Push method, device, computer equipment and storage medium based on trust network

By analyzing product review information to construct a user trust matrix and trusted user clusters, the cold start problem in collaborative filtering recommendation algorithms is solved, resulting in more accurate product recommendations.

CN109360058BActive Publication Date: 2024-11-22PING AN TECH (SHENZHEN) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201811191704.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2018-10-12
Publication Date
2024-11-22
Estimated Expiration
2038-10-12

AI Technical Summary

Technical Problem

Existing collaborative filtering-based recommendation algorithms suffer from a cold start problem when calculating the similarity between users, which affects the accuracy of the recommendation process.

Method used

By analyzing product review information, the mutual review relationships between users are determined, the user trust matrix is ​​updated, a cluster of trusted users is formed, and products are recommended and pushed to target users based on the ratings of trusted users.

Benefits of technology

It enables the construction of trusted user clusters through explicit trust relationships between users, improving the accuracy and precision of recommendations and solving the cold start problem.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN109360058B_ABST
    Figure CN109360058B_ABST
Patent Text Reader

Abstract

The present invention discloses a push method, device, computer equipment and storage medium based on a trust network. The method analyzes product review information to determine whether there is a mutual review relationship between users, so as to obtain the user trust matrix stored at the current moment, obtain the target user corresponding to the row vector selected from the current trust matrix, and obtain the trust users whose trust values ​​are before the ranking threshold according to the size of each trust value in the row vector to form a trust user cluster; according to the score row vector corresponding to each trust user in the trust user cluster in the user-score matrix, obtain the trust score value of each product by the trust user to form a product recommendation row vector; obtain a product recommendation list from the product recommendation row vector, and push the product recommendation list to the receiving end corresponding to the target user. The method adopts intelligent recommendation technology to realize the determination of the trust user cluster through the mutual review relationship between users, and makes accurate recommendations to the target user according to the products recommended by the trust users.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of information push technology, and in particular to a push method, device, computer equipment and storage medium based on a trust network. Background Art

[0002] At present, online shopping on online shopping malls based on the Internet has become more and more frequent. When these online shopping malls recommend products to users, they usually use recommendation algorithms based on collaborative filtering (the principle of collaborative filtering algorithm is that users like products that have been liked by users with similar interests. For example, if your friends like the movie Harry Potter I, then it will be recommended to you. This is the simplest collaborative filtering algorithm based on users). When using the recommendation algorithm based on collaborative filtering, the data set used is mainly based on the user-product rating matrix, and then the similarity between users is calculated, which belongs to the invisible trust network. However, there is a cold start problem when using the rating matrix to calculate the similarity between users, which affects the recommendation process. Summary of the invention

[0003] The embodiments of the present invention provide a push method, apparatus, computer device and storage medium based on a trust network, aiming to solve the problem in the prior art that a recommendation algorithm based on collaborative filtering needs to use a scoring matrix to calculate the similarity between users, which easily leads to a cold start of the recommendation operation, thereby affecting the recommendation process.

[0004] In a first aspect, an embodiment of the present invention provides a push method based on a trust network, which includes:

[0005] If the product review information at the current moment is compared with the product review information at the previous moment to determine whether there is a new mutual review relationship between users, the current trust value is obtained according to the review content corresponding to the new mutual review relationship between users;

[0006] The user trust matrix stored at the last moment is updated according to the current trust value to obtain the user trust matrix stored at the current moment; wherein each value in the user trust matrix stored at the last moment and the user trust matrix stored at the current moment represents the trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located;

[0007] Obtaining a target user corresponding to a row vector selected from the user trust matrix stored at the current moment, and according to the values ​​of each trust value in the row vector corresponding to the target user, obtaining trust users whose trust values ​​are before a preset ranking threshold to form a trust user cluster;

[0008] According to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix, obtaining the trust score value of each product by the trusted user to form a product recommendation row vector; and

[0009] A product recommendation list is obtained from the product recommendation row vector, and the product recommendation list is pushed to a receiving end corresponding to the target user.

[0010] In a second aspect, an embodiment of the present invention provides a push device based on a trust network, which includes:

[0011] A current trust value acquisition unit, used to compare the product review information at the current moment with the product review information at the previous moment to determine whether there is a new mutual review relationship between users, and to acquire the current trust value according to the review content corresponding to the new mutual review relationship between users;

[0012] A user trust matrix updating unit, configured to update the user trust matrix stored at the last moment according to the current trust value to obtain the user trust matrix stored at the current moment; wherein each value in the user trust matrix stored at the last moment and the user trust matrix stored at the current moment represents a trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located;

[0013] A trusted user cluster acquisition unit, configured to acquire a target user corresponding to a row vector selected from the user trust matrix stored at the current moment, and acquire trusted users whose trust values ​​are before a preset ranking threshold according to the trust values ​​in the row vector corresponding to the target user to form a trusted user cluster;

[0014] A recommendation row vector acquisition unit, configured to acquire the trust score value of each product by the trusted user according to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix to form a product recommendation row vector;

[0015] The push unit is used to obtain a product recommendation list from the product recommendation row vector, and push the product recommendation list to a receiving end corresponding to a target user.

[0016] In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the push method based on the trust network described in the first aspect is implemented.

[0017] In a fourth aspect, an embodiment of the present invention 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 processor executes the push method based on the trust network described in the first aspect above.

[0018] The embodiment of the present invention provides a push method, device, computer equipment and storage medium based on a trust network. The method determines whether there is a mutual comment relationship between users by parsing product review information, so as to adjust and obtain the user trust matrix stored at the current moment, obtain the target user corresponding to the row vector selected from the user trust matrix stored at the current moment, and obtain the trust users whose trust values ​​are before the preset ranking threshold according to the size of each trust value in the row vector corresponding to the target user to form a trust user cluster; according to the score row vector corresponding to each trust user in the trust user cluster in the user-score matrix, obtain the trust score value of each product by the trust user to form a product recommendation row vector; obtain the product recommendation list from the product recommendation row vector, and push the product recommendation list to the receiving end corresponding to the target user. The method adopts intelligent recommendation technology to realize the determination of the trust user cluster through the mutual comment relationship between users, and makes accurate recommendations to the target user according to the products recommended by the trust user. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0020] Figure 1 A schematic diagram of a push method based on a trust network provided by an embodiment of the present invention;

[0021] Figure 2 A schematic diagram of a sub-process of a push method based on a trust network provided by an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of another sub-process of the push method based on the trust network provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of another sub-process of the push method based on the trust network provided by an embodiment of the present invention;

[0024] Figure 5 A schematic block diagram of a push device based on a trust network provided by an embodiment of the present invention;

[0025] Figure 6 A schematic block diagram of subunits of a push device based on a trust network provided by an embodiment of the present invention;

[0026] Figure 7 A schematic block diagram of another subunit of the push device based on the trust network provided by an embodiment of the present invention;

[0027] Figure 8 A schematic block diagram of another subunit of the push device based on the trust network provided by an embodiment of the present invention;

[0028] Fig. 9 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0031] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0032] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0033] See also Figure 1 , Figure 1 It is a flow chart of a push method based on a trust network provided in an embodiment of the present invention. The push method based on a trust network is applied to a management server. The method is executed by application software installed in the management server. The management server is an enterprise terminal for performing push based on a trust network.

