Duration prediction method, device and electronic device

Abstract

The present application discloses a duration prediction method, device and electronic device. The method includes: obtaining the information type of the target information, and determining the first feature corresponding to the target information based on the information type; obtaining the user type of the target user, and determining the feature source of the second feature corresponding to the target user based on the user type, and obtaining the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs; based on the first feature and the second feature, predicting the duration required for the target user to understand the target information. Therefore, the present application can automatically and accurately predict the required duration based on the first feature and the second feature, without manual intervention, saving labor costs, shortening the prediction time, ensuring the accuracy of the duration prediction results, and improving the efficiency, accuracy and reliability of the duration prediction process.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a duration prediction method, device and electronic device. Background Art

[0002] In recent years, with the explosive growth of Internet data, various resources have also grown exponentially. However, due to the differences in personality, education and other background information between individuals, different people take different amounts of time to understand different information. In particular, when faced with information such as finance and architecture, different people take different amounts of time to understand different information. Therefore, it is of great value to accurately predict the amount of time different people take to understand different information.

[0003] However, the duration prediction methods in related technologies often rely on manual prediction, which will inevitably lead to technical problems such as low efficiency, low accuracy, and poor reliability in the duration prediction process. Therefore, how to improve the efficiency and accuracy of the duration prediction process has become one of the important research directions. Summary of the invention

[0004] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0005] To this end, the first purpose of the present application is to propose a duration prediction method to solve the technical problems of low efficiency, low accuracy and poor reliability existing in the process of related duration prediction methods.

[0006] The second objective of the present invention is to provide a duration prediction device.

[0007] A third objective of the present invention is to provide an electronic device.

[0008] A fourth object of the present invention is to provide a computer-readable storage medium.

[0009] In order to achieve the above-mentioned purpose, the first aspect embodiment of the present application provides a duration prediction method, which includes the following steps: obtaining the information type of target information, and determining a first feature corresponding to the target information based on the information type; obtaining the user type of the target user, and determining the feature source of the second feature corresponding to the target user based on the user type, and obtaining the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs; based on the first feature and the second feature, predicting the time required for the target user to understand the target information.

[0010] In addition, the duration prediction method according to the above embodiment of the present application may also have the following additional technical features:

[0011] According to one embodiment of the present application, determining the first feature corresponding to the target information based on the information type includes: if the information type identifies the target information as new information, obtaining the information similarity between the target information and the seed information; based on the information similarity, obtaining the first feature corresponding to the target information; if the information type identifies the target information as old information, obtaining the feature of the target information itself as the first feature.

[0012] According to an embodiment of the present application, obtaining the first feature corresponding to the target information based on the information similarity includes: determining whether there is information similarity greater than a first similarity threshold; if there is information similarity greater than the first similarity threshold, selecting similar information of the target information from seed information corresponding to the information similarity greater than the first similarity threshold; and obtaining a feature of the similar information as the first feature.

[0013] According to an embodiment of the present application, selecting similar information of the target information from seed information corresponding to information similarity greater than the first similarity threshold includes: selecting seed information with the greatest information similarity among the seed information corresponding to information similarity greater than the first similarity threshold as the similar information.

[0014] According to an embodiment of the present application, the method further includes: if there is no information similarity greater than the first similarity threshold, identifying the target information group to which the target information belongs based on the characteristics of the target information itself; and acquiring the characteristics of the target information group as the first characteristic.

[0015] According to one embodiment of the present application, determining the feature source of the second feature corresponding to the target user based on the user type includes: if the user type identifies the target user as a new user, obtaining the user similarity between the target user and the seed user; based on the user similarity, obtaining the feature source of the second feature corresponding to the target user; if the user type identifies the target user as an old user, obtaining the feature of the target information itself as the second feature.

[0016] According to an embodiment of the present application, obtaining the feature source of the second feature corresponding to the target user based on the user similarity includes: determining whether there is a user similarity greater than a second similarity threshold; if there is a user similarity greater than the second similarity threshold, selecting similar seed users of the target user from seed users corresponding to the user similarities greater than the second similarity threshold; and using the seed similar users as the feature source of the target user, wherein the features of the similar users are used as the second features.

[0017] According to an embodiment of the present application, selecting a similar user to the target user from seed users corresponding to user similarities greater than the second similarity threshold includes: selecting a seed user with the greatest user similarity among the seed users corresponding to user similarities greater than the second similarity threshold as the similar seed user.

[0018] According to an embodiment of the present application, the method further includes: if there is no user similarity greater than the second similarity threshold, identifying the target population to which the target user belongs based on the characteristics of the target user itself; and using the target population as a feature source for the target user, wherein the characteristics of the target population are used as the second features.

[0019] According to one embodiment of the present application, predicting the time required for the target user to understand the target information based on the first feature and the second feature includes: obtaining a target prediction time model corresponding to the target user based on the user type and the information type; inputting the first feature and the second feature into the target prediction time model to predict the time required for the target user to understand the target information.

[0020] According to one embodiment of the present application, the target prediction duration model corresponding to the target user is obtained based on the user type and the information type, including: if the information type identifies the target information as new information and the user type identifies the target user as a new user, then obtaining the information similarity between the target information and the seed information, and obtaining the user similarity between the target user and the seed user; based on the user similarity and the information similarity, obtaining the target prediction duration model corresponding to the target user.

[0021] According to one embodiment of the present application, the target prediction duration model corresponding to the target user is obtained based on the user type and the information type, including: if the information type identifies the target information as new information and the user type identifies the target user as an old user, then obtaining the information similarity between the target information and the seed information; based on the information similarity, obtaining the target prediction duration model corresponding to the target user.

[0022] According to one embodiment of the present application, the target prediction duration model corresponding to the target user is obtained based on the user type and the information type, including: if the information type identifies the target information as old information and the user type identifies the target user as a new user, then obtaining the user similarity between the target user and the seed user; based on the user similarity, obtaining the target prediction duration model corresponding to the target user.

[0023] According to one embodiment of the present application, obtaining the information type of the target information includes: obtaining the cumulative number of times the target information is sent; if the cumulative number of times the target information is sent is greater than or equal to a first preset value, determining that the information type of the target information is a first information type, wherein the first information type represents that the target information is old information; if the cumulative number of times the target information is sent is less than the first preset value, determining that the information type of the target information is a second information type, wherein the second information type represents that the target information is new information.

[0024] According to one embodiment of the present application, obtaining the user type of the target user includes: obtaining the cumulative behavior data of the target user; if the cumulative behavior data is greater than or equal to a second preset value, determining that the user type of the target user is a first user type, wherein the first user type characterizes that the target user is an old user; if the cumulative behavior data is less than the second preset value, determining that the user type of the target user is a second user type, wherein the second user type characterizes that the target user is a new user.

[0025] According to one embodiment of the present application, the method also includes: obtaining first behavior feature data of the first candidate user; for each first candidate user, obtaining a score value of each first behavior feature data, and obtaining a first target score value of the first candidate user based on the score value of each behavior feature data; sorting the first target score values ​​of the first candidate users in descending order, and selecting the seed users from the first candidate users in order, wherein the similar seed user is one of the seed users.

[0026] According to one embodiment of the present application, the method also includes: obtaining a first static feature and a first behavioral feature data of a second candidate user; performing a clustering operation based on the first static feature and the first behavioral feature data to divide the second candidate users to form a candidate group, wherein the target group is a group of people in the candidate group.

[0027] According to one embodiment of the present application, the method also includes: obtaining second behavior characteristic data of the first candidate information; for each first candidate information, obtaining a scoring value of each second behavior characteristic data, and based on the scoring value of each second behavior characteristic data, obtaining a second target scoring value of the first candidate information; sorting the second target scoring values ​​of the first candidate information in order from large to small, and selecting the seed information from the first candidate information in order.

[0028] According to one embodiment of the present application, the method also includes: obtaining a second static feature and a second behavioral feature data of the second candidate information; performing a clustering operation based on the second static feature and the second behavioral feature data to divide the second candidate information to form a candidate information group.

[0029] The first aspect of the present application provides a duration prediction method, which can obtain the information type of target information, determine the first feature corresponding to the target information based on the information type, obtain the user type of the target user, and determine the feature source of the second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs, and then predict the duration required for the target user to understand the target information based on the first feature and the second feature to achieve duration prediction, so that the required duration can be automatically and accurately predicted based on the first feature and the second feature without manual intervention, thereby saving labor costs, shortening prediction time, ensuring the accuracy of the duration prediction results, and improving the efficiency, accuracy and reliability of the duration prediction process.

