Application recommendation method and device, terminal equipment and computer readable storage medium

By calculating application similarity and user interest, and recommending applications, the problem of low accuracy of application recommendation in the prior art is solved, and the efficiency of user selection is improved.

CN118939866APending Publication Date: 2024-11-12GUANGDONG XIAOTIANCAI TECH CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202310532271.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-11
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing application recommendation methods have low accuracy due to data sparseness, especially when the number of applications in the application store increases, the efficiency of users selecting target applications is affected.

Method used

The application is recommended by calculating the application similarity between the downloaded and the undownloaded application and determining the user interest of the undownloaded application based on the user interest and application similarity of the downloaded application.

Benefits of technology

It improves the accuracy of application recommendations, avoids data sparseness issues, ensures the accuracy of user interest in unpaved applications, and enhances user selection efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118939866A_ABST
    Figure CN118939866A_ABST
Patent Text Reader

Abstract

The invention is suitable for the technical field of software, and provides an application recommendation method and device, terminal equipment and a computer readable storage medium, and the method comprises the steps that the application similarity of target applications is calculated, and the target applications comprise downloaded applications and undownloaded applications; determining the user interestingness of the downloaded application; according to the user interestingness of the downloaded application and the application similarity, determining the user interestingness of the application which is not downloaded; and determining recommended applications according to the user interestingness of the applications which are not downloaded. Through the method, the accuracy of the determined recommended application can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the field of software technology, and in particular, relates to an application recommendation method, apparatus, terminal device, and computer-readable storage medium. Background Art

[0002] As the number of applications in the app store gradually increases, users will have more choices in the application pool and the difficulty of selection will also increase.

[0003] In order to improve the efficiency of users selecting target applications, collaborative filtering methods can be used to recommend applications to users. In this method, the score of the application is calculated by counting the historical installation records of the application by the user, and then the recommended application is determined from all applications (i.e., applications that have been installed by the user and applications that have not been installed by the user) based on the calculated score. However, due to the serious data sparsity problem of the above method, the accuracy of the recommended application is low. Summary of the invention

[0004] The embodiments of the present application provide an application recommendation method, apparatus, terminal device, and computer-readable storage medium, which can solve the problem of low accuracy of determined recommended applications.

[0005] In a first aspect, an embodiment of the present application provides an application recommendation method, comprising:

[0006] Calculating application similarity of target applications, wherein the target applications include downloaded applications and non-downloaded applications;

[0007] Determining user interest in the downloaded application;

[0008] determining the user interest of the application that has not been downloaded according to the user interest of the downloaded application and the application similarity;

[0009] Determine a recommended application according to the user interest in the application that has not been downloaded.

[0010] In a second aspect, an embodiment of the present application provides an application recommendation device, including:

[0011] An application similarity calculation module, used to calculate application similarity of target applications, wherein the target applications include downloaded applications and non-downloaded applications;

[0012] A first user interest determination module, used to determine the user interest of the downloaded application;

[0013] A second user interest determination module, configured to determine the user interest of the application that has not been downloaded according to the user interest of the downloaded application and the application similarity;

[0014] The recommended application determination module is used to determine the recommended application according to the user interest of the application that has not been downloaded.

[0015] In a third aspect, an embodiment of the present application provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the first aspect when executing the computer program.

[0016] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.

[0017] In a fifth aspect, an embodiment of the present application provides a computer program product, which, when executed on a terminal device, enables the terminal device to execute the method described in the first aspect above.

[0018] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0019] In the embodiment of the present application, since the recommended application is determined based on the user interest of the application that has not been downloaded, that is, the recommended application is not determined based only on the user interest of the application that has been downloaded, there will be no data sparsity problem when the recommended application is determined by the above method, thereby improving the accuracy of the determined recommended application. At the same time, since the user interest of the application is related to the characteristics of the application itself, that is, the application similarity of two applications is related to the user interest of the two applications, therefore, determining the user interest of the application that has not been downloaded based on the user interest of the downloaded application and the application that has not been downloaded, and the application similarity between the downloaded application and the application that has not been downloaded can ensure the accuracy of the determined user interest of the application that has not been downloaded. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0021] Figure 1 It is a flowchart of an application recommendation method provided by an embodiment of the present application;

[0022] Figure 2 is a structural diagram of an application recommendation device provided by an embodiment of the present application;

[0023] Figure 3 It is a structural diagram of a terminal device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0024] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0025] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0026] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0027] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0028] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the phrases "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. appearing in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways.