[0034] like Figure 1 As shown, the method includes steps S110 to S150.

[0035] S110: If the product review information at the current moment is compared with the product review information at the previous moment to determine whether a new mutual review relationship exists between users, the current trust value is obtained according to the review content corresponding to the new mutual review relationship between users.

[0036] In this embodiment, after the management server crawls the product review information corresponding to each web page in the preset URL address list, it can determine the trust relationship between users based on whether there is a review relationship between users, and obtain the current trust value based on the review content corresponding to the newly added mutual review relationship between users. Such user relationships are explicit. For example, in product reviews, user A affirms the comments of user B, which shows that user A trusts user B, or in the community space, user A affirms the content of user B. Based on this explicit trust network, combined with existing historical data, the cold start problem can be better solved.

[0037] In one embodiment, if Figure 2 As shown, step S110 includes:

[0038] S111, obtaining the comment content corresponding to the newly added mutual comment relationship between users, and performing word segmentation on the comment content to obtain a word segmentation result;

[0039] S112, obtaining a word vector corresponding to each comment keyword in the word segmentation result through the Word2Vec model, and obtaining a text vector from the word vector corresponding to each comment keyword in the word segmentation result;

[0040] S113, using the text vector as the input of the pre-trained naive Bayes model to obtain a sentiment recognition result corresponding to the comment content; wherein, if the sentiment recognition result is a positive sentiment result, the sentiment recognition result value is 1, if the sentiment recognition result is a negative sentiment result, the sentiment recognition result value is -1, and if the sentiment recognition result is a neutral sentiment result, the sentiment recognition result value is 0;

[0041] S114, multiply the emotion recognition result by the attenuation coefficient to obtain the current trust value; where the attenuation coefficient is e -λ(t-t0) , λ is a preset adjustment parameter with a value range of (0,1), and t-t0 is the comment time interval corresponding to the mutual comment relationship between users.

[0042] In this embodiment, by parsing the product review information, it is mainly to determine whether there is a mutual review relationship between users. If there is a review relationship between two users, the review content is judged as positive, neutral, and negative. When performing emotion recognition on the review content, the review content is first segmented and then converted into a text vector, and the text vector is used as the input of the naive Bayes model to obtain the emotion recognition result. However, in order to take into account the influence of time decay on the review relationship between users, it is necessary to multiply the emotion recognition result by an attenuation coefficient (which can also be understood as a time decay factor) to obtain a more objective and accurate current trust value. Among them, Word2Vec is a model for learning semantic knowledge in an unsupervised manner from a large amount of text corpus.

[0043] If the current trust value of user i to user j is recorded as a ij , then a ij + = Sentiment ij *e -λ(t-t0) , that is, the trust value at the current moment is adjusted automatically based on the trust value at the previous moment (for example, if user i comments on user j's comment every once in a while, then each comment will have an impact on the current trust value), where Sentiment ij Sentiment represents the emotion recognition result obtained by performing emotion recognition on the comment made by user i on user j. If the emotion recognition result is a positive emotion result, then Sentiment ij =1, if the emotion recognition result is a negative emotion result, Sentiment ij = -1, if the emotion recognition result is a neutral emotion result, Sentiment ij =0.

[0044] For example, if the preset adjustment parameter λ=0.5, user 1 adds a comment below user 2's comment, and the comment is a positive comment, and the difference between user 1's comment time t and user 2's comment time t0 is 1 day, then user 1's current trust value for user 2 is recorded as a 12 =0+1*e -0.5*(1-0) =0.607; User 1 adds a comment below User 3’s comment, and the comment is a negative comment. The difference between User 1’s comment time t and User 3’s comment time t0 is 1 day. Then User 1’s current trust value for User 3 is recorded as a 13 =0-1*e -0.5*(1-0) =-0.607; User 1 adds a comment below User 4's comment, and the comment is a neutral comment. The difference between User 1's comment time t and User 4's comment time t0 is 1 day. Then User 1's current trust value for User 4 is recorded as a 15 =0-0*e -0.5*(1-0) =0.

[0045] Using the emotion recognition results corresponding to the comments between users as the basis for calculating the trust value between users is an explicit rather than implicit way of building a trust network, which can solve the problem of cold start in the recommendation process.

[0046] In one embodiment, if Figure 4 As shown, step S111 includes:

[0047] S1111, extracting candidate words from the review content in order from left to right;

[0048] S1112, searching the pre-stored dictionary for a probability value corresponding to each candidate word, and recording the left neighbor word of each candidate word;

[0049] S1113, calculating and obtaining the cumulative probability of each candidate word, and obtaining the cumulative probabilities of the multiple left neighboring words corresponding to each candidate word, if there is a left neighboring word with the maximum cumulative probability among the multiple left neighboring words of each candidate word, the left neighboring word with the maximum cumulative probability is used as the best left neighboring word corresponding to the candidate word;

[0050] S1114. Taking the end word of the comment content as the starting point, output the best left neighbor word corresponding to each candidate word from right to left in order to obtain the word segmentation result.

[0051] In this embodiment, when the review content is segmented, the segmentation is performed using a segmentation method based on a probability statistics model. For example, let C = C1C2...Cm, C is the Chinese character string to be segmented, let W = W1W2...Wn, W is the segmentation result, Wa, Wb,...Wk are all possible segmentation schemes of C. Then, the segmentation model based on probability statistics is a segmentation model that can find the target word string W so that W satisfies: P(W|C) = MAX(P(Wa|C), P(Wb|C)...P(Wk|C)), and the word string W obtained by the above segmentation model is the word string with the maximum estimated probability.

