A service recommendation method, device, electronic device and readable storage medium

Abstract

The present invention relates to a service recommendation method, device, electronic device and readable storage medium. By collecting and organizing user service rating source data and service tag source data, the rating of the user for the service tags is obtained; according to the user service rating source data, it is determined whether the user is a first user or a second user; if the user service rating source data is sparse, it is determined as the second user, otherwise it is the first user; a rating prediction method using improved similarity calculation is adopted to obtain the predicted rating of the first user for the unused service; and a missing item rating prediction method using clustering is adopted to obtain the predicted rating of the second user for the unused service. That is, combined with the actual situations of different users, more accurate recommendation services are provided, and at the same time, the user cold start problem caused by data sparsity is solved, and the recommendation coverage rate is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of service recommendation, and in particular, to a service recommendation method, apparatus, electronic device, and readable storage medium. Background Art

[0002] Service recommendation technology has been widely applied to various software products to provide users with personalized services. Recommendation based on collaborative filtering, as the most widely used and successful recommendation technology, first finds groups with interests similar to those of the target user, and predicts the preference degree of the target user for candidate services according to the service selection situations of users within the group, and finally recommends the service with the highest preference degree to the user. Recommendation based on collaborative filtering can discover potential services of interest to users and has a high degree of personalization, but there are problems of user cold start and recommended content not meeting user expectations.

[0003] Regarding the user cold start problem in the recommendation system, relevant scholars fill the original data set, use the average user rating or average service rating in the rating matrix to fill in the missing items, so as to solve the user cold start problem. Although such methods improve the data sparsity problem intuitively, the prediction results of the missing items have a large deviation.

[0004] Regarding the problem that the recommended content in the recommendation system does not meet user expectations, relevant scholars first use methods such as Pearson correlation coefficient, improved cosine similarity, and Jaccard similarity coefficient to find similar neighbors, and then calculate the user's preference for services according to the neighbors to complete the recommendation. However, the above-mentioned recommendation methods have low accuracy in evaluation indicators such as MAE (Mean Absolute Error) and NDCG (Normalized Discounted Cumulative Gain).

[0005] Based on this, the present invention proposes a service recommendation method, apparatus, electronic device, and readable storage medium for solving cold start and improving accuracy, aiming to solve the above problems of the collaborative filtering recommendation technology. Summary of the Invention

[0006] To solve the deficiencies of the prior art, the present invention proposes a service recommendation method, apparatus, electronic device, and readable storage medium.

[0007] To achieve the above purposes, the technical solutions adopted by the present invention include:

[0008] According to the first aspect of the present invention, a service recommendation method is disclosed, including the following steps:

[0009] Collect and organize the source data of user service ratings and the source data of service tags, and obtain the ratings of users for service tags;

[0010] Judge whether the user is a first user or a second user according to the source data of the user service score; if the source data of the user service score is sparse, judge it as the second user, otherwise it is the first user;

[0011] Use the score prediction method with improved similarity calculation to obtain the predicted score of the first user for the unused service;

[0012] Use the missing item score prediction method based on clustering to obtain the predicted score of the second user for the unused service.

[0013] Furthermore, the step of using the score prediction method with improved similarity calculation to obtain the predicted score of the first user for the unused service includes the following sub-steps:

[0014] Calculate the user similarity in the service label dimension based on the Euclidean distance;

[0015] Calculate the user similarity in the service dimension based on the Pearson correlation coefficient;

[0016] Combine the user similarity in the service label dimension and the user similarity in the service dimension to calculate the final user similarity, determine the neighbor users of the first user, and calculate the predicted score of the first user for the unused service based on the scores of the neighbor users for the service.

[0017] Furthermore, the step of using the missing item score prediction method based on clustering to obtain the predicted score of the second user for the unused service includes the following sub-steps:

[0018] Perform K-means clustering on the users according to the scores of the users for the service labels to obtain user clusters with similar preferences;

[0019] Calculate the predicted score of the second user for the unused service based on the user clusters.

[0020] Furthermore, it also includes determining the value of the number of clusters according to the number of service labels.

[0021] Furthermore, the value of the number of service labels is the same as the value of the number of clusters.