[0030] In order to achieve the above-mentioned purpose, the second aspect embodiment of the present application provides a duration prediction device, which includes: a first determination module, used to obtain the information type of target information, and determine the first feature corresponding to the target information based on the information type; a second determination module, used to obtain the user type of the target user, and determine the feature source of the second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs; a prediction module, used to predict the time required for the target user to understand the target information based on the first feature and the second feature.

[0031] According to one embodiment of the present application, the first determination module includes: a first acquisition submodule, used to obtain the information similarity between the target information and the seed information if the information type identifies the target information as new information; a second acquisition submodule, used to obtain the first feature corresponding to the target information based on the information similarity; and a third acquisition submodule, used to obtain the feature of the target information itself as the first feature if the information type identifies the target information as old information.

[0032] According to one embodiment of the present application, the second acquisition submodule includes: a first judgment unit, used to judge whether there is information similarity greater than a first similarity threshold; a first selection unit, used to select similar information of the target information from seed information corresponding to the information similarity greater than the first similarity threshold if there is information similarity greater than the first similarity threshold; a first determination unit, used to obtain a feature of the similar information as the first feature.

[0033] According to an embodiment of the present application, the first selection unit includes: a first determination subunit, configured to select seed information with the greatest information similarity among the seed information corresponding to the information similarity greater than the first similarity threshold as the similarity information.

[0034] According to an embodiment of the present application, it also includes: a first identification unit, used to identify the target information group to which the target information belongs based on the characteristics of the target information itself if there is no information similarity greater than the first similarity threshold; and a second determination unit, used to obtain the characteristics of the target information group as the first characteristic.

[0035] According to one embodiment of the present application, the second determination module includes: a fourth acquisition submodule, used to obtain the user similarity between the target user and the seed user if the user type identifies the target user as a new user; a fifth acquisition submodule, used to obtain the feature source of the second feature corresponding to the target user based on the user similarity; and a sixth acquisition submodule, used to obtain the feature of the target information itself as the second feature if the user type identifies the target user as an old user.

[0036] According to one embodiment of the present application, the fifth acquisition submodule includes: a second judgment unit, used to judge whether there is a user similarity greater than a second similarity threshold; a second selection unit, used to select similar users of the target user from seed users corresponding to the user similarities greater than the second similarity threshold if there is a user similarity greater than the second similarity threshold; a third determination unit, used to use the similar users as the feature source of the target user, wherein the features of the similar users are used as the second features.

[0037] According to an embodiment of the present application, the second selection unit includes: a second determination subunit, configured to select a seed user with the greatest user similarity among seed users corresponding to user similarities greater than the second similarity threshold as the similar seed user.

[0038] According to an embodiment of the present application, it also includes: a second identification unit, used to identify the target population to which the target user belongs based on the characteristics of the target user itself if there is no user similarity greater than the second similarity threshold; a fourth determination unit, used to use the target population as the feature source of the target user, wherein the characteristics of the target population are used as the second feature.

[0039] According to one embodiment of the present application, the prediction module includes: a seventh acquisition submodule, used to obtain a target prediction duration model corresponding to the target user based on the user type and the information type; a prediction submodule, used to input the first feature and the second feature into the target prediction duration model to predict the time required for the target user to understand the target information.

[0040] According to one embodiment of the present application, the seventh acquisition submodule includes: a fifth acquisition unit, used to obtain the information similarity between the target information and the seed information, and to obtain the user similarity between the target user and the seed user if the information type identifies the target information as new information and the user type identifies the target user as a new user; a sixth acquisition unit, used to obtain the target prediction duration model corresponding to the target user based on the user similarity and the information similarity.

[0041] According to one embodiment of the present application, the seventh acquisition submodule includes: a seventh acquisition unit, used to obtain the information similarity between the target information and the seed information if the information type identifies the target information as new information and the user type identifies the target user as an old user; an eighth acquisition unit, used to obtain the target prediction duration model corresponding to the target user based on the information similarity.

[0042] According to one embodiment of the present application, the seventh acquisition submodule includes: a ninth acquisition unit, used to obtain the user similarity between the target user and the seed user if the information type identifies the target information as old information and the user type identifies the target user as a new user; and a tenth acquisition unit, used to obtain the target prediction duration model corresponding to the target user based on the user similarity.

[0043] According to one embodiment of the present application, the first determination module includes: an eighth acquisition submodule, used to acquire the cumulative number of times the target information is sent; a first determination submodule, used to determine that the information type of the target information is a first information type if the cumulative number of times the target information is sent is greater than or equal to a first preset value, wherein the first information type indicates that the target information is old information; and a second determination submodule, used to determine that the information type of the target information is a second information type if the cumulative number of times the target information is sent is less than the first preset value, wherein the second information type indicates that the target information is new information.

[0044] According to one embodiment of the present application, the second determination module includes: a ninth acquisition submodule, used to acquire the cumulative behavior data of the target user; a third determination submodule, used to determine that the user type of the target user is a first user type if the cumulative behavior data is greater than or equal to a second preset value, wherein the first user type characterizes that the target user is an old user; and a fourth determination submodule, used to determine that the user type of the target user is a second user type if the cumulative behavior data is less than the second preset value, wherein the second user type characterizes that the target user is a new user.

[0045] According to one embodiment of the present application, the device also includes a first selection module, which is used to: obtain first behavior feature data of a first candidate user; for each first candidate user, obtain a score value of each first behavior feature data, and obtain a first target score value of the first candidate user based on the score value of each behavior feature data; sort the first target score values ​​of the first candidate users in descending order, and select the seed users from the first candidate users in order, wherein the similar seed user is one of the seed users.

[0046] According to one embodiment of the present application, the device also includes a first division module, which is used to: obtain a first static feature and a first behavioral feature data of a second candidate user; perform a clustering operation based on the first static feature and the first behavioral feature data, and divide the second candidate users to form a candidate group, wherein the target group is a group in the candidate group.

[0047] According to one embodiment of the present application, the device also includes a second selection module, which is used to: obtain second behavior feature data of the first candidate information; for each first candidate information, obtain a score value of each second behavior feature data, and obtain a second target score value of the first candidate information based on the score value of each second behavior feature data; sort the second target score values ​​of the first candidate information in order from large to small, and select the seed information from the first candidate information in order.

[0048] According to one embodiment of the present application, the device also includes a second division module, which is used to: obtain second static features and second behavioral feature data of second candidate information; perform clustering operations based on the second static features and the second behavioral feature data to divide the second candidate information to form a candidate information group.

[0049] The second aspect of the present application provides a duration prediction device, which can obtain the information type of target information, determine the first feature corresponding to the target information based on the information type, obtain the user type of the target user, and determine the feature source of the second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs, and then predict the duration required for the target user to understand the target information based on the first feature and the second feature to achieve duration prediction, so that the required duration can be automatically and accurately predicted based on the first feature and the second feature without manual intervention, thereby saving labor costs, shortening prediction time, ensuring the accuracy of the duration prediction results, and improving the efficiency, accuracy and reliability of the duration prediction process.

[0050] In order to achieve the above-mentioned purpose, the third aspect embodiment of the present application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, when the processor executes the program, it implements the duration prediction method as described in any one of the embodiments of the first aspect of the present application.

