CN104915842A - Electronic commerce transaction monitoring method based on internet transaction data - Google Patents
Electronic commerce transaction monitoring method based on internet transaction data Download PDFInfo
- Publication number
- CN104915842A CN104915842A CN201510306742.1A CN201510306742A CN104915842A CN 104915842 A CN104915842 A CN 104915842A CN 201510306742 A CN201510306742 A CN 201510306742A CN 104915842 A CN104915842 A CN 104915842A
- Authority
- CN
- China
- Prior art keywords
- transaction
- monitoring
- association
- index
- aggregation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The invention provides an electronic commerce transaction monitoring method based on internet transaction data, which comprises the steps of: carrying out reputation parameter analysis on transaction records, and establishing a reputation information set; screening the reputation information set to obtain a problematic transaction set and extract an index aggregation based on reputation parameters, and establishing assembly correlation based on the reputation parameters; establishing a monitoring model and acquiring a monitoring aggregation according to the index aggregation based on the reputation parameters and a result aggregation based on the assembly correlation, and taking the monitoring aggregation as reference for monitoring and pre-warning electronic commerce transactions; and tracking new transaction records at regular intervals or when in need, extracting new assembly correlation results, and adding the new assembly correlation results into the monitoring aggregation, so as to improve prediction accuracy of the monitoring model at regular intervals. The electronic commerce transaction monitoring method aims at features of internet transactions, carries out problem identification and transaction early warning based on a neural network algorithm, reduces false alarm rate by realizing self-learning, self-adaption and self-defending performance of an active intelligent defense model, and improves accuracy and coverage of the internet transaction monitoring.
Description
Technical field
The present invention relates to communication technical field, particularly a kind of e-commerce transaction monitoring method based on internet business data.
Background technology
Along with the fast development of internet business, comprise mobile electric business, various new marketing model such as microblogging electricity business, micro-shop etc. continues to bring out, social concerns degree is more and more higher, has higher requirement also to the supervision in market., also there is the insoluble problem in some markets self in the development of internet business, the credible trading environment of shortage ecommerce is Important Problems wherein, becomes the Main Bottleneck affecting e-commerce development.Specifically, internet business transaction supervision Problems existing and difficulty mainly comprise: 1. main body difficulty is determined; 2. quoting ability is weak; 3. internet business has the feature such as cross-region and disguise; 4. current appraisement system is lack of standardization.This all causes being difficult to carry out of monitoring the abnormal behaviour of internet business.
The domestic and international abnormal behaviour for the internet business research of excavating at present is also few, is substantially also the research concentrating on mechanism and preventive measure and legal means, and what truly may be used for that intelligent algorithm or digging technology realization give warning in advance is less.Correlation technique is specific as follows:
1) by the betting model of product quality, the inherent mechanism of specification improvement is carried out to reduce network fraud behavior to relating to the changeability of product content, differentiation, the not easily identity of network trading identity and the network fraud of market participator to contents such as online product utility evaluations.
2) comprise for the strategy of the transaction data analyzing and processing of C2C transaction platform proof analysis is carried out to the transaction data of transaction platform, the aspect such as concept, Crack cause of fraud is sorted out, so induction and conclusion electricity business conclude the business transaction platform take precautions against and process swindle principal element and type to carry out e-commerce transaction monitoring.
3) based on the outlier algorithm of hypergraph model, outlier is monitored by the support degree of membership and scale deviation that calculate each outlier, by effectively finding that the outlier in high latitude spatial data just can process numerical attribute and category attribute, and then analyzing and processing is carried out to transaction data.
Above research is substantially all based on mechanism policy type, for the application scenarios of internet business, the generation of problem transaction necessarily type afterwards, after being all generally the problematic transaction of appearance, produced by litigant and lose and take corresponding treatment measures again after complaining, therefore can not meet the guarantee to transaction security early warning.
In the current trust for internet business and prestige algorithm, warning index only considers the relevant index of prestige, does not consider the relevance between transaction agent, and the maximum feature of problem transaction swindling main body is just that target is extensive.Therefore, not only to carry out discovery to problem trading activity to excavate, more will by classifying and predicting the feature of the interrelated relation between user of finding and problem trading activity, just can predict contingent problem trading activity and carry out early warning and then reduce the risk of internet business.
