An online shopping false marketing information identification method based on emotional divergence

Abstract

The application belongs to the field of false information identification, and particularly relates to an online shopping false marketing information identification method based on emotional divergence, which comprises the following steps: obtaining user online shopping comment data and preprocessing the data to obtain preprocessed data; obtaining the subject emotional dependence factor of each comment text by adopting a unit emotional factor representation method; calculating the overall emotional dependence factor of each kind of commodity according to the subject emotional dependence factor, and adjusting and updating the overall emotional dependence factor of each kind of commodity by adopting a game theory; constructing a network relationship graph according to the overall emotional dependence factor, and obtaining the user divergence degree through the network relationship graph; obtaining the monthly average release amount and the monthly average coverage rate of user evaluation, and fusing the user divergence degree, the monthly average release amount and the monthly average coverage rate to obtain a user emotional diversity factor; and processing the user emotional diversity factor by adopting a sigmoid function to obtain an identification result.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of false information recognition, and particularly relates to an online shopping false marketing information recognition method based on emotional divergence. BACKGROUND

[0002] False marketing information mainly refers to false information misleading consumers in commercial marketing activities, which is widespread in social life. In order to improve the sales of their own goods, merchants promote their products in the form of virtual buyers. In the information age, the proliferation of online false marketing information has become an unavoidable phenomenon, so the recognition of false information has received high attention and become a hot topic in the academic field. It can be seen that the research on online false marketing information recognition is very important.

[0003] In recent years, many experts and scholars have done a lot of research in the field of false information recognition. According to different research emphases, these methods can be roughly divided into three categories: recognition based on information content, recognition based on user features, and recognition based on information transmission characteristics.

[0004] In the research process, we found an interesting phenomenon: false marketing information in online shopping platforms is often evaluation information issued by virtual buyers led by merchants. Through further research, we found that there is a significant difference in the emotional divergence between the evaluation information published by virtual buyers of merchants and actual buyers of customers. For daily consumer goods in life, the evaluation information published by actual buyers of customers is more obviously divergent in emotional tendency than that by virtual buyers of merchants. We combine the emotional divergence in user comments and the current status of false information recognition research, and find that there are still several challenges in false marketing information recognition based on emotional divergence:

[0005] 1. The concealment of user evaluation emotion. For some user evaluation information, the emotional types contained are rich, and there is no obvious emotional tendency judgment mark. Therefore, the overall emotional tendency contained cannot be accurately judged and classified. Therefore, how to accurately perceive and judge the emotion of the evaluation content is a very challenging task.

[0006] 2. Uncertainty of emotional divergence degree measurement. The emotional divergence degree between different evaluations published by the same user may be affected by various factors, and the emotional divergence degree is not accurate in quantization. Therefore, it is difficult to design a model to consider the influence of various factors to accurately quantify the emotional divergence degree between different evaluations.

[0007] 3. Evaluate the computational complexity of information. For shopping information in shopping platform, it has the characteristics of massive data, various types and high complexity, and often needs to consume a lot of time in the process of data processing. Therefore, how to improve the efficiency of false marketing information identification makes the identification more accurate still has the challenge. SUMMARY

[0008] To solve the above problems, the application provides an online shopping false marketing information identification method based on emotional divergence, comprising the following steps:

[0009] S1. Obtain user online shopping comment data and preprocess to obtain preprocessed data; the preprocessed data includes a user evaluation set for different goods, and each evaluation set for a good includes multiple comment texts, and all the comment texts are arranged in chronological order according to the publishing time;

[0010] S2. Based on the preprocessed data, the subject emotional dependence factor of each comment text is obtained by using the unit emotional factor representation method;

[0011] S3. The overall emotional dependence factor of each kind of goods is calculated according to the subject emotional dependence factor, and the overall emotional dependence factor of each kind of goods is adjusted and updated by using the game theory;

[0012] S4. The network relationship graph is constructed according to the overall emotional dependence factor, and the user divergence degree is obtained through the network relationship graph;

[0013] S5. Obtain the monthly average publishing amount and the monthly average coverage rate of user evaluation, and fuse the user divergence degree, the monthly average publishing amount and the monthly average coverage rate to obtain the user emotional diversity factor;

[0014] S6. The user emotional diversity factor is processed by using the sigmoid function to obtain the identification result.