[0029] Embodiment 1:

[0030] At present, the usage rate of mobile terminals (such as tutoring machines) is getting higher and higher, and there are more and more applications that can be installed. When users are faced with a large number of applications in the application pool, it will be more difficult for them to choose.

[0031] In order to reduce the difficulty of user application selection and improve the efficiency of application selection, applications in the application pool are usually screened and recommended to users based on the screening results.

[0032] In the process of screening applications recommended to users, when recommending applications to users through collaborative filtering methods, since the matrix used by this method when determining the recommended applications is determined based on the score determined by whether the application has been installed (wherein, a valid score exists only when the application has been installed by the user, and no valid score exists when the application has not been installed by the user), therefore, this method has a serious data sparsity problem, resulting in low accuracy of the recommended applications.

[0033] In order to improve the accuracy of recommended applications, an embodiment of the present application provides an application recommendation method. In the application recommendation method, the application similarity between the downloaded application and the non-downloaded application is calculated, and then the user interest of the second user in the non-downloaded application is calculated based on the application similarity and the first user's user interest rating for the downloaded application, and finally the application recommended to the second user is determined based on the user interest of the non-downloaded application.

[0034] The application recommendation method provided in the embodiment of the present application is described below with reference to the accompanying drawings.

[0035] Figure 1 A schematic diagram of a process flow of an application recommendation method provided in an embodiment of the present application is shown, and is described in detail as follows:

[0036] Step S11, calculating the application similarity of the target application, where the target application includes the downloaded application and the non-downloaded application.

[0037] Specifically, the target application is an application in the application pool. Since there are a large number of applications in the application pool, there are usually applications that have been downloaded and applications that have not been downloaded in the application pool. It should be pointed out that the "downloaded" and "not downloaded" here can be for the same user or for different users. For example, when targeting the same user A, the application similarity refers to the calculation of the application similarity between the applications in the application pool that have been downloaded by user A and the applications that have not been downloaded by user A. For another example, when targeting different users, the application similarity may refer to the calculation of the application similarity between the applications in the application pool that have been downloaded by one or more users and the applications that have not been downloaded by any user.

[0038] In the embodiment of the present application, the application similarity of the two applications may be calculated based on the description information of the two applications or the comment information of each user on the two applications.

[0039] In some embodiments, in order to obtain the application similarity between the downloaded applications and the non-downloaded applications in a timely manner, the application similarity is calculated according to a preset interval time, wherein the preset interval time can be set according to actual conditions, such as 24 hours.

[0040] In some embodiments, the application similarity calculation may also be performed after a new application is added to the application pool, so as to obtain the similarity between the new application and the downloaded application in a more timely manner.

[0041] Step S12: determining the user interest level of the downloaded application.

[0042] The user corresponding to the downloaded application may be the user to whom the application is currently recommended, or may be another user, which is not limited here.

[0043] In the embodiment of the present application, the user's interest in the application downloaded by the user can be determined based on the user's comment information on the application downloaded by the user or the user's behavior, etc. For example, assuming that user A downloads application A, and the user A's comment information on application A is "very good", it indicates that the user A's user interest in application A is very high.

[0044] In some embodiments, the user interest level may be expressed by a score, such as a user interest level of 60 points. The user interest level may also be expressed by a level, such as a user interest level of 5.

[0045] In some embodiments, after an application is downloaded by multiple users, the user interest of the application is determined based on the user interest of the multiple users in the application. For example, the maximum user interest is selected from the user interest of multiple users in the same application, and the maximum user interest is used as the user interest of the application. For another example, the average value of the user interest is determined based on the user interest of multiple users in the same application, and the average value is used as the user interest of the application. For another example, the identity of the user to whom the application is to be recommended is determined in advance, and different interest weights are assigned to different users who have downloaded the application based on the determined user identity and the identities of different users who have downloaded the application (such as setting a larger interest weight for the application used by the user whose identity is the same as the identity of the user to whom the application is to be recommended, so as to improve the accuracy of the obtained user interest), and then the user interest of the downloaded application is determined based on the user interest and the interest weight. Specifically, assuming that an application needs to be recommended to user C, user A and user B have both downloaded application A, and the identity of user A is the same as that of user C (such as both are teachers). Assuming that the interest weight of user A for the user interest corresponding to application A is weight X, and the interest weight of user B for the user interest corresponding to application A is weight Y, then weight X is set greater than weight Y, and then the user interest in application A is determined based on weight X, the user interest corresponding to weight X, weight Y, and the user interest corresponding to weight Y.

