User-based Recommendation Method, Device, Equipment and Computer Readable Storage Medium

Abstract

The present invention discloses a user-based recommendation method, device, equipment and computer-readable storage medium. The method includes: determining user similarity according to the obtained data set, sorting the user similarity, and determining a neighbor cluster according to the sorting result; determining a rating deviation degree based on the rating criteria of each user in the neighbor cluster, and determining a predicted rating based on the rating deviation degree; recommending the item to be recommended to the user to be recommended based on the predicted rating. Thus, the predicted rating of the user to be recommended for the item to be recommended is predicted based on the rating criteria of each user, improving the accuracy of the predicted rating of each user, and further improving the accuracy of the recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of collaborative filtering recommendation in financial technology (Fintech), and particularly to a user-based recommendation method, device, equipment, and computer-readable storage medium. Background Art

[0002] With the development of computer technology, more and more technologies are applied in the financial field. The traditional financial industry is gradually transforming into financial technology (Fintech). However, due to the security and real-time requirements of the financial industry, higher requirements are also put forward for technologies.

[0003] Currently, the user-based collaborative filtering algorithm has been widely applied in life. However, in the traditional user-based collaborative filtering algorithm, when predicting ratings, only the difference between the ratings of similar users for an item and the average rating is considered as the preference deviation of the user, without considering that the subjective rating criteria of each user are inconsistent. This will ultimately lead to inaccurate predicted ratings for each user and low recommendation accuracy. Summary of the Invention

[0004] The present invention provides a user-based recommendation method, device, equipment, and computer-readable storage medium, aiming to improve the accuracy of predicted ratings for each user and the accuracy of recommendations.

[0005] To achieve the above object, the present invention provides a user-based recommendation method, and the method includes:

[0006] Determine user similarity according to the obtained data set, sort the user similarity, and determine a neighbor cluster according to the sorting result;

[0007] Determine a rating deviation degree based on the rating criteria of each user in the neighbor cluster, and determine a predicted rating based on the rating deviation degree;

[0008] Recommend the item to be recommended to the user to be recommended based on the predicted rating.

[0009] Preferably, the determining a rating deviation degree based on the rating criteria of each user in the neighbor cluster and determining a predicted rating based on the rating deviation degree includes:

[0010] Determine a first rating deviation amount of the item to be recommended by the rated users who have rated the item to be recommended from the neighbor cluster;

[0011] Determine a first rating deviation degree of the first rating deviation amount based on the rating criteria of the rated users, where the rating criteria of the rated users include an average rating, a rating upper limit, and a rating lower limit;

[0012] Determine the change trend of the user preference degree and the second scoring deviation degree of the to-be-recommended user for the to-be-recommended item based on the first scoring deviation degree and the user similarity;

[0013] Determine the second scoring deviation amount of the to-be-recommended user for the to-be-recommended item based on the second scoring deviation degree and the average score of the to-be-recommended user;

[0014] Determine the predicted score based on the correlation trend, the second scoring deviation amount, and the average score of the to-be-recommended user.

[0015] Preferably, the first scoring deviation degree for determining the first scoring deviation amount based on the scoring criteria of the rated users includes:

[0016] Determine the first scoring deviation amount of the rated users in the neighbor cluster for the to-be-recommended item, where the first scoring deviation amount is the difference between the actual score of the rated user for the to-be-recommended item and the first average score of the corresponding rated user;

[0017] Based on the scoring deviation amount, the score upper limit, and the score lower limit of the corresponding rated user, determine the first scoring deviation degree of the corresponding rated user for the to-be-recommended item.

[0018] Preferably, the determining the predicted score based on the correlation trend, the second scoring deviation amount, and the average score of the to-be-recommended user includes:

[0019] If the correlation trend is positive, determine the sum of the average score of the to-be-recommended user and the second scoring deviation amount as the predicted score;

[0020] If the correlation trend is negative, determine the difference between the average score of the to-be-recommended user and the second scoring deviation amount as the predicted score.

[0021] Preferably, the obtaining the user similarity according to the obtained data set, sorting the user similarity, and determining the neighbor cluster according to the sorting result includes:

[0022] Obtain a data set including a user-item scoring matrix, and determine the user similarity based on the data set;

[0023] Sort the user similarity in descending order, and determine a preset number of users with common scoring items as the neighbor cluster according to the sorting result.