Summary of the invention
The object of the present invention is to provide a kind of e-commerce transaction monitoring method based on internet business data, to solve cannot predict contingent problem trading activity exactly and carrying out the problem of early warning existing for the existing trust for internet business and prestige algorithm.
For achieving the above object, the invention provides a kind of e-commerce transaction monitoring method based on internet business data, comprise the steps:
S1: obtain transaction record, and carry out prestige Parameter analysis according to transaction record, and set up reputation information collection;
S2: screening is carried out to reputation information collection and obtains problem transaction set, sample analysis is carried out to problem transaction set, extract the index set based on prestige parameter, and set up the aggregation association based on prestige parameter, obtain the results set of aggregation association;
S3: set up monitoring model based on the index set of prestige parameter and the results set of aggregation association according to described and obtain monitoring set, and this monitoring set is monitored and early warning e-commerce transaction as a reference, be specially: this monitoring set is met to the index of such as current e-commerce transaction, then judge that this e-commerce transaction is problem transaction;
Wherein, described aggregation association is specially and current transaction is related to information and the problem index set related to of concluding the business and associate.
Preferably, described prestige Parameter analysis is divided into the analysis based on B2C or C2C and the analysis based on B2B; Wherein, the analysis based on B2C or C2C comprises: analyzing influence executes the factor of letter side to the trust of trading object, and the factor that analyzing influence internet business is trusted; Analysis based on B2B comprises: the factor of the trust between analyzing influence tissue.
Preferably, impact is executed the factor of letter side to the trust of trading object and is comprised: to the trust of the other side, based on the trust controlled, for the expectation of potential income and himself is for the attitude of risk;
The factor affecting internet business trust comprises: technical factor, environmental factor, commercial factors and individual factor;
The factor of the trust between impact tissue comprises: appreciable supervision, appreciable certification, appreciable legal restraint, appreciable feedback and appreciable Writing Standards.
Preferably, what described index set comprised in registration behavior, operation behavior, popularization behavior and complained behavior is one or more.
Preferably, described aggregation association comprises IP association, essential information associates and the industry number of commodity association, and wherein, described IP association comprises direct IP association and associates with IP is advanced; The association of described essential information is specially the information that mailbox in the log-on message of concluding the business and relating to, responsible official, telephone number, shop network address or other log-on messages and confirmed problem conclude the business and carries out direct correlation; The industry number of described commodity association is the degree of association index of commodity and industry.
Preferably, described aggregation association also comprises commodity association, described commodity association be, registration essential information identical in IP information identical the association of product.
Preferably, also comprise after described step S3:
S4: regularly or when needed follow the trail of new transaction record and extract new aggregation association result, and new aggregation association result is added described monitoring set.
Preferably, described step S4 specifically comprises:
S41: follow the trail of new transaction record and set up trading activity information set in conjunction with described reputation information collection;
S42: by sorting algorithm to described trading activity information set carry out first time prediction, obtain suspicion problem transaction first time suspicion class information and predict the outcome collection A;
S43: filter the essential information in new transaction record and analyze, obtains new transaction essential information collection, and new transaction essential information collection is collected A carry out associating computing with described predicting the outcome, and obtains second time suspicion class information and the collection B that predicts the outcome;
S44: by setting threshold value will predict the outcome collection B be categorized as excessive risk problem transaction group and low-risk problem conclude the business group, excessive risk problem transaction group associates with the aggregation association result in step S3 again, obtain new aggregation association result, and add described monitoring set, to upgrade this monitoring set.
Preferably, neural network algorithm is adopted to set up described monitoring model, be specially: using the described index set based on prestige parameter as Input Monitor Connector index set X, (0, 1) random number between is as the initial weight set W of monitoring, the warning index Y of the trading activity exported is as monitoring result, wherein, the results set of aggregation association constantly adjusts the error of output and actual value in the process of training sample data, and feed back to by output layer the network that input layer carries out the multilayer feed-forward of backpropagation, until the learning error of algorithm no longer obviously reduces.
Preferably, the nominal error function setting each weights in described initial weight set W is e, and given computational accuracy ε and maximum study number of times M, Input Monitor Connector index set X comprise described based on the arbitrary one or more index in the index set of prestige parameter; When algorithm performs, when error satisfies condition or reached maximum study number of times, export the warning index of trading activity.