[0052] That is, for a substring S to be segmented, take out all candidate words w1, w2, ..., wi, ..., wn in order from left to right; find the probability value P(wi) of each candidate word in the dictionary, and record all the left neighbor words of each candidate word; calculate the cumulative probability of each candidate word, and compare and obtain the best left neighbor word of each candidate word; if the current word wn is the last word of the string S, and the cumulative probability P(wn) is the largest, then wn is the terminal word of S; starting from wn, in order from right to left, output the best left neighbor word of each word in turn, that is, the segmentation result of S.

[0053] S120. Update the user trust matrix stored at the previous moment according to the current trust value to obtain the user trust matrix stored at the current moment; wherein, in the user trust matrix stored at the previous moment and the user trust matrix stored at the current moment, each value represents the trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located.

[0054] In this embodiment, the user trust matrix represents the trust value between users. The horizontal axis and the vertical axis in the user trust matrix are both user lists. The user trust matrix A n*n Middle ijRefers to the trust value of user i to user j. All values ​​in the user trust matrix stored at the initial moment are 0. After that, it is re-judged every time period T whether user i has new comments on user j to determine whether to adjust the trust value of user i to user j. The user trust matrix stored at the last moment is the management server crawling the product review information corresponding to each web page in the preset URL address list at the last moment (such as Z1 hour Z2 minutes on X month Y day 2018). It can judge the trust relationship between users based on whether there is a comment relationship between users, and obtain the trust value at the last moment based on the comment content corresponding to the mutual comment relationship between users.

[0055] If the time interval T between the previous moment and the current moment (e.g., Z3:04:00 on X / Y / 2018) is the same as the time period T between the previous moment and the current moment (e.g., Z3:04:00 on X / Y / 2018), after crawling the product review information corresponding to each web page in the preset URL address list, it is possible to determine whether the trust relationship between users has changed based on whether there is a new review relationship between users, and obtain the trust value at the current moment based on the review content corresponding to the mutual review relationship between users. That is, the current trust value of user i to user j is recorded as a ij , then a ij + = Sentiment ij *e -λ(t-t0) That is, the trust value at the current moment is adjusted automatically based on the trust value at the previous moment (for example, if user i comments on user j's comment every once in a while, then each comment will have an impact on the current trust value).

[0056] S130, obtaining a target user corresponding to a row vector selected from the user trust matrix stored at the current moment, and according to the sizes of the trust values ​​in the row vector corresponding to the target user, obtaining trust users whose trust values ​​are before a preset ranking threshold to form a trust user cluster.

[0057] In this embodiment, for example, the user trust matrix stored at the current moment is as follows:

[0058]

[0059] The row vector selected by the user trust matrix stored at the current moment is the row vector of the first row, the target user corresponding to the row vector of the first row is user 1, and each value in the row vector of the first row represents the trust value between user 1 and other users. If the preset ranking threshold is 3, the trust values ​​before the third place (i.e. the trust values ​​of the top two) are 3 and 1, corresponding to user 2 and user 4 respectively, and the trusted user cluster includes user 2 and user 4.

[0060] S140 . According to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix, obtain the trust score value of each product by the trusted user to form a product recommendation row vector.

[0061] In this embodiment, the user-rating matrix represents the user's rating of the project (the project can be understood as a specific commodity). The horizontal axis of the user-rating matrix is ​​the project, the vertical axis is the user, and the value in the middle is the user i's rating of the project j. For example, the user-rating matrix S is a 4×6 matrix, such as:

[0062]

[0063] The row vectors in the first row of the user-rating matrix S represent the ratings of user 1 for items 1 to 5 respectively, the row vectors in the second row represent the ratings of user 2 for items 1 to 5 respectively, and the row vectors in the third row represent the ratings of user 3 for items 1 to 5 respectively.

[0064] For example, if it is determined that the trusted user cluster includes user 2 and user 4, then the score row vectors of user 2 and user 4 are obtained in the user-score matrix, that is, the row vector of the second row and the row vector of the fourth row. By comprehensively considering the trust value between the target user and each user in the trusted user cluster, as well as the score value of each product by the trusted user, the product recommendation row vector can be calculated.

[0065] In one embodiment, if Figure 3 As shown, step S140 includes:

[0066] S141, obtaining a scoring matrix composed of the scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix;

[0067] S142, obtaining a trusted user row vector composed of the magnitudes of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user;

[0068] S143: Multiply the trusted user row vector by the rating matrix to obtain a product recommendation row vector.

[0069] In this implementation, for example, if the selected trusted user cluster includes user 2 and user 4, the score row vectors of user 2 and user 4 are obtained in the user-score matrix, that is, the row vector of the second row and the row vector of the fourth row. Then, the score matrix composed of the scores between user 2 and user 4 for each item (that is, each product) is as follows:

[0070]

[0071] In one embodiment, step S141 specifically includes: obtaining the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix, and arranging them according to the order of the row numbers in the score row vector corresponding to each trusted user in the user-score matrix to obtain the score matrix. According to the order in which each row of each user in the user-score matrix appears, the score row vector corresponding to each trusted user in the user-score matrix is ​​obtained in sequence. The above method can accurately obtain the score matrix composed of the score row vector corresponding to each trusted user, which is convenient for the subsequent calculation of the user's trust score value for each product.

[0072] The trusted user row vector composed of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user is

[31] .

[0073] In one embodiment, step S142 specifically includes: obtaining trust users whose trust values ​​are before a preset ranking threshold, arranging the trust users whose trust values ​​are before the ranking threshold in the order of column numbers in the row vector corresponding to the target user, and obtaining a trust user row vector. According to the order in which the columns of the above-mentioned users appear in the row vector corresponding to the target user, the trust values ​​of the above-mentioned users in the row vector corresponding to the target user are sequentially obtained to form a trust user row vector. The above method can accurately obtain the trust user row vector, which is convenient for the subsequent calculation of the trust score value of the user for each product.

[0074] Multiply the trusted user row vector by the rating matrix, that is,

[0075]

[0076] That is, the product recommendation row vector is [2 13 8 13 7]. The products with higher scores in this product recommendation row vector can be used as the products recommended by the trust group to the target user. By multiplying the trust user row vector with the score matrix to obtain the product recommendation row vector, the comprehensive score of each product given by the trust user can be effectively referenced as a reference indicator for product recommendation to the target user.

[0077] S150 , obtaining a product recommendation list from the product recommendation row vector, and pushing the product recommendation list to a receiving end corresponding to the target user.

