CN116934418A - Abnormal order detection and early warning method, system, equipment and storage medium - Google Patents
Abnormal order detection and early warning method, system, equipment and storage medium Download PDFInfo
- Publication number
- CN116934418A CN116934418A CN202310715693.1A CN202310715693A CN116934418A CN 116934418 A CN116934418 A CN 116934418A CN 202310715693 A CN202310715693 A CN 202310715693A CN 116934418 A CN116934418 A CN 116934418A
- Authority
- CN
- China
- Prior art keywords
- order
- abnormal
- data
- commodity
- promotion
- 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.)
- Granted
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 243
- 238000001514 detection method Methods 0.000 title claims abstract description 102
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000012216 screening Methods 0.000 claims abstract description 37
- 238000004364 calculation method Methods 0.000 claims abstract description 18
- 230000005856 abnormality Effects 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 19
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 claims description 7
- 238000012790 confirmation Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 4
- 238000013075 data extraction Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 241000283690 Bos taurus Species 0.000 description 9
- 238000012217 deletion Methods 0.000 description 3
- 230000037430 deletion Effects 0.000 description 3
- 208000012260 Accidental injury Diseases 0.000 description 2
- 208000014674 injury Diseases 0.000 description 2
- 206010043417 Therapeutic response unexpected Diseases 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0633—Lists, e.g. purchase orders, compilation or processing
- G06Q30/0635—Processing of requisition or of purchase orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The application discloses a detection and early warning method, a system, equipment and a storage medium for abnormal orders, wherein the method comprises the steps of screening and removing order lists of promotion periods of a large promotion node to obtain order transaction data of the promotion periods; according to the key information and the order transaction data, extracting order key data of each promotion period through order association degree calculation, and summarizing to obtain order key data of the large promotion node; inputting order key data of the large promotion node into an order anomaly detection model to obtain an order anomaly detection result; extracting all abnormal order users according to the order abnormality detection result, screening all abnormal order users according to commodity purchase data of all abnormal order users, and obtaining final abnormal order users; and carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user. The embodiment realizes the rapid detection and early warning of the abnormal order placed by the large promotion node, and improves the accuracy of abnormal order detection.
Description
Technical Field
The present application relates to the field of order detection, and in particular, to a method, system, device, and storage medium for detecting and early warning of abnormal orders.
Background
The e-commerce large promotion node refers to a large promotion which is regularly held on an e-commerce platform. These activities are typically performed at specific time nodes, such as 618, double 11, double 12, etc. In addition to providing more offers and discounts to consumers, the e-commerce facilitation node may also promote the development of the e-commerce industry. However, the large volume of orders and the large number of order users for the large promotion nodes also present some problems and challenges. For example, due to the great preference of the large promotion node, the original price is used for buying some sales or out-of-stock products, then the price is sold at high price to earn the difference, a large number of cattle orders are appeared, the sales market is seriously disturbed, the rights and interests of merchants and consumers are damaged, besides the cattle orders, the situation that professional counterfeiters maliciously place orders can appear in the orders of the large promotion node, and the credit of the merchants is seriously damaged due to malicious complaints after ordering. Therefore, detection of abnormal orders such as cattle orders and professional dummy orders is of great importance to ensure legal rights of merchants and consumers.
In the prior art, account information is verified through address information and order information, so that a cattle user is prevented from purchasing a product by using false account information, but all abnormal orders under a large promotion node cannot be accurately judged only through a single account verification method. The existing training data obtained based on historical orders are trained to generate an order detection model, the order detection model is called to automatically detect the orders to be detected, the relevance among the orders is ignored only by a model prediction method, the detection of a small number of orders is effective, and accurate results cannot be obtained quickly for identifying a large number of orders.
Disclosure of Invention
The application provides a detection and early warning method, a system, equipment and a storage medium for abnormal orders, which are used for realizing rapid detection and early warning of abnormal orders placed by a large promotion node and improving the detection accuracy of the abnormal orders.
