server receives the package recommendation request, obtains the user feature information of the user, and selects multiple recommendation models from the model pool, uses the multiple recommendation models to independently analyze the user feature information, outputs multiple candidate service packages, and determines the corresponding service package for the user from the multiple candidate service packages. Although the above scenario takes the telecom service package recommendation scenario as an example, it can be understood that the application scenario of the technical solution provided in the embodiment of the present invention is not limited to the above scenario, and can be extended to any scenario where content information recommendation is required.

[0077] Figure 3 FIG. 2 is a flow chart of an information recommendation method provided by another embodiment of the present invention. Figure 3 As shown, the method includes:

[0078] Step S301: Obtain user characteristic information of the target user.

[0079] Step S302: Perform field item detection on the user characteristic information to determine whether there are missing field items in the user characteristic information.

[0080] Step S303: when it is determined that there are missing fields in the user characteristic information, a filling method corresponding to the missing fields is determined according to the attributes of the missing fields.

[0081] Step S304: fill in the missing field items of the user characteristic information according to the filling method.

[0082] Step S305: Filter out multiple recommendation models from the pre-built model pool, input the user feature information as input data into the multiple recommendation models respectively, and obtain candidate service packages output by each of the recommendation models.

[0083] Step S306: Determine the weight corresponding to each of the recommendation models.

[0084] Step S307: Calculate the score of each candidate service package according to the weight corresponding to each recommendation model.

[0085] Step S308: Determine the service package corresponding to the target user according to the scores of the candidate service packages.

[0086] Among them, steps S301 to S305 can refer to Figure 1 The illustrated embodiments are not described in detail here to avoid repetition.

[0087] For step S30, the weights corresponding to each recommendation model can be used to indicate the accuracy of the recommendation results output by the recommendation model. The larger the weight, the greater the accuracy of the recommendation results output by the recommendation model. For example, the weight of model A is 0.8, indicating that the accuracy of the recommendation results output by model A is 0.8. The weight of model B is 0.7, indicating that the accuracy of the recommendation results output by model B is 0.7. The accuracy of the recommendation results output by model A is higher than that of model B. Among them, the weights corresponding to each recommendation model can be updated according to historical recommendation results and user-selected results. If the recommendation results output by the recommendation model are the same as the user-selected results, the weight of the recommendation model can be increased. If the recommendation results output by the recommendation model are different from the user-selected results, the weight of the recommendation model can be reduced.

[0088] For step S307, the weighted sum of each candidate service package can be calculated according to the weight corresponding to each recommendation model, and the weighted sum can be used as the score of the candidate service package. For example, the weight of model A is 0.8, the weight of model B is 0.7, the weight of model C is 0.85, and the weight of model D is 0.8. The candidate service package output by model A is package 1, the candidate service package output by model B is package 1, the candidate service package output by model C is package 3, and the candidate service package output by model A is package 5. The score of package 1 is 0.8*100+0.7*100=150, the score of package 2 is 0.85*100=85, and the score of package 3 is 0.8*100=80, so the service package recommended to the user is package 1.

[0089] The information recommendation method provided by the embodiment of the present invention can comprehensively consider each candidate service package according to the weight of each recommendation model, so as to determine a more accurate service package to be recommended.

[0090] Figure 4 FIG. 2 is a flow chart showing an information recommendation method provided by another embodiment of the present invention. Figure 4 As shown, the method includes:

[0091] Step S401: Obtain user characteristic information of the target user.

[0092] Step S402: Perform field item detection on the user characteristic information to determine whether there are any missing field items in the user characteristic information.

[0093] Step S403: when it is determined that there are missing fields in the user characteristic information, a filling method corresponding to the missing fields is determined according to the attributes of the missing fields.

[0094] Step S404: fill in the missing field items of the user characteristic information according to the filling method.

[0095] Step S405: Filter out multiple recommendation models from the pre-built model pool, input the user feature information as input data into the multiple recommendation models respectively, and obtain candidate service packages output by each of the recommendation models.