[0024] Preferably, after obtaining the data set including the user-item scoring matrix, it further includes:

[0025] Determine the to-be-recommended user and the to-be-recommended item from the neighbor cluster.

[0026] Preferably, recommending the item to be recommended to the user to be recommended based on the predicted score includes:

[0027] Comparing the predicted score with the average score of the user to be recommended;

[0028] If the predicted score is greater than the average score of the user to be recommended, recommend the item to be recommended to the user to be recommended.

[0029] In addition, to achieve the above object, the present invention further provides a user-based recommendation device, and the user-based recommendation device includes:

[0030] A determination module, configured to determine user similarity according to the obtained data set, sort the user similarity, and determine a neighbor cluster according to the sorting result;

[0031] A scoring module, configured to determine a scoring deviation degree based on the scoring criteria of each user in the neighbor cluster, and determine a predicted score based on the scoring deviation degree;

[0032] A recommendation module, configured to recommend the item to be recommended to the user to be recommended based on the predicted score.

[0033] In addition, to achieve the above object, the present invention further provides a user-based recommendation device, and the user-based recommendation device includes a processor, a memory, and a user-based recommendation program stored in the memory. When the user-based recommendation program is run by the processor, the steps of the above-mentioned user-based recommendation method are implemented.

[0034] In addition, to achieve the above object, the present invention further provides a computer-readable storage medium, and a user-based recommendation program is stored on the computer-readable storage medium. When the user-based recommendation program is run by a processor, the steps of the above-mentioned user-based recommendation method are implemented.

[0035] Compared with the prior art, the present invention provides a user-based recommendation method, device, equipment, and computer-readable storage medium. The user similarity is determined according to the obtained data set, the user similarity is sorted, and a neighbor cluster is determined according to the sorting result; a scoring deviation degree is determined based on the scoring criteria of each user in the neighbor cluster, and a predicted score is determined based on the scoring deviation degree; the item to be recommended is recommended to the user to be recommended based on the predicted score. Thus, the predicted score of the user to be recommended for the item to be recommended is predicted based on the scoring criteria of each user, the accuracy of the predicted score of each user is improved, and the accuracy of the recommendation is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1It is a schematic diagram of the hardware structure of a user-based recommendation device involved in various embodiments of the present invention;

[0037] Figure 2 It is a schematic flowchart of the first embodiment of the user-based recommendation method of the present invention;

[0038] Figure 3 It is a schematic diagram of the functional modules of the first embodiment of the user-based recommendation device of the present invention.

[0039] The realization of the purpose, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed Embodiments

[0040] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] The user-based recommendation device mainly involved in the embodiments of the present invention refers to a network connection device capable of implementing network connection, and the user-based recommendation device may be a server, a cloud platform, etc.

[0042] Refer to Figure 1 , Figure 1 is a schematic diagram of the hardware structure of a user-based recommendation device involved in various embodiments of the present invention. In the embodiments of the present invention, the user-based recommendation device may include a processor 1001 (such as a Central Processing Unit, CPU), a communication bus 1002, an input port 1003, an output port 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components; the input port 1003 is used for data input; the output port 1004 is used for data output, and the memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001. Those skilled in the art can understand that Figure 1 the hardware structure shown in does not constitute a limitation to the present invention, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0043] Continue to refer to Figure 1 , Figure 1 the memory 1005 as a readable computer-readable storage medium in may include an operating system, a network communication module, an application program module, and a user-based recommendation program. In Figure 1Among them, the network communication module is mainly used to connect to the server and communicate with the server for data; while the processor 1001 can call the user-based recommendation program stored in the memory 1005 and execute the user-based recommendation method provided by the embodiments of the present invention.

[0044] The embodiments of the present invention provide a user-based recommendation method. The collaborative filtering algorithm is a relatively well-known and commonly used recommendation algorithm. It discovers the user's preference bias based on the mining of the user's historical behavior data, predicts the products that the user may like, and then makes recommendations. That is, the common functions such as "Guess You Like" and "People Who Bought This Item Also Like". Its main implementation includes: recommending to you based on people who have the same preferences as you; recommending similar items based on the items you like; and making comprehensive recommendations based on the above conditions. Therefore, it can be concluded that the commonly used collaborative filtering algorithms are divided into two types, the user-based collaborative filtering algorithm, and the item-based collaborative filtering algorithm.

[0045] Based on the user-based collaborative filtering algorithm, this embodiment provides an improved collaborative difference algorithm based on user rating criteria.