In order to solve the above technical problems, an embodiment of the present application provides a method for detecting and early warning an abnormal order, including:
dividing promotion time of the large promotion node in time intervals to obtain a plurality of promotion time intervals, screening and removing order lists of the promotion time intervals to obtain order transaction data of the promotion time intervals; the order list of each promotion period is obtained by arranging the orders of each promotion period in a list form according to the orders of each promotion period obtained in a background management database;
according to the key information and the order transaction data of each promotion period, extracting order key data of each promotion period through order association degree calculation, and summarizing the corresponding order key data of each promotion period to obtain order key data of a large promotion node; the key information comprises a receiving address, a user ID, user consumption habits and preferences, a trade commodity price and commodity trade quantity;
inputting order key data of the large promotion node into an order anomaly detection model to obtain an order anomaly detection result; the order anomaly detection model is obtained through neural network training through a historical order data set;
according to the order anomaly detection result, all the anomaly order users are extracted, the historical order information of all the anomaly order users is counted to obtain commodity purchase data of all the anomaly order users, and according to the commodity purchase data of all the anomaly order users, all the anomaly order users are screened to obtain final anomaly order users;
and carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user.
According to the embodiment of the application, the promotion time of the large promotion node is divided into a plurality of promotion time periods, the order list of each promotion time period is screened and removed, and order transaction data of each promotion time period is obtained; according to the key information and the order transaction data of each promotion period, extracting order key data of each promotion period through order association degree calculation, and summarizing the corresponding order key data of each promotion period to obtain order key data of a large promotion node; inputting order key data of the large promotion node into an order anomaly detection model to obtain an order anomaly detection result; according to the order anomaly detection result, all the anomaly order users are extracted, the historical order information of all the anomaly order users is counted to obtain commodity purchase data of all the anomaly order users, and according to the commodity purchase data of all the anomaly order users, all the anomaly order users are screened to obtain final anomaly order users; and carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user. And (3) selecting order key data placed by the large promotion node through order screening and removing and order association degree calculation, and predicting based on the order key data, so that the detection precision of order anomaly detection is improved. And then, according to the order abnormality detection result and the history order information of the abnormal order user, further analyzing and determining the final abnormal order user, avoiding the accidental injury to the normal user and improving the accuracy of abnormal order detection. Finally, the order key data of the large promotion node and the final abnormal order users are combined to perform abnormal order early warning, so that abnormal orders of the large promotion node are accurately found, and the rights and interests of consumers and merchants are effectively maintained.
As a preferred scheme, the order list of each promotion period is screened and removed to obtain the order transaction data of each promotion period, which specifically comprises:
taking order basic transaction information as a header, arranging order lists in a current promotion period to obtain each row of the current order list, and assigning indexes of each row of the current order list to obtain the row index of the current order list;
searching data meeting preset screening conditions in the current order list to obtain reject data; the preset screening conditions comprise that the list element value is null, the order is unpaid, the invalid order is invalid, and the order is cancelled;
searching a row index corresponding to the reject data through a preset searching function, and obtaining data to be deleted according to the row index corresponding to the reject data and the current order list;
deleting the line data to be deleted from the current order list through a preset deleting function, and obtaining order transaction data of the current promotion period according to the deleted current order list.
According to the embodiment of the application, invalid data in the order list is screened and removed, effective transaction data is obtained, the processing is performed based on the effective transaction data, the order detection time is shortened, abnormal orders can be detected conveniently and rapidly, parallel processing of each promotion period can be realized, the concurrence pressure of a server is reduced, and the processing speed of abnormal detection is improved.
As a preferred scheme, according to the key information and the order transaction data of each promotion period, the order key data of each promotion period is extracted through the calculation of the order association degree, specifically:
acquiring user information data and commodity transaction data in the current order transaction data according to the order transaction data of the current promotion number segment;
performing iterative operation processing on the user information data and commodity transaction data to generate an order association probability map;
determining a correlation degree value between orders in current order transaction data according to the order correlation probability map, and extracting orders meeting a preset probability threshold to obtain correlation orders;
according to the association order, carrying out association screening on the order transaction data of the current promotion period to obtain association order transaction data;
and carrying out data extraction on the related order transaction data according to the key information to obtain the order key data of the current promotion period.
According to the embodiment of the application, the related orders are found out through the calculation of the degree of association of the orders, the key data of the orders are extracted, related false orders such as cattle, professional shavings and the like are rapidly extracted and obtained, a large amount of related abnormal order data are effectively simplified according to the key information, the key data of the orders are extracted, rapid abnormal information detection is facilitated, abnormal orders are efficiently obtained, and the accuracy of abnormal order detection is improved.