[0046] It should be noted that the above step S12 can be executed in parallel with step S11, or can be executed after step S11, which is not limited here.

[0047] Step S13, determining the user interest in the application that has not been downloaded according to the user interest in the downloaded application and the application similarity.

[0048] Specifically, the user interest of the application that has not been downloaded can be determined based on the product of the user interest of the downloaded application and the application similarity between the downloaded application and the application that has not been downloaded. The corresponding relationship between the application similarity interval and the decrease ratio of the user interest can also be preset. After the application similarity is determined, the decrease ratio of the user interest corresponding to the application similarity interval is determined according to the application similarity interval in which the application similarity is located. Finally, the user interest of the application that has not been downloaded is determined based on the user interest of the downloaded application and the decrease ratio of the user interest.

[0049] Step S14: determining recommended applications according to the user interest in the above-mentioned applications that have not been downloaded.

[0050] Specifically, the user interest of each non-downloaded application may be ranked from large to small, and recommended applications may be selected from the non-downloaded applications according to the ranking result, such as selecting the top 5 ranked applications as recommended applications.

[0051] In the embodiment of the present application, the application similarity of the target application including the downloaded application and the non-downloaded application is calculated, and the user interest of the downloaded application is determined, and the user interest of the non-downloaded application is determined based on the user interest and application similarity of the downloaded application, and the recommended application is determined based on the user interest of the non-downloaded application. Since the recommended application is determined based on the user interest of the non-downloaded application, that is, the recommended application is not determined only based on the user interest of the downloaded application, there will be no data sparsity problem when determining the recommended application by the above method, thereby improving the accuracy of the determined recommended application. At the same time, since the user interest of the application is related to the characteristics of the application itself, that is, the application similarity of two applications is related to the user interest of the two applications, therefore, determining the user interest of the non-downloaded application based on the user interest of the downloaded application and the non-downloaded application and the application similarity between the downloaded application and the non-downloaded application can ensure the accuracy of the determined user interest of the non-downloaded application.

[0052] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0053] Embodiment 2:

[0054] In some embodiments, the application similarity of the two applications is determined according to the description information of the two applications. In this case, the above step S11 includes:

[0055] A1. Obtain description information of the downloaded application and description information of the non-downloaded application.

[0056] The description information of the application usually includes information such as the function and content of the application. For example, the following may be the description information of an application: Every child's favorite hide-and-seek game is here! Sailor Fette is hiding in the jungle or in the desert rocks, can you find him? A hide-and-seek game with many cute animals, you deserve it!

[0057] A2. Calculate the application similarity between the downloaded application and the non-downloaded application based on the description information of the downloaded application and the description information of the non-downloaded application to obtain the application similarity of the target application.

[0058] Specifically, the application similarity between the two applications can be calculated by analyzing the repetition of words in the description information of the two applications, or by converting the description information of the two applications into vectors and then calculating the similarity of the two vectors, and using the calculated similarity as the application similarity between the two applications.

[0059] Since the description information of an application includes information describing the application, the description information of the application can reflect the characteristics of the application, so when the application similarity of two applications is calculated based on the description information of the two applications, the accuracy of the obtained application similarity can be guaranteed.

[0060] In some embodiments, the description information is converted into vectors and then the similarity between the vectors is calculated. In this case, the above step A2 includes:

[0061] A21. Convert the description information of the downloaded application into a first text vector according to target description information, wherein the target description information includes the description information of the downloaded application and the description information of the non-downloaded application.

[0062] Specifically, the description information of the downloaded applications can be converted into a first text vector based on the term frequency–inverse document frequency (TF-IDF) method of the vector space model, or a statistical method, or a neural network method. For example, assuming that the description information of the downloaded applications (one description information is one text) is converted into a first text vector based on the TF-IDF method, the words included in all texts are: w 1 , 2, …, n , then each text can be represented by an n-dimensional vector, that is, P = (P 1 , 2, …, n ). Let n1 be the number of times w 1 appears in text 1, N be the total number of words in all texts, m be the number of texts containing w 1 in all other texts, and M be the total number of texts. Then the calculation method of TF is:

[0063]

[0064] Considering that some common words such as "you" and "of" appear frequently and the calculated TF value is very high but the actual meaning is very low, a calculation weight should be introduced to weaken it. Considering that these words appear in many texts, the inverse document frequency weight, that is, IDF, is introduced. The calculation formula is:

[0065]

[0066] Furthermore, the calculation formula of P i (i = 1, 2 … n) is:

[0067] P i = TF * IDF.