E-commerce transaction monitoring method based on internet business data provided by the invention is passed through: first, carry out prestige Parameter analysis according to transaction record, set up reputation information collection; Secondly, screening is carried out to reputation information collection and obtains problem transaction set and sample analysis is carried out to it, extract the index set based on prestige parameter, and set up the aggregation association based on prestige parameter, obtain the results set of aggregation association; Again according to setting up monitoring model based on the index set of prestige parameter and the results set of aggregation association and obtaining monitoring set, and this monitoring set is monitored and early warning e-commerce transaction as a reference.Finally, regularly or when needed follow the trail of new transaction record and extract new aggregation association result, and new aggregation association result being added monitoring set, regularly to provide the precision of prediction of monitoring model.The method is for internet business feature, based on neural network algorithm identification problem transaction early warning, by training the problem transaction data obtained thus realizing self study, the self-adaptation of initiatively intelligence defence model and reduce rate of false alarm from defence, contingent problem trading activity can be predicted exactly and carry out early warning, improve accuracy and the coverage of internet business monitoring.
The present invention adopts the artificial intelligence process method of neural network to carry out analyzing and processing for the data message of internet business process, realize analyzing the ecommerce of internet business, monitoring and early warning by long-term intelligent self study, self-adaptation, realize utilizing infotech and artificial intelligence to realize assessing the transaction risk degree of e-commerce transaction, and then build internet business ecommerce credible transaction guarantee service platform, promote Internet business subjective basis information disclosure, merchandise news is open, electronic mark is open, credit information is open.
Accompanying drawing explanation
Fig. 1 is the flow chart of steps of the inventive method;
Fig. 2 is B2C and C2C trust model structural representation in the preferred embodiment of the present invention;
Fig. 3 is B2B research model structural representation in the preferred embodiment of the present invention;
The neural network algorithm process flow diagram of Fig. 4 for adopting in the preferred embodiment of the present invention;
Fig. 5 is the optimization monitoring model process schematic diagram of Behavior-based control in the preferred embodiment of the present invention and trust.
Embodiment
For better the present invention being described, hereby with a preferred embodiment, and accompanying drawing is coordinated to elaborate to method provided by the invention, specific as follows:
As shown in Figure 1, a kind of e-commerce transaction monitoring method based on internet business data provided by the invention, comprises the steps:
S1: obtain transaction record, and carry out prestige Parameter analysis according to transaction record, and set up reputation information collection.
Wherein, prestige Parameter analysis is divided into the analysis based on B2C or C2C and the analysis based on B2B.According to the consumption habit in B2C and C2C transaction and transaction characteristics, set up the research model of B2C and C2C transaction, as shown in Figure 2, the analysis based on B2C or C2C comprises: analyzing influence executes the factor of letter side to the trust of trading object, and the factor that analyzing influence internet business is trusted.Can draw according to trust model, impact is executed the factor of letter side to the trust of trading object and is mainly comprised: to the trust of the other side, based on the trust controlled, for the expectation of potential income and himself is for the attitude of risk.
And the factor affecting internet business trust comprises: technical factor, environmental factor, commercial factors and individual factor; Wherein:
Technical factor: comprise the construction technique of electric business's platform, the stability of network security technology and other authentication techniques and Third-party payment, reliability and security, and to refresh corresponding time, both artistic and practical, user-interaction experience etc. all kinds of to Web Hosting with run relevant technology type factor in website;
Environmental factor: electric business's platform and the confidence level of third party intermediary, the integrity degree of authentication service and standardization, and the integrality of relevant laws and regulations and system;
Commercial factors: comprise the popularity of electric business's platform, product enterprise provides the quality of products & services, the smooth degree of trade degree after sale, logistics and treat the attitude of client, the interdynamic factor etc. with client;
Individual factor: comprise the consumer individual degree of concern to privacy, the degree of belief to other consumer evaluations and dependency degree, and itself is for the ability to bear of risk, and the factor such as the impression of online transaction and experience.
Be different from the prestige analysis of B2C and C2C, the prestige parameter main manifestations of B2B type ecommerce be based on the trust of prestige, institution based trust and based on the trust of approval tissue between trust; In the present embodiment, trusting relationship when guaranteeing that B2B concludes the business based on the mechanism of system between enterprise is specially: the analysis based on B2B comprises the factor of the trust between analyzing influence tissue.As shown in Figure 3, the factor of the trust between impact tissue comprises: appreciable supervision, appreciable certification, appreciable legal restraint, appreciable feedback and appreciable Writing Standards.