[0015] The beneficial effects of the application are as follows:

[0016] The application is aimed at the problem that user emotion has concealment and cannot be accurately perceived. From the word angle, the unit emotional factor representation method is proposed to obtain the subject emotional dependence factor of each comment text. In this method, each evaluation text is divided into small units, the unit emotional intensity of each small unit is obtained according to the word emotional intensity of each emotional word, the unit emotional dependence factor of each small unit is obtained by analyzing the mutual influence between small units, and the subject emotional dependence factor of the evaluation unit subject is calculated according to the unit emotional dependence factor, so that the emotional tendency in the evaluation content can be better mined, and the classification can be more accurately determined.

[0017] Since the emotional divergence degree between different evaluations published by users can be influenced by multiple factors, and the emotional divergence degree is not accurate in quantification, the emotional divergence degree is measured by constructing a network relationship graph.

[0018] The user is divided into a virtual buyer of a merchant and an actual buyer of a customer, and the false marketing information is identified according to the emotional divergence degree in the shopping evaluation, so that the accuracy of the false marketing information identification is improved. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A framework diagram of a false marketing information identification method based on user evaluation emotional divergence is provided in the present application.

[0020] Figure 2 A user evaluation difference moment evaluation information emotional game diagram is provided in the present application.

[0021] Figure 3 A user emotional divergence degree quantification diagram for different goods is provided in the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0023] The present application provides an online shopping false marketing information identification method based on emotional divergence, as shown in Figure 1 The method comprises the following steps:

[0024] S1. Obtain user online shopping comment data and perform preprocessing to obtain preprocessed data; the preprocessed data comprises an evaluation set of different goods by the user, and each evaluation set of goods comprises multiple comment texts, and all the comment texts are arranged in chronological order according to the publishing time.

[0025] S2. Based on the preprocessed data, the subject emotional dependence factor of each comment text is obtained by using a unit emotional factor representation method.

[0026] Preferably, the present application provides a unit emotional factor representation method, referred to as UEFR method. The method uses text mining technology to split the evaluation text into multiple small units, then uses emotional dependence factors to represent the emotional tendency of the small units, and finally integrates the emotional dependence factors of each small unit to obtain the overall emotional tendency of the evaluation text, and reveals the possible implicit emotional tendency in the evaluation sentence.

[0027] In step S2, the subject sentiment dependency factor of each comment text is obtained through the UEFR method.

[0028] Specifically, a review text is taken as an evaluation unit body, and step S2 processes an evaluation unit body, including:

[0029] S21. Using sentence boundary detection technology to divide the evaluation unit into multiple small units;

[0030] S22. For each small unit, the word sentiment intensity of each sentiment word in the small unit is calculated using the BosonNLP sentiment dictionary, and the unit sentiment intensity of the small unit is obtained by adding up all the word sentiment intensities, which is expressed as

[0031]

[0032] Among them, Eins(w i ) represents the i-th sentiment word w in the small unit i The intensity of the associated emotion, Nunw(w i ) represents the i-th sentiment word w in the small unit i The number of associated negative words, SentiIns(w i ) represents the sentiment value of the i-th sentiment word itself; the present invention divides the intensity of the sentiment expressed by the adverbs, QualDeg(w i ) represents the i-th sentiment word w in the small unit i the degree level of the associated adverbs;

[0033] The unit sentiment intensity of a small unit is defined as

[0034]

[0035] Among them, Uein(v i ) represents a small unit v i Unit emotional intensity, Nuw(v i ) represents a small unit v i The number of sentiment words.

[0036] S22. Define the sentiment intensity of the conjunction between each two adjacent small units as follows

[0037]

[0038] Then the mutual influence between two adjacent small units is quantified as

[0039] Emi(vi,vj)=Emco(vi,vj)×log2(Uein(vi)×Uein(vj))

[0040] wherein co(v i ,v j ) represents a conjunction between the small unit v i and the small unit v j ; co(v i ,v j ) ∈ Empr(c i ) represents that the conjunction between the small unit v i and the small unit v j belongs to an emotion progressive conjunction, co(v i ,v j ) ∈ Empa(c i ) represents that the conjunction between the small unit v i and the small unit v j belongs to an emotion turning conjunction, co(v i ,v j ) ∈ Emtr(c i ) represents that the conjunction between the small unit v i and the small unit v j belongs to an emotion substitution conjunction; Emco(v i ,v j ) represents an emotion intensity of the conjunction between the small unit v i and the small unit v j , Uein(v i ) represents a unit emotion intensity of the small unit v i , and Emi(v i ,v j ) represents a mutual influence degree between the small unit v i and the small unit v j .