[0022] Furthermore, obtain the score of the user for the service label through formula I:

[0023]

[0024] In formula I: a u is the score of user u for service label a, I u,a is the set of services that user u has scored and includes service label a, |I u,a | is the number of elements in the service set; r u,i is the score of user u for service i.

[0025] Further, the method for calculating the user similarity in the service label dimension based on the Euclidean distance is shown in Formula II and Formula III:

[0026]

[0027]

[0028] In Formula II and III: Sim(u, v) L is the similarity between user u and user v in the service label dimension, Dis(max) and Dis(min) are the maximum and minimum Euclidean distances of the user respectively, Dis(u, v) is the Euclidean distance between user u and user v, and L u,v is the set of service labels for which both user u and user v have ratings, and a u is the rating given by user u to service label a.

[0029] Further, the method for calculating the user similarity in the service dimension based on the Pearson correlation coefficient is shown in Formula IV:

[0030]

[0031] In Formula IV: Sim(u, v) S is the similarity between user u and user v in the service dimension, I u,v is the set of services that have been rated by both user u and user v, r u,i is the rating given by user u to service i, is the average rating of user u.

[0032] Further, the method for calculating the predicted rating of the first user for an unused service based on the ratings of neighbor users for the service is shown in Formula V:

[0033]

[0034] In Formula V: P(r u,i ) is the predicted rating of user u for service i, is the average rating of user u, N u is the neighbor set of user u, and r v,i is the rating of service i given by user v within the neighbor set.

[0035] Further, the method for calculating the predicted rating of the second user for an unused service based on the user cluster is shown in Formula VI:

[0036]

[0037] In Formula VI: P(r u,i ) is the predicted rating of user u for service i; rv,i The rating of service i by user v within the user cluster, C is the cluster to which user u belongs, and |C| represents the number of users in the cluster who have rated service i.

[0038] According to the second aspect of the present invention, a service recommendation device is disclosed, including:

[0039] An acquisition module, configured to collect and organize the source data of user service ratings and the source data of service tags, and obtain the ratings of users for service tags;

[0040] A judgment module, configured to judge whether the user is a first user or a second user according to the source data of user service ratings; if the source data of user service ratings is sparse, judge it as a second user, otherwise as a first user;

[0041] A first calculation module, configured to obtain the predicted ratings of the first user for un-used services by using a rating prediction method with improved similarity calculation;

[0042] A second calculation module, configured to obtain the predicted ratings of the second user for un-used services by using a missing item rating prediction method based on clustering.

[0043] According to the third aspect of the present invention, an electronic device is disclosed, including:

[0044] One or more processors;

[0045] A memory;

[0046] One or more applications, wherein the one or more applications are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs are configured to: execute the service recommendation method as described above.

[0047] According to the fourth aspect of the present invention, a computer-readable storage medium is disclosed, and the computer-readable storage medium is used to store computer instructions, which when running on a computer, enable the computer to execute the service recommendation method as described above.

[0048] The beneficial effects of the present invention are:

[0049] By using the service recommendation method, device, electronic device and readable storage medium of the present invention, the method first judges whether the user is a first user or a second user according to the sparsity of the source data of user service ratings, and obtains the predicted ratings of the first user for un-used services by using a rating prediction method with improved similarity calculation; and obtains the predicted ratings of the second user for un-used services by using a missing item rating prediction method based on clustering. That is, by combining the actual situations of different users, more accurate recommendation services are provided, and at the same time, the user cold start problem caused by data sparsity is solved, and the recommendation coverage rate is improved. Description of the Drawings

[0050] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of embodiments in conjunction with the accompanying drawings, where:

[0051] Figure 1 It is a schematic flowchart of a service recommendation method provided for an embodiment of the present application.

[0052] Figure 2 It is a schematic flowchart of a scoring prediction method for improving similarity calculation provided for an embodiment of the present application.

[0053] Figure 3 It is a schematic flowchart of a missing item scoring prediction method for clustering provided for an embodiment of the present application.

[0054] Figure 4 It is a schematic structural diagram of a service recommendation device provided for an embodiment of the present application.