[0051] In order to achieve the above-mentioned purpose, the fourth aspect embodiment of the present application provides a computer-readable storage medium, and when the program is executed by a processor, it implements the duration prediction method as described in any one of the embodiments of the first aspect of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a flow chart of a duration prediction method disclosed in an embodiment of the present application;

[0053] Figure 2 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0054] Figure 3 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0055] Figure 4 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0056] Figure 5 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0057] Figure 6 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0058] Figure 7 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0059] Figure 8 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0060] Fig. 9 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0061] Fig.10 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0062] Fig.11 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0063] Fig.12 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0064] Fig.13 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0065] Fig.14 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0066] Fig.15 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0067] Fig.16 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0068] Fig.17 A schematic flow chart of a duration prediction method disclosed in another embodiment of the present application;

[0069] Fig.18 A flowchart of a process of obtaining a target prediction duration model according to different acquisition bases;

[0070] Fig.19 A schematic diagram of the structure of a duration prediction device disclosed in one embodiment of the present application;

[0071] Fig. 20 A schematic diagram of the structure of a duration prediction device disclosed in another embodiment of the present application;

[0072] Fig.21 A schematic diagram of the structure of a duration prediction device disclosed in another embodiment of the present application;

[0073] Fig. 22 A schematic diagram of the structure of a duration prediction device disclosed in another embodiment of the present application;

[0074] Fig.23 A schematic diagram of the structure of a duration prediction device disclosed in another embodiment of the present application;

[0075] Fig.24 A schematic diagram of the structure of a duration prediction device disclosed in another embodiment of the present application;

[0076] Fig.25 A schematic diagram of the structure of a duration prediction device disclosed in another embodiment of the present application;

[0077] Fig.26 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0078] In order to better understand the above technical solution, exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0079] It should be understood that the "and / or" involved in the embodiments of the present application describes the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0080] A duration prediction method, device, and electronic device proposed according to an embodiment of the present application are described below with reference to the accompanying drawings.

[0081] Figure 1 A flowchart of a duration prediction method disclosed in an embodiment of the present application is shown.

[0082] like Figure 1 As shown, the duration prediction method proposed in the embodiment of the present application specifically includes the following steps:

[0083] S101. Acquire the information type of target information, and determine a first feature corresponding to the target information based on the information type.

[0084] The target information may be any content, such as articles, videos, audios, etc.

[0085] The information type may include: a first information type and a second information type. It should be noted that the first information type may indicate that the target information is new information or old information, and the first information type and the second information type are inconsistent and may be set according to actual conditions.

[0086] In an embodiment of the present application, the information type of the target information can be acquired, and the first feature corresponding to the target information can be determined according to a pre-set first feature determination strategy.

[0087] S102. Obtain a user type of the target user, and determine a feature source of a second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, seed users similar to the target user, and a target population to which the target user belongs.

[0088] The target user may be any registered user.

[0089] The user type may include: a first user type and a second user type. It should be noted that the first user type may indicate that the target user is a new user or an old user, and the first user type and the second user type are inconsistent and may be set according to actual conditions.

[0090] In the embodiment of the present application, the user type of the target user may be acquired, and the second feature corresponding to the target user may be determined according to a preset second feature determination strategy.

[0091] S103: predicting the time required for the target user to understand the target information based on the first feature and the second feature.

[0092] It should be noted that the specific method for predicting the time required for the target user to understand the target information based on the first feature and the second feature is not limited in this application and can be selected according to actual conditions. Optionally, the first feature and the second feature can be used as inputs to a pre-trained target prediction time model to obtain a prediction result of the time required for the target user to understand the target information.

[0093] Therefore, the duration prediction method proposed in the present application can obtain the information type of the target information, determine the first feature corresponding to the target information based on the information type, obtain the user type of the target user, and determine the feature source of the second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs, and then predict the time required for the target user to understand the target information based on the first feature and the second feature to achieve duration prediction, so that the required duration can be automatically and accurately predicted based on the first feature and the second feature without manual intervention, saving labor costs, shortening prediction time, ensuring the accuracy of the duration prediction results, and improving the efficiency, accuracy and reliability of the duration prediction process.

[0094] It should be noted that in the present application, when attempting to determine the first feature corresponding to the target information and the second feature corresponding to the target user, it can be determined whether feature replacement is needed based on the user type of the target user or the information type of the target information.

[0095] For example, for old users, their accumulated features are usually more. In this case, their own features can be used without feature replacement. For new users, their accumulated features are usually less. In this case, using their own features cannot meet the requirements of accuracy and so feature replacement is required.

[0096] It should be noted that in the present application, when attempting to determine the first feature corresponding to the target information based on the information type, the corresponding first feature can be determined by matching the first feature determination strategy for different types of target information.

[0097] As a possible implementation, Figure 2 As shown, based on the above embodiment, the specific process of determining the first feature corresponding to the target information based on the information type includes the following steps:

[0098] S201. If the information type indicates that the target information is new information, obtain the information similarity between the target information and the seed information.

[0099] Among them, the seed information can be determined through pre-screening.

[0100] As a possible implementation method, the characteristic dimensions of the seed information may be determined. The characteristic dimensions of the seed information may include but are not limited to the following dimensions: the number of user behaviors on the content, such as the number of exposures, clicks, purchases, transactions, etc., the total time spent browsing the content, the last browsing time on the information, the total transaction amount on the information, etc.

[0101] Furthermore, a score may be assigned to each feature dimension, wherein one feature dimension corresponds to one score, such as any score between 1 and 5 on a 5-point scale, and the average of the scores of all feature dimensions is taken as the final score of the information.

[0102] Furthermore, the scores of all information may be arranged in descending order, and information in a first preset ranking may be selected as seed information. The first preset ranking may be set according to actual conditions, for example, may be set to 1, 3, 10, etc.

[0103] S202: Based on the information similarity, obtain a first feature corresponding to the target information.

[0104] As a possible implementation, Figure 3 As shown, based on the above embodiment, the specific process of obtaining the first feature corresponding to the target information based on the information similarity in the above step S202 includes the following steps:

[0105] S301: Determine whether there is information similarity greater than a first similarity threshold.

[0106] The first similarity threshold can be set according to actual conditions.

[0107] Optionally, the first similarity between the target information and each seed information can be calculated respectively, and then each first similarity can be compared with the first similarity threshold respectively, and when it is identified that any first similarity is greater than the first similarity threshold, it indicates that there is information similarity greater than the first similarity threshold, and step S302 can be executed.

[0108] S302: If there is information similarity greater than a first similarity threshold, select similar information of the target information from seed information corresponding to the information similarity greater than the first similarity threshold.

[0109] Optionally, seed information with the greatest information similarity among the seed information corresponding to the information similarity greater than the first similarity threshold may be selected as the similar information.

[0110] It should be noted that if there is no information similarity greater than the first similarity threshold, the feature of the target information group can be obtained as the first feature.

[0111] As a possible implementation, Figure 4 As shown, based on the above embodiment, the following steps are specifically included:

[0112] S401: If there is no information similarity greater than a first similarity threshold, identifying the target information group to which the target information belongs based on the characteristics of the target information itself.

[0113] It should be noted that the present application does not limit the specific method of identifying the target information group to which the target information belongs based on the characteristics of the target information itself, and it can be selected according to actual conditions.

[0114] As a possible approach, target information groups can be identified through clustering.

[0115] Optionally, information features can be constructed. Information features can include: static features and dynamic features. Static features can include but are not limited to the following features: category, business scenario, applicable population, content theme, content interest point, involved activity type, price factor and other features; dynamic features can include but are not limited to the following features: total number of information browsing, total browsing time, total transaction amount, total number of people, last browsing time and other features.

[0116] Furthermore, the constructed information features can be used as inputs to a pre-trained information clustering model for clustering processing. The information clustering model can be selected according to the actual situation. For example, any model such as KNN (K-NearestNeighbor, also known as K-proximity clustering), KMEANS (K-Means, also known as k-means clustering), DBSCAN (Density-Based Spatial Clustering of Applications with Noise, also known as density clustering) can be selected. In this case, the entire seed information will be clustered into n categories. At the same time, the values ​​of each dimension of the center of each category can be calculated.

[0117] S402: Acquire a feature of the target information group as a first feature.

[0118] In the embodiment of the present application, after identifying the target information group to which the target information belongs, the first feature of the target information may be replaced by the feature of the target information group.

[0119] S303: Acquire a feature of similar information as a first feature.

[0120] In the embodiment of the present application, after selecting similar information to the target information, the feature of the similar information may be used as the first feature of the target information.

[0121] S203: If the information type indicates that the target information is old information, a feature of the target information itself is obtained as a first feature.

[0122] In the embodiment of the present application, if the information type identifies the target information as old information, it means that its own features can be used without feature replacement, and the features of the target information itself can be directly obtained as the first feature.

[0123] Therefore, the duration prediction method proposed in the present application can determine whether feature replacement is required based on the user type of the target user or the information type of the target information. Optionally, feature replacement is not required for old users, but feature replacement is required for new users, so as to ensure that the determined first feature is more accurate, further improving the efficiency, accuracy and reliability of the duration prediction process.

[0124] It should be noted that in the present application, when attempting to determine the second feature corresponding to the target user based on the user type, the corresponding second feature can be determined by matching the second feature determination strategy for different types of target users.