[0041] S23. A unit emotion dependence factor of each small unit is obtained according to the unit emotion intensity and the mutual influence degree, as shown below:

[0042]

[0043] wherein Nuc(v i ) represents a number of evaluations of the user under the product corresponding to the small unit v i , and Nutl(v i ) represents a total number of evaluations of the user.

[0044] A subject emotion dependence factor of the evaluation unit subject is calculated according to the unit emotion dependence factor.

[0045] Specifically, in order to express the emotional tendency of the evaluation unit subject, first, the emotional tendency relationship between the evaluation unit subject and the small unit is expressed as follows:

[0046]

[0047] wherein, Esd(e n ) denotes the subject sentiment dependence factor of the nth evaluation text e n published by the user, Esd(v i ) denotes the unit sentiment dependence factor of the unit v n in the evaluation text e i , and Nu(e n ) denotes the number of units in the evaluation text e n .

[0048] S3. Calculate the overall sentiment dependence factor of each kind of commodity according to the subject sentiment dependence factor, and adjust and update the overall sentiment dependence factor of each kind of commodity by using the game theory.

[0049] Specifically, the overall sentiment dependence factor of each kind of commodity is calculated according to the subject sentiment dependence factor, and is expressed as

[0050]

[0051] wherein, denotes the overall sentiment dependence factor of the commodity m i , Esd(e n ) denotes the subject sentiment dependence factor of the nth evaluation text e n , N denotes the number of evaluation texts corresponding to the commodity m i , t1 denotes the time when the user publishes the first evaluation text of the commodity m i , t n denotes the time when the user publishes the nth evaluation text of the commodity m i , and w denotes the half-life time. The half-life function is introduced to simulate the change of the importance of the evaluation of the same kind of commodity over time.

[0052] Specifically, as shown in Figure 2 , the process of adjusting and updating the overall sentiment dependence factor of any kind of commodity by using the game theory in step S3 includes:

[0053] S41. Define the game strategy including the user evaluation support and the user evaluation opposition, and for each evaluation text corresponding to the commodity, if the subject sentiment dependence factor thereof is positive, it is considered that the evaluation text belongs to the user evaluation support, otherwise, it is considered that the evaluation text belongs to the user evaluation opposition;

[0054] S42. Define the revenue function of the user evaluation support and the user evaluation opposition, and is expressed as

[0055] Rev sup (e n)=Pa×Esd(e n ), Esd(e n )>0

[0056] Rev obj (e n )=Pb×Esd(e n ), Esd(e n )<0

[0057] Among them, Esd(e n ) represents the nth evaluation text e n The subject emotion dependency factor, Pa, Pb respectively represent the proportion of user evaluation support and user evaluation opposition in user product evaluation, Pa + Pb = 1; Rev sup (e n ) indicates the revenue when the nth review text belongs to the user review support case, Rev obj (e n ) indicates the revenue when the nth review text belongs to the user's objection review;

[0058] S43. Calculate the emotional influence of each comment text based on the profit function, where

[0059] If the comment text is e n If it belongs to the user evaluation support situation, its emotional influence is expressed as Mut sup (e n ), the calculation formula is

[0060]

[0061] If the comment text is e n If the user's evaluation is against the situation, its emotional influence is expressed as Mut obj (e n ), the calculation formula is

[0062]

[0063] S44. Update the overall emotional dependency factor based on the emotional influence, expressed as

[0064]

[0065] Among them, x, y are adjustment coefficients, N sup Indicates the number of evaluation texts that belong to user evaluation support, N obj Indicates the number of evaluation texts that are user-reviewed objections, N sup +N obj =N, N represents product m i The number of corresponding evaluation texts; the overall sentiment dependence factor after adjustment, the overall sentiment dependence factor before adjustment.

[0066] We represent the user's sentiment tendency for a single product according to the overall sentiment dependence factor after adjustment by the product evaluation set, and provide a basis for quantifying the user's sentiment divergence for multiple products.