[0068] Calculate all P i of a text to obtain the vector corresponding to this text.

[0069] In some embodiments, the above target description information includes the description information of multiple downloaded applications and the description information of multiple undownloaded applications, such as including the description information of all downloaded applications in the application pool and the description information of all undownloaded applications in this application pool. Since the target description information includes the description information of multiple applications, it can ensure that the number of description information participating in the vector conversion is large enough, thereby improving the accuracy of the converted vector (such as the first text vector).

[0070] In some embodiments, considering that the description information of an application usually contains some useless symbols or words irrelevant to the content theme, it is necessary to construct a stop word library suitable for the application description, such as constructing a stop word library composed of some numbers, English letters, advertising words, contact information, etc., and then filtering the stop words of the description information of the application according to the constructed stop word library, and finally converting the vector according to the filtering result. For example, assuming that the description information of an application is "Every child's favorite hide-and-seek game is here! Sailor Fit is hiding in the jungle or in the desert rocks, can you find him? A hide-and-seek game with lots of cute animals, you deserve it!", the description information of the application is processed by stuttering and word segmentation to obtain words "Every / child's / favorite / hide-and-seek / game / is / here! / Sailor / Fit / is / hiding / in / the / jungle / or / desert / rocks / , / can / you / find / him / ", and the above words are filtered by stop words, and the result obtained is "Children's / like / hide-and-seek / game / Sailor / Fit / hides / in / the / jungle / desert / rocks / and / finds / lots / of / cute / animal / hide-and-seek / game".

[0071] A22. Convert the description information of the application that has not been downloaded into a second text vector according to the target description information.

[0072] The conversion method of the second text vector is similar to the conversion method of the first text vector, which will not be described in detail here.

[0073] A23. Perform dimensionality reduction processing on the first text vector and the second text vector.

[0074] The dimensionality reduction process can be implemented by any of the following algorithms: Singular Value Decomposition (SVD), Principal Component Analysis (PCA), and Linear Discriminant Analysis (LDA).

[0075] Specifically, when the dimensions of the first text vector and the second text vector are high, the first text vector and the second text vector are subjected to dimensionality reduction processing (i.e., the dimensions of the vectors are reduced after the dimensionality reduction processing), which can reduce the amount of calculation required for calculating the application similarity. Of course, if the dimensions of the first text vector and the second text vector are low (such as less than a preset dimensionality threshold), there is no need to perform dimensionality reduction processing, and step A24 is directly executed.

[0076] In an embodiment of the present application, when the SVD algorithm is used for dimensionality reduction processing, the matrix to be reduced in dimension is first determined based on the first text vector and the second text vector, and then the matrix to be reduced in dimension is subjected to singular value decomposition, and the first text vector after dimensionality reduction processing and the second text vector after dimensionality reduction processing are determined based on the result of the singular value decomposition.

[0077] Of course, if there are multiple first text vectors and multiple second text vectors, the matrix to be reduced in dimension is determined according to the multiple first text vectors and multiple second text vectors. For example, assuming that the matrix to be reduced in dimension is an m×n matrix A, then the SVD of the matrix A is defined as: A=UΣV T .

[0078] Where U is an m×m matrix, Σ is an m×n matrix, and all elements of the matrix Σ are 0 except for the elements on the main diagonal, and each element on the main diagonal is called a singular value, that is:

[0079]

[0080] Among them, Σ 0 =diog(σ1,σ2,…,σr), and σ1≥σ2…≥σr>0.

[0081] V is an n×n matrix. Both U and V are unitary matrices, that is, U T U=I,V T V = I. Right singular matrix V T By calculating A T The eigenvector of A is obtained as:

[0082] (A T A)v i =λ i v i .

[0083] A T All eigenvectors of A are stretched into an n×n matrix V, which is the V matrix in the SVD formula.

[0084] The left singular matrix U is calculated by AA T The eigenvectors of are obtained:

[0085] (AA T ) i =λ i u i .

[0086] AA T All eigenvectors of T The matrix U is the U matrix in the SVD formula.