[0125] As a possible implementation, Figure 5 As shown, based on the above embodiment, the specific process of determining the feature source of the second feature corresponding to the target user based on the user type includes the following steps:

[0126] S501: If the user type identifies the target user as a new user, obtain the user similarity between the target user and the seed user.

[0127] Among them, seed users can be determined through pre-screening.

[0128] As a possible implementation method, the characteristic dimensions of seed users can be determined. The characteristic dimensions of seed users may include but are not limited to the following dimensions: the number of user behaviors on the application (Application, APP for short), such as the number of exposures, clicks, purchases, transactions, etc., the total duration of the behavior, the last browsing time on the APP, the total transaction amount on the APP, etc.

[0129] Furthermore, a score may be assigned to each feature dimension, wherein one feature dimension corresponds to one score, such as any score between 1 and 5 on a 5-point scale, and the average of the scores of all feature dimensions is taken as the final score of the user.

[0130] Furthermore, the scores of all users may be arranged in descending order, and information in the second preset ranking may be selected as seed information. The second preset ranking may be set according to actual conditions, for example, may be set to 1, 3, 5, etc.

[0131] S502: Based on the user similarity, obtain a feature source of a second feature corresponding to the target user.

[0132] As a possible implementation, Figure 6As shown, based on the above embodiment, the specific process of obtaining the feature source of the second feature corresponding to the target user based on the user similarity in the above step S502 includes the following steps:

[0133] S601: Determine whether there is a user similarity greater than a second similarity threshold.

[0134] The second similarity threshold can be set according to actual conditions.

[0135] Optionally, the second similarity between the target user and each seed user may be calculated respectively, and then each second similarity may be compared with a second similarity threshold respectively, and when any second similarity is identified to be greater than the second similarity threshold, it indicates that there is a user similarity greater than the second similarity threshold, and step S602 may be executed.

[0136] S602: If there is a user similarity greater than a second similarity threshold, select a similar seed user to the target user from seed users corresponding to the user similarities greater than the second similarity threshold.

[0137] Optionally, a seed user with the greatest user similarity among the seed users corresponding to the user similarities greater than the second similarity threshold may be selected as a similar seed user.

[0138] It should be noted that if there is no user similarity greater than the second similarity threshold, the characteristics of the target population can be obtained as the second characteristics.

[0139] As a possible implementation, Figure 7 As shown, based on the above embodiment, the following steps are specifically included:

[0140] S701: If there is no user similarity greater than a second similarity threshold, identify the target group to which the target user belongs based on the characteristics of the target user.

[0141] It should be noted that the present application does not limit the specific method of identifying the target population to which the target user belongs based on the characteristics of the target user himself, and it can be selected according to actual conditions.

[0142] As a possible approach, target populations can be identified through clustering.

[0143] Optionally, user features can be constructed. User features can include: static features and dynamic features. Static features can include but are not limited to the following features: gender, age, city, education level, occupation, income stratification, etc.; dynamic features can include but are not limited to the following features: number of views, viewing time, transaction amount, last viewing time, etc. on each content category.

[0144] Furthermore, the constructed user features can be used as inputs to a pre-trained user clustering model to perform clustering operations. The information clustering model can be selected according to the actual situation. For example, any model such as KNN, KMEANS, DBSCAN, etc. can be selected. In this case, the entire seed user will be clustered into m categories. At the same time, the values ​​of each dimension of the center of each category can be calculated.

[0145] S702: Use the target population as a feature source of the target users, wherein the feature of the target population is used as a second feature.

[0146] In the embodiment of the present application, after identifying the target population to which the target user belongs, the second feature of the target user may be replaced by the feature of the target population.

[0147] S603: Use seed similar users as feature sources of target users, wherein the features of similar users are used as second features.

[0148] In the embodiment of the present application, after selecting similar users of the target user, the characteristics of the similar users may be used as the second characteristics of the target user.

[0149] S503: If the user type identifies the target user as an old user, obtain a feature of the target information itself as a second feature.

[0150] In the embodiment of the present application, if the user type identifies the target user as an old user, it means that the target user's own features can be used without feature replacement, and the target user's own features can be directly obtained as the second features.

[0151] Therefore, the duration prediction method proposed in the present application can determine whether feature replacement is required based on the user type of the target user or the information type of the target information. Optionally, feature replacement is not required for old users, but feature replacement is required for new users, so that the determined second feature can be ensured to be more accurate, further improving the efficiency, accuracy and reliability of the duration prediction process.

[0152] Furthermore, in the present application, when attempting to predict the time required for the target user to understand the target information based on the first feature and the second feature, the required time can be predicted based on the target prediction time model.

[0153] As a possible implementation, Figure 8 As shown, based on the above embodiment, the specific process of predicting the time required for the target user to understand the target information according to the first feature and the second feature includes the following steps:

[0154] S801. Obtain a target prediction duration model corresponding to a target user based on user type and information type.

[0155] It should be noted that the combination of different user types and different information types corresponds to different methods for obtaining the target prediction duration model. In other words, the methods for obtaining the target prediction duration model are different for new information and new users, new information and old users, and old information and new users.

[0156] It should be noted that multiple trained target prediction duration models are pre-set in the present application. After determining the user type and information type, the user type and information type can be comprehensively considered to select a matching target prediction duration model based on different criteria.

[0157] The following explains the process of obtaining the target prediction duration model corresponding to the target user based on the user type and information type for new information and new users, new information and old users, and old information and new users respectively.

[0158] As a possible implementation method for new information and new users, Fig. 9 As shown, based on the above embodiment, the specific process of obtaining the target prediction duration model corresponding to the target user based on the user type and the information type in the above step S801 includes the following steps:

[0159] S901. If the information type identifies the target information as new information and the user type identifies the target user as a new user, obtain information similarity between the target information and the seed information, and obtain user similarity between the target user and the seed user.

[0160] In an embodiment of the present application, if the information type identifies the target information as new information and the user type identifies the target user as a new user, it means that user similarity and information similarity need to be replaced. Then, the information similarity between the target information and the seed information can be obtained, and the user similarity between the target user and the seed user can be obtained.

[0161] S902: Obtain a target prediction duration model corresponding to the target user according to user similarity and information similarity.

[0162] In an embodiment of the present application, if the information type identifies the target information as new information and the user type identifies the target user as a new user, the target prediction duration model corresponding to the target user can be obtained based on user similarity and information similarity.

[0163] As a possible implementation method for new information and old users, Fig.10As shown, based on the above embodiment, the specific process of obtaining the target prediction duration model corresponding to the target user based on the user type and the information type in the above step S801 includes the following steps:

[0164] S1001. If the information type identifies the target information as new information and the user type identifies the target user as an old user, then obtain information similarity between the target information and the seed information.

[0165] In the embodiment of the present application, if the information type identifies the target information as new information and the user type identifies the target user as an old user, indicating that information similarity replacement is required, then only the information similarity between the target information and the seed information may be obtained.

[0166] S1002. Obtain a target prediction duration model corresponding to the target user according to information similarity.

[0167] In the embodiment of the present application, if the information type identifies the target information as new information and the user type identifies the target user as an old user, the target prediction duration model corresponding to the target user can be obtained based on the information similarity.

[0168] As a possible implementation method for old information and new users, Fig.11 As shown, based on the above embodiment, the specific process of obtaining the target prediction duration model corresponding to the target user based on the user type and the information type in the above step S801 includes the following steps:

[0169] S1101: If the information type identifies the target information as old information and the user type identifies the target user as a new user, obtain the user similarity between the target user and the seed user.

[0170] In the embodiment of the present application, if the information type identifies the target information as old information and the user type identifies the target user as a new user, indicating that user similarity replacement is required, only the user similarity between the target user and the seed user may be obtained.

[0171] S1102: Obtain a target prediction duration model corresponding to the target user according to user similarity.

[0172] In the embodiment of the present application, if the information type identifies the target information as old information and the user type identifies the target user as a new user, the target prediction duration model corresponding to the target user can be obtained based on the user similarity.

[0173] S802: Input the first feature and the second feature into a target prediction duration model to predict the duration required for the target user to understand the target information.

[0174] In an embodiment of the present application, the first feature and the second feature may be used as inputs into the obtained matching target prediction duration model to obtain a prediction result of the duration required by the target user to understand the target information.