[0067] S4. Construct a network relationship graph according to the overall sentiment dependence factor, and obtain the user divergence through the network relationship graph.

[0068] Preferably, the present application provides a product evaluation graph quantification model EGR. This model quantifies the correlation between multiple evaluation sentiments based on game theory, and quantifies the divergence degree between multiple product evaluation sentiment tendencies based on graph closeness centrality theory, as shown in Figure 3 .

[0069] Specifically, step S4 specifically includes:

[0070] S41. In order to reflect the mutual relationship between the overall sentiment dependence factors of two different products, the difference degree between different products is defined as

[0071]

[0072] wherein Dis(m i ,m j ) represents the difference degree between product m i and product m j , and represents the overall sentiment dependence factor of product m i .

[0073] S42. Each product is regarded as a node, and the edge weight value between each two nodes is assigned as the difference degree between two products, to obtain a network relationship graph. Based on this, we can quantify the user sentiment divergence degree between different products by measuring the tightness of the network relationship graph.

[0074] S43. According to the closeness centrality theory of the graph, the closeness centrality of each node is calculated, and the average closeness centrality is calculated through the closeness centrality of all nodes.

[0075] Specifically, the calculation formula of the closeness centrality of each node is

[0076]

[0077] wherein Clc(i) represents the closeness centrality of node i; M represents the number of nodes, i.e. the number of products; and Hvc(m i ) represents the overall sentiment dependence factor of product m iThe heat value of the product, which reflects the importance of the product in the user's heart in an objective way, based on which we can obtain the average closeness centrality Clcto of all nodes in the figure as a whole:

[0078]

[0079] Through the above formula, we can evaluate the closeness of the network relationship diagram as a whole, and then obtain the divergence degree of the user's emotional tendency to different products, so we take Clcto as an important index for quantifying divergence.

[0080] S44. Calculate the user divergence degree according to the average closeness centrality, which is represented as

[0081]

[0082] Through the above calculation, we can obtain the emotional divergence degree of the user, which is an important basis for dividing the user into a virtual buyer of the merchant and an actual buyer of the user.

[0083] S5. Obtain the monthly average publishing quantity and the monthly average coverage rate of the user evaluation, and fuse the user divergence degree, the monthly average publishing quantity and the monthly average coverage rate to obtain a user emotional diversity factor.

[0084] S6. Process the user emotional diversity factor by using a sigmoid function to obtain a recognition result.

[0085] Preferably, the present application proposes a user evaluation feature association recognition model BUAI. The model introduces Yager combination rule and combines the evaluation frequency of the user and the time distribution of the evaluation to accurately classify the user, thereby reducing the calculation complexity of the evaluation information. In addition, the model also proposes a user emotional diversity factor to divide the user into a virtual buyer of the merchant and an actual buyer of the customer to distinguish the false marketing information.

[0086] Obtain the monthly average publishing quantity Anp(u i ) of the user evaluation. At the same time, obtain the monthly average coverage rate Mcr(u i ) of the user evaluation by the following method, and the specific calculation is as follows:

[0087]

[0088] Dir(u i ) represents the average number of days per month (i.e. the user has an average of how many days to publish the evaluation in a month) involved in the user evaluation, and Dam represents the average number of days per month;

[0089] Combine the above multiple features together for comprehensive evaluation according to the Yager combination rule, and introduce a user emotional diversity factor at the same time, and the specific calculation is as follows:

[0090] Usd(u i ) = w1 x Div(u i ) + w2 x Anp(u i ) + w3 x Mcr(u i )

[0091] a1, a2, a3 are credibility weights, and a1 + a2 + a3 = 1 ∩ a1, a2, a3 ∈ (0, 1).

[0092] The sigmoid function is combined to predict the probability value of the obtained combined feature model. The BUAI model can be represented by the formula:

[0093] P = sigmoid [Usd(u i )]

[0094] wherein, is an activation function. We denote the output of the model as P(n, fuser), which is specifically defined as follows:

[0095]

[0096] wherein, θ represents a recognition result, if θ = 1, it indicates that the current user is judged as a virtual buyer of a merchant, that is, the information published by the user is false marketing information; if θ = 0, it indicates that the current user is judged as an actual buyer of a customer, that is, the information published by the user is real evaluation information; P(n, f|user) represents the output of the sigmoid function, max[P(n, f|user)] represents the maximum value of the output of the sigmoid function, and min[P(n, f|user)] represents the minimum value of the output of the sigmoid function.