[0096] Step S406: Determine the recommendation value corresponding to the candidate service packages output by each recommendation model, and determine the service package corresponding to the target user.

[0097] Among them, steps S401 to S405 can refer to Figure 1 The illustrated embodiments are not described in detail here to avoid repetition.

[0098] For step S406, the recommendation value of the candidate service package can be used to describe the popularity, economy or practicality of the package. The higher the recommendation value, the higher the popularity or the more practical the package. In this embodiment, the candidate service package with the highest recommendation value can be determined as the service package corresponding to the target user. If there are two candidate service packages with the highest recommendation value, the two candidate service packages can be used as the service packages corresponding to the target user, or the candidate service package with a lower rate can be used as the service package corresponding to the target user. The recommendation value of the service package can be set manually, or it can be determined based on the ratio of the historical recommendation number to the historical selection number of the service package. For example, if the historical recommendation number is n1 and the historical selection number is n2, n2 / n1 is used as the score of the service package.

[0099] In an optional embodiment, determining the recommendation value corresponding to the candidate service packages output by each recommendation model includes: according to the user tag corresponding to the target user; determining the recommendation value of each candidate service package under the user tag.

[0100] A user tag is an attribute description of a certain dimension feature of a user. Optionally, user tags can be maintained and set according to gender, age, nature of work, region (e.g., first-tier cities, second-tier cities, etc.), user preferences (e.g., preferred call duration, traffic or cost, etc.). The user tag corresponding to the target user can be one or more. The recommended values ​​of each service package under each user tag can be the same or different. For example, compared with other service packages, service package 1 has a longer call duration, and the recommended value under the preferred call duration tag is higher than the preferred traffic tag. For another example, the tariff of service package 1 is the same as that of service package 2, and service package 1 and service package 2 have the same recommended value under the preferred cost tag. In this embodiment, the recommended values ​​of each candidate service package under each user tag corresponding to the target user can be summed, and the sum can be used as the score of the candidate service package. The recommended value of each candidate service package under each user tag can be determined according to the ratio of the number of historical recommendations to the number of historical selections of the candidate service package under the user tag.

[0101] Figure 5 FIG. 1 is a schematic diagram showing the structure of an information recommendation device provided by an embodiment of the present invention. Figure 4 As shown, the information recommendation device 500 includes:

[0102] An acquisition module 501 is used to acquire user characteristic information of a target user;

[0103] The processing module 502 is used to select multiple recommendation models from the pre-built model pool, input the user feature information as input data into the multiple recommendation models respectively, and obtain the candidate service packages output by each of the recommendation models;

[0104] The screening module 503 is used to comprehensively analyze the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user.

[0105] The information recommendation device provided in the embodiment of the present invention uses multiple recommendation models to analyze the user characteristic information of the target user, obtains multiple candidate service packages, and comprehensively analyzes the multiple candidate service packages to determine the target service package that matches the target user. The method introduces multiple recommendation models, fully utilizes the advantages of each recommendation model, and makes up for the defects of a single model, thereby improving the accuracy of the recommendation as a whole, and then comprehensively analyzes multiple candidate service packages to determine the target service package that matches the target user, providing users with more accurate package recommendation services.

[0106] In an optional embodiment, the device further includes a preprocessing module, which is used to:

[0107] Performing field item detection on the user characteristic information to determine whether there are missing field items in the user characteristic information;

[0108] In the case where it is determined that there are missing field items in the user characteristic information, determining a filling method corresponding to the missing field item according to the attributes of the missing field item;

[0109] According to the filling method, the missing field items of the user characteristic information are filled.

[0110] In an optional embodiment, the screening module is used to: determine the weights corresponding to each of the recommendation models; calculate the scores of each candidate service package based on the weights corresponding to each of the recommendation models; and determine the service package corresponding to the target user based on the scores of each candidate service package.

[0111] In an optional embodiment, the screening module is used to: determine the recommendation value corresponding to the candidate service packages output by each recommendation model, and determine the service package corresponding to the target user.

[0112] In an optional embodiment, the screening module is used to: determine the recommended value of each of the candidate service packages under the user tag according to the user tag corresponding to the target user.