[0078] In this embodiment, after obtaining the product recommendation row vector, the comprehensive score of each trusted user in the trusted user cluster for each product can be obtained. At this time, the score whose score ranking is before the preset ranking value (such as the preset ranking value of 4) among the scores corresponding to the product recommendation row vector can be selected, and then the product information corresponding to the score whose score ranking is before the preset ranking value among the scores can be obtained, and the product information can be formed into a product recommendation list and pushed to the receiving end corresponding to the target user. The method of calculating the product recommendation row vector to obtain the product recommendation list can effectively improve the accuracy of the recommendation.

[0079] For example, among the five items recommended by the above trust group (user 2 and user 4) to user 1, the top three items after sorting by scores in descending order are item 2, item 3 and item 4. At this time, the above three items are used as the item recommendation row vector.

[0080] This method uses intelligent recommendation technology to determine the trusted user cluster through the mutual comment relationship between users, and makes accurate recommendations to target users based on the products recommended by the trusted users.

[0081] The embodiment of the present invention further provides a push device based on a trust network, and the push device based on a trust network is used to execute any embodiment of the push method based on a trust network. Figure 5 , Figure 5 1 is a schematic block diagram of a push device based on a trust network provided by an embodiment of the present invention. The push device 100 based on a trust network can be configured in a management server.

[0082] like Figure 5 As shown, the push device 100 based on the trust network includes a current trust value acquisition unit 110 , a user trust matrix update unit 120 , a trusted user cluster acquisition unit 130 , a recommendation row vector acquisition unit 140 , and a push unit 150 .

[0083] The current trust value acquisition unit 110 is used to compare the product review information at the current moment with the product review information at the previous moment to determine whether there is a new mutual review relationship between users, and to acquire the current trust value according to the review content corresponding to the new mutual review relationship between users.

[0084] In this embodiment, after the management server crawls the product review information corresponding to each web page in the preset URL address list, it can determine the trust relationship between users based on whether there is a review relationship between users, and obtain the current trust value based on the review content corresponding to the newly added mutual review relationship between users. Such user relationships are explicit. For example, in product reviews, user A affirms the comments of user B, which shows that user A trusts user B, or in the community space, user A affirms the content of user B. Based on this explicit trust network, combined with existing historical data, the cold start problem can be better solved.

[0085] In one embodiment, if Figure 6 As shown, the current trust value acquisition unit 110 includes:

[0086] The word segmentation unit 111 is used to obtain the comment content corresponding to the newly added mutual comment relationship between users, and segment the comment content to obtain a word segmentation result;

[0087] A text vector acquisition unit 112 is used to acquire a word vector corresponding to each comment keyword in the word segmentation result through a Word2Vec model, and acquire a text vector from the word vector corresponding to each comment keyword in the word segmentation result;

[0088] The emotion recognition unit 113 is used to use the text vector as the input of the pre-trained naive Bayes model to obtain the emotion recognition result corresponding to the comment content; wherein, if the emotion recognition result is a positive emotion result, the emotion recognition result value is 1, if the emotion recognition result is a negative emotion result, the emotion recognition result value is -1, and if the emotion recognition result is a neutral emotion result, the emotion recognition result value is 0;

[0089] The trust value calculation unit 114 is used to multiply the emotion recognition result by the attenuation coefficient to obtain the current trust value; wherein the attenuation coefficient is e -λ(t-t0) , λ is a preset adjustment parameter with a value range of (0,1), and t-t0 is the comment time interval corresponding to the mutual comment relationship between users.

[0090] In this embodiment, by parsing the product review information, it is mainly to determine whether there is a mutual review relationship between users. If there is a review relationship between two users, the review content is judged as positive, neutral, and negative. When performing emotion recognition on the review content, the review content is first segmented and then converted into a text vector, and the text vector is used as the input of the naive Bayes model to obtain the emotion recognition result. However, in order to take into account the influence of time decay on the review relationship between users, it is necessary to multiply the emotion recognition result by an attenuation coefficient (which can also be understood as a time decay factor) to obtain a more objective and accurate current trust value.

[0091] If the current trust value of user i to user j is recorded as a ij , then a ij + = Sentiment ij *e -λ(t-t0) , that is, the trust value at the current moment is adjusted automatically based on the trust value at the previous moment (for example, if user i comments on user j's comment every once in a while, then each comment will have an impact on the current trust value), where Sentiment ij Sentiment represents the emotion recognition result obtained by performing emotion recognition on the comment made by user i on user j. If the emotion recognition result is a positive emotion result, then Sentiment ij =1, if the emotion recognition result is a negative emotion result, Sentiment ij = -1, if the emotion recognition result is a neutral emotion result, Sentiment ij =0.

[0092] For example, if the preset adjustment parameter λ=0.5, user 1 adds a comment below user 2's comment, and the comment is a positive comment, and the difference between user 1's comment time t and user 2's comment time t0 is 1 day, then user 1's current trust value for user 2 is recorded as a 12 =0+1*e -0.5*(1-0) =0.607; User 1 adds a comment below User 3’s comment, and the comment is a negative comment. The difference between User 1’s comment time t and User 3’s comment time t0 is 1 day. Then User 1’s current trust value for User 3 is recorded as a 13 =0-1*e -0.5*(1-0) =-0.607; User 1 adds a comment below User 4's comment, and the comment is a neutral comment. The difference between User 1's comment time t and User 4's comment time t0 is 1 day. Then User 1's current trust value for User 4 is recorded as a 15 =0-0*e -0.5*(1-0) =0.

[0093] Using the emotion recognition results corresponding to the comments between users as the basis for calculating the trust value between users is an explicit rather than implicit way of building a trust network, which can solve the problem of cold start in the recommendation process.

[0094] In one embodiment, if Figure 8 As shown, the word segmentation unit 111 includes:

[0095] The candidate word selection unit 1111 is used to extract candidate words from the comment content in order from left to right;

[0096] The initial left neighbor word acquisition unit 1112 is used to query the probability value corresponding to each candidate word in the pre-stored dictionary and record the left neighbor word of each candidate word;

[0097] The best left neighbor word acquisition unit 1113 is used to calculate and acquire the cumulative probability of each candidate word, and acquire the cumulative probabilities of multiple left neighbor words corresponding to each candidate word. If there is a left neighbor word with a cumulative probability that is the maximum value among the cumulative probabilities of multiple left neighbor words among the multiple left neighbor words of each candidate word, the left neighbor word with the maximum value in the cumulative probability is used as the best left neighbor word corresponding to the candidate word;

[0098] The word segmentation result output unit 1114 is used to take the terminal word of the comment content as the starting point, and output the best left neighbor word corresponding to each candidate word from right to left in order to obtain the word segmentation result.