[0175] Therefore, the duration prediction method proposed in the present application can determine the basis for obtaining the matching target prediction duration model by combining different user types and different information types, and then obtain the corresponding target prediction duration model based on the determined basis, so that the matching target prediction duration model can be determined more accurately for different user types and different information types, further improving the efficiency, accuracy and reliability of the duration prediction process.

[0176] It should be noted that, in the present application, when attempting to obtain the information type of the target information, the information type can be obtained based on a pre-set information acquisition strategy.

[0177] As a possible implementation, Fig.12 As shown, based on the above embodiment, the specific process of obtaining the information type of the target information includes the following steps:

[0178] S1201. Obtain the cumulative number of times the target information is sent.

[0179] In the embodiment of the present application, the cumulative number of times the target information is sent can be obtained based on the information sending history records in the database.

[0180] The cumulative number of transmissions may be the sum of all transmissions after the target information is established, or the sum of the transmissions of the target information within the first preset time period. The first preset time period may be set according to actual conditions, for example, 1 week, 3 months, etc.

[0181] S1202: If the cumulative number of transmissions is greater than or equal to a first preset value, determine that the information type of the target information is a first information type, wherein the first information type indicates that the target information is old information.

[0182] The first preset value may be set according to actual conditions, for example, it may be set to 100 times, 10,000 times, etc.

[0183] For example, the cumulative number of times the target information is sent is 15,000 times, and the first preset value is 10,000 times. In this case, the cumulative number of times the target information is sent is greater than the first preset value, and the information type of the target information is determined to be the first information type, wherein the first information type indicates that the target information is old information.

[0184] S1203: If the cumulative number of transmissions is less than a first preset value, determine that the information type of the target information is a second information type, wherein the second information type indicates that the target information is new information.

[0185] For example, the cumulative number of times the target information is sent is 6,000 times, and the first preset value is 10,000 times. In this case, the cumulative number of times the target information is sent is less than the first preset value, and the information type of the target information is determined to be the second information type, wherein the second information type indicates that the target information is new information.

[0186] Therefore, the duration prediction method proposed in the present application can obtain the cumulative number of times the target information is sent. If the cumulative number of times the target information is sent is greater than or equal to a first preset value, the information type of the target information is determined to be the first information type; if the cumulative number of times the target information is sent is less than the first preset value, the information type of the target information is determined to be the second information type, so that new information and old information can be automatically distinguished more accurately and efficiently, further improving the efficiency, accuracy and reliability of the duration prediction process.

[0187] It should be noted that, in the present application, when attempting to obtain the user type of the target user, the user type may be obtained based on a pre-set acquisition strategy.

[0188] As a possible implementation, Fig.13 As shown, based on the above embodiment, the specific process of obtaining the user type of the target user includes the following steps:

[0189] S1301. Obtain cumulative behavior data of target users.

[0190] In the embodiment of the present application, the accumulated behavior data of the target user can be obtained based on the user behavior history records in the database.

[0191] The accumulated behavior data may be the sum of all behavior data of the target user after registration, or the sum of behavior data of the target user within a second preset time period. The second preset time period may be set according to actual conditions, for example, 1 week, 3 months, etc.

[0192] S1302: If the accumulated behavior data is greater than or equal to a second preset value, determine that the user type of the target user is a first user type, wherein the first user type indicates that the target user is an old user.

[0193] The second preset value may be set according to actual conditions, for example, it may be set to 100 times, 500 times, etc.

[0194] For example, the accumulated behavior data of the target user is 1000 times, and the second preset value is 500 times. In this case, the accumulated behavior data is greater than the second preset value, and the user type of the target user is determined to be the first user type, wherein the first user type indicates that the target user is an old user.

[0195] S1303: If the accumulated behavior data is less than a second preset value, determine that the user type of the target user is a second user type, wherein the second user type indicates that the target user is a new user.

[0196] For example, the cumulative behavior data of the target user is 50 times, and the second preset value is 500 times. In this case, the cumulative behavior data is less than the second preset value, and the user type of the target user is determined to be the second user type, wherein the second user type indicates that the target user is a new user.

[0197] Therefore, the duration prediction method proposed in the present application can obtain the cumulative behavior data of the target user. If the cumulative behavior data is greater than or equal to the second preset value, the user type of the target user is determined to be the first user type; if the cumulative behavior data is less than the second preset value, the user type of the target user is determined to be the second user type, so that new users and old users can be automatically distinguished more accurately and efficiently, further improving the efficiency, accuracy and reliability of the duration prediction process.

[0198] Furthermore, in the present application, seed users and seed information may be selected separately.

[0199] The following explains the process of selecting seed users and seed information respectively.

[0200] As a possible implementation method for selecting seed users, Fig.14 As shown, based on the above embodiment, the following steps are specifically included:

[0201] S1401. Obtain first behavior characteristic data of a first candidate user.

[0202] S1402: For each first candidate user, obtain a score value of each first behavior characteristic data, and based on the score value of each behavior characteristic data, obtain a first target score value of the first candidate user.

[0203] S1403: Sort the first target rating values ​​of the first candidate users in descending order, and select seed users from the first candidate users in order, wherein the similar seed user is one of the seed users.

[0204] As a possible implementation method for selecting seed information, Fig.15As shown, based on the above embodiment, the following steps are specifically included:

[0205] S1501. Obtain second behavior characteristic data of first candidate information.

[0206] S1502: For each piece of first candidate information, obtain a score value of each piece of second behavior feature data, and based on the score value of each piece of second behavior feature data, obtain a second target score value of the first candidate information.

[0207] S1503: Sort the second target score values ​​of the first candidate information in descending order, and select seed information from the first candidate information in order.

[0208] Furthermore, in the present application, candidate groups and candidate information groups can be formed by division respectively.

[0209] The processes of forming a candidate group and a candidate information group are explained below respectively.

[0210] As a possible implementation method for forming a candidate group, Fig.16 As shown, based on the above embodiment, the following steps are specifically included:

[0211] S1601. Obtain first static feature and first behavioral feature data of a second candidate user.

[0212] S1602: Perform a clustering operation based on the first static feature and the first behavioral feature data to divide the second candidate users into candidate groups, wherein the target group is a group in the candidate group.

[0213] As a possible implementation method for forming a candidate information group, Fig.17 As shown, based on the above embodiment, the following steps are specifically included:

[0214] S1701. Obtain second static feature and second behavioral feature data of second candidate information.

[0215] S1702: Perform a clustering operation based on the second static feature and the second behavior feature data to divide the second candidate information to form a candidate information group.

[0216] The process of selecting seed users and seed information and forming candidate groups and candidate information groups through division is explained below with reference to specific examples.

[0217] It should be noted that in this application, the target prediction duration model is pre-trained. As a possible implementation method, the candidate prediction duration models to be trained can be determined according to the pre-set number of models, and multiple candidate prediction duration models to be trained can be trained respectively to generate trained candidate prediction duration models. In this case, when the matching model is determined according to the model acquisition basis, the target prediction duration models obtained are all trained models.

[0218] Optionally, information samples and user samples may be acquired.

[0219] Among them, information samples can be screened based on seed information.

[0220] Among them, user samples can be screened based on seed users.

[0221] In the embodiment of the present application, based on the seed information and the seed user, information samples and user samples can be obtained according to the first preset screening number and the second preset screening number to serve as the original data for training the candidate prediction duration model to be trained. The first preset screening number and the second preset screening number can be set according to actual conditions.

[0222] Furthermore, feature and label information may be obtained based on information samples and user samples.

[0223] Among them, characteristics may include: user characteristics, information characteristics, and user information relationship characteristics.

[0224] Among them, user characteristics may include: user static characteristics and user dynamic characteristics. Among them, user static characteristics may include but are not limited to the following characteristics: gender, age, city, education level, occupation, income level and other characteristics; user dynamic characteristics may include but are not limited to the following characteristics: number of views, viewing time, transaction amount, last viewing time and other characteristics in each content category.

[0225] Among them, information features may include: information static features and information dynamic features. Among them, information static features may include but are not limited to the following features: category, business scenario, applicable population, content theme, content interest point, involved activity type, price factor and other features; information dynamic features may include but are not limited to the following features: total number of views of the content, total viewing time, total transaction amount, total number of people, last viewing time and other features.