[0046] Refer to Figure 2 , Figure 2 is a schematic flowchart of the first embodiment of the user-based recommendation method of the present invention.

[0047] In this embodiment, the user-based recommendation method is applied to a user-based recommendation device, and the method includes:

[0048] Step S101, determine the user similarity according to the obtained data set, sort the user similarity, and determine the neighbor cluster according to the sorting result;

[0049] Specifically, obtain a data set including a user-item rating matrix, and determine the user similarity based on the data set; in this embodiment, the data set is a user-item (user-item) rating matrix R(m,n), where m represents the number of users and n represents the number of items, and R ij represents the rating of user i for item j. The rating can be represented by binary or other methods such as a 5-point system or a 10-point system. Usually, the data set used is the Movielens (Power Lens) data set, and the rating system is a 5-point system. Table 1 shows a user-item rating matrix.

[0050] Table 1 User-Item Rating Matrix

[0051]

[0052]

[0053] In linear algebra, the Pearson correlation coefficient is usually used to describe the degree of linear correlation between two variables, and the specific expression is as follows:

[0054]

[0055] In the collaborative filtering algorithm, this formula can also be used to represent the similarity between users. Let Sim(u, v) represent the user similarity, then:

[0056]

[0057] In the above formula, respectively represent the average ratings of user u and user v, respectively represent the ratings of user u and user v for item i.

[0058] After determining the user similarity, sort the user similarity in descending order, and determine a preset number of users with common rating items as the neighbor cluster according to the sorting result.

[0059] Sort according to the user similarity from large to small, and determine the Top-n neighbor cluster U = {u1, u2... u n} of the user v to be recommended, where Sim(u1, v) > Sim(u2, v) >... > Sim(u n , v), where the value of n can be 1 - len(U u ), where len(U u ) represents the size of the set of all users who have common rating items with user u), and the final range of n is determined by comparing the prediction errors of multiple training samples.

[0060] Step S102, determine the rating deviation degree based on the rating criteria of each user in the neighbor cluster, and determine the predicted rating based on the rating deviation degree;

[0061] After obtaining the data set, determine the user to be recommended and the item to be recommended from the neighbor cluster. Generally, there are missing data in the data set, that is, not all users have rated each item. Therefore, a user who has not rated a certain item can be determined as the user to be recommended, and the corresponding item can be determined as the item to be rated. And predict the predicted rating of the user to be recommended for the item to be recommended.

[0062] Determine the first rating deviation amount of the rated users who have rated the item to be recommended from the neighbor cluster for the item to be recommended; in this embodiment, the absolute value of the difference between the rating of the rated user for the item to be recommended and the average rating of the rated user is determined as the first rating deviation amount of the rated user for the item to be recommended.

[0063] Denote the first rating deviation amount as α, the rated user as u2, and the rating of the rated user u2 for the item to be recommended as R u2j , and denote the average rating of the rated user as Then

[0064]

[0065] Understandably, the rating R of the rated user u2 for the item to be recommended u2j may be greater than the average rating R u2j or may be less than And each user has their own rating upper limit and rating lower limit. For example, if the rating standard is 0 - 5 points, but some users may never give 0 points or the highest score, and users may habitually set their ratings between 2 - 4 points. Therefore, the first rating deviation degree of the first rating deviation amount can be determined according to the rating upper limit and rating lower limit of each user.

[0066] That is, based on the rating standard of the rated user, determine the first rating deviation degree of the first rating deviation amount, where the rating standard of the rated user includes the average rating, the rating upper limit, and the rating lower limit;

[0067] Specifically, determine the first rating deviation amount of the rated users in the neighbor cluster for the item to be recommended, where the first rating deviation amount is the difference between the actual rating of the rated user for the item to be recommended and the first average rating of the corresponding rated user;

[0068] Based on the rating deviation amount and the rating upper limit and rating lower limit of the corresponding rated user, determine the first rating deviation degree of the corresponding rated user for the item to be recommended. In this embodiment, denote the rating upper limit of each user as R MAX , and denote the rating lower limit as R MIN .