[0099] In this embodiment, when the review content is segmented, the segmentation is performed using a segmentation method based on a probability statistics model. For example, let C = C1C2...Cm, C is the Chinese character string to be segmented, let W = W1W2...Wn, W is the segmentation result, Wa, Wb,...Wk are all possible segmentation schemes of C. Then, the segmentation model based on probability statistics is a segmentation model that can find the target word string W so that W satisfies: P(W|C) = MAX(P(Wa|C), P(Wb|C)...P(Wk|C)), and the word string W obtained by the above segmentation model is the word string with the maximum estimated probability.

[0100] That is, for a substring S to be segmented, take out all candidate words w1, w2, ..., wi, ..., wn in order from left to right; find the probability value P(wi) of each candidate word in the dictionary, and record all the left neighbor words of each candidate word; calculate the cumulative probability of each candidate word, and compare and obtain the best left neighbor word of each candidate word; if the current word wn is the last word of the string S, and the cumulative probability P(wn) is the largest, then wn is the terminal word of S; starting from wn, in order from right to left, output the best left neighbor word of each word in turn, that is, the segmentation result of S.

[0101] The user trust matrix updating unit 120 is used to update the user trust matrix stored at the previous moment according to the current trust value to obtain the user trust matrix stored at the current moment; wherein, in the user trust matrix stored at the previous moment and the user trust matrix stored at the current moment, each value represents the trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located.

[0102] In this embodiment, the user trust matrix represents the trust value between users. The horizontal axis and the vertical axis in the user trust matrix are both user lists. The user trust matrix A n*n Middle ijRefers to the trust value of user i to user j. All values ​​in the user trust matrix stored at the initial moment are 0. After that, it is re-judged every time period T whether user i has new comments on user j to determine whether to adjust the trust value of user i to user j. The user trust matrix stored at the last moment is the management server crawling the product review information corresponding to each web page in the preset URL address list at the last moment (such as Z1 hour Z2 minutes on X month Y day 2018). It can judge the trust relationship between users based on whether there is a comment relationship between users, and obtain the trust value at the last moment based on the comment content corresponding to the mutual comment relationship between users.

[0103] If the time interval T between the previous moment and the current moment (e.g., Z3:04:00 on X / Y / 2018) is the same as the time period T between the previous moment and the current moment (e.g., Z3:04:00 on X / Y / 2018), after crawling the product review information corresponding to each web page in the preset URL address list, it is possible to determine whether the trust relationship between users has changed based on whether there is a new review relationship between users, and obtain the trust value at the current moment based on the review content corresponding to the mutual review relationship between users. That is, the current trust value of user i to user j is recorded as a ij , then a ij + = Sentiment ij *e -λ(t-t0) That is, the trust value at the current moment is adjusted automatically based on the trust value at the previous moment (for example, if user i comments on user j's comment every once in a while, then each comment will have an impact on the current trust value).

[0104] The trusted user cluster acquisition unit 130 is used to acquire the target user corresponding to the row vector selected from the user trust matrix stored at the current moment, and acquire the trusted users whose trust values ​​are before a preset ranking threshold according to the trust values ​​in the row vector corresponding to the target user to form a trusted user cluster.

[0105] In this embodiment, for example, the user trust matrix stored at the current moment is as follows:

[0106]

[0107] The row vector selected by the user trust matrix stored at the current moment is the row vector of the first row, the target user corresponding to the row vector of the first row is user 1, and each value in the row vector of the first row represents the trust value between user 1 and other users. If the preset ranking threshold is 3, the trust values ​​before the third place (i.e. the trust values ​​of the top two) are 3 and 1, corresponding to user 2 and user 4 respectively, and the trusted user cluster includes user 2 and user 4.

[0108] The recommendation row vector acquisition unit 140 is used to acquire the trust score value of each product of the trusted user according to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix to form a product recommendation row vector.

[0109] In this embodiment, the user-rating matrix represents the user's rating of the project (the project can be understood as a specific commodity). The horizontal axis of the user-rating matrix is ​​the project, the vertical axis is the user, and the value in the middle is the user i's rating of the project j. For example, the user-rating matrix S is a 4×6 matrix, such as:

[0110]

[0111] The row vectors in the first row of the user-rating matrix S represent the ratings of user 1 for items 1 to 5 respectively, the row vectors in the second row represent the ratings of user 2 for items 1 to 5 respectively, and the row vectors in the third row represent the ratings of user 3 for items 1 to 5 respectively.

[0112] For example, if it is determined that the trusted user cluster includes user 2 and user 4, then the score row vectors of user 2 and user 4 are obtained in the user-score matrix, that is, the row vector of the second row and the row vector of the fourth row. By comprehensively considering the trust value between the target user and each user in the trusted user cluster, as well as the score value of each product by the trusted user, the product recommendation row vector can be calculated.

[0113] In one embodiment, if Figure 7 As shown, the recommended row vector acquisition unit 140 includes:

[0114] The rating matrix acquisition unit 141 is used to acquire a rating matrix composed of rating row vectors corresponding to each trusted user in the trusted user cluster in the user-rating matrix;

[0115] A trusted user row vector acquisition unit 142 is used to acquire a trusted user row vector composed of the magnitudes of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user;

[0116] The matrix calculation unit 143 is used to multiply the trusted user row vector by the rating matrix to obtain a product recommendation row vector.

[0117] In this implementation, for example, if the selected trusted user cluster includes user 2 and user 4, the score row vectors of user 2 and user 4 are obtained in the user-score matrix, that is, the row vector of the second row and the row vector of the fourth row. Then, the score matrix composed of the scores between user 2 and user 4 for each item (that is, each product) is as follows:

[0118]

[0119] In one embodiment, the rating matrix acquisition unit 141 is further used to: obtain the rating row vectors corresponding to each trusted user in the trusted user cluster in the user-rating matrix, and arrange them according to the order of the row numbers in the rating row vectors corresponding to each trusted user in the user-rating matrix to obtain the rating matrix. According to the order in which the rows of each user in the user-rating matrix appear, the rating row vectors corresponding to each trusted user in the user-rating matrix are sequentially obtained. The above method can accurately obtain the rating matrix composed of the rating row vectors corresponding to each trusted user, which is convenient for the subsequent calculation of the user's trust rating value for each product.

[0120] The trusted user row vector composed of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user is

[31] .