[0087] Furthermore, the eigenvalue matrix is ​​equal to the square of the singular value matrix, that is, the singular value calculation formula is:

[0088]

[0089] The singular values ​​are arranged from large to small in the singular value matrix Σ, and the effective information of the data is concentrated on the first k singular values, that is:

[0090]

[0091] The above method can realize the dimensionality reduction processing of text vectors.

[0092] In the embodiment of the present application, the value of k is related to the dimension of the matrix to be reduced. When the dimension of the matrix to be reduced is large, the value of k is also large. For example, assuming that the dimension of the matrix to be reduced is less than 1000, the value of k can be 8 or 16. In some embodiments, when A=UΣV is calculated T Afterwards, the value of k is determined based on the difference between two adjacent singular values. For example, after the difference between two adjacent singular values ​​is greater than a preset difference threshold, the value of k is determined based on the dimension corresponding to the two singular values. Since the SVD algorithm is a feature matrix decomposition method, after decomposition, the more information is contained in the front singular values, therefore, only the front k eigenvalues ​​extracted can be retained to retain a large amount of original information. In addition, since the difference between two adjacent singular values ​​is large, it indicates that the subsequent singular values ​​will become smaller and smaller, and the smaller the singular value, the smaller the information contained. Therefore, the value of k is determined based on the difference between two adjacent singular values, so that the vector after dimensionality reduction not only retains a large amount of original information of the text, but also can realize dimensionality reduction processing of the text vector.

[0093] A24. Calculate the application similarity between the downloaded application and the non-downloaded application based on the first text vector after dimensionality reduction processing and the second text vector after dimensionality reduction processing.

[0094] Specifically, the application similarity between vectors can be calculated using cosine distance or Euclidean distance. For example, when calculating application similarity based on cosine distance, assuming that the first text vector after dimensionality reduction is α, the second text vector after dimensionality reduction is β, and the application similarity is R, then:

[0095]

[0096] Among them, R is also called cosine distance, R∈[0,1]. The larger R is, the greater the vector similarity is, that is, the greater the correlation between the vectors is.

[0097] Since the calculation between vectors is more accurate than the calculation between texts, converting the description information of the application into the corresponding vector before calculating the application similarity is beneficial to improving the accuracy of the calculated application similarity. At the same time, since the first text vector and the second text vector are also subjected to dimensionality reduction processing before calculating the application similarity, the dimension of the application similarity calculation is reduced, which is beneficial to improving the efficiency of calculating the application similarity.

[0098] Embodiment three:

[0099] Based on the above embodiment 1 and / or the above embodiment 2, in some embodiments, the above step S12 includes:

[0100] B1. Obtain the user behavior for the downloaded application, where the user behavior includes at least one of the following: the download behavior corresponding to the user's secondary account, the download behavior corresponding to the user's primary account, and the activation behavior corresponding to the user.

[0101] The permissions of the user's secondary account are less than the permissions of the user's primary account. For example, as a parent, the user can log in to his primary account on a mobile phone or a tutoring machine, and the user's child can log in to the user's secondary account on a tutoring machine.

[0102] When a child or parent downloads an application on a tutoring machine, the user behavior is: the download behavior corresponding to the slave account. When a parent downloads an application on a mobile phone for a tutoring machine, the user behavior is: the download behavior corresponding to the main account.

[0103] When a user (such as a child or parent) activates a downloaded application from an account or a primary account, it indicates that the user has activated the application.

[0104] B2. Calculate the rating of the downloaded application according to the user behavior and the weight of the user behavior.

[0105] Specifically, the number of weights of user behaviors is equal to the number of items included in the user behaviors. For example, assuming that the user behaviors include: the download behaviors corresponding to the user's secondary account, the download behaviors corresponding to the user's primary account, and the activation behaviors corresponding to the user, the weights of user behaviors include: the weights of the download behaviors corresponding to the secondary account, the weights of the download behaviors corresponding to the primary account, and the weights of the activation behaviors.

[0106] The weights of the above items can be set according to actual conditions and are not limited here.

[0107] In some embodiments, considering that the authority of the master account is greater than that of the slave account, the weight of the download behavior corresponding to the master account is set to be greater than the weight of the download behavior corresponding to the slave account.

[0108] B3. Determine the user interest in the downloaded application based on the rating.

[0109] Among them, the rating of the application is positively correlated with the user interest in the application. The higher the rating of the application, the higher the user interest in the application.

[0110] In the embodiment of the present application, considering that a user can apply for a main account and a slave account, and both the main account and the slave account can perform download actions, therefore, when obtaining user behavior, the download behavior corresponding to the slave account and / or the download behavior corresponding to the main account are considered, which can make the obtained user behavior more accurate, and thus ensure that the user interest determined based on the user behavior is also more accurate.