As a preferred scheme, iterative operation processing is carried out on the user information data and commodity transaction data to generate an order association probability map, which specifically comprises the following steps:
adding edges between user information data and commodity transaction data, and establishing an initial mapping relation to form an initial association relation diagram;
judging whether the association relation taking commodity transaction data as a public end exists between the user information data or not based on the initial association relation graph, and if so, establishing an adjacent relation between the user information data and the commodity transaction data;
according to the iterative adjacency relationship, the association probability among the user information data, the association probability among the commodity transaction data and the association probability among the user information data and the commodity transaction data are obtained;
and according to each adjacent relation and each association probability, adjusting the initial association relation graph to form an order association probability graph.
As a preferred scheme, the order anomaly detection model is obtained by training a neural network through a historical order data set, specifically:
acquiring order key data of a plurality of historical large promotion nodes, labeling abnormal orders of the order key data of each historical large promotion node, and constructing a historical order data set according to the labeled order key data of each historical large promotion node;
and inputting the historical order data set into the initial order anomaly detection model for training of the neural network, and stopping training of the neural network when a preset training ending condition is met, so as to obtain the order anomaly detection model.
According to the embodiment of the application, based on the order key data, the abnormal order to the current large promotion node is determined through the neural network prediction, so that the prediction accuracy and the detection speed are effectively improved.
As a preferred scheme, according to commodity purchase data of all abnormal order users, screening all abnormal order users to obtain final abnormal order users, specifically:
extracting the commodity quantity and commodity names of all abnormal order users according to commodity purchase data of all abnormal order users;
according to the commodity quantity of the current abnormal order users, the commodity names of the current abnormal order users are ordered in a reverse order to obtain a current commodity purchase ordering table, and according to the preset quantity condition, the current commodity purchase ordering table is subjected to data rejection to obtain abnormal commodity data corresponding to the current abnormal order users; the commodity names and the commodity quantity in the current commodity purchase ordering list are used as the header;
counting the abnormal commodity data corresponding to each abnormal order user, and clustering commodity names in the abnormal commodity data corresponding to each abnormal order user to obtain commodity categories of each abnormal order user;
counting the total number of commodities corresponding to commodity categories of each abnormal order user, screening out abnormal order users meeting preset abnormal user conditions from all the abnormal order users according to the commodity categories of each abnormal order user and the total number of commodities corresponding to commodity categories, and obtaining a final abnormal order user; the preset abnormal user condition is that the commodity category is larger than a first preset value, and the total number of commodities corresponding to the commodity category is larger than a second preset value.
By implementing the embodiment of the application, whether the user really buys the abnormal user such as the cattle or the professional dummy is further judged according to the historical order of the abnormal user through the historical order data, so that the normal purchase of the consumer is avoided from being accidentally injured, and the accuracy of abnormal order detection is improved.
As a preferred scheme, according to order key data of the large promotion node and a final abnormal order user, abnormal order early warning is carried out, specifically:
matching the order key data of the large promotion node with a final abnormal order user to obtain abnormal order information;
marking an abnormal order according to the abnormal order information, generating abnormal early warning information according to the abnormal order, and sending the abnormal early warning information to a corresponding background management account so that the marked abnormal order is displayed in the corresponding background management account.
In order to solve the same technical problem, the embodiment of the application also provides a detection and early warning system for abnormal orders, which comprises: the system comprises an order data screening module, a related key order extraction module, an abnormal order detection module, an abnormal user confirmation module and an abnormal order early warning module;
the order data screening module is used for dividing the promotion time of the large promotion node into time intervals to obtain a plurality of promotion time intervals, and screening and removing the order list of each promotion time interval to obtain order transaction data of each promotion time interval; the order list of each promotion period is obtained by arranging the orders of each promotion period in a list form according to the orders of each promotion period obtained in a background management database;
the related key order extraction module is used for extracting order key data of each promotion period through order related degree calculation according to the key information and the order transaction data of each promotion period, and summarizing the corresponding order key data of each promotion period to obtain order key data of a large promotion node; the key information comprises a receiving address, a user ID, user consumption habits and preferences, a trade commodity price and commodity trade quantity;
the abnormal order detection module is used for inputting order key data of the large promotion node into an order abnormal detection model to obtain an order abnormal detection result; the order anomaly detection model is obtained through neural network training through a historical order data set;
the abnormal user confirmation module is used for extracting all abnormal order users according to the order abnormal detection result, counting the historical order information of all abnormal order users to obtain commodity purchase data of all abnormal order users, and screening all abnormal order users according to the commodity purchase data of all abnormal order users to obtain final abnormal order users;
the abnormal order early warning module is used for carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user.
