Method, system, electronic device and storage medium for constructing user interest model

Abstract

The present application relates to a method, system, electronic device, and storage medium for constructing a user interest model, wherein the method comprises: obtaining user data and calibrating the observation time on the user data to obtain user observation data; then, inputting the user observation data into a data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes features of the observation time, features of a preset time after the observation time, a completion identifier, and a response identifier of a preset time after the observation time; finally, training a temporal difference learning model using the user observation change data set to obtain a trained temporal difference learning model. Through the present application, the problem of inadequate description of interest concentration changes when predicting user interests is solved, and the accuracy of the description of user interest concentration is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, system, electronic device, and storage medium for constructing a user interest model. Background Art

[0002] Currently, user interest modeling is generally performed using a binary classification model. The modeling process is as follows: 1. Randomly select a portion of user data; 2. Randomly assign observation times to these users; 3. Based on each user's observation time, label a response within 30 days of the observation time as 1, and label a response as 0; 4. Perform feature association and feature processing on these users; 5. Use machine learning algorithms such as logistic regression, decision trees, gradient boosting trees, or deep learning to build a binary classification model; 6. Use the probability output of the binary classification model as a representation of user interest.

[0003] However, in the related art, there are still some problems with the solution of realizing user interest prediction through the above process. On the one hand, the observation time of the existing solution only reflects the interest concentration 30 days after the observation time point, and lacks fine-grained characterization. For example, the interest concentration on the first and second days after the observation point is actually different, and the existing solution does not characterize it; on the other hand, the existing solution predicts the possibility of response 30 days after the observation time based on the characteristics corresponding to the observation time. This method cannot infer the true interest concentration based on the changes in characteristics within these 30 days. For example, the characteristics one day after the observation point are slightly different from the characteristics of the observation point, and the network model of the existing solution cannot capture this subtle change.

[0004] Currently, in related technologies, there is a problem of inadequate description of changes in interest concentration when predicting user interests, and no effective solution has been proposed. Summary of the Invention

[0005] The embodiments of the present application provide a method, system, electronic device, and storage medium for constructing a user interest model to at least solve the problem in the related art of not carefully depicting changes in interest concentration when predicting user interests.

[0006] In a first aspect, an embodiment of the present application provides a method for constructing a user interest model, the method comprising:

[0007] Acquiring user data and calibrating observation time on the user data to obtain user observation data;

[0008] Inputting the user observation data into a data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes a feature of the observation time, a feature of a preset time after the observation time, a completion identifier, and a response identifier of the preset time after the observation time;

[0009] The temporal difference learning model is trained using the user observation change data set to obtain a trained temporal difference learning model.

[0010] In some embodiments, after inputting the user observation data into a data processing module, the method includes:

[0011] The user observation data is copied, and the observation time in the copied user observation data is gradually increased to obtain a user observation data set.

[0012] In some embodiments, after obtaining the user observation dataset, the method includes:

[0013] Obtaining an observation response time, and responding to a corresponding field of the observation time in the user observation data set using the observation response time to generate first associated data;

[0014] Extracting features of each user observation data in the user observation data set to obtain features of the observation time, performing feature backtracking on the features of the observation time through the first associated data to generate features of a preset time after the observation time.

[0015] In some embodiments, generating the first associated data includes:

[0016] In the case where there is a response at a preset time after the observation time, the completion identifier in the user observation change data set at the observation time is true, and the response identifier at the preset time after the observation time is 1;

[0017] In the case where there is no response at a preset time after the observation time, the completion identifier in the user observation change data set at the observation time is false, and the response identifier at the preset time after the observation time is 0;

[0018] Or when the preset time after the observation time exceeds the preset time, the completion identifier in the user observation change data set under the observation time is true. At this time, if there is a response at the preset time after the observation time, the response identifier of the preset time after the observation time is 1, otherwise it is 0.

[0019] In some embodiments, training the temporal difference learning model using the user observation change dataset includes:

[0020] Data is randomly extracted from the user observation change data set, and different values are input into the training module of the temporal difference learning model according to the truth or falsity of the completion identifier in the data. The temporal difference learning model is trained and the model parameters are updated until the model converges.