S2: screening is carried out to reputation information collection and obtains problem transaction set, sample analysis is carried out to problem transaction set, extract the index set based on prestige parameter, and set up the aggregation association based on prestige parameter, obtain the results set of aggregation association.
Wherein, be choose from 4 main aspects to the index selected by sample training set, namely index set comprises the one or more content in registration behavior, operation behavior, popularization behavior and complained behavior.In the present embodiment, by the sample analysis of concluding the business to particular problem, extract the monitoring index set of problem transaction.The index of monitoring is see following form one:
Table one index system structure
Aggregation association is specially and current transaction is related to information and the problem index set related to of concluding the business and associate.And aggregation association comprises IP association, essential information associates and the industry number of commodity association., registration essential information identical in IP information is identical, aggregation association also comprises the operation of commodity association.
Particularly, IP association: comprise direct IP association and associate with IP is advanced.Direct correlation is obtained by the server log record of the electric business's platform Web of access.Disguise by IP mapping techniques is accessed, then by Trimmed sums IP mapping is carried out on the advanced basis being associated in the Apache daily record of collection of IP, thus the regional information of the suspicion of acquisition problem transaction, carry out regional interrelation.This association finally obtains the IP of the problem transaction monitored.
Essential information associates: the information of concluding the business for the information such as the mailbox in the log-on message that internet business relates to, responsible official, telephone number, shop network address and confirmed problem carries out direct correlation, contributes to the suspicion grade finding transaction.This association finally obtains the essential information of the problem transaction monitored.
Commodity association: as assisting of essential information association, registration essential information identical in IP information is similar, the degree of association of product also can as one of judgement information of problem transaction association.This association finally obtains the merchandise news of the problem transaction monitored
The industry number of commodity association: this index is used for the degree of association index of commodity and industry, the industry number of self commodity association of Main Analysis goods providers.This association finally obtains the commodity industry of the problem transaction monitored.
Each in above-mentioned aggregation association associates the monitoring result obtained, the results set of common composition aggregation association.
S3: according to setting up monitoring model based on the index set of prestige parameter and the results set of aggregation association and obtaining monitoring set, and this monitoring set is monitored and early warning e-commerce transaction as a reference, be specially: this monitoring set is met to the index of such as current e-commerce transaction, then judge that this e-commerce transaction is problem transaction, thus carry out early warning.Specific as follows:
Adopt neural network algorithm to set up monitoring model, the process of Modling model is see Fig. 4.Be specially:
First, by the initial weight set W of the random number between (0,1) as monitoring, the nominal error function of each weights in setting initial weight set W is e, given computational accuracy ε and maximum study number of times M;
Secondly, from table one data, a random selecting kth input amendment is as Input Monitor Connector index (input vector): x (k)=(x
1(k), x
2(k) ..., x
n(k)), set trading activity warning index as exporting monitoring index (desired output vector): d
o(k)=(d
1(k), d
2(k) ..., d
q(k)).In the present embodiment, by the index set based on prestige parameter as Input Monitor Connector index set X, and the warning index Y of the trading activity exported is as monitoring result.Input Monitor Connector index set X can comprise based on the arbitrary one or more index in the index set of prestige parameter; Choose (see form one) from four main aspects to sample training collection and selected index in the present embodiment, amount to 26 indexs, namely input layer number is n=26, comprises four large classes: registration behavior: the essential information in the shop of electric business's platform registration; Operation behavior: the situation that logs in of electric business's platform member and inquiry situation; Popularization behavior: its product of retail shop position or service buy popularization behavior; Complained behavior: complained number of times record.
In this algorithm, hidden layer nodes is h, that is, hidden layer input vector hi=(hi
1, hi
2..., hi
p), hidden layer output vector ho=(ho
1, ho
2..., ho
p), output layer nodes is o, that is, output layer input vector yi=(yi
1, yi
2..., yi
p), output layer output vector yo=(yo
1, yo
2..., yo
p).Determine input layer and hidden layer respectively, the weight matrix that links between hidden layer with output layer is W
h, W
oand threshold values b
h, b
o.The neural network of a m layer, for given sample set X
i(i=1,2 ..., n), if the i of kth layer neuronic input summation is expressed as
output summation is
i-th neuronic weight coefficient from a jth neuron of kth-1 layer to kth layer is W
ij, each neuronic excitation function is f (), then the relation of each variable can be expressed as formula (1), (2):
The quadratic sum of definition desired output and the actual difference exported as error function such as formula shown in (3):
Y
ibe the expectation value of output unit, m layer is output layer,
actual output.BP algorithm adopts the steepest descending method in nonlinear programming, by the negative gradient direction power of amendment coefficient of error function e.