[0069] If the rating R of the rated user u2 for the item to be recommended j u2j is greater than the average rating then represent the first rating deviation degree β of the corresponding rated user for the item to be recommended as:

[0070]

[0071] If the rating R of the item j to be recommended given by the rated user u2 u2j is less than or equal to the average rating then the first rating deviation degree β of the corresponding rated user to the item to be recommended is expressed as:

[0072]

[0073] Based on R u2j and the size relationship, the first rating deviation degree β can be specifically expressed as:

[0074]

[0075] For example, the average rating of user A is 4 points. Suppose his rating for item I is 3 points and his rating for item J is 5 points. When calculating the first rating deviation amount, the deviation amounts of both item I and J are 1. However, it is obvious that 5 points is the highest score that A can rate, and it can be considered the most favorite degree in terms of preference, and the deviation degree can be considered 100%. For item I, 3 points is not the lowest score, and the lowest score that the user can rate is 0 points. Therefore, when evaluating the first rating deviation degree, the range within which the rating can fluctuate also needs to be considered.

[0076] Based on the first rating deviation degree and user similarity, determine the change trend of user preference degree and the second rating deviation degree of the user to be recommended for the item to be recommended;

[0077] Express the change trend of the user preference degree as γ, then

[0078]

[0079] where, u1 represents the user to be recommended, and u2 represents the rated user of the item to be recommended. The change trend γ of the user preference degree can reflect to a certain extent the rating deviation degree of the user to be recommended u1 for the item to be recommended j and the change trend of the preference degrees of user u1 and user u2.

[0080] Furthermore, express the second rating deviation degree as δ, then

[0081]

[0082] Based on the second rating deviation degree and the average rating of the user to be recommended, determine the second rating deviation amount of the user to be recommended for the item to be recommended;

[0083] In this embodiment, the product of the second scoring deviation degree δ and the average score of the user to be recommended is determined as the second scoring deviation amount, and the second scoring deviation amount is represented as μ. Then

[0084]

[0085] After determining the second scoring deviation amount δ, the predicted score is determined based on the correlation trend, the second scoring deviation amount, and the average score of the user to be recommended.

[0086] If the correlation trend is positive, the sum of the average score of the user to be recommended and the second scoring deviation amount is determined as the predicted score.

[0087] If the correlation trend is negative, the difference between the average score of the user to be recommended and the second scoring deviation amount is determined as the predicted score.

[0088] The predicted score is represented as P u1j , then

[0089]

[0090] Step S103: Recommend the item to be recommended to the user to be recommended based on the predicted score.

[0091] Predict the predicted preference degree of the user to be recommended for the item to be recommended based on the predicted score and the average score of the user to be recommended; generally, if the predicted score is greater than or equal to the average score of the user to be predicted, it is determined that the preference degree is high; otherwise, if the predicted score is less than the average score of the user to be predicted, it is determined that the preference degree is low. Compare the size of the predicted score and the average score of the user to be recommended; if the predicted score is greater than the average score of the user to be recommended, recommend the item to be recommended to the user to be recommended.

[0092] For example, the average score of user A is 3 points, then those higher than 3 points are considered to have a relatively high preference degree, and those lower than 3 points are considered to have a relatively low preference degree. The average score of user B is 3.5 points, then those with a predicted score higher than 3.5 points are classified as having a high preference degree, and those with a predicted score lower than 3.5 points have a low preference degree.

[0093] Suppose the scores of both user A and user B for an item are 5 points. In the traditional algorithm, when calculating the predicted score, the deviation degrees of user A and user B are inconsistent. However, in reality, the highest score is only 5 points. Considering that the score of 5 points is actually the highest value of the preference degree that user A and user B can reflect, it should be considered that the preference deviation degrees of the two users are the same.

[0094] For another example, the rating of a certain item by User A is 4.5 points, and the rating of User B is 5 points. In traditional algorithms, when predicting ratings, the deviation degrees of User A and User B are the same. However, in reality, the highest score is only 5 points, and 5 points is actually the highest value of the preference level that User B can show. But User A's rating is only 4.5 points. Obviously, this item is not the one with the highest preference for User A. Therefore, it should be considered that the preference deviations of the two users are different.

[0095] Through the above solution, this embodiment determines the user similarity according to the obtained data set, sorts the user similarity, and determines the neighbor cluster according to the sorting result; determines the rating deviation degree based on the rating criteria of each user in the neighbor cluster, and determines the predicted rating based on the rating deviation degree; recommends the item to be recommended to the user to be recommended based on the predicted rating. Thus, the predicted rating of the user to be recommended for the item to be recommended is predicted based on the rating criteria of each user, improving the accuracy of the predicted rating of each user and further improving the accuracy of the recommendation.

[0096] In addition, this embodiment also provides a user-based recommendation device. Refer to Figure 3 , Figure 3 which is a schematic diagram of the functional modules of the first embodiment of the user-based recommendation device of the present invention.