[0097] In the present application, unless otherwise explicitly specified and limited, the terms “mounting”, “setting”, “connecting”, “fixing”, “rotating” and the like should be understood in a broad sense, for example, can be fixed connection, or detachable connection, or integrated; can be mechanical connection, or electrical connection; can be directly connected, or indirectly connected through an intermediate medium; can be the internal communication of two elements or the interaction relationship between two elements, unless otherwise explicitly limited, the above-mentioned terms in the present application can be understood according to the specific meaning of the above-mentioned terms in the present application by the person skilled in the art according to the specific situation.

[0098] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for identifying false marketing information in online shopping based on emotional divergence, characterized in that, The method comprises the following steps: S1. Obtain shopping review data on a user line and pre-process the data to obtain pre-processed data; The pre-processed data comprises a set of evaluations of different commodities by a user, and each set of evaluations of a commodity comprises a plurality of comment texts, wherein all the comment texts are arranged in chronological order of publishing time; S2. Based on the pre-processed data, a unit sentiment factor representation method is used to obtain a subject sentiment dependence factor of each comment text; A comment text is taken as a subject of an evaluation unit, and step S2 processes the subject of the evaluation unit, comprising: S21. The subject of the evaluation unit is divided to obtain a plurality of small units; S22. For each small unit, the word sentiment intensity of each sentiment word in the small unit is calculated, and all the word sentiment intensities are added to obtain the unit sentiment intensity of the small unit; S22. The mutual influence degree between each two adjacent small units is calculated, which is represented as Emi(v i ,v j ) = Emco(v i ,v j ) x log2(Uein(v i ) x Uein(v j )) where co(v i ,v j ) represents the conjunction between the small unit v i and the small unit v j ; co(v i ,v j ) ∈ Empr(c i ) represents that the conjunction between the small unit v i and the small unit v j belongs to the emotional progressive conjunction, co(v i ,v j ) ∈ Empa(c i ) represents that the conjunction between the small unit v i and the small unit v j belongs to the emotional turning conjunction, co(v i ,v j ) ∈ Emtr(c i ) represents that the conjunction between the small unit v i and the small unit v j belongs to the emotional substitution conjunction; Emco(v i ,v j ) represents the emotional intensity of the conjunction between the small unit v i and the small unit v j , Uein(v i ) represents the unit emotional intensity of the small unit v i , and Emi(v i ,v j ) represents the mutual influence degree between the small unit v i and the small unit v j . S23. The unit sentiment dependence factor of each small unit is obtained according to the unit sentiment intensity and the mutual influence degree, and the subject sentiment dependence factor of the evaluation unit is calculated according to the unit sentiment dependence factor; S3. The overall sentiment dependence factor of each kind of commodity is calculated according to the subject sentiment dependence factor, and the overall sentiment dependence factor of each kind of commodity is adjusted and updated using game theory; S4. A network relationship diagram is constructed according to the overall sentiment dependence factor, and a user divergence degree is obtained through the network relationship diagram; S5. The monthly average publishing amount and the monthly average coverage rate of user evaluations are obtained, and a user sentiment diversity factor is obtained by fusing the user divergence degree, the monthly average publishing amount and the monthly average coverage rate; S6. The user sentiment diversity factor is processed using a sigmoid function to obtain a recognition result. 2.The method of claim 1, wherein the method further comprises: determining a sentiment of the product review; and determining whether the product review is a false marketing information based on the determined sentiment. The overall sentiment dependence factor of each kind of commodity is calculated according to the subject sentiment dependence factor, which is represented as wherein, Esd(m, e i ) represents the overall sentiment dependence factor of the product m n , Esd(e n ) represents the subject sentiment dependence factor of the nth evaluation text e i , N represents the number of corresponding evaluation texts of the product m i , t1 represents the time when the user posts the first evaluation text of the product m n , and tn represents the time when the user posts the nth evaluation text of the product m i , wherein, t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m , t represents the time when the user posts the nth evaluation text of the product m 3.The method of claim 1, wherein the method further comprises: determining a sentiment of the product review; and determining whether the product review is a false marketing information based on the determined sentiment. The process of step S3 for adjusting and updating the overall sentiment dependence factor of any kind of commodity using game theory comprises: S31. Two game strategies including user evaluation support and user evaluation opposition are defined, and for each evaluation text corresponding to the commodity, if the subject sentiment dependence factor thereof is positive, the evaluation text is considered to belong to user evaluation support, otherwise, the evaluation text is considered to belong to user evaluation opposition; S32. The revenue functions of the two situations of user evaluation support and user evaluation opposition are defined, which are represented as Rev sup (e n )=Pa×Esd(e n ), Esd(e n )>0 Rev obj (e n )=Pb×Esd(e n ), Esd(e n ) < 0 Wherein, Esd(e n ) represents the nth evaluation text e n subject emotion dependence factor, Pa, Pb represents the proportion of user evaluation support and user evaluation opposition in user commodity evaluation respectively, Pa+Pb=1; Rev sup (e n ) represents the nth evaluation text belongs to the user evaluation support case income, Rev obj (e n ) denotes the n-th evaluation text belongs to the user's evaluation of the benefits of the case against; S33. The sentiment influence of each comment text is calculated according to the revenue function, wherein If the review text e n belongs to the user evaluation support case, its sentiment influence is represented as Mut sup (e n ), and the calculation formula is If the review text e n belongs to the user evaluation opposition case, its sentiment influence is represented as Mut obj (e n ), and the calculation formula is S34. The overall sentiment dependence factor is adjusted and updated according to the sentiment influence, which is represented as Among them, x, y are adjustment coefficients, N sup Indicates the number of evaluation texts that belong to user evaluation support, N obj Indicates the number of evaluation texts that are user-reviewed objections, N sup +N obj =N, N represents product m i The number of corresponding evaluation texts; represents the overall emotional dependence factor after adjustment, represents the overall emotional dependence factor before adjustment.