In order to solve the same technical problems, the embodiment of the application also provides computer equipment, which comprises a processor and a memory, wherein the memory is used for storing a computer program, and the computer program is executed by the processor to realize the detection and early warning method of the abnormal order.
In order to solve the same technical problem, the embodiment of the application also provides a computer readable storage medium which stores a computer program, and the computer program realizes the detection and early warning method of the abnormal order when being executed by a processor.
Drawings
Fig. 1: the application provides a flow diagram of one embodiment of an abnormal order detection and early warning method;
fig. 2: an order correlation degree calculation flow chart of one embodiment of the abnormal order detection and early warning method provided by the application;
fig. 3: the application provides a structural schematic diagram of an embodiment of an abnormal order detection and early warning system.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Example 1
Referring to fig. 1, a flow chart of an abnormal order detection and early warning method according to an embodiment of the present application is shown. The detection and early warning method is suitable for abnormal order detection of the large promotion node, the order key data is obtained through order screening and removing and correlation degree calculation, and abnormal orders and early warning under the large promotion node are rapidly detected through model prediction and abnormal user screening, so that the accuracy of abnormal order detection is improved. The detection early warning method comprises steps 101 to 105, wherein the steps are as follows:
step 101: dividing promotion time of the large promotion node in time intervals to obtain a plurality of promotion time intervals, screening and removing order lists of the promotion time intervals to obtain order transaction data of the promotion time intervals; the order list of each promotion period is obtained by arranging the orders of each promotion period in a list form according to the orders of each promotion period obtained in a background management database.
In this embodiment, since the promotion time of the large promotion node has an uncertain line, the time interval is divided according to the promotion time of a specific large promotion node, and specific time interval dividing methods include, but are not limited to, dividing by hours, dividing by days and dividing by weeks, if the promotion time of the large promotion node is shorter, the whole large promotion node can be used as one promotion time interval, or divided into two promotion time intervals. Meanwhile, the order comprises basic information data of the order and specific data values of the order, and the data processing is carried out in a two-dimensional form of an order list, so that the data reading and the further processing are facilitated.
Optionally, the order list of each promotion period is screened and removed to obtain the order transaction data of each promotion period, which specifically includes: taking order basic transaction information as a header, arranging order lists in a current promotion period to obtain each row of the current order list, and assigning indexes of each row of the current order list to obtain the row index of the current order list; searching data meeting preset screening conditions in the current order list to obtain reject data; the preset screening conditions comprise that the list element value is null, the order is unpaid, the invalid order is invalid, and the order is cancelled; searching a row index corresponding to the reject data through a preset searching function, and obtaining data to be deleted according to the row index corresponding to the reject data and the current order list; deleting the line data to be deleted from the current order list through a preset deleting function, and obtaining order transaction data of the current promotion period according to the deleted current order list.
In this embodiment, the order base transaction information includes all information of orders such as commodity name, commodity, shop name, shop preference, shipping charge information, order number, receiving address, user ID, user consumption habit and preference, transaction commodity price, commodity transaction quantity, and the like. When the order list is screened and removed, a corresponding search function and a deletion function can be selected according to the derived order list, wherein the search function comprises an index function, a find function and a custom search function, and the deletion function comprises a remove function, a pop function and a custom deletion function.
According to the embodiment of the application, invalid data in the order list is screened and removed, effective transaction data is obtained, the processing is performed based on the effective transaction data, the order detection time is shortened, abnormal orders can be detected conveniently and rapidly, parallel processing of each promotion period can be realized, the concurrence pressure of a server is reduced, and the processing speed of abnormal detection is improved.
Step 102: according to the key information and the order transaction data of each promotion period, extracting order key data of each promotion period through order association degree calculation, and summarizing the corresponding order key data of each promotion period to obtain order key data of a large promotion node; the key information comprises a receiving address, a user ID, user consumption habits and preferences, a commodity price and commodity transaction quantity.
Optionally, according to the key information and the order transaction data of each promotion period, the order key data of each promotion period is extracted through order association degree calculation, and the order association degree calculation flow chart, as shown in fig. 2, specifically includes steps 201 to 205, and each step specifically includes the following steps:
step 201: and acquiring user information data and commodity transaction data in the current order transaction data according to the order transaction data of the current promotion number segment.