[0226] Among them, user information relationship characteristics may include but are not limited to the following characteristics: the number of times a user browses the information, the browsing time, the number of transactions, the transaction amount and other characteristics, or the number of times a user browses the information category, the browsing time, the number of transactions, the transaction amount and other characteristics.

[0227] The tag information may include, but is not limited to, the following information: the total time that a user spends consulting on a piece of information. For example, the total time that user A spends on telephone communication with information A from the first time he calls customer service to the time he successfully purchases the information.

[0228] Furthermore, the features and labels of the information samples and user samples may be input into the candidate prediction duration model to be trained, and the candidate prediction duration model to be trained may be trained to generate a trained candidate prediction duration model.

[0229] In the embodiment of the present application, the candidate prediction duration model to be trained can be trained according to a variety of regression models based on machine learning (ML), such as XGBOOST, Random Forest, etc. Further, the trained model can be output to save the results in the form of files.

[0230] It should be noted that the training process is consistent for different candidate prediction duration models to be trained, and the features and label information input during the training process can be adjusted accordingly for different situations.

[0231] It should be noted that due to the combination of different user types and different information types, different methods of obtaining the target prediction duration model correspond to each other. In this case, the target prediction duration models obtained based on different bases are also different.

[0232] For example, Fig.18 As shown, after determining the basis for obtaining the target prediction duration model based on information similarity and user similarity, the target prediction duration model can be obtained based on the following different acquisition bases.

[0233] If the acquisition basis is the seed user and the seed information, the target prediction duration model can be determined as the corresponding user-information model.

[0234] If the acquisition basis is user category and seed information, the target prediction duration model can be determined as the corresponding user category-information model.

[0235] If the acquisition basis is the seed user and the information category, the target prediction duration model can be determined as the corresponding user-information category model.

[0236] If the acquisition basis is a user category and an information category, the target prediction duration model can be determined as a corresponding user category-information category model.

[0237] In summary, in this application, a matching algorithm can be used to predict the time required for the target user to understand the target information for different algorithm scenarios. Optionally, for the case with the most similar seed user and the most similar seed information, a user-information model is used for prediction; for the case with a user category and the most similar seed information, a user category-information model is used for prediction; for the case with the most similar seed user and information category, a user-information category model is used for prediction; for the case with a user category and information category, a user category-information category model is used for prediction, which further improves the efficiency, accuracy and reliability of the duration prediction process.

[0238] Based on the same application concept, an embodiment of the present application also provides a device corresponding to a duration prediction method.

[0239] Fig.19 A schematic diagram of the structure of a duration prediction device provided in an embodiment of the present application.

[0240] like Fig.19 As shown, the duration prediction device 1000 includes: a first determination module 110, a second determination module 120 and a prediction module 130.

[0241] A first determination module 110, configured to obtain an information type of target information, and determine a first feature corresponding to the target information based on the information type;

[0242] A second determination module 120 is used to obtain a user type of a target user, and determine a feature source of a second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, seed users similar to the target user, and a target group to which the target user belongs;

[0243] The prediction module 130 is used to predict the time required for the target user to understand the target information based on the first feature and the second feature.

[0244] According to one embodiment of the present application, Fig. 20 As shown, Fig.19 The first determining module 110 in the embodiment comprises:

[0245] A first acquisition submodule 111 is configured to acquire information similarity between the target information and the seed information if the information type indicates that the target information is new information;

[0246] A second acquisition submodule 112 is used to acquire a first feature corresponding to the target information based on the information similarity;

[0247] The third acquisition submodule 113 is configured to acquire a feature of the target information itself as the first feature if the information type indicates that the target information is old information.

[0248] According to one embodiment of the present application, the second acquisition submodule 112 includes:

[0249] The first judging unit 1121 is used to judge whether there is information similarity greater than a first similarity threshold;

[0250] A first selection unit 1122 is configured to select similar information of the target information from seed information corresponding to the information similarity greater than the first similarity threshold if there is information similarity greater than the first similarity threshold;

[0251] The first determining unit 1123 is configured to obtain a feature of the similar information as the first feature.

[0252] The first selection unit 1122 includes:

[0253] The first determining subunit 11221 is configured to select seed information having the greatest information similarity from among the seed information corresponding to information similarity greater than the first similarity threshold as the similar information.

[0254] According to an embodiment of the present application, the second acquisition submodule 112 further includes:

[0255] A first identification unit 1124 is configured to identify the target information group to which the target information belongs based on the characteristics of the target information itself if there is no information similarity greater than the first similarity threshold;

[0256] The second determining unit 1125 is configured to obtain a feature of the target information group as the first feature.

[0257] According to one embodiment of the present application, Fig.21 As shown, Fig.19 The second determining module 120 in the embodiment comprises:

[0258] A fourth acquisition submodule 121 is configured to acquire a user similarity between the target user and a seed user if the user type identifies the target user as a new user;

[0259] A fifth acquisition submodule 122, configured to acquire a second feature corresponding to the target user based on the user similarity;

[0260] The sixth acquisition submodule 123 is configured to acquire a feature of the target information itself as the second feature if the user type identifies the target user as an old user.

[0261] The fifth acquisition submodule 122 includes:

[0262] The second determination unit 1221 is used to determine whether there is a user similarity greater than a second similarity threshold;

[0263] A second selection unit 1222 is configured to select a similar user to the target user from seed users corresponding to the user similarities greater than the second similarity threshold if there is a user similarity greater than the second similarity threshold;

[0264] The third determining unit 1223 is configured to use the similar users as feature sources of the target user, wherein the features of the similar users are used as the second features.

[0265] The second selection unit 1222 includes:

[0266] The second determining subunit 12221 is configured to select a seed user with the greatest user similarity among the seed users corresponding to user similarities greater than the second similarity threshold as the similar seed user.

[0267] The fifth acquisition submodule 122 further includes:

[0268] The second identification unit 1224 is configured to identify the target group to which the target user belongs based on the characteristics of the target user if there is no user similarity greater than the second similarity threshold;

[0269] The fourth determining unit 1225 is configured to use the target group as a feature source of the target user, wherein the feature of the target group is used as the second feature.

[0270] According to one embodiment of the present application, Fig. 22 As shown, Fig.19 The prediction module 130 in the embodiment comprises:

[0271] A seventh acquisition submodule 131 is used to acquire a target prediction duration model corresponding to the target user based on the user type and the information type;

[0272] The prediction submodule 132 is used to input the first feature and the second feature into the target prediction time model to predict the time required for the target user to understand the target information.

[0273] The seventh acquisition submodule 131 includes:

[0274] A fifth acquiring unit 1311 is configured to acquire information similarity between the target information and the seed information, and acquire user similarity between the target user and the seed user if the information type identifies the target information as new information and the user type identifies the target user as a new user;

[0275] A sixth acquisition unit 1312 is used to acquire a target prediction duration model corresponding to the target user according to the user similarity and the information similarity;

[0276] A seventh acquisition unit 1313 is configured to acquire information similarity between the target information and the seed information if the information type identifies that the target information is new information and the user type identifies that the target user is an old user;

[0277] An eighth acquisition unit 1314 is configured to acquire a target prediction duration model corresponding to the target user according to the information similarity;

[0278] A ninth acquisition unit 1315, configured to acquire a user similarity between the target user and the seed user if the information type identifies that the target information is old information and the user type identifies that the target user is a new user;

[0279] The tenth obtaining unit 1316 is used to obtain the target prediction duration model corresponding to the target user according to the user similarity.

[0280] According to one embodiment of the present application, Fig.23 As shown, Fig.19 The first determining module 110 in the embodiment comprises:

[0281] An eighth acquisition submodule 114 is used to acquire the cumulative number of times the target information is sent;

[0282] A first determining submodule 115 is configured to determine that the information type of the target information is a first information type if the cumulative number of transmissions is greater than or equal to a first preset value, wherein the first information type indicates that the target information is old information;

[0283] The second determining submodule 116 is configured to determine that the information type of the target information is a second information type if the cumulative number of transmissions is less than the first preset value, wherein the second information type indicates that the target information is new information.

[0284] According to one embodiment of the present application, Fig.24 As shown, Fig.19 The second determining module 120 in the embodiment comprises:

[0285] A ninth acquisition submodule 124, used to acquire the accumulated behavior data of the target user;

[0286] A third determination submodule 125 is configured to determine that the user type of the target user is a first user type if the accumulated behavior data is greater than or equal to a second preset value, wherein the first user type indicates that the target user is an old user;

[0287] The fourth determination submodule 126 is configured to determine that the user type of the target user is a second user type if the accumulated behavior data is less than the second preset value, wherein the second user type indicates that the target user is a new user.