[0055] Figure 5 It is a schematic structural diagram of an electronic device provided for an embodiment of the present application. Detailed Embodiments

[0056] The embodiments of the present application will be described in detail below. Examples of each embodiment are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and should not be construed as a limitation of the present application.

[0057] Those skilled in the art of the present technology can understand that, unless specifically stated otherwise, the singular forms "a", "an" and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present application means the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. The term "and / or" used herein includes all or any unit and all combinations of one or more of the associated listed items.

[0058] To make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below in conjunction with the accompanying drawings.

[0059] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. These several specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0060] An embodiment of the present application provides a service recommendation method, as Figure 1 shown, the method may include the following steps:

[0061] Step S1: Collect and organize the user service rating source data and service tag source data to obtain the user's rating of the service tag;

[0062] Step S2: Determine whether the user is a first user or a second user according to the user service rating source data; if the user service rating source data is sparse, it is determined as the second user, otherwise it is the first user;

[0063] Step S3: Use a rating prediction method that improves similarity calculation to obtain the predicted rating of the first user for the unused service.

[0064] Step S4: Use a clustering-based missing item rating prediction method to obtain the predicted rating of the second user for the unused service.

[0065] In the embodiment of the present application, preferably, if the user service rating source data is relatively sparse and the user cold start phenomenon occurs, that is, when the rating prediction method that improves similarity calculation cannot be used, it is determined as the second user, otherwise it is the first user.

[0066] Specifically, in the step S1, the user's rating of the service tag is obtained through formula I:

[0067]

[0068] In formula I: a u is the rating of the service tag a by the user u, I u,a is the set of services that the user u has rated and contains the service tag a, |I u,a | is the number of elements in the service set; r u,i is the rating of the service i by the user u.

[0069] As an embodiment, Figure 2 shows a schematic flowchart of the specific implementation manner of using the rating prediction method that improves similarity calculation provided by the embodiment of the present application to obtain the predicted rating of the first user for the unused service.

[0070] As Figure 2 shown, the method includes the following sub-steps:

[0071] Step S31: Calculate the user similarity in the service tag dimension based on the Euclidean distance;

[0072] Specifically, the calculation method is shown in formula II and formula III:

[0073]

[0074]

[0075] In Formulas II and III: Sim(u, v) L is the similarity between user u and user v in the service label dimension, Dis(max) and Dis(min) are the maximum and minimum Euclidean distances of the user respectively, Dis(u, v) is the Euclidean distance between user u and user v, and L u,v is the set of service labels for which both user u and user v have ratings, and a u is the rating given by user u to service label a.

[0076] Step S32: Calculate the user similarity in the service dimension based on the Pearson correlation coefficient;

[0077] Specifically, the calculation method is as shown in Formula IV:

[0078]

[0079] In Formula IV: Sim(u, v) S is the similarity between user u and user v in the service dimension, and I u,v is the set of services that have been rated by both user u and user v, and r u,i is the rating given by user u to service i, is the average rating of user u.

[0080] Step S33: Combine the user similarity in the service label dimension and the user similarity in the service dimension to calculate the final user similarity, determine the neighbor users of the first user, and calculate the predicted rating of the first user for the un - used service based on the ratings of the neighbor users for the service.

[0081] Specifically, the calculation method is as shown in Formula V:

[0082]

[0083] In Formula V: P(r u,i ) is the predicted rating of user u for service i, is the average rating of user u, and N u is the set of neighbors of user u, and r v,i is the rating of service i given by user v within the set of neighbors.

[0084] As an embodiment, Figure 3 shows a schematic flow diagram of the specific implementation manner of the missing item rating prediction method for clustering provided by the embodiment of the present application to obtain the predicted rating of the second user for the un - used service.

[0085] As Figure 3 shown, the method includes the following sub - steps:

[0086] Step S41: Perform K-means clustering on users according to their ratings of service tags to obtain user clusters with similar preferences.

[0087] Optionally, determine the value of the number of clusters according to the number of service tags.

[0088] Specifically, the number of service tags is the same as the value of the number of clusters.