[0097] In this embodiment, the user-based recommendation device is a virtual device stored in the memory 1005 of the user-based recommendation device shown in Figure 1 to implement all the functions of the user-based recommendation program: used to determine the user similarity according to the obtained data set, sort the user similarity, and determine the neighbor cluster according to the sorting result; used to determine the rating deviation degree based on the rating criteria of each user in the neighbor cluster, and determine the predicted rating based on the rating deviation degree; used to recommend the item to be recommended to the user to be recommended based on the predicted rating.

[0098] Specifically, the user-based recommendation device includes:

[0099] Determination module 10, configured to determine the user similarity according to the obtained data set, sort the user similarity, and determine the neighbor cluster according to the sorting result;

[0100] Rating module 20, configured to determine the rating deviation degree based on the rating criteria of each user in the neighbor cluster, and determine the predicted rating based on the rating deviation degree;

[0101] Recommendation module 30, configured to recommend the item to be recommended to the user to be recommended based on the predicted rating.

[0102] Further, the rating module is further configured to:

[0103] Determine a first rating deviation amount of the item to be recommended by the rated users who have rated the item to be recommended from the neighbor cluster;

[0104] Determine a first rating deviation degree of the first rating deviation amount based on the rating criteria of the rated users, where the rating criteria of the rated users include average rating, rating upper limit, and rating lower limit;

[0105] Determine the change trend of user preference degree and a second rating deviation degree of the user to be recommended for the item to be recommended based on the first rating deviation degree and user similarity;

[0106] Determine a second rating deviation amount of the user to be recommended for the item to be recommended based on the second rating deviation degree and the average rating of the user to be recommended;

[0107] Determine the predicted rating based on the correlation trend, the second rating deviation amount, and the average rating of the user to be recommended.

[0108] Further, the rating module is further configured to:

[0109] Determine a first rating deviation amount of the rated users in the neighbor cluster for the item to be recommended, where the first rating deviation amount is the difference between the actual rating of the rated user for the item to be recommended and the first average rating of the corresponding rated user;

[0110] Determine a first rating deviation degree of the corresponding rated user for the item to be recommended based on the rating deviation amount, the rating upper limit, and the rating lower limit of the corresponding rated user.

[0111] Further, the rating module is further configured to:

[0112] If the correlation trend is positive correlation, determine the sum of the average rating of the user to be recommended and the second rating deviation amount as the predicted rating;

[0113] If the correlation trend is negative correlation, determine the difference between the average rating of the user to be recommended and the second rating deviation amount as the predicted rating.

[0114] Further, the determination module is further configured to:

[0115] Obtain a data set including a user-item rating matrix, and determine the user similarity based on the data set;

[0116] Sort the user similarity in descending order, and determine a preset number of users with common rated items as the neighbor cluster according to the sorting result.

[0117] Further, the determining module is further configured to:

[0118] Determine the users to be recommended and the items to be recommended from the neighbor clusters.

[0119] Further, the recommending module is further configured to:

[0120] Compare the predicted score with the average score of the user to be recommended;

[0121] If the predicted score is greater than the average score of the user to be recommended, recommend the item to be recommended to the user to be recommended.

[0122] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a user-based recommendation program is stored. When the user-based recommendation program is run by a processor, the steps of the user-based recommendation method described above are implemented, which will not be elaborated here.

[0123] Compared with the prior art, a user-based recommendation method, device, equipment and computer-readable storage medium provided by the present invention include: determining user similarity according to the obtained data set, sorting the user similarity, and determining neighbor clusters according to the sorting result; determining a score deviation degree based on the scoring criteria of each user in the neighbor clusters, and determining a predicted score based on the score deviation degree; recommending the item to be recommended to the user to be recommended based on the predicted score. Thus, the predicted score of the user to be recommended for the item to be recommended is predicted based on the scoring criteria of each user, improving the accuracy of the predicted score of each user, and further improving the accuracy of the recommendation.

[0124] It should be noted that, in this article, the terms "including", "comprising" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or system. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or system including the element.

[0125] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0126] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium as described above (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions for causing a terminal device to execute the methods described in various embodiments of the present invention.