[0288] According to one embodiment of the present application, Fig.25 As shown, the duration prediction device 1000 also includes: a first selection module 140, which is used to: obtain the first behavior feature data of the first candidate user; for each first candidate user, obtain the score value of each first behavior feature data, and obtain the first target score value of the first candidate user based on the score value of each behavior feature data; sort the first target score values ​​of the first candidate users in descending order, and select the seed users from the first candidate users in order, wherein the similar seed user is one of the seed users.

[0289] According to one embodiment of the present application, Fig.25 As shown, the duration prediction device 1000 also includes: a first division module 150, used to: obtain the first static feature and the first behavioral feature data of the second candidate user; perform a clustering operation based on the first static feature and the first behavioral feature data, and divide the second candidate users to form a candidate group, wherein the target group is a group in the candidate group.

[0290] According to one embodiment of the present application, Fig.25 As shown, the duration prediction device 1000 also includes: a second selection module 160, which is used to: obtain second behavior feature data of the first candidate information; for each first candidate information, obtain the score value of each second behavior feature data, and obtain the second target score value of the first candidate information based on the score value of each second behavior feature data; sort the second target score values ​​of the first candidate information in descending order, and select the seed information from the first candidate information in order.

[0291] According to one embodiment of the present application, Fig.25As shown, the duration prediction device 1000 also includes: a second division module 170, which is used to: obtain the second static feature and the second behavioral feature data of the second candidate information; perform clustering operation based on the second static feature and the second behavioral feature data, and divide the second candidate information to form a candidate information group.

[0292] Therefore, the duration prediction device proposed in the present application can obtain the information type of the target information, determine the first feature corresponding to the target information based on the information type, obtain the user type of the target user, and determine the feature source of the second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, similar seed users to the target user, and the target population to which the target user belongs, and then predict the duration required for the target user to understand the target information based on the first feature and the second feature to achieve duration prediction, so that the required duration can be automatically and accurately predicted based on the first feature and the second feature without manual intervention, thereby saving labor costs, shortening prediction time, ensuring the accuracy of the duration prediction results, and improving the efficiency, accuracy and reliability of the duration prediction process.

[0293] Based on the same application concept, an embodiment of the present application also provides an electronic device.

[0294] Fig.26 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig.26 As shown, the electronic device 3000 includes a memory 310, a processor 320, and a computer program stored in the memory 310 and executable on the processor 320. When the processor executes the program, the aforementioned duration prediction method is implemented.

[0295] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0296] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0297] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0298] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0299] It should be noted that in the claims, any reference signs placed between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The present application may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0300] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0301] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A duration prediction method, characterized in that: include: Acquire an information type of target information, and determine a first feature corresponding to the target information based on the information type, where the information type of the target information includes new information and old information; Acquire a user type of a target user, and determine a feature source of a second feature corresponding to the target user based on the user type, and acquire the second feature according to the feature source, wherein the feature source includes the target user, similar seed users of the target user, and a target group to which the target user belongs, and the user type of the target user includes a new user and an old user; According to the first feature and the second feature, a target prediction duration model is used to predict the duration required for the target user to understand the target information.

2. The duration prediction method according to claim 1, characterized in that: The determining, based on the information type, a first feature corresponding to the target information includes: If the information type identifies the target information as new information, obtaining information similarity between the target information and the seed information; Based on the information similarity, obtaining a first feature corresponding to the target information; If the information type identifies that the target information is old information, a feature of the target information itself is acquired as the first feature.

3. The duration prediction method according to claim 2, characterized in that: The acquiring, based on the information similarity, a first feature corresponding to the target information includes: Determining whether there is information similarity greater than a first similarity threshold; If there is information similarity greater than the first similarity threshold, selecting similar information of the target information from the seed information corresponding to the information similarity greater than the first similarity threshold; A feature of the similar information is acquired as the first feature.

4. The duration prediction method according to claim 3, characterized in that: The selecting similar information of the target information from the seed information corresponding to the information similarity greater than the first similarity threshold includes: The seed information having the greatest information similarity among the seed information corresponding to the information similarity greater than the first similarity threshold is selected as the similar information.

5. The duration prediction method according to claim 3, characterized in that: Also includes: If there is no information similarity greater than the first similarity threshold, identifying the target information group to which the target information belongs based on the characteristics of the target information itself; A feature of the target information group is acquired as the first feature.

6. The duration prediction method according to claim 1, characterized in that: The determining, based on the user type, a feature source of the second feature corresponding to the target user includes: If the user type identifies the target user as a new user, obtaining the user similarity between the target user and the seed user; Based on the user similarity, obtaining a feature source of a second feature corresponding to the target user; If the user type identifies the target user as an old user, a feature of the target user itself is obtained as the second feature.

7. The duration prediction method according to claim 6, characterized in that: The acquiring, based on the user similarity, a feature source of a second feature corresponding to the target user includes: Determining whether there is a user similarity greater than a second similarity threshold; If there is a user similarity greater than the second similarity threshold, selecting a similar seed user to the target user from seed users corresponding to the user similarities greater than the second similarity threshold; The similar seed users are used as feature sources of the target users, wherein the features of the similar seed users are used as the second features.

8. The duration prediction method according to claim 7, characterized in that: The selecting a similar seed user to the target user from seed users corresponding to user similarities greater than the second similarity threshold comprises: A seed user with the greatest user similarity among the seed users corresponding to the user similarities greater than the second similarity threshold is selected as the similar seed user.

9. The duration prediction method according to claim 7, characterized in that: Also includes: If there is no user similarity greater than the second similarity threshold, identifying the target group to which the target user belongs based on the characteristics of the target user; The target population is used as a feature source of the target users, wherein the features of the target population are used as the second features.

10. The duration prediction method according to any one of claims 1 to 9, characterized in that: The predicting, based on the first feature and the second feature, the time required for the target user to understand the target information includes: Based on the user type and the information type, obtaining a target prediction duration model corresponding to the target user; The first feature and the second feature are input into the target prediction duration model to predict the duration required for the target user to understand the target information.

11. The duration prediction method according to claim 10, characterized in that: The acquiring, based on the user type and the information type, a target prediction duration model corresponding to the target user includes: If the information type identifies the target information as new information and the user type identifies the target user as a new user, then obtaining information similarity between the target information and the seed information, and obtaining user similarity between the target user and the seed user; A target prediction duration model corresponding to the target user is obtained according to the user similarity and the information similarity.

12. The duration prediction method according to claim 10, characterized in that: The acquiring, based on the user type and the information type, a target prediction duration model corresponding to the target user includes: If the information type identifies the target information as new information and the user type identifies the target user as an old user, obtaining information similarity between the target information and the seed information; According to the information similarity, a target prediction duration model corresponding to the target user is obtained.

13. The duration prediction method according to claim 10, characterized in that: The acquiring, based on the user type and the information type, a target prediction duration model corresponding to the target user includes: If the information type identifies that the target information is old information and the user type identifies that the target user is a new user, obtaining a user similarity between the target user and the seed user; According to the user similarity, a target prediction duration model corresponding to the target user is obtained.

14. The duration prediction method according to claim 1, characterized in that: The information types of the target information to be acquired include: Obtain the cumulative number of times the target information is sent; If the cumulative number of transmissions is greater than or equal to a first preset value, determining that the information type of the target information is a first information type, wherein the first information type indicates that the target information is old information; If the accumulated number of transmissions is less than the first preset value, it is determined that the information type of the target information is a second information type, wherein the second information type indicates that the target information is new information.

15. The duration prediction method according to claim 1, characterized in that: The method of obtaining the user type of the target user includes: Obtaining the accumulated behavior data of the target user; If the accumulated behavior data is greater than or equal to a second preset value, determining that the user type of the target user is a first user type, wherein the first user type indicates that the target user is an old user; If the accumulated behavior data is less than the second preset value, it is determined that the user type of the target user is a second user type, wherein the second user type indicates that the target user is a new user.