[0111] In some embodiments, considering that the frequency of application downloads and / or the frequency of application activation can reflect the user interest in the application, the user interest in the application is calculated based on the ratings of the application on different dates, so as to ensure the accuracy of the obtained user interest. In this case, the above step B2 includes:

[0112] For each of the preset dates, the score of the downloaded application on the above date is calculated based on the user behavior on the above date and the weight of the above user behavior. The above preset date includes multiple dates. For example, assuming that the current date is February 1, the 14 days before February 1 can be used as preset dates. Specifically, for the same application, the user behavior of the application is deduplicated according to the date, and then the score of the application is calculated based on the deduplicated user data to ensure the accuracy of the score of the application corresponding to each date. For example, if user A downloaded application A on January 25, and user B also downloaded application A on January 25, after deduplicating the user's download behavior, only the download behavior of user A downloading application A can be retained, or only the download behavior of user B downloading application A can be retained.

[0113] Correspondingly, the above step B3 includes:

[0114] The user interest in the downloaded application is determined according to the scores corresponding to the respective dates and the weights corresponding to the respective dates, wherein the closer the date is to the date on which the user behavior is obtained, the greater the weight corresponding to the date.

[0115] In some embodiments, the weight corresponding to the date is less than 1.

[0116] Since the closer the date is to the date when the user behavior is obtained, the greater the weight corresponding to the date is, so when the user behavior is the same, it can be ensured that the closer the date is to the date when the user behavior is obtained, the higher the score corresponding to the date is. In actual situations, the user behavior closer to the current date is more valuable for reference, so the user interest determined by the above method is more accurate.

[0117] In order to more clearly describe the process of determining the user interest in a downloaded application, a description is given below in conjunction with a specific application example.

[0118] Assume that the scoring rules are as follows: The download behavior score S of a user downloading an application using a tablet on a certain day pad_tall is w1 (i.e., the weight of the download behavior corresponding to the account), and whether there is download behavior is recorded as B pad_ (0 or 1), the activation behavior score is S active is w2 (i.e., the weight of the activation behavior), and whether there is an activation behavior is recorded as B active (0 or 1), the parent's download behavior score is S parent_stall is w3 points (i.e., the weight of the download behavior corresponding to the main account), and whether there is a parent-side download behavior is recorded as B parent_

[0119] (0 or 1), calculate the total score of an application on that day day , the specific formula can be:

[0120] Score day =Max(w1×B pad_ ,w3×B parent_ )+(w2×B active ).

[0121] Among them, w1 is 3, w2=1, w3=10.

[0122] Take the data of the 14 days before the specific day, remove duplicates of user behaviors by day, and decay the score based on the number of days J since the user behavior, multiply it by the decay factor D to get the total app score. The formula is as follows:

[0123]

[0124] Among them, D=0.93.

[0125] Embodiment 4:

[0126] In some embodiments, the above step S13 includes:

[0127] For any application that has not been downloaded, the user interest level of the application that has not been downloaded is determined according to the user interest levels of each of the downloaded applications and the application similarity between each of the downloaded applications and the application that has not been downloaded.

[0128] Specifically, multiple intermediate interest values ​​can be calculated based on the user interest of each downloaded application and the application similarity between each downloaded application and the application that has not been downloaded, and then the user interest of the application that has not been downloaded is determined based on the multiple intermediate interest values. For example, assuming that the current user has Q apps with behavior data, and the score of the i-th app is S i , the app to be scored is the jth app: app j (The app that will determine the user's interest level) i and app j The similarity is R ij , if the user interest is expressed in scores, the score of the jth app is:

[0129]

[0130] Since the higher the application similarity between two applications, the closer the corresponding user interests of the two applications are, therefore, in an embodiment of the present application, when the user interest of each application that has not been downloaded is determined based on the user interests of multiple downloaded applications and the similarities between the multiple downloaded applications and the non-downloaded applications, the user interest of each downloaded application can be more comprehensively and effectively utilized, thereby improving the accuracy of the determined user interest of the non-downloaded applications.

[0131] Embodiment five:

[0132] Corresponding to the application recommendation method described in each method embodiment above, Figure 2 A structural block diagram of an application recommendation device provided in an embodiment of the present application is shown. For ease of explanation, only the parts related to the embodiment of the present application are shown.