[0021] In some embodiments, inputting different values into the training module of the temporal difference learning model according to the authenticity of the completion identifier in the data includes:

[0022] When the completion identifier in the data is true, the feature of the observation time is used as input, and the response identifier of the preset time after the observation time is used as the value to be fitted, which is input into the training module of the temporal difference learning model;

[0023] When the completion identifier in the data is false, the characteristics of the preset time after the observation time are input into the prediction module of the time difference learning model, and the first parameter is output. A custom calculation is performed based on the first parameter to obtain the second parameter. The characteristics of the observation time are then used as input, and the second parameter is used as the value to be fitted, and input into the training module of the time difference learning model.

[0024] In a second aspect, an embodiment of the present application provides a system for constructing a user interest model, the system comprising:

[0025] An acquisition module is used to acquire user data and mark the observation time on the user data to obtain user observation data;

[0026] a data processing module, configured to input the user observation data into a data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes a feature of the observation time, a feature of a preset time after the observation time, a completion identifier, and a response identifier of the preset time after the observation time;

[0027] The training module is used to train the temporal difference learning model using the user observation change data set to obtain a trained temporal difference learning model.

[0028] In some embodiments, after the user observation data is input into the data processing module,

[0029] The data processing module is further configured to copy the user observation data and gradually increase the observation time in the copied user observation data to obtain a new user observation data set.

[0030] In a third aspect, an embodiment of the present application provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the method for constructing a user interest model as described in the first aspect above is implemented.

[0031] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, which, when executed by a processor, implements the method for constructing a user interest model as described in the first aspect above.

[0032] Compared with the related art, the method for constructing a user interest model provided in the embodiment of the present application obtains user data and calibrates the observation time on the user data to obtain user observation data; then, the user observation data is input into the data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes the characteristics of the observation time, the characteristics of the preset time after the observation time, the completion identifier and the response identifier of the preset time after the observation time; finally, the time difference learning model is trained by the user observation change data set to obtain a trained time difference learning model.

[0033] This application simulates the changes in user interest concentration through iteration in the time direction, and can learn the feature differences at different times to obtain a more detailed representation of interest changes. It solves the problem of unclear characterization of interest concentration changes when predicting user interests, and improves the accuracy of the characterization of user interest concentration. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0035] Figure 1 Schematic diagram of the application environment of the method for constructing a user interest model according to an embodiment of the present application;

[0036] Figure 2 is a flowchart of a method for constructing a user interest model according to an embodiment of the present application;

[0037] Figure 3 is a schematic diagram of the temporal difference learning model process according to an embodiment of the present application;

[0038] Figure 4 This is a structural block diagram of a system constructed according to a user interest model according to an embodiment of the present application;

[0039] Figure 5 is a structural block diagram of a system constructed according to another user interest model according to an embodiment of the present application;

[0040] Figure 6 Schematic diagram of the internal structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application. In addition, it can also be understood that although the efforts made in this development process may be complex and lengthy, for ordinary technicians in the field related to the contents disclosed in the present application, some changes such as design, manufacturing or production based on the technical contents disclosed in the present application are only conventional technical means and should not be understood as the contents disclosed in the present application being insufficient.

[0042] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments unless there is a conflict.

[0043] Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by a person of ordinary skill in the technical field to which this application belongs. The words "one", "a", "the" and the like used in this application do not indicate a limit on quantity and may indicate the singular or plural. The terms "include", "comprise", "have" and any variations thereof used in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units that are inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The word "multiple" used in this application means greater than or equal to two. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. The terms "first", "second", "third" and the like involved in this application are merely used to distinguish similar objects and do not represent a specific ordering of the objects.