Mahalanobis distance in employing machine learning herein weighs the difference in a certain sample between institute's directed quantity, for l vectorial X
1~ X
l, establish the most reasonable vectorial X
ksample training is launched as BP neural network standard output.Covariance matrix is designated as S, vectorial X
iwith X
jbetween mahalanobis distance definition such as formula shown in (4):
In covariance matrix, each element is the covariance Cov (X, Y) between each vector element, Cov (X, Y)=E{ [X-E (X)] [Y-E (Y)] }, wherein E is mathematical expectation.
Wherein, the results set of aggregation association constantly adjusts the error e of output and actual value in the process of training sample data, and feed back to the network that input layer carries out the multilayer feed-forward of backpropagation, until the learning error of algorithm no longer obviously reduces by output layer.
When algorithm performs, when error satisfies condition or reached maximum study number of times, export the warning index of trading activity.Model is monitored current e-commerce transaction after setting up, if its index meets the monitoring set of this model, then judges that this e-commerce transaction is problem transaction, thus carries out early warning.
S4: regularly or when needed follow the trail of new transaction record and extract new aggregation association result, and new aggregation association result is added described monitoring set.
This step S4 detailed process as shown in Figure 5, comprising:
S41: follow the trail of new transaction record and set up trading activity information set in conjunction with reputation information collection.
S42: by sorting algorithm to new trading activity information set carry out first time prediction, obtain suspicion problem transaction first time suspicion class information and predict the outcome collection A.
S43: filter the essential information in new transaction record and analyze, obtains new transaction essential information collection, and new transaction essential information collection is collected A carry out associating computing with described predicting the outcome, and obtains second time suspicion class information and the collection B that predicts the outcome; Wherein, filter analysis is classified according to incidence relation exactly, and classification registers the large class of behavior, operation behavior, popularization behavior and complained behavior four exactly.
S44: by setting threshold value will predict the outcome collection B be categorized as excessive risk problem transaction group and low-risk problem conclude the business group, excessive risk problem transaction group associates with the aggregation association result in step S3 again, obtain new aggregation association result, and add above-mentioned monitoring set, to upgrade this monitoring set, the operation of this step can expand estimation range and precision further, thus improves monitoring model.Wherein, namely after twice prediction, its suspicion ranking score is higher than the data group of threshold value for the transaction of excessive risk problem, and this kind of user has greater probability and is judged as that problem is concluded the business.Low-risk problem transaction group be then twice prediction after score lower than the data group of threshold value, the result of this low-risk problem transaction group can be used as the collection C that predicts the outcome and carries out reference, also can not consider that it directly abandons the impact of monitoring set.
The present invention is based on the analyzing and processing of neural network recognization internet business data, mining analysis is carried out from the Subjective and Objective of e-commerce transaction and intersubjective association various dimensions, the omnibearing behavior to process of exchange, infer the behavioral characteristic of problem transaction, the self study and the self-adaptation that realize transaction Early-warning Model improve precision of prediction and coverage rate.The present invention takes into account the equality of both parties, and in the angle of e-commerce platform, neutrality is analyzed with objective problem transaction, monitored and early warning relatively, realizes the e-commerce transaction real-time risk assessment to internet business data.
The above; be only the specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, any those skilled in the art is in the technical scope that the present invention discloses; the distortion do the present invention or replacement, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of described claim.
Claims (10)
1., based on an e-commerce transaction monitoring method for internet business data, it is characterized in that, comprise the steps:
S1: obtain transaction record, and carry out prestige Parameter analysis according to transaction record, and set up reputation information collection;
S2: screening is carried out to reputation information collection and obtains problem transaction set, sample analysis is carried out to problem transaction set, extract the index set based on prestige parameter, and set up the aggregation association based on prestige parameter, obtain the results set of aggregation association;
S3: set up monitoring model based on the index set of prestige parameter and the results set of aggregation association according to described and obtain monitoring set, and this monitoring set is monitored and early warning e-commerce transaction as a reference, be specially: this monitoring set is met to the index of such as current e-commerce transaction, then judge that this e-commerce transaction is problem transaction;
Wherein, described aggregation association is specially and current transaction is related to information and the problem index set related to of concluding the business and associate.