In this embodiment, a certain correlation exists between orders of most cattle and counterfeiters, the correlation of the orders is that goods purchased by users, and the correlation between order data can be effectively obtained by user information data and commodity transaction data in order transaction data. And extracting order key data of each promotion period, and taking the order transaction data of the unprocessed promotion period as the current order transaction data required to be processed.
Step 202: and carrying out iterative operation processing on the user information data and the commodity transaction data to generate an order association probability map.
Optionally, step 202 specifically includes: adding edges between user information data and commodity transaction data, and establishing an initial mapping relation to form an initial association relation diagram; judging whether the association relation taking commodity transaction data as a public end exists between the user information data or not based on the initial association relation graph, and if so, establishing an adjacent relation between the user information data and the commodity transaction data; according to the iterative adjacency relationship, the association probability among the user information data, the association probability among the commodity transaction data and the association probability among the user information data and the commodity transaction data are obtained; and according to each adjacent relation and each association probability, adjusting the initial association relation graph to form an order association probability graph.
In this embodiment, an edge is added between user information data and commodity transaction data, an initial mapping relationship is established, an initial association relationship diagram is formed, and the specific edge adding process is as follows: according to the user information data, generating user nodes of an initial association relation graph, then according to commodity type data in commodity transaction data, generating input nodes of the initial association relation graph, and according to commodity quantity data in commodity transaction data, constructing initial edges between the user nodes and the input nodes, and further constructing the initial association relation graph. The value of the initial edge represents an initial value of an association probability between the user node and the input node, and the initial value of the association probability between the user node and the input node is equal to a ratio of the current commodity purchase quantity of the user node corresponding to the input node to the maximum commodity purchase quantity in the data set. In the iteration process of establishing the adjacency relation, if the association relation taking a commodity transaction as a public end does not exist between certain user information, the adjacency relation between the user information and the commodity transaction does not need to be established. The initial association relation graph is adjusted by deleting the multimedia nodes and the input nodes which do not have intermediate nodes in the initial association relation graph according to each adjacent relation and each association probability, and the order association probability graph is obtained by adjusting the initial association relation graph.
Step 203: and determining the association degree value between orders in the current order transaction data according to the order association probability map, and extracting orders meeting a preset probability threshold value to obtain association orders.
In this embodiment, the association probability between orders is used as the association degree value between orders, and the preset probability threshold can be adjusted according to actual situations.
Step 204: and carrying out relevance screening on the order transaction data of the current promotion period according to the relevance order to obtain relevance order transaction data.
In this embodiment, the relevance filtering is to delete the non-relevance orders in the order transaction data of the current promotion period, so as to obtain the relevant order transaction data.
Step 205: and carrying out data extraction on the related order transaction data according to the key information to obtain the order key data of the current promotion period.
According to the embodiment of the application, the related orders are found out through the calculation of the degree of association of the orders, the key data of the orders are extracted, related false orders such as cattle, professional shavings and the like are rapidly extracted and obtained, a large amount of related abnormal order data are effectively simplified according to the key information, the key data of the orders are extracted, rapid abnormal information detection is facilitated, abnormal orders are efficiently obtained, and the accuracy of abnormal order detection is improved.
Step 103: inputting order key data of the large promotion node into an order anomaly detection model to obtain an order anomaly detection result; the order anomaly detection model is obtained through neural network training through a historical order data set.
Optionally, the order anomaly detection model is obtained by training a neural network through a historical order data set, specifically: acquiring order key data of a plurality of historical large promotion nodes, labeling abnormal orders of the order key data of each historical large promotion node, and constructing a historical order data set according to the labeled order key data of each historical large promotion node; and inputting the historical order data set into the initial order anomaly detection model for training of the neural network, and stopping training of the neural network when a preset training ending condition is met, so as to obtain the order anomaly detection model.
In this embodiment, when order key data of a plurality of large historic promotion nodes are obtained, orders of the large historic promotion nodes are obtained from a background management database, order lists of the large historic promotion nodes are screened and removed to obtain order transaction data of the large historic promotion nodes, then the order transaction data of the large historic promotion nodes are calculated through order association degree, and the order key data of the large historic promotion nodes are extracted, wherein a specific implementation process is the same as that in steps 101-102, and details are omitted. When the abnormal order is marked, whether the order is abnormal or not can be judged according to human experience, and the abnormal order is marked. The initial order anomaly detection model may be at least one or a combination of a CNN network model, RNN network model, LTSM network model, DNN network model, or other such as random forest algorithm model.