[0044] The method for constructing a user interest model provided in this application can be applied to Figure 1 In the application environment shown, Figure 1 Schematic diagram of the application environment of the method for constructing a user interest model according to an embodiment of the present application. Figure 1 As shown. The terminal 11 communicates with the server 10 through a network. The server 10 obtains user data and calibrates the observation time on the user data to obtain user observation data; then, the user observation data is input into the data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes the characteristics of the observation time, the characteristics of the preset time after the observation time, the completion identifier and the response identifier of the preset time after the observation time; finally, the time difference learning model is trained by the user observation change data set to obtain a trained time difference learning model. The terminal 11 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers and portable wearable devices, and the server 10 can be implemented as an independent server or a server cluster consisting of multiple servers.

[0045] This application embodiment provides a method for constructing a user interest model. Figure 2 is a flowchart of a method for constructing a user interest model according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:

[0046] Step S201: Obtain user data and mark the observation time on the user data to obtain user observation data;

[0047] In this embodiment, some user data user are randomly obtained, and the observation time ds is calibrated on these user data to obtain user observation data (user, ds);

[0048] Step S202: Input the user observation data into the data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes a feature of the observation time, a feature of the preset time after the observation time, a completion identifier, and a response identifier of the preset time after the observation time. It should be noted that the time interval between the observation time and the preset time after the observation time is user-defined. For example, if the observation time is set to t, then the preset time after the observation time can be set to t+n.

[0049] Preferably, the user observation data (user, ds) obtained above is input into the data processing module. In this embodiment, before processing the user data field and the observation time field, the user observation data is first copied, and the observation time in the copied user observation data is gradually increased in sequence to obtain a user observation data set. For example, the user observation data (user, ds) is copied 30 times, and the observation time ds in the copied user observation data is gradually increased in sequence. Assuming that the observation time is gradually increased by 1 day, a new observation time is obtained. For example, the input user observation data is (user, 20210701), copied 30 times, and the observation time is gradually increased by 1 day in sequence to obtain the user observation data set: (user, 20210701), (user, 20210702)......(user, 20210730). It should be noted that the minimum observation time corresponding to each user is min_ds.

[0050] Furthermore, data processing is performed based on the user data field and observation time field in the user observation dataset (user, ds), and the output is a user observation change dataset (user, ds, state, done, reward, next_state). The data in the user observation change dataset includes the characteristic state at the observation time, the characteristic next_state at a preset time after the observation time, the completion identifier done, and the response identifier reward at a preset time after the observation time. Preferably, the specific process of data processing in this embodiment is as follows:

[0051] S1. Obtain the observation response time, and use the observation response time to respond to the corresponding field of the observation time in the user observation data set obtained above, to generate the first associated data. Specifically, the judgment logic in the first associated data is: if there is a response within a preset time after the observation time, the completion identifier in the user observation change data set under the observation time is true, and the response identifier of the preset time after the observation time is 1; if there is no response within the preset time after the observation time, the completion identifier in the user observation change data set under the observation time is false, and the response identifier of the preset time after the observation time is 0; or if the preset time after the observation time exceeds the preset time, the completion identifier in the user observation change data set under the observation time is true. In this case, if there is a response within the preset time after the observation time, the response identifier of the preset time after the observation time is 1, otherwise it is 0. For example, let the observation time be t, and the preset time after the observation time be t+1. Then, if there is a response at t+1=20210715, the completion identifier done in the user observation change data set at the observation time t=20210714 is true (true) and the response identifier reward=1 at the preset time t+1=20210715 after the observation time; and if there is no response at 20210715, the completion identifier done is false (false) and the response identifier reward=0 at the preset time t+1=20210715 after the observation time; or, if the preset time t+1=2021071 5. When the preset time is exceeded, for example, the preset time t+1=20210715 after the observation time is min_ds+30, that is, 20210715 is more than one month, then the completion identifier done in the user observation change dataset at the observation time t=20210714 is true (true). At this time, if there is a response at t+1=20210715, the response identifier reward=1 at 20210715; if there is no response at t+1=20210715, the response identifier reward=0 at 20210715;

[0052] S2, extract the features of each user observation data in the user observation data set (user, ds), obtain the feature state of the observation time, and perform feature backtracking on the features of the observation time through the first associated data obtained in the above S1 to generate the feature next_state of the preset time after the observation time.