[0113] In an optional embodiment, the screening module is used to: count the historical recommendation times and historical selection times of each of the candidate service packages under the user tag; and determine the recommendation value of each of the candidate service packages under the user tag based on the ratio of the historical recommendation times to the historical selection times.

[0114] The above device can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the information recommendation method provided by the embodiment of the present invention.

[0115] Figure 6 FIG. 1 is a schematic diagram showing the structure of an electronic device according to an embodiment of the present invention. Figure 6 As shown, the electronic device includes:

[0116] Processor 601, communication interface 602, memory 603 and communication bus 604, wherein processor 601, communication interface 602, memory 603 communicate with each other via communication bus 604.

[0117] Memory 603, used for storing computer programs;

[0118] The processor 601 is used to execute the program stored in the memory 603 to implement the following steps:

[0119] Obtain user characteristic information of target users;

[0120] Filtering out multiple recommendation models from a pre-built model pool, inputting the user feature information as input data into the multiple recommendation models respectively, and obtaining candidate service packages output by each of the recommendation models;

[0121] Comprehensively analyze the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user.

[0122] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0123] The communication interface is used for communication between the above terminal and other devices.

[0124] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0125] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0126] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions. When the instructions are executed on a computer, the computer executes the information recommendation method described in any one of the above embodiments.

[0127] In another embodiment of the present invention, a computer program product including instructions is provided. When the computer program product is executed on a computer, the computer executes the information recommendation method described in any one of the above embodiments.

[0128] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0129] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article 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, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0130] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0131] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. An information recommendation method, characterized in that: include: Obtain user characteristic information of target users; Filtering out multiple recommendation models from a pre-built model pool, inputting the user feature information as input data into the multiple recommendation models respectively, and obtaining candidate service packages output by each of the recommendation models; Comprehensively analyze the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user.

2. The method according to claim 1, characterized in that Before inputting the user characteristic information as input data into the plurality of recommendation models respectively, the method further includes: Performing field item detection on the user characteristic information to determine whether there are missing field items in the user characteristic information; In the case where it is determined that there are missing field items in the user characteristic information, determining a filling method corresponding to the missing field item according to the attributes of the missing field item; According to the filling method, the missing field items of the user characteristic information are filled.

3. The method according to claim 1, characterized in that The comprehensively analyzing the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user includes: Determine the weight corresponding to each of the recommendation models; Calculate the score of each candidate service package according to the weight corresponding to each of the recommendation models; According to the scores of the candidate service packages, the service package corresponding to the target user is determined.

4. The method according to claim 1 or 3, characterized in that: The comprehensively analyzing the candidate service packages output by the multiple recommendation models to determine the service package corresponding to the target user includes: Determine the recommendation value corresponding to the candidate service packages output by each recommendation model, and determine the service package corresponding to the target user.

5. The method according to claim 4, characterized in that The determining of the recommendation value corresponding to the candidate service packages output by each recommendation model includes: According to the user tag corresponding to the target user; Determine the recommended value of each of the candidate service packages under the user tag.

6. The method according to claim 5, characterized in that The determining of the recommended value of each of the candidate service packages under the user tag includes: Counting the number of historical recommendations and selections of each candidate service package under the user tag; According to the ratio of the historical recommendation times to the historical selection times, the recommendation value of each of the candidate service packages under the user tag is determined.

7. The method according to claim 1, characterized in that The user characteristic information of the target user includes one or more of the following: historical package usage data, consumption habit data and basic information data of the target user.

8. An information recommendation device, characterized in that: include: An acquisition module is used to acquire user characteristic information of a target user; A processing module, used to select multiple recommendation models from a pre-built model pool, input the user feature information as input data into the multiple recommendation models respectively, and obtain candidate service packages output by each of the recommendation models; The screening module is used to comprehensively analyze the candidate service packages output by the multiple recommendation models and determine the service package corresponding to the target user.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, configured to implement the method according to any one of claims 1 to 7 when executing a program stored in a memory.

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

Images