4. The method for identifying false online shopping marketing information based on sentiment divergence according to claim 1, characterized in that: Step S4 specifically comprises: S41. The difference degree between different commodities is calculated, which is represented as wherein Dis(m i ,m j ) represents the difference between the product m i and the product m j , represents the overall sentiment dependence factor of the product m i ; S42. Each kind of commodity is regarded as a node, and the edge weight value between each two nodes is assigned as the difference degree between the two commodities to obtain a network relationship diagram; S43. The proximity centrality of each node is calculated according to the proximity centrality theory of the graph, and the average proximity centrality is calculated through the proximity centralities of all the nodes; S44. The user divergence degree is calculated according to the average proximity centrality. 5.The method of claim 4, wherein the method further comprises: The calculation formula of step S43 is Wherein, Clc(i) represents the closeness centrality of node i; M represents the number of nodes, i.e. the number of commodities; Clcto represents the average closeness centrality, Hvc(m i ) represents the heat value of commodity m i . 6.The method of claim 1, wherein the method further comprises: determining a sentiment of the product review; and determining whether the product review is a fake marketing information based on the sentiment of the product review. The user sentiment diversity factor is represented as Usd(u i ) = a1 x Div(u i ) + a2 x Anp(u i ) + a3 x Mcr(u i ) wherein Usd(u i ) represents the user sentiment diversity factor, Div(u i ) represents the user divergence, Anp(u i ) represents the average monthly release quantity of user reviews, Mcr(u i ) represents the average monthly coverage rate of user reviews; Clcto represents the average closeness centrality, Dir(u i ) represents the average monthly days involved in user reviews, Dam represents the average monthly days; a1, a2, a3 are credibility weights, and a1+a2+a3=1∩a1,a2,a3∈(0,1). 7.The method of claim 1, wherein the method further comprises: determining a sentiment of the product review; and determining whether the product review is a fake marketing information based on the sentiment of the product review. In step S6, the recognition result is represented as Wherein, theta indicates the identification result, if theta=1, it indicates that the current user is judged as a virtual buyer of the merchant, if theta=0, it indicates that the current user is judged as an actual buyer of the customer; P(n,f|user) indicates the sigmoid function output, max[P(n,f|user)] indicates the maximum value of the sigmoid function output, and min[P(n,f|user)] indicates the minimum value of the sigmoid function output.

Images