[0089] Step S42: Calculate the predicted rating of the second user for unused services based on the user cluster.

[0090] Specifically, the calculation method is as shown in Formula VI:

[0091]

[0092] In Formula VI: P(r u,i ) is the predicted rating of user u for service i; r v,i is the rating of user v within the user cluster for service i, C is the cluster to which user u belongs, and |C| represents the number of users in the cluster who have rated service i.

[0093] More preferably, taking the Movielens dataset commonly used in recommendation systems as an example, it not only contains user item ratings, but also contains user personal information and movie type information, as shown in Table 1 and Table 2:

[0094] Table 1 Source data of user service ratings

[0095]

[0096] Table 2 Source data of service tags

[0097]

[0098] In Table 1, the numbers are the rating data of users for services, ranging from "1 to 5", where a score of "0" means the user has not used the service; in Table 2, "1" means the service belongs to this type, and "0" means the service does not belong to this type.

[0099] By analyzing user service ratings and service tag information, the rating of the user for the service tag is calculated according to Formula I, as shown in Table 3 specifically:

[0100] Table 3 Ratings of users for service tags

[0101]

[0102]

[0103] Exemplarily, select U3 as the target user. Determine that it is the first user, and use the scoring prediction method with improved similarity calculation shown in the embodiments of the present invention to perform scoring prediction. According to the data in Table 1 and Table 3, calculate the user similarity in the service label dimension and the user similarity in the service dimension respectively using Formula II and Formula IV shown above. The calculation results are shown in Table 4 and Table 5:

[0104] Table 4 User Similarity in Service Label Dimension

[0105]

[0106] Table 5 User Similarity in Service Dimension

[0107]

[0108] Use Formula V shown above to perform scoring prediction on candidate services (i.e., unused services) I1 and I4 of target user U3. At the same time, use the prediction results of the collaborative filtering recommendation method based on the dual neighbor selection strategy in the prior art as a comparative example for comparison. The calculation results are shown in Table 6.

[0109] Table 6 Comparison of Prediction Results of Recommendation Accuracy

[0110]

[0111] In the literature, the predicted scores of candidate services I1 and I4 by the collaborative filtering recommendation method based on the dual neighbor selection strategy are 4.61 and 2.79 respectively, while the true scores of I1 and I4 are 4 and 3.5. Therefore, the average misdetection is 0.66; the scores of candidate services I1 and I4 by the scoring prediction method with improved similarity calculation proposed by the present invention are 4.43 and 2.86 respectively, and the average error is 0.535. The error of the method shown in the present invention is significantly smaller than the error of the collaborative filtering recommendation method in the prior art. Therefore, the recommendation accuracy has been significantly improved.

[0112] Exemplarily, select U4 as the target user. Since the source data of the service scores of this target user is sparse, it is impossible to use the scoring prediction method with improved similarity calculation for scoring prediction. Determine that it is the second user, and use the missing item scoring prediction method of clustering shown in the embodiments of the present invention to obtain the predicted scores of U4 for unused services.

[0113] According to the user scoring data for service labels shown in Table 3, perform K-means clustering on users. Since there are three service labels a, b, and c, select the number of clusters K to be 3, and initially select users U1, U3, and U4 who have higher scores for the three service labels as the clustering centers.

[0114] Taking the data in Table 3, the K value, and the initial clustering centers as inputs, and through iterative calculation of the K-means clustering algorithm, the result after user cluster division can be obtained. The users included in the cluster to which the target user U4 belongs are U3 and U4.

[0115] The formula VI is used to predict the missing score items of the target user U4. To reflect the processing effect of the method of the present invention on the user cold start problem, at the same time, the prediction result of the traditional user-based collaborative filtering method is used as a comparative example for comparison. The calculation results are shown in Table 7.

[0116] Table 7 Comparison of prediction results for the user cold start problem

[0117]

[0118] It can be seen from the dataset in Table 7 that due to data sparsity, the user-based collaborative filtering method cannot calculate the similarity between the target user U4 and other users, resulting in the inability to predict the scores of the target user U4 for services I1, I2, and I3, and the user cold start phenomenon occurs. The missing item score prediction method based on user service label score clustering of the present invention can find other neighboring users with similar preferences through the cluster method, and can predict the scores of the target user U4 for services I2 and I3, effectively solving the user cold start problem and improving the recommendation coverage rate.