2. the e-commerce transaction monitoring method based on internet business data according to claim 1, is characterized in that, described prestige Parameter analysis is divided into the analysis based on B2C or C2C and the analysis based on B2B; Wherein, the analysis based on B2C or C2C comprises: analyzing influence executes the factor of letter side to the trust of trading object, and the factor that analyzing influence internet business is trusted; Analysis based on B2B comprises: the factor of the trust between analyzing influence tissue.
3. the e-commerce transaction monitoring method based on internet business data according to claim 2, it is characterized in that, impact is executed the factor of letter side to the trust of trading object and is comprised: to the trust of the other side, based on the trust controlled, for the expectation of potential income and himself is for the attitude of risk;
The factor affecting internet business trust comprises: technical factor, environmental factor, commercial factors and individual factor;
The factor of the trust between impact tissue comprises: appreciable supervision, appreciable certification, appreciable legal restraint, appreciable feedback and appreciable Writing Standards.
4. the e-commerce transaction monitoring method based on internet business data according to claim 1, is characterized in that, it is one or more that described index set comprises in registration behavior, operation behavior, popularization behavior and complained behavior.
5. the e-commerce transaction monitoring method based on internet business data according to claim 1, it is characterized in that, described aggregation association comprises IP association, essential information associates and the industry number of commodity association, and wherein, described IP association comprises direct IP association and associates with IP is advanced; The association of described essential information is specially the information that mailbox in the log-on message of concluding the business and relating to, responsible official, telephone number, shop network address or other log-on messages and confirmed problem conclude the business and carries out direct correlation; The industry number of described commodity association is the degree of association index of commodity and industry.
6. the e-commerce transaction monitoring method based on internet business data according to claim 5, it is characterized in that, described aggregation association also comprises commodity association, described commodity association be, registration essential information identical in IP information identical the association of product.
7. the e-commerce transaction monitoring method based on internet business data according to claim 1, is characterized in that, also comprise after described step S3:
S4: regularly or when needed follow the trail of new transaction record and extract new aggregation association result, and new aggregation association result is added described monitoring set.
8. the e-commerce transaction monitoring method based on internet business data according to claim 7, it is characterized in that, described step S4 specifically comprises:
S41: follow the trail of new transaction record and set up trading activity information set in conjunction with described reputation information collection;
S42: by sorting algorithm to described trading activity information set carry out first time prediction, obtain suspicion problem transaction first time suspicion class information and predict the outcome collection A;
S43: filter the essential information in new transaction record and analyze, obtains new transaction essential information collection, and new transaction essential information collection is collected A carry out associating computing with described predicting the outcome, and obtains second time suspicion class information and the collection B that predicts the outcome;
S44: by setting threshold value will predict the outcome collection B be categorized as excessive risk problem transaction group and low-risk problem conclude the business group, excessive risk problem transaction group associates with the aggregation association result in step S3 again, obtain new aggregation association result, and add described monitoring set, to upgrade this monitoring set.
9. the e-commerce transaction monitoring method based on internet business data according to claim 1, it is characterized in that, neural network algorithm is adopted to set up described monitoring model, be specially: using the described index set based on prestige parameter as Input Monitor Connector index set X, (0, 1) random number between is as the initial weight set W of monitoring, the warning index Y of the trading activity exported is as monitoring result, wherein, the results set of aggregation association constantly adjusts the error of output and actual value in the process of training sample data, and feed back to by output layer the network that input layer carries out the multilayer feed-forward of backpropagation, until the learning error of algorithm no longer obviously reduces.