[0053] Through the above steps, after the user observation data (user, ds) is input into the data processing module, the user observation change data set (user, ds, state, done, reward, next_state) is finally obtained. Table 1 is the user observation change data set obtained after processing by the data processing module, as shown in Table 1 below.

[0054] Table 1

[0055] user ds done reward state next_state 123 20210701 FALSE 0 [0,2,1,1,...] [0,2,1,2,...] 123 20210702 FALSE 0 [0,2,1,2,...] [0,2,1,3,...] 123 20210703 FALSE 0 [0,2,1,3,...] [0,2,1,4,...] ... ... FALSE 0 123 20210712 FALSE 0 [0,2,1,12,...] [0,2,1,13,...] 123 20210713 FALSE 0 [0,2,1,13,...] [0,2,1,14,...] 123 20210714 TRUE 1 [0,2,1,14,...] [0,2,1,15,...] 456 20210706 FALSE 0 [6,1,3,1...] [7,1,3,1...] 456 20210707 FALSE 0 [7,1,3,1...] [8,1,3,1...] ... ... FALSE 0 456 20210803 FALSE 0 [3,1,3,1...] [4,1,3,1...] 456 20210804 TRUE 0 [4,1,3,1...] [5,1,3,1...]

[0056] Step S203: training the temporal difference learning model using the user observation change data set to obtain a trained temporal difference learning model.

[0057] Preferably, in this embodiment, batch_size pieces of data are randomly extracted from the user observation change data set (user, ds, state, done, reward, next_state), and then different values are input into the training module of the temporal difference learning model according to the truth or falsehood of the completion identifier done in these data to train the temporal difference learning model and update the model parameters until the model converges; specifically, Figure 3 This is a flow chart of the temporal difference learning model according to an embodiment of the present application. Figure 3 As shown, when the completion identifier done in each randomly selected data is true, the feature state of the observation time is input, and the response identifier reward at the preset time after the observation time is used as the value to be fitted, and input into Figure 3 In the training module of the time difference learning model; or when the completion identifier done in each randomly extracted data is false, the feature next_state of the preset time after the observation time is input into Figure 3 In the prediction module of the time difference learning model, the first parameter next_pred is output, and the second parameter reward+γ*next_pred is calculated by a custom algorithm based on the first parameter next_pred. Then, the feature state of the observation time is input, and the second parameter is used as the value to be fitted, and input into Figure 3In the training module of the temporal difference learning model, finally, the above process is repeated using the data in the user observation change dataset to train the temporal difference learning model, update the model parameters, and obtain a trained temporal difference learning model until the model converges. For example, assuming that the current observation time is t = 20210701, and there is a response at t + 4 = 20210705, then the completion identifier done in the user observation change dataset under the observation time 20210704 is true (true) and the response identifier reward = 1 at the preset time 20210705 after the observation time. At this time, it can be seen from step S203 that when the completion identifier done in the data is true (true), the feature state of the observation time 20210704 is used as input, and the response identifier reward = 1 at the preset time (20210705) after the observation time is used as the value to be fitted and input into the training module of the temporal difference learning model. Next, since there is no response at the observation time 20210704, according to step S202, the completion identifier done in the user observation change data set under 20210703 is false (false) and the response identifier reward = 0 at the preset time after the observation time (20210704). At this time, it can be seen from step S203 that the feature next_state of the preset time after the observation time (20210704) is input into the prediction module of the time difference learning model, and the first parameter next_pred = γ*next_pred is output, and the second parameter reward + γ*next_pred is calculated based on the first parameter next_pred. Then, the feature state of the observation time (20210703) is used as input, and the second parameter reward + γ*next_pred is used as the value to be fitted, and input into the training module of the time difference learning model. Similarly, since there is no response at observation time 20210702 and observation time 20210701, the estimated first parameter at observation time 20210702 is γ*(γ*next_pred), and the estimated first parameter at observation time 20210701 is γ*(γ*(γ*next_pred)). The second parameters calculated based on the above estimated first parameters are reward+γ*(γ*next_pred) and reward+γ*(γ*(γ*next_pred)), respectively. The feature state at observation time (20210702 and 20210701) is used as input, and the second parameters obtained above are used as the values to be fitted and input into the training module of the temporal difference learning model. The temporal difference learning model is trained by repeating the above process, and the model parameters are updated until the model converges to obtain a trained model.It should be noted that, since γ<1, the user interest prediction using the above-trained model can estimate that the interest change curve is actually in a decaying state according to time, which can better simulate the daily interest changes.