[0121] In one embodiment, the trusted user row vector acquisition unit 142 is further used to: acquire trusted users whose trust values ​​are before a preset ranking threshold, and arrange the trusted users whose trust values ​​are before the ranking threshold in the order of the column numbers in the row vector corresponding to the target user to obtain the trusted user row vector. According to the order in which the above users appear in each column in the row vector corresponding to the target user, the trust values ​​in the row vector corresponding to the target user are sequentially acquired to form the trusted user row vector. The above method can accurately acquire the trusted user row vector, which is convenient for the subsequent calculation of the user's trust score value for each product.

[0122] Multiply the trusted user row vector by the rating matrix, that is,

[0123]

[0124] That is, the product recommendation row vector is [2 13 8 13 7]. The products with higher scores in this product recommendation row vector can be used as the products recommended by the trust group to the target user. By multiplying the trust user row vector with the score matrix to obtain the product recommendation row vector, the comprehensive score of each product given by the trust user can be effectively referenced as a reference indicator for product recommendation to the target user.

[0125] The push unit 150 is used to obtain a product recommendation list from the product recommendation row vector, and push the product recommendation list to a receiving end corresponding to a target user.

[0126] In this embodiment, after obtaining the product recommendation row vector, the comprehensive score of each trusted user in the trusted user cluster for each product can be obtained. At this time, the score whose score ranking is before the preset ranking value (such as the preset ranking value of 4) among the scores corresponding to the product recommendation row vector can be selected, and then the product information corresponding to the score whose score ranking is before the preset ranking value among the scores can be obtained, and the product information can be formed into a product recommendation list and pushed to the receiving end corresponding to the target user. The method of calculating the product recommendation row vector to obtain the product recommendation list can effectively improve the accuracy of the recommendation.

[0127] For example, among the five items recommended by the above trust group (user 2 and user 4) to user 1, the top three items after sorting by scores in descending order are item 2, item 3 and item 4. At this time, the above three items are used as the item recommendation row vector.

[0128] The device uses intelligent recommendation technology to determine the trusted user cluster through the mutual comment relationship between users, and makes accurate recommendations to target users based on the products recommended by the trusted users.

[0129] The above-mentioned push device based on the trust network can be implemented in the form of a computer program, and the computer program can be Fig. 9 Runs on the computer device shown.

[0130] See also Fig. 9 , Fig. 9 It is a schematic block diagram of a computer device provided by an embodiment of the present invention.

[0131] See also Fig. 9 The computer device 500 includes a processor 502 , a memory and a network interface 505 connected via a system bus 501 , wherein the memory may include a non-volatile storage medium 503 and an internal memory 504 .

[0132] The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, the processor 502 may execute a push method based on a trust network.

[0133] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

[0134] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute the push method based on the trust network.

[0135] The network interface 505 is used for network communication, such as providing data information transmission, etc. Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device 500 to which the solution of the present invention is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0136] The processor 502 is used to run a computer program 5032 stored in the memory to implement the following functions: if the product review information at the current moment is compared with the product review information at the previous moment to determine whether there is a new mutual review relationship between users, the current trust value is obtained according to the review content corresponding to the new mutual review relationship between users; the user trust matrix stored at the previous moment is updated according to the current trust value to obtain the user trust matrix stored at the current moment; wherein, in the user trust matrix stored at the previous moment and the user trust matrix stored at the current moment, each value represents the trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located; the target user corresponding to the row vector selected from the user trust matrix stored at the current moment is obtained, and according to the size of each trust value in the row vector corresponding to the target user, the trust users whose trust values ​​are before the preset ranking threshold are obtained to form a trust user cluster; according to the score row vector corresponding to each trust user in the trust user cluster in the user-score matrix, the trust score value of each product of the trust user is obtained to form a product recommendation row vector; the product recommendation list is obtained from the product recommendation row vector, and the product recommendation list is pushed to the receiving end corresponding to the target user.

[0137] In one embodiment, when the processor 502 executes the step of obtaining the current trust value based on the comment content corresponding to the newly added mutual comment relationship between users, the processor 502 performs the following operations: obtaining the comment content corresponding to the newly added mutual comment relationship between users, and segmenting the comment content to obtain a segmentation result; obtaining the word vector corresponding to each comment keyword in the segmentation result through the Word2Vec model, and obtaining the text vector from the word vector corresponding to each comment keyword in the segmentation result; using the text vector as the input of the pre-trained naive Bayes model to obtain the emotion recognition result corresponding to the comment content; wherein, if the emotion recognition result is a positive emotion result, the emotion recognition result value is 1, if the emotion recognition result is a negative emotion result, the emotion recognition result value is -1, and if the emotion recognition result is a neutral emotion result, the emotion recognition result value is 0; multiplying the emotion recognition result by the attenuation coefficient to obtain the current trust value; wherein the attenuation coefficient is e -λ(t-t0) , λ is a preset adjustment parameter with a value range of (0,1), and t-t0 is the comment time interval corresponding to the mutual comment relationship between users.

[0138] In one embodiment, when the processor 502 executes the step of obtaining the trust score value of each product by the trusted user to form a product recommendation row vector based on the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix, the processor 502 performs the following operations: obtaining a scoring matrix composed of the score row vectors corresponding to each trusted user in the trusted user cluster in the user-score matrix; obtaining a trust user row vector composed of the size of each trust value in the row vector corresponding to the target user in the trusted user cluster; multiplying the trust user row vector by the score matrix to obtain a product recommendation row vector.

[0139] In one embodiment, when the processor 502 executes the step of segmenting the comment content to obtain the segmentation result, the processor 502 performs the following operations: extracting candidate words from the comment content in order from left to right; querying the probability value corresponding to each candidate word in a pre-stored dictionary, and recording the left neighbor word of each candidate word; calculating and obtaining the cumulative probability of each candidate word, and obtaining the cumulative probability of each of the multiple left neighbor words corresponding to each candidate word, if there is a left neighbor word with a cumulative probability that is the maximum value among the cumulative probabilities of the multiple left neighbor words of each candidate word, the left neighbor word with the maximum value in the cumulative probability is used as the best left neighbor word corresponding to the candidate word; taking the end word of the comment content as the starting point, outputting the best left neighbor word corresponding to each candidate word from right to left in sequence to obtain the segmentation result.

[0140] In one embodiment, when the processor 502 executes the step of obtaining the scoring matrix composed of the scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix, it performs the following operations: obtaining the scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix, and arranging them in the order of the row numbers in the scoring row vectors corresponding to each trusted user in the user-scoring matrix to obtain the scoring matrix.