[0127] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structure or process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A user-based recommendation method, characterized in that, The method includes: Determine the user similarity Sim(u1, u2) according to the obtained dataset, sort the user similarity, and determine the neighbor cluster according to the sorting result; Determine the absolute value of the difference between the rating of the item to be recommended by the rated users in the neighbor cluster and the average rating of the rated users as the first rating deviation amount of the item to be recommended; If the rating of the item to be recommended by the rated user is greater than the average rating, determine the first rating deviation degree β of the first rating deviation amount based on the first rating deviation amount, the rating of the item to be recommended by the rated user, and the upper rating limit; If the rating of the item to be recommended by the rated user is less than or equal to the average rating, determine the first rating deviation degree β of the first rating deviation amount based on the first rating deviation amount, the rating of the item to be recommended by the rated user, and the lower rating limit; Determine the change trend of user preference degree based on the first scoring deviation β and user similarity Sim(u1, u2) where n represents the number of items, and determine the second scoring deviation of the item to be recommended for the user to be recommended based on the change trend γ of the user preference degree The change trend of the user preference degree is used to reflect the change trend of the preference degrees of the user to be recommended and the users who have scored the item to be recommended; Determine the second rating deviation amount of the item to be recommended for the user to be recommended based on the second rating deviation degree and the average rating of the user to be recommended; If the change trend of the user preference degree is greater than 0, determine the sum of the average rating of the user to be recommended and the second rating deviation amount as the predicted rating; If the change trend of the user preference degree is less than 0, determine the difference between the average rating of the user to be recommended and the second rating deviation amount as the predicted rating; Recommend the item to be recommended to the user to be recommended based on the predicted rating.

2. The method according to claim 1, wherein The step of determining the user similarity according to the obtained dataset, sorting the user similarity, and determining the neighbor cluster according to the sorting result includes: Obtain a dataset including a user-item rating matrix, and determine the user similarity based on the dataset; Sort the user similarity in descending order, and determine a preset number of users with common rated items as the neighbor cluster according to the sorting result.

3. The method according to claim 2, wherein After obtaining the dataset including the user-item rating matrix, it further includes: Determine the user to be recommended and the item to be recommended from the neighbor cluster.

4. The method according to any one of claims 1-3, characterized in that, The above step of recommending the item to be recommended to the user to be recommended based on the predicted rating includes: Compare the size of the predicted rating and the average rating of the user to be recommended; If the predicted rating is greater than the average rating of the user to be recommended, recommend the item to be recommended to the user to be recommended.

5. A user-based recommendation device, characterized in that, The user-based recommendation device includes: A determination module, configured to determine the user similarity Sim(u1, u2) according to the obtained dataset, sort the user similarity, and determine the neighbor cluster according to the sorting result; A scoring module, which is used to determine the absolute value of the difference between the score of the item to be recommended by the rated users in the neighbor cluster and the average score of the rated users as the first scoring deviation amount of the item to be recommended; if the score of the rated user for the item to be recommended is greater than the average score, then based on the first scoring deviation amount, the score of the rated user for the item to be recommended, and the score upper limit, determine the first scoring deviation degree β of the first scoring deviation amount; if the score of the rated user for the item to be recommended is less than or equal to the average score, then based on the first scoring deviation amount, the score of the rated user for the item to be recommended, and the score lower limit, determine the first scoring deviation degree β of the first scoring deviation amount; determine the change trend of the user preference degree based on the first scoring deviation degree β and the user similarity Sim(u1,u2) where n represents the number of items, and determine the second scoring deviation degree of the item to be recommended for the user to be recommended based on the change trend γ of the user preference degree The change trend of the user preference degree is used to reflect the change trend of the preference degrees of the user to be recommended and the rated users of the item to be recommended; determine the second scoring deviation amount of the item to be recommended for the user to be recommended based on the second scoring deviation degree and the average score of the user to be recommended; if the change trend of the user preference degree is greater than 0, then determine the sum of the average score of the user to be recommended and the second scoring deviation amount as the predicted score; if the change trend of the user preference degree is less than 0, then determine the difference between the average score of the user to be recommended and the second scoring deviation amount as the predicted score; A recommendation module, configured to recommend the item to be recommended to the user to be recommended based on the predicted rating.

6. A user-based recommendation device, characterized in that, The user-based recommendation device includes a processor, a memory, and a user-based recommendation program stored in the memory. When the user-based recommendation program is run by the processor, it implements the steps of the user-based recommendation method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, A recommendation program based on users is stored on the computer-readable storage medium. When the recommendation program based on users is run by a processor, the steps of the recommendation method based on users according to any one of claims 1-4 are implemented.

Images