[0119] In the embodiment of the present application, finally, the candidate services for the target user U4 are scored, predicted, and sorted, and the services with higher scores are selected to complete the recommendation.

[0120] The embodiment of the present application provides a service recommendation method. By collecting and organizing the user service score source data and service label source data, the scores of users for service labels are obtained; according to the user service score source data, it is determined whether the user is a first user or a second user; if the user service score source data is sparse, it is determined as the second user, otherwise it is the first user; a score prediction method using improved similarity calculation is used to obtain the predicted scores of the first user for unused services; and a missing item score prediction method using clustering is used to obtain the predicted scores of the second user for unused services, that is, combining the actual situations of different users, providing a more accurate recommendation service, at the same time solving the user cold start problem caused by data sparsity and improving the recommendation coverage rate.

[0121] Based on the same principle as the Figure 1 method shown, Figure 4 FIG. 25 is a structural schematic diagram of a service recommendation device provided by an embodiment of the present application. The device 20 includes: an acquisition module 201, a judgment module 202, a first calculation module 203, and a second calculation module 204, where,

[0122] An acquisition module 201, configured to collect and organize user service rating source data and service tag source data, and obtain the ratings of users for service tags;

[0123] A judgment module 202, configured to judge whether a user is a first user or a second user according to the user service rating source data; if the user service rating source data is sparse, judge as the second user, otherwise as the first user;

[0124] A first calculation module 203, configured to obtain the predicted ratings of the first user for unused services by using a rating prediction method that improves similarity calculation;

[0125] A second calculation module 204, configured to obtain the predicted ratings of the second user for unused services by using a missing item rating prediction method based on clustering.

[0126] It can be understood that the above-mentioned modules of the service recommendation device in this embodiment have the functions of implementing the corresponding steps of the service recommendation method in the above-mentioned embodiment. This function can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The above modules can be software and / or hardware, and the above-mentioned modules can be implemented separately or integrated by multiple modules. For the function descriptions of the above-mentioned modules of the service recommendation device, reference can be specifically made to the corresponding descriptions of the service recommendation method in the above-mentioned embodiment. Their implementation principles are similar, and the beneficial effects achieved are similar, so they will not be elaborated here.

[0127] An embodiment of the present application provides an electronic device, as Figure 5 shown, Figure 5 The electronic device 30 shown includes: a processor 3001 and a memory 3003. Among them, the processor 3001 and the memory 3003 are connected, such as connected through a bus 3002. Further, the electronic device 30 may further include a transceiver 3004. It should be noted that in practical applications, the transceiver 3004 is not limited to one, and the structure of the electronic device 30 does not constitute a limitation to the embodiment of the present application.

[0128] Among them, the processor 3001 is applied in the embodiments of the present application and is used to implement the methods shown in the above embodiments. The transceiver 3004 includes a receiver and a transmitter. The processor 3001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in connection with the disclosure of the present application. The processor 3001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc. The bus 3002 may include a path for transmitting information between the above components. The bus 3002 may be a PCI bus or an EISA bus, etc. The bus 3002 may be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 5 only a thick line is used to represent it in Figure 5 , but it does not mean that there is only one bus or one type of bus. The memory 3003 may be a ROM or other type of static storage device that can store static information and instructions, a RAM, or other type of dynamic storage device that can store information and instructions. It may also be an EEPROM, a CD-ROM, or other optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0129] Optionally, the memory 3003 is used to store the application program code for executing the solution of the present application, and is controlled by the processor 3001 to execute. The processor 3001 is used to execute the application program code stored in the memory 3003 to implement the service recommendation method provided in any embodiment of the present application.

[0130] The electronic device provided in the embodiments of the present application is applicable to any embodiment of the above method, and will not be elaborated here.

[0131] The embodiments of the present application provide a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the service recommendation method shown in the above method embodiments. This will not be elaborated here.