[0141] In one embodiment, when the processor 502 executes the step of obtaining a trusted user row vector composed of the sizes of trust values ​​of the trusted user cluster in the row vector corresponding to the target user, the processor 502 performs the following operations: obtaining trusted users whose trust values ​​are before a preset ranking threshold, and arranging the trusted users whose trust values ​​are before the ranking threshold in the order of column numbers in the row vector corresponding to the target user to obtain a trusted user row vector.

[0142] Those skilled in the art will understand that Fig. 9The embodiments of the computer device shown in the figure do not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structure and function of the memory and the processor are the same as those of the embodiment of the present invention. Fig. 9 The embodiments shown are consistent and will not be described again here.

[0143] It should be understood that in the embodiment of the present invention, the processor 502 may be a central processing unit (CPU), and the processor 502 may also be 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. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0144] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium may be a non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, wherein the computer program implements the following steps when executed by a processor: if the product review information at the current moment is compared with the product review information at the previous moment to determine whether there is a newly added mutual review relationship between users, the current trust value is obtained according to the review content corresponding to the newly added mutual review relationship between users; the user trust matrix stored at the previous moment is updated according to the current trust value to obtain the user trust matrix stored at the current moment; wherein, in the user trust matrix stored at the previous moment and the user trust matrix stored at the current moment, each value represents the trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located; the target user corresponding to the row vector selected from the user trust matrix stored at the current moment is obtained, and according to the size of each trust value in the row vector corresponding to the target user, the trust users whose trust values ​​are before a preset ranking threshold are obtained to form a trust user cluster; according to the score row vector corresponding to each trust user in the trust user cluster in the user-score matrix, the trust score value of each product of the trust user is obtained to form a product recommendation row vector; and a product recommendation list is obtained from the product recommendation row vector, and the product recommendation list is pushed to the receiving end corresponding to the target user.

[0145] In one embodiment, the method of obtaining the current trust value based on the comment content corresponding to the newly added mutual comment relationship between users includes: obtaining the comment content corresponding to the newly added mutual comment relationship between users, and segmenting the comment content to obtain a segmentation result; obtaining the word vector corresponding to each comment keyword in the segmentation result through the Word2Vec model, and obtaining the text vector from the word vector corresponding to each comment keyword in the segmentation result; using the text vector as the input of the pre-trained naive Bayes model to obtain the emotion recognition result corresponding to the comment content; wherein, if the emotion recognition result is a positive emotion result, the emotion recognition result value is 1, if the emotion recognition result is a negative emotion result, the emotion recognition result value is -1, and if the emotion recognition result is a neutral emotion result, the emotion recognition result value is 0; multiplying the emotion recognition result by the attenuation coefficient to obtain the current trust value; wherein the attenuation coefficient is e -λ(t-t0) , λ is a preset adjustment parameter with a value range of (0,1), and t-t0 is the comment time interval corresponding to the mutual comment relationship between users.

[0146] In one embodiment, obtaining the trust score value of each product by the trusted user to form a product recommendation row vector based on the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix includes: obtaining a score matrix composed of the score row vectors corresponding to each trusted user in the trusted user cluster in the user-score matrix; obtaining a trust user row vector composed of the size of each trust value in the row vector corresponding to the target user in the trusted user cluster; and multiplying the trust user row vector by the score matrix to obtain a product recommendation row vector.

[0147] In one embodiment, the word segmentation of the comment content to obtain the word segmentation result includes: extracting candidate words from the comment content in order from left to right; searching for the probability value corresponding to each candidate word in a pre-stored dictionary, and recording the left neighbor word of each candidate word; calculating and obtaining the cumulative probability of each candidate word, and obtaining the cumulative probability of each of the multiple left neighbor words corresponding to each candidate word, if there is a left neighbor word with a cumulative probability that is the maximum value among the cumulative probabilities of the multiple left neighbor words of each candidate word, the left neighbor word with the maximum value in the cumulative probability is used as the best left neighbor word corresponding to the candidate word; taking the end word of the comment content as the starting point, outputting the best left neighbor word corresponding to each candidate word in sequence from right to left to obtain the word segmentation result.

[0148] In one embodiment, obtaining a scoring matrix composed of scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix includes: obtaining scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix, and arranging them in order of row numbers in the scoring row vectors corresponding to each trusted user in the user-scoring matrix to obtain a scoring matrix.

[0149] In one embodiment, obtaining a trusted user row vector composed of the sizes of trust values ​​of the trusted user cluster in the row vector corresponding to the target user includes: obtaining trusted users whose trust values ​​are before a preset ranking threshold, arranging the trusted users whose trust values ​​are before the ranking threshold in the order of column numbers in the row vector corresponding to the target user, and obtaining the trusted user row vector.

[0150] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the above-described equipment, devices and units can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. Those of ordinary skill in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are executed 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 exceed the scope of the present invention.

[0151] In the several embodiments provided by the present invention, it should be understood that the disclosed equipment, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. Units with the same function may also be combined into one unit. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices or units, or may be an electrical, mechanical or other form of connection.

[0152] 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 embodiments of the present invention.

[0153] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0154] 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 storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), disk or optical disk and other media that can store program codes.