10. the e-commerce transaction monitoring method based on internet business data according to claim 9, it is characterized in that, the nominal error function setting each weights in described initial weight set W is e, given computational accuracy ε and maximum study number of times M, Input Monitor Connector index set X comprise described based on the arbitrary one or more index in the index set of prestige parameter; When algorithm performs, when error satisfies condition or reached maximum study number of times, export the warning index of trading activity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510306742.1A CN104915842A (en) | 2015-06-04 | 2015-06-04 | Electronic commerce transaction monitoring method based on internet transaction data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510306742.1A CN104915842A (en) | 2015-06-04 | 2015-06-04 | Electronic commerce transaction monitoring method based on internet transaction data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104915842A true CN104915842A (en) | 2015-09-16 |
Family
ID=54084888
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510306742.1A Pending CN104915842A (en) | 2015-06-04 | 2015-06-04 | Electronic commerce transaction monitoring method based on internet transaction data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104915842A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106920109A (en) * | 2017-02-21 | 2017-07-04 | 福建师范大学福清分校 | Recognition methods, system and e-commerce system for ecommerce wash sale |
| CN107958341A (en) * | 2017-12-12 | 2018-04-24 | 阿里巴巴集团控股有限公司 | Risk Identification Method and device and electronic equipment |
| CN108985638A (en) * | 2018-07-25 | 2018-12-11 | 腾讯科技（深圳）有限公司 | A kind of customer investment methods of risk assessment and device and storage medium |
| CN109074559A (en) * | 2016-02-22 | 2018-12-21 | 塔塔咨询服务有限公司 | System and method for complaint and Prestige Management in multiparty data market |
| CN109242625A (en) * | 2018-09-06 | 2019-01-18 | 成都数银科技有限公司 | A kind of big data monitoring and managing method |
| CN109242499A (en) * | 2018-09-19 | 2019-01-18 | 中国银行股份有限公司 | A kind of processing method of transaction risk prediction, apparatus and system |
| CN109389270A (en) * | 2017-08-09 | 2019-02-26 | 菜鸟智能物流控股有限公司 | Logistics object determination method and device and machine readable medium |
| CN109426965A (en) * | 2017-08-16 | 2019-03-05 | 上海诺悦智能科技有限公司 | A kind of suspicious trade network recognition methods based on machine learning |
| CN109658643A (en) * | 2018-12-06 | 2019-04-19 | 中国工商银行股份有限公司 | Payment risk alarm method, server and intelligence POS |
| WO2019090519A1 (en) * | 2017-11-08 | 2019-05-16 | Paypal, Inc. | Robust and adaptive artificial intelligence modeling |
| TWI673666B (en) * | 2016-07-01 | 2019-10-01 | 香港商阿里巴巴集團服務有限公司 | Method and device for data risk control |
| CN112308718A (en) * | 2020-11-23 | 2021-02-02 | 许昌学院 | Electronic commerce transaction monitoring method based on internet data |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060271460A1 (en) * | 2005-05-31 | 2006-11-30 | Ebay Inc. | Method and system to provide user created social networks in a distributed commerce system |
| CN102117459A (en) * | 2010-01-06 | 2011-07-06 | 阿里巴巴集团控股有限公司 | Risk control system and method |
| CN102915508A (en) * | 2012-09-13 | 2013-02-06 | 吴启波 | System and method for fiduciary loan risk control based on transaction records |
-
2015
- 2015-06-04 CN CN201510306742.1A patent/CN104915842A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060271460A1 (en) * | 2005-05-31 | 2006-11-30 | Ebay Inc. | Method and system to provide user created social networks in a distributed commerce system |
| CN102117459A (en) * | 2010-01-06 | 2011-07-06 | 阿里巴巴集团控股有限公司 | Risk control system and method |
| CN102915508A (en) * | 2012-09-13 | 2013-02-06 | 吴启波 | System and method for fiduciary loan risk control based on transaction records |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109074559A (en) * | 2016-02-22 | 2018-12-21 | 塔塔咨询服务有限公司 | System and method for complaint and Prestige Management in multiparty data market |
| US11074350B2 (en) | 2016-07-01 | 2021-07-27 | Advanced New Technologies Co., Ltd. | Method and device for controlling data risk |
| TWI673666B (en) * | 2016-07-01 | 2019-10-01 | 香港商阿里巴巴集團服務有限公司 | Method and device for data risk control |
| CN106920109A (en) * | 2017-02-21 | 2017-07-04 | 福建师范大学福清分校 | Recognition methods, system and e-commerce system for ecommerce wash sale |
| CN109389270B (en) * | 2017-08-09 | 2022-11-04 | 菜鸟智能物流控股有限公司 | Logistics object determination method and device and machine readable medium |