[0058] In some embodiments, after obtaining a trained temporal difference learning model, this embodiment uses the trained temporal difference learning model to predict the unprocessed user observation data and outputs a user interest prediction value.

[0059] Through the above steps S201 to S203, the embodiment of the present application inputs the features at time t+1 into the network to obtain the predicted value at time t+1, and uses this to estimate the interest at time t, which is equivalent to adjusting the model obtained according to the future state to obtain an estimated value that conforms to the current state. In this process, there is a game process between the future network and the current network, so that the model can capture the subtle differences in the time of interest changes, so as to obtain a better representation of the interest changes. It solves the problem of unclear description of interest concentration changes when predicting user interests, and improves the accuracy of the description of user interest concentration.

[0060] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0061] This embodiment also provides a system for constructing a user interest model, which is used to implement the above-mentioned embodiments and preferred implementation methods, and will not be repeated here. As used below, the terms "module", "unit", "sub-unit", etc. can be a combination of software and / or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceivable.

[0062] Figure 4 This is a structural block diagram of a system constructed based on the user interest model of an embodiment of the present application. Figure 4 As shown, the system includes an acquisition module 41, a data processing module 42 and a training module 43:

[0063] The acquisition module 41 is used to acquire user data and calibrate the observation time on the user data to obtain user observation data; the data processing module 42 is used to input the user observation data into the data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes the characteristics of the observation time, the characteristics of the preset time after the observation time, the completion identifier and the response identifier of the preset time after the observation time; the training module 43 is used to train the time difference learning model through the user observation change data set to obtain a trained time difference learning model.

[0064] Through the above system, the present application simulates the changes in user interest concentration through iteration in the time direction, and can learn the feature differences at different times to obtain a more detailed representation of interest changes, thereby solving the problem of unclear depiction of interest concentration changes when predicting user interests and improving the accuracy of the depiction of user interest concentration.

[0065] In some embodiments, the system further comprises a prediction module, Figure 5 This is a structural block diagram of a system constructed according to another user interest model embodiment of the present application. Figure 5 As shown, the system includes an acquisition module 41, a data processing module 42, a training module 43, and a prediction module 51. After obtaining the trained temporal difference learning model, the prediction module 51 is used to predict the unprocessed user observation data using the trained temporal difference learning model and output a user interest prediction value.

[0066] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be repeated here.

[0067] Furthermore, it should be noted that each of the aforementioned modules can be either a functional module or a program module, and can be implemented via software or hardware. For modules implemented via hardware, each of the aforementioned modules can be located in the same processor; or each of the aforementioned modules can be located in different processors in any combination.

[0068] This embodiment further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0069] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0070] In addition, in conjunction with the user interest prediction method in the above embodiments, the present application embodiment may provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, it implements any of the user interest prediction methods in the above embodiments.

[0071] In one embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, memory, a network interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for predicting user interests is implemented. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen. The input device of the computer device may be a touch layer covering the display screen, or may be a key, trackball, or touchpad provided on the computer device housing, or may be an external keyboard, touchpad, or mouse.

[0072] In one embodiment, Figure 6 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application, such as Figure 6 As shown, an electronic device is provided, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 6 As shown. The electronic device includes a processor, a network interface, an internal memory, and a non-volatile memory connected via an internal bus, wherein the non-volatile memory stores an operating system, a computer program, and a database. The processor is used to provide computing and control capabilities, the network interface is used to communicate with an external terminal via a network connection, the internal memory is used to provide an environment for the operation of the operating system and the computer program, and when the computer program is executed by the processor, a method for predicting user interests is implemented. The database is used to store data.