[0133] Reference Figure 2 The application recommendation device 2 includes: an application similarity calculation module 21, a first user interest determination module 22, a second user interest determination module 23, and a recommended application determination module 24. Among them:

[0134] The application similarity calculation module 21 is used to calculate the application similarity of target applications, where the target applications include downloaded applications and non-downloaded applications.

[0135] The first user interest level determination module 22 is used to determine the user interest level of the downloaded application.

[0136] The second user interest determination module 23 is used to determine the user interest of the application that has not been downloaded according to the user interest of the downloaded application and the application similarity.

[0137] The recommended application determination module 24 is used to determine the recommended applications according to the user interest in the above-mentioned applications that have not been downloaded.

[0138] In the embodiment of the present application, since the recommended application is determined based on the user interest of the application that has not been downloaded, that is, the recommended application is not determined based only on the user interest of the application that has been downloaded, there will be no data sparsity problem when the recommended application is determined by the above method, thereby improving the accuracy of the determined recommended application. At the same time, since the user interest of the application is related to the characteristics of the application itself, that is, the application similarity of two applications is related to the user interest of the two applications, therefore, determining the user interest of the application that has not been downloaded based on the user interest of the downloaded application and the application that has not been downloaded, and the application similarity between the downloaded application and the application that has not been downloaded can ensure the accuracy of the determined user interest of the application that has not been downloaded.

[0139] In some embodiments, the application similarity calculation module 21 includes:

[0140] The description information acquisition unit is used to acquire the description information of the downloaded application and the description information of the non-downloaded application.

[0141] The application similarity calculation unit is used to calculate the application similarity between the downloaded application and the non-downloaded application according to the description information of the downloaded application and the description information of the non-downloaded application, so as to obtain the application similarity of the target application.

[0142] In some embodiments, the application similarity calculation unit includes:

[0143] The first text vector generating unit is used to convert the description information of the downloaded application into a first text vector according to target description information, wherein the target description information includes the description information of the downloaded application and the description information of the non-downloaded application.

[0144] The second text vector generating unit is used to convert the description information of the application that has not been downloaded into a second text vector according to the target description information.

[0145] The dimension reduction unit is used to perform dimension reduction processing on the first text vector and the second text vector.

[0146] The dimensionality reduction post-processing unit is used to calculate the application similarity between the downloaded application and the non-downloaded application based on the first text vector after dimensionality reduction processing and the second text vector after dimensionality reduction processing.

[0147] In some embodiments, the first user interest determination module 22 includes:

[0148] The user behavior acquisition unit is used to acquire the user behavior of the user for the above-mentioned downloaded application, and the above-mentioned user behavior includes at least one of the following: the download behavior corresponding to the above-mentioned user's slave account, the download behavior corresponding to the above-mentioned user's master account and the activation behavior corresponding to the above-mentioned user.

[0149] The rating calculation unit is used to calculate the rating of the downloaded application according to the user behavior and the weight of the user behavior.

[0150] The user interest calculation unit of the downloaded application is used to determine the user interest of the downloaded application according to the score.

[0151] In some embodiments, the weight of the download behavior corresponding to the master account is greater than the weight of the download behavior corresponding to the slave account.

[0152] In some embodiments, the score calculation unit includes:

[0153] For each of the preset dates, the score of the downloaded application on the date is calculated according to the user behavior on the date and the weight of the user behavior.

[0154] Correspondingly, the user interest calculation unit of the downloaded application includes:

[0155] The user interest in the downloaded application is determined according to the scores corresponding to the respective dates and the weights corresponding to the respective dates, wherein the closer the date is to the date on which the user behavior is obtained, the greater the weight corresponding to the date.

[0156] In some embodiments, the recommended application determination module 24 includes:

[0157] For any application that has not been downloaded, the user interest level of the application that has not been downloaded is determined according to the user interest levels of each of the downloaded applications and the application similarity between each of the downloaded applications and the application that has not been downloaded.

[0158] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0159] Embodiment six:

[0160] Figure 3 This is a schematic diagram of the structure of a terminal device provided in an embodiment of the present application. Figure 3 As shown, the terminal device 3 of this embodiment includes: at least one processor 30 ( Figure 3 Only one processor is shown in the figure), a memory 31, and a computer program 32 stored in the above-mentioned memory 31 and executable on at least one of the above-mentioned processors 30, wherein the above-mentioned processor 30 implements the steps in any of the above-mentioned method embodiments when executing the above-mentioned computer program 32.