| CN109389270A (en) * | 2017-08-09 | 2019-02-26 | 菜鸟智能物流控股有限公司 | Logistics object determination method and device and machine readable medium |
| CN109426965A (en) * | 2017-08-16 | 2019-03-05 | 上海诺悦智能科技有限公司 | A kind of suspicious trade network recognition methods based on machine learning |
| CN111566683A (en) * | 2017-11-08 | 2020-08-21 | 贝宝公司 | Robust and adaptive artificial intelligence modeling |
| WO2019090519A1 (en) * | 2017-11-08 | 2019-05-16 | Paypal, Inc. | Robust and adaptive artificial intelligence modeling |
| CN107958341A (en) * | 2017-12-12 | 2018-04-24 | 阿里巴巴集团控股有限公司 | Risk Identification Method and device and electronic equipment |
| CN108985638A (en) * | 2018-07-25 | 2018-12-11 | 腾讯科技（深圳）有限公司 | A kind of customer investment methods of risk assessment and device and storage medium |
| CN108985638B (en) * | 2018-07-25 | 2020-07-24 | 腾讯科技（深圳）有限公司 | User investment risk assessment method and device and storage medium |
| CN109242625A (en) * | 2018-09-06 | 2019-01-18 | 成都数银科技有限公司 | A kind of big data monitoring and managing method |
| CN109242499A (en) * | 2018-09-19 | 2019-01-18 | 中国银行股份有限公司 | A kind of processing method of transaction risk prediction, apparatus and system |
| CN109658643B (en) * | 2018-12-06 | 2021-04-27 | 中国工商银行股份有限公司 | Payment risk warning method, server and intelligent POS (point of sale) |
| CN109658643A (en) * | 2018-12-06 | 2019-04-19 | 中国工商银行股份有限公司 | Payment risk alarm method, server and intelligence POS |
| CN112308718A (en) * | 2020-11-23 | 2021-02-02 | 许昌学院 | Electronic commerce transaction monitoring method based on internet data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104915842A (en) | Electronic commerce transaction monitoring method based on internet transaction data | |
| Valmohammadi et al. | Using interpretive structural modeling and fuzzy analytical process to identify and prioritize the interactive barriers of e-commerce implementation | |
| Ogwueleka et al. | Neural network and classification approach in identifying customer behavior in the banking sector: A case study of an international bank | |
| Sharma et al. | Artificial neural networks: Applications in management | |
| Gabralla et al. | Computational modeling of crude oil price forecasting: A review of two decades of research | |
| CN108492001A (en) | A method of being used for guaranteed loan network risk management | |
| Abdou et al. | Would credit scoring work for Islamic finance? A neural network approach | |
| Kuzmenko et al. | An approach to managing innovation to protect financial sector against cybercrime | |
| Zhang et al. | A hybrid model considering cointegration for interval‐valued pork price forecasting in China | |
| Harini et al. | Levarging Blockchain for Transparency in Agriculture Supply Chain Management Using IoT and Machine Learning | |
| Kayakuş et al. | Estimation the Number of Visitor of E-Commerce Website by Artificial Neural Networks During Covid19 in Turkey. | |
| Wang et al. | E-commerce credit risk assessment based on fuzzy neural network | |
| Ju et al. | Stress test for a technology credit guarantee fund based on survival analysis | |
| Adewusi et al. | Application of artificial neural network to loan recovery prediction | |
| Rao et al. | Risk evaluation and forecast behavior analysis of supply chain financing based on blockchain | |
| Qian et al. | The implementation of leisure tourism enterprise management system based on deep learning | |
| Sugumar et al. | A technique to stock market prediction using fuzzy clustering and artificial neural networks | |
| Olaniyan et al. | Sentiment and stock market volatility predictive modelling—A hybrid approach | |
| Li et al. | Real-Time Prediction of Cross-Border e-Commerce Spike Performance Based on Neural Network and Decision Tree | |
| Peng | Research on credit risk identification of Internet financial enterprises based on big data | |
| Wang et al. | Innovative risk early warning model based on internet of things under big data technology | |
| Pang et al. | Wt model & applications in loan platform customer default prediction based on decision tree algorithms | |
| Xie | Spatial aggregation analysis and systemic risk prevention of internet finance based on neural network model | |
| Prathyusha et al. | Detecting frauds in online auction system | |
| Carmona et al. | Learning of weighted multi-layer networks via dynamic social spaces, with application to financial interbank transactions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150916 |
|
| RJ01 | Rejection of invention patent application after publication |