16. The duration prediction method according to claim 1, characterized in that: The method further comprises: Acquire first behavior characteristic data of a first candidate user; For each first candidate user, obtaining a score value of each first behavior feature data, and obtaining a first target score value of the first candidate user based on the score value of each behavior feature data; The first target rating values ​​of the first candidate users are sorted in descending order, and the seed users are selected from the first candidate users in order, wherein the similar seed user is one of the seed users.

17. The duration prediction method according to claim 1, characterized in that: The method further comprises: Acquire first static feature and first behavioral feature data of a second candidate user; A clustering operation is performed based on the first static feature and the first behavioral feature data to divide the second candidate users to form a candidate group, wherein the target group is a group of people in the candidate group.

18. The duration prediction method according to claim 2, characterized in that: The method further comprises: Acquire second behavior characteristic data of the first candidate information; For each first candidate information, obtaining a score value of each second behavior feature data, and obtaining a second target score value of the first candidate information based on the score value of each second behavior feature data; The second target score values ​​of the first candidate information are sorted in descending order, and the seed information is selected from the first candidate information in order.

19. The duration prediction method according to claim 1, characterized in that: The method further comprises: Acquire second static feature and second behavioral feature data of second candidate information; A clustering operation is performed based on the second static feature and the second behavioral feature data to divide the second candidate information to form a candidate information group.

20. A duration prediction device, characterized in that: include: A first determination module is used to obtain an information type of target information, and determine a first feature corresponding to the target information based on the information type, wherein the information type of the target information includes new information and old information; a second determination module, configured to obtain a user type of a target user, determine a feature source of a second feature corresponding to the target user based on the user type, and obtain the second feature according to the feature source, wherein the feature source includes the target user, similar seed users of the target user, and a target group to which the target user belongs, and the user type of the target user includes a new user and an old user; A prediction module is used to predict the time required for the target user to understand the target information using a target prediction time model based on the first feature and the second feature.

21. The duration prediction device according to claim 20, characterized in that: The first determining module includes: A first acquisition submodule, configured to acquire information similarity between the target information and the seed information if the information type identifies the target information as new information; A second acquisition submodule, configured to acquire a first feature corresponding to the target information based on the information similarity; The third acquisition submodule is configured to acquire a feature of the target information itself as the first feature if the information type identifies that the target information is old information.

22. The duration prediction device according to claim 21, characterized in that: The second acquisition submodule includes: A first judging unit, configured to judge whether there is information similarity greater than a first similarity threshold; A first selection unit, configured to select similar information of the target information from seed information corresponding to the information similarity greater than the first similarity threshold if there is information similarity greater than the first similarity threshold; The first determining unit is configured to obtain a feature of the similar information as the first feature.

23. The duration prediction device according to claim 22, characterized in that: The first selection unit includes: The first determining subunit is configured to select seed information having the greatest information similarity from among the seed information corresponding to information similarities greater than the first similarity threshold as the similar information.

24. The duration prediction device according to claim 22, characterized in that: Also includes: a first identification unit, configured to identify the target information group to which the target information belongs based on the characteristics of the target information itself if there is no information similarity greater than the first similarity threshold; The second determining unit is used to obtain a feature of the target information group as the first feature.

25. The duration prediction device according to claim 20, characterized in that: The second determining module includes: a fourth acquisition submodule, configured to acquire a user similarity between the target user and a seed user if the user type identifies the target user as a new user; A fifth acquisition submodule, configured to acquire a feature source of a second feature corresponding to the target user based on the user similarity; The sixth acquisition submodule is configured to acquire a feature of the target user itself as the second feature if the user type identifies the target user as an old user.

26. The duration prediction device according to claim 25, characterized in that: The fifth acquisition submodule comprises: A second determination unit, configured to determine whether there is a user similarity greater than a second similarity threshold; A second selection unit is configured to select a similar seed user of the target user from seed users corresponding to the user similarities greater than the second similarity threshold if there is a user similarity greater than the second similarity threshold; The third determining unit is configured to use the similar seed users as feature sources of the target users, wherein the features of the similar seed users are used as the second features.

27. The duration prediction device according to claim 26, characterized in that: The second selection unit includes: The second determining subunit is configured to select a seed user with the greatest user similarity among seed users corresponding to user similarities greater than the second similarity threshold as the similar seed user.

28. The duration prediction device according to claim 26, characterized in that: Also includes: A second identification unit is used to identify the target group to which the target user belongs based on the characteristics of the target user if there is no user similarity greater than the second similarity threshold; The fourth determining unit is configured to use the target population as a feature source of the target user, wherein the feature of the target population is used as the second feature.

29. The duration prediction device according to any one of claims 20 to 28, characterized in that: The prediction module comprises: A seventh acquisition submodule, configured to acquire a target prediction duration model corresponding to the target user based on the user type and the information type; The prediction submodule is used to input the first feature and the second feature into the target prediction time model to predict the time required for the target user to understand the target information.

30. The duration prediction device according to claim 29, characterized in that: The seventh acquisition submodule comprises: a fifth acquisition unit, configured to acquire information similarity between the target information and the seed information, and acquire user similarity between the target user and the seed user, if the information type identifies the target information as new information and the user type identifies the target user as a new user; The sixth acquisition unit is used to acquire the target prediction duration model corresponding to the target user according to the user similarity and the information similarity.

31. The duration prediction device according to claim 29, characterized in that: The seventh acquisition submodule comprises: a seventh acquisition unit, configured to acquire information similarity between the target information and the seed information if the information type identifies that the target information is new information and the user type identifies that the target user is an old user; An eighth acquisition unit is used to acquire a target prediction duration model corresponding to the target user according to the information similarity.

32. The duration prediction device according to claim 29, characterized in that: The seventh acquisition submodule comprises: a ninth acquisition unit, configured to acquire a user similarity between the target user and the seed user if the information type identifies that the target information is old information and the user type identifies that the target user is a new user; The tenth acquisition unit is used to acquire the target prediction duration model corresponding to the target user according to the user similarity.

33. The duration prediction device according to claim 20, characterized in that: The first determining module includes: An eighth acquisition submodule is used to acquire the cumulative number of times the target information is sent; A first determination submodule, configured to determine that the information type of the target information is a first information type if the cumulative number of transmissions is greater than or equal to a first preset value, wherein the first information type indicates that the target information is old information; The second determination submodule is used to determine that the information type of the target information is a second information type if the cumulative number of sending times is less than the first preset value, wherein the second information type indicates that the target information is new information.

34. The duration prediction device according to claim 20, characterized in that: The second determining module includes: A ninth acquisition submodule, used to acquire the accumulated behavior data of the target user; A third determination submodule, configured to determine that the user type of the target user is a first user type if the accumulated behavior data is greater than or equal to a second preset value, wherein the first user type indicates that the target user is an old user; The fourth determination submodule is configured to determine that the user type of the target user is a second user type if the accumulated behavior data is less than the second preset value, wherein the second user type indicates that the target user is a new user.

35. The duration prediction device according to claim 20, characterized in that: The device also includes a first selection module, which is used to: Acquire first behavior characteristic data of a first candidate user; For each first candidate user, obtaining a score value of each first behavior feature data, and obtaining a first target score value of the first candidate user based on the score value of each behavior feature data; The first target rating values ​​of the first candidate users are sorted in descending order, and the seed users are selected from the first candidate users in order, wherein the similar seed user is one of the seed users.

36. The duration prediction device according to claim 20, characterized in that: The device further includes a first dividing module, configured to: Acquire first static feature and first behavioral feature data of a second candidate user; A clustering operation is performed based on the first static feature and the first behavioral feature data to divide the second candidate users to form a candidate group, wherein the target group is a group of people in the candidate group.

37. The duration prediction device according to claim 21, characterized in that: The device also includes a second selection module, which is used to: Acquire second behavior characteristic data of the first candidate information; For each first candidate information, obtaining a score value of each second behavior feature data, and obtaining a second target score value of the first candidate information based on the score value of each second behavior feature data; The second target score values ​​of the first candidate information are sorted in descending order, and the seed information is selected from the first candidate information in order.

38. The duration prediction device according to claim 20, characterized in that: The device further includes a second dividing module, configured to: Acquire second static feature and second behavioral feature data of second candidate information; A clustering operation is performed based on the second static feature and the second behavioral feature data to divide the second candidate information to form a candidate information group.

39. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the duration prediction method according to any one of claims 1 to 19 is implemented.

40. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the duration prediction method according to any one of claims 1 to 19 is implemented.

Images