[0132] It should be understood that although the steps in the flowchart of the accompanying drawings are shown in sequence according to the indication of the arrows, these steps are not necessarily executed in the sequence indicated by the arrows. Unless there is a clear indication in this article, there is no strict sequence restriction for the execution of these steps, and they can be executed in other sequences. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments, and their execution sequence is not necessarily in sequence, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0133] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A service recommendation method, characterized in that, It includes the following steps: Collect and organize the user service rating source data and service label source data, and obtain the user's rating of the service label; Judge whether the user is a first user or a second user according to the user service rating source data; If the user service rating source data is sparse, judge it as a second user, otherwise as a first user; Adopt a rating prediction method with improved similarity calculation to obtain the predicted rating of the first user for the unused service; Adopt a missing item rating prediction method based on clustering to obtain the predicted rating of the second user for the unused service; The method of adopting a rating prediction method with improved similarity calculation to obtain the predicted rating of the first user for the unused service includes the following sub-steps: Calculate the user similarity in the service label dimension based on the Euclidean distance; Calculate the user similarity in the service dimension based on the Pearson correlation coefficient; Combine the user similarity in the service label dimension and the user similarity in the service dimension to calculate the final user similarity, determine the neighbor users of the first user, and calculate the predicted rating of the first user for the unused service based on the ratings of the neighbor users for the service.

2. The method according to claim 1, wherein The method of adopting a missing item rating prediction method based on clustering to obtain the predicted rating of the second user for the unused service includes the following sub-steps: Perform K-means clustering on the users according to the ratings of the users for the service labels to obtain user clusters with similar preferences; Calculate the predicted rating of the second user for the unused service based on the user clusters.

3. The method according to claim 2, wherein It also includes determining the value of the clustering number according to the number of service labels.

4. The method according to claim 2, wherein Obtain the user's rating of the service label through formula I: In Formula I: a u is the score given by user u to service label a, and I u,a is the set of services that user u has scored and contains service label a, and |I u,a | is the number of elements in the service set; r u,i is the score given by user u to service i.

5. The method according to claim 4, wherein The method of calculating the user similarity in the service label dimension based on the Euclidean distance is shown in formula II and formula III: In Formulas II and III: Sim(u, v) L is the similarity between user u and user v in the service label dimension. Dis(max) and Dis(min) are the maximum and minimum Euclidean distances of the users respectively, Dis(u, v) is the Euclidean distance between user u and user v, L u,v is the set of service labels for which both user u and user v have ratings, a u is the rating of user u for service label a, a v is the rating of user v for service label a.

6. The method according to any one of claims 4 or 5, characterized in that, The method of calculating the predicted rating of the second user for the unused service based on the user clusters is shown in formula VI: In Formula VI: P(r u,i ) is the predicted score of user u for service i; r v,i is the score of user v within the user cluster for service i, C is the cluster to which user u belongs, and |C| represents the number of users in the cluster who have scored service i.

7. A service recommendation device, characterized in that, It includes: An acquisition module, used to collect and organize the user service rating source data and service label source data, and obtain the user's rating of the service label; A judgment module, used to judge whether the user is a first user or a second user according to the user service rating source data; If the user service rating source data is sparse, judge it as a second user, otherwise as a first user; A first calculation module, used to adopt a rating prediction method with improved similarity calculation to obtain the predicted rating of the first user for the unused service; A second calculation module, used to adopt a missing item rating prediction method based on clustering to obtain the predicted rating of the second user for the unused service; The method of adopting a rating prediction method with improved similarity calculation to obtain the predicted rating of the first user for the unused service includes the following sub-steps: Calculate the user similarity in the service label dimension based on the Euclidean distance; Calculate the user similarity in the service dimension based on the Pearson correlation coefficient; Combine the user similarity in the service label dimension and the user similarity in the service dimension to calculate the final user similarity, determine the neighbor users of the first user, and calculate the predicted rating of the first user for the unused service based on the ratings of the neighbor users for the service.

8. An electronic device, characterized in that, It includes: One or more processors; A memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the service recommendation method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store computer instructions, which, when run on a computer, cause the computer to execute the service recommendation method according to any one of claims 1 to 6.

Images