[0161] The terminal device 3 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The terminal device may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will appreciate that Figure 3 It is only an example of the terminal device 3 and does not constitute a limitation on the terminal device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0162] The processor 30 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0163] In some embodiments, the memory 31 may be an internal storage unit of the terminal device 3, such as a hard disk or memory of the terminal device 3. In other embodiments, the memory 31 may also be an external storage device of the terminal device 3, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 3. Further, the memory 31 may also include both an internal storage unit and an external storage device of the terminal device 3. The memory 31 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 31 may also be used to temporarily store data that has been output or is to be output.

[0164] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0165] An embodiment of the present application also provides a network device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any of the above-mentioned method embodiments when executing the computer program.

[0166] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0167] An embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0168] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0169] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0170] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0171] In the embodiments provided in the present application, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0172] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0173] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. An application recommendation method, characterized in that: include: Calculating application similarity of target applications, wherein the target applications include downloaded applications and non-downloaded applications; Determining user interest in the downloaded application; determining the user interest of the application that has not been downloaded according to the user interest of the downloaded application and the application similarity; Determine a recommended application according to the user interest in the application that has not been downloaded.

2. The application recommendation method according to claim 1, characterized in that: The calculating the application similarity of the target application includes: Obtaining description information of the downloaded application and description information of the non-downloaded application; The application similarity between the downloaded application and the non-downloaded application is calculated according to the description information of the downloaded application and the description information of the non-downloaded application to obtain the application similarity of the target application.

3. The application recommendation method according to claim 2, characterized in that: The calculating the application similarity between the downloaded application and the non-downloaded application according to the description information of the downloaded application and the description information of the non-downloaded application includes: Converting the description information of the downloaded application into a first text vector according to target description information, wherein the target description information includes the description information of the downloaded application and the description information of the non-downloaded application; Converting the description information of the application that has not been downloaded into a second text vector according to the target description information; Performing dimensionality reduction processing on the first text vector and the second text vector; The application similarity between the downloaded application and the non-downloaded application is calculated based on the first text vector after the dimensionality reduction processing and the second text vector after the dimensionality reduction processing.

4. The application recommendation method according to any one of claims 1 to 3, characterized in that: The determining the user interest level of the downloaded application includes: Acquire a user behavior of the user for the downloaded application, the user behavior including at least one of the following: a download behavior corresponding to the user's secondary account, a download behavior corresponding to the user's primary account, and an activation behavior corresponding to the user; Calculating a score of the downloaded application according to the user behavior and a weight of the user behavior; The user interest in the downloaded application is determined according to the rating.

5. The application recommendation method according to claim 4, characterized in that: The weight of the download behavior corresponding to the master account is greater than the weight of the download behavior corresponding to the slave account.

6. The application recommendation method according to claim 4, characterized in that: The calculating the score of the downloaded application according to the user behavior and the weight of the user behavior includes: For each of the preset dates, calculating the score of the downloaded application on the date according to the user behavior on the date and the weight of the user behavior; The determining the user interest level of the downloaded application according to the rating includes: The user interest in the downloaded application is determined according to the scores corresponding to the respective dates and the weights corresponding to the respective dates, wherein the closer the date is to the date on which the user behavior is obtained, the greater the weight corresponding to the date.

7. The application recommendation method according to any one of claims 1 to 3, characterized in that: The determining the user interest of the application that has not been downloaded according to the user interest of the downloaded application and the application similarity includes: For any application that has not been downloaded, the user interest level of the application that has not been downloaded is determined according to the user interest levels of each of the downloaded applications and the application similarities between each of the downloaded applications and the application that has not been downloaded.

8. An application recommendation device, characterized in that: include: An application similarity calculation module, used to calculate application similarity of target applications, wherein the target applications include downloaded applications and non-downloaded applications; A first user interest determination module, used to determine the user interest of the downloaded application; A second user interest determination module, configured to determine the user interest of the application that has not been downloaded according to the user interest of the downloaded application and the application similarity; The recommended application determination module is used to determine the recommended application according to the user interest of the application that has not been downloaded.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

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

Citation Information

Patent Citations

  • Information recommendation method and device, computer equipment and storage medium

    CN110363580A

  • Application recommendation method, application recommendation device and terminal device

    CN110688582A

  • Information recommendation method and device and electronic equipment

    CN113268655A

  • Method and device for determining application account

    WO2018196546A1