[0155] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A push method based on a trust network, characterized in that: include: If the product review information at the current moment is compared with the product review information at the previous moment to determine whether there is a new mutual review relationship between users, the current trust value is obtained according to the review content corresponding to the new mutual review relationship between users; The user trust matrix stored at the last moment is updated according to the current trust value to obtain the user trust matrix stored at the current moment; wherein each value in the user trust matrix stored at the last moment and the user trust matrix stored at the current moment represents the trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located; Obtaining a target user corresponding to a row vector selected from the user trust matrix stored at the current moment, and according to the values ​​of each trust value in the row vector corresponding to the target user, obtaining trust users whose trust values ​​are before a preset ranking threshold to form a trust user cluster; According to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix, obtaining the trust score value of each product by the trusted user to form a product recommendation row vector; and Obtain a product recommendation list from the product recommendation row vector, and push the product recommendation list to a receiving end corresponding to the target user; The obtaining of the current trust value according to the comment content corresponding to the newly added mutual comment relationship between users includes: Obtaining the comment content corresponding to the newly added mutual comment relationship between users, and segmenting the comment content to obtain a segmentation result; The word vector corresponding to each comment keyword in the word segmentation result is obtained through the Word2Vec model, and the text vector is obtained from the word vector corresponding to each comment keyword in the word segmentation result; The text vector is used as the input of the pre-trained naive Bayes model to obtain the sentiment recognition result corresponding to the comment content; wherein, if the sentiment recognition result is a positive sentiment result, the sentiment recognition result value is 1; if the sentiment recognition result is a negative sentiment result, the sentiment recognition result value is -1; if the sentiment recognition result is a neutral sentiment result, the sentiment recognition result value is 0; Multiply the emotion recognition result by the attenuation coefficient to get the current trust value; the attenuation coefficient is , is a preset adjustment parameter with a value range of (0,1), t-t0 is the comment time interval corresponding to the mutual comment relationship between users; The emotion recognition result is multiplied by the attenuation coefficient to obtain the current trust value, including: The current trust value of user i to user j is recorded as a ij ,but Among them, Sentiment ij Sentiment represents the emotion recognition result obtained by performing emotion recognition on the comment made by user i on user j. If the emotion recognition result is a positive emotion result, then Sentiment ij =1, if the emotion recognition result is a negative emotion result, Sentiment ij =-1, if the emotion recognition result is a neutral emotion result, Sentiment ij =0; The step of obtaining the trust score value of each product by the trusted user according to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix to form a product recommendation row vector includes: Obtaining a scoring matrix composed of the scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix; Obtaining a trusted user row vector composed of the magnitudes of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user; Multiplying the trusted user row vector by the rating matrix to obtain a product recommendation row vector; The user-rating matrix represents the user's rating of the product, the horizontal axis of the user-rating matrix is ​​the product, and the vertical axis is the user; The user trust matrix represents the trust value between users, and the horizontal axis and the vertical axis in the user trust matrix are both user lists.

2. The push method based on the trust network according to claim 1, characterized in that: The word segmentation of the comment content to obtain the word segmentation result includes: Extract candidate words from the review content in order from left to right; Query the probability value corresponding to each candidate word in the pre-stored dictionary, and record the left neighbor word of each candidate word; Calculate and obtain the cumulative probability of each candidate word, and obtain the cumulative probabilities of multiple left neighboring words corresponding to each candidate word. If there is a left neighboring word with the maximum cumulative probability among the multiple left neighboring words of each candidate word, the left neighboring word with the maximum cumulative probability is taken as the best left neighboring word corresponding to the candidate word. Taking the end word of the comment content as the starting point, output the best left neighbor word corresponding to each candidate word from right to left to obtain the word segmentation result.

3. The push method based on the trust network according to claim 1, characterized in that: The step of obtaining a scoring matrix composed of the scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix includes: The score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix is ​​obtained, and the score row vectors corresponding to each trusted user in the user-score matrix are arranged in the order of the row numbers to obtain a score matrix.

4. The push method based on a trust network according to claim 1, characterized in that: The step of obtaining a trusted user row vector composed of the magnitudes of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user includes: Trust users whose trust values ​​are before a preset ranking threshold are obtained, and the trust users whose trust values ​​are before the ranking threshold are arranged in the order of column numbers in the row vector corresponding to the target user to obtain a trust user row vector.

5. A push device based on a trust network, characterized in that: include: A current trust value acquisition unit, used to compare the product review information at the current moment with the product review information at the previous moment to determine whether there is a new mutual review relationship between users, and to acquire the current trust value according to the review content corresponding to the new mutual review relationship between users; A user trust matrix updating unit, configured to update the user trust matrix stored at the last moment according to the current trust value to obtain the user trust matrix stored at the current moment; wherein each value in the user trust matrix stored at the last moment and the user trust matrix stored at the current moment represents a trust value of the user corresponding to the row where the value is located to the user corresponding to the column where the value is located; A trusted user cluster acquisition unit, configured to acquire a target user corresponding to a row vector selected from the user trust matrix stored at the current moment, and acquire trusted users whose trust values ​​are before a preset ranking threshold according to the trust values ​​in the row vector corresponding to the target user to form a trusted user cluster; A recommendation row vector acquisition unit, configured to acquire the trust score value of each product by the trusted user according to the score row vector corresponding to each trusted user in the trusted user cluster in the user-score matrix to form a product recommendation row vector; A push unit, configured to obtain a product recommendation list from the product recommendation row vector, and push the product recommendation list to a receiving end corresponding to a target user; The current trust value acquisition unit includes: A word segmentation unit is used to obtain the comment content corresponding to the newly added mutual comment relationship between users, and segment the comment content to obtain a word segmentation result; A text vector acquisition unit is used to acquire a word vector corresponding to each comment keyword in the word segmentation result through a Word2Vec model, and acquire a text vector from the word vector corresponding to each comment keyword in the word segmentation result; The emotion recognition unit is used to use the text vector as the input of the pre-trained naive Bayes model to obtain the emotion recognition result corresponding to the comment content; wherein, if the emotion recognition result is a positive emotion result, the emotion recognition result value is 1, if the emotion recognition result is a negative emotion result, the emotion recognition result value is -1, and if the emotion recognition result is a neutral emotion result, the emotion recognition result value is 0; The trust value calculation unit is used to multiply the emotion recognition result by the attenuation coefficient to obtain the current trust value; the attenuation coefficient is , is a preset adjustment parameter with a value range of (0,1), t-t0 is the comment time interval corresponding to the mutual comment relationship between users; The emotion recognition result is multiplied by the attenuation coefficient to obtain the current trust value, including: The current trust value of user i to user j is recorded as a ij ,but Among them, Sentiment ij Sentiment represents the emotion recognition result obtained by performing emotion recognition on the comment made by user i on user j. If the emotion recognition result is a positive emotion result, then Sentiment ij =1, if the emotion recognition result is a negative emotion result, Sentiment ij =-1, if the emotion recognition result is a neutral emotion result, Sentiment ij =0; The recommended row vector acquisition unit is specifically used for: Obtaining a scoring matrix composed of the scoring row vectors corresponding to each trusted user in the trusted user cluster in the user-scoring matrix; Obtaining a trusted user row vector composed of the magnitudes of the trust values ​​of the trusted user cluster in the row vector corresponding to the target user; Multiplying the trusted user row vector by the rating matrix to obtain a product recommendation row vector; The user-rating matrix represents the user's rating of the product, the horizontal axis of the user-rating matrix is ​​the product, and the vertical axis is the user; The user trust matrix represents the trust value between users, and the horizontal axis and the vertical axis in the user trust matrix are both user lists.

6. A computer 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 push method based on the trust network as claimed in any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the push method based on a trust network according to any one of claims 1 to 4.