[0073] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0074] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, which can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0075] Those skilled in the art should understand that the various technical features of the above-described embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0076] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for constructing a user interest model, characterized in that: The method comprises: Acquiring user data and calibrating observation time on the user data to obtain user observation data; Inputting the user observation data into a data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes a feature of the observation time, a feature of a preset time after the observation time, a completion identifier, and a response identifier of the preset time after the observation time; Data is randomly extracted from the user observation change data set, and different values are input into the training module of the temporal difference learning model according to the truth or falsity of the completion identifier in the data. The temporal difference learning model is trained and the model parameters are updated until the model converges to obtain a trained temporal difference learning model. After obtaining the trained temporal difference learning model, the user observation data that has not been processed is predicted by the trained temporal difference learning model to output a user interest prediction value.

2. The method according to claim 1, characterized in that After the user observation data is input into a data processing module, the method includes: Copying the user observation data, and gradually increasing the observation time in the copied user observation data to obtain a user observation data set; Data processing is performed according to the user data field and the observation time field in the user observation data set, and the user observation change data set is output, wherein each data in the user observation change data set includes a feature of the observation time, a feature of a preset time after the observation time, the completion identifier, and a response identifier of the preset time after the observation time.

3. The method according to claim 2, characterized in that After obtaining the user observation data set, the data processing process is as follows: Obtaining an observation response time, and responding to a corresponding field of the observation time in the user observation data set using the observation response time to generate first associated data; Extracting features of each user observation data in the user observation data set to obtain features of the observation time, performing feature backtracking on the features of the observation time through the first associated data to generate features of a preset time after the observation time.

4. The method according to claim 3, characterized in that The generating of the first associated data includes: In the case where there is a response at a preset time after the observation time, the completion identifier in the user observation change data set at the observation time is true, and the response identifier at the preset time after the observation time is 1; In the case where there is no response at a preset time after the observation time, the completion identifier in the user observation change data set at the observation time is false, and the response identifier at the preset time after the observation time is 0; Or when the preset time after the observation time exceeds the preset duration, the completion identifier in the user observation change data set under the observation time is true. At this time, if there is a response within the preset time after the observation time, the response identifier of the preset time after the observation time is 1, otherwise it is 0.

5. The method according to claim 1, wherein Inputting different values into the training module of the temporal difference learning model according to the authenticity of the completion identifier in the data includes: When the completion identifier in the data is true, the feature of the observation time is used as input, and the response identifier of the preset time after the observation time is used as the value to be fitted, which is input into the training module of the temporal difference learning model; When the completion identifier in the data is false, the characteristics of the preset time after the observation time are input into the prediction module of the time difference learning model, and the first parameter is output. A custom calculation is performed based on the first parameter to obtain the second parameter. The characteristics of the observation time are then used as input, and the second parameter is used as the value to be fitted, and input into the training module of the time difference learning model.

6. A system for constructing a user interest model, characterized in that: The system comprises: An acquisition module is used to acquire user data and mark the observation time on the user data to obtain user observation data; a data processing module, configured to input the user observation data into a data processing module to obtain a user observation change data set, wherein each data in the user observation change data set includes a feature of the observation time, a feature of a preset time after the observation time, a completion identifier, and a response identifier of the preset time after the observation time; A training module is used to randomly extract data from the user observation change data set, input different values into the training module of the temporal difference learning model according to the truth or falsity of the completion identifier in the data, train the temporal difference learning model, update the model parameters until the model converges, and obtain a trained temporal difference learning model. After obtaining the trained temporal difference learning model, the user observation data that has not been processed is predicted by the trained temporal difference learning model, and the user interest prediction value is output.

7. The system according to claim 6, characterized in that After the user observation data is input into the data processing module, The data processing module is further configured to copy the user observation data and gradually increase the observation time in the copied user observation data to obtain a new user observation data set; The data processing module is further configured to perform data processing based on the user data field and the observation time field in the user observation data set, and output a user observation change data set.

8. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the method for constructing a user interest model according to any one of claims 1 to 5.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method for constructing a user interest model according to any one of claims 1 to 5 when running.

Images