By implementing the embodiment of the application, the neural network model can make feasible and good-effect analysis on a large-scale data source in a relatively short time, and has stronger robustness and fault tolerance. And the neural network prediction is carried out based on the order key data to determine an abnormal order to the current large promotion node, so that the prediction accuracy and the detection speed are effectively improved.
Step 104: and extracting all abnormal order users according to the order abnormality detection result, counting the historical order information of all abnormal order users to obtain commodity purchase data of all abnormal order users, and screening all abnormal order users according to the commodity purchase data of all abnormal order users to obtain the final abnormal order users.
Optionally, screening all abnormal order users according to commodity purchase data of all abnormal order users to obtain final abnormal order users, specifically: extracting the commodity quantity and commodity names of all abnormal order users according to commodity purchase data of all abnormal order users; according to the commodity quantity of the current abnormal order users, the commodity names of the current abnormal order users are ordered in a reverse order to obtain a current commodity purchase ordering table, and according to the preset quantity condition, the current commodity purchase ordering table is subjected to data rejection to obtain abnormal commodity data corresponding to the current abnormal order users; the commodity names and the commodity quantity in the current commodity purchase ordering list are used as the header; counting the abnormal commodity data corresponding to each abnormal order user, and clustering commodity names in the abnormal commodity data corresponding to each abnormal order user to obtain commodity categories of each abnormal order user; counting the total number of commodities corresponding to commodity categories of each abnormal order user, screening out abnormal order users meeting preset abnormal user conditions from all the abnormal order users according to the commodity categories of each abnormal order user and the total number of commodities corresponding to commodity categories, and obtaining a final abnormal order user; the preset abnormal user condition is that the commodity category is larger than a first preset value, and the total number of commodities corresponding to the commodity category is larger than a second preset value.
By implementing the embodiment of the application, whether the user really buys the abnormal user such as the cattle or the professional dummy is further judged according to the historical order of the abnormal user through the historical order data, so that the normal purchase of the consumer is avoided from being accidentally injured, and the accuracy of abnormal order detection is improved.
Step 105: and carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user.
Optionally, step 105 specifically includes: matching the order key data of the large promotion node with a final abnormal order user to obtain abnormal order information; marking an abnormal order according to the abnormal order information, generating abnormal early warning information according to the abnormal order, and sending the abnormal early warning information to a corresponding background management account so that the marked abnormal order is displayed in the corresponding background management account.
According to the embodiment of the application, the promotion time of the large promotion node is divided into a plurality of promotion time periods, the order list of each promotion time period is screened and removed, and order transaction data of each promotion time period is obtained; according to the key information and the order transaction data of each promotion period, extracting order key data of each promotion period through order association degree calculation, and summarizing the corresponding order key data of each promotion period to obtain order key data of a large promotion node; inputting order key data of the large promotion node into an order anomaly detection model to obtain an order anomaly detection result; according to the order anomaly detection result, all the anomaly order users are extracted, the historical order information of all the anomaly order users is counted to obtain commodity purchase data of all the anomaly order users, and according to the commodity purchase data of all the anomaly order users, all the anomaly order users are screened to obtain final anomaly order users; and carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user. And (3) selecting order key data placed by the large promotion node through order screening and removing and order association degree calculation, and predicting based on the order key data, so that the detection precision of order anomaly detection is improved. And then, according to the order abnormality detection result and the history order information of the abnormal order user, further analyzing and determining the final abnormal order user, avoiding the accidental injury to the normal user and improving the accuracy of abnormal order detection. Finally, the order key data of the large promotion node and the final abnormal order users are combined to perform abnormal order early warning, so that abnormal orders of the large promotion node are accurately found, and the rights and interests of consumers and merchants are effectively maintained.
Example two
Correspondingly, referring to fig. 3, fig. 3 is a schematic structural diagram of a second embodiment of an abnormal order detection and early warning system provided by the application. As shown in fig. 3, the abnormal order detection and early warning system includes an order data screening module 301, an associated key order extraction module 302, an abnormal order detection module 303, an abnormal user confirmation module 304, and an abnormal order early warning module 305;
the order data screening module 301 is configured to divide the promotion time of the large promotion node into time slots to obtain a plurality of promotion time slots, and screen and reject the order list of each promotion time slot to obtain order transaction data of each promotion time slot; the order list of each promotion period is obtained by arranging the orders of each promotion period in a list form according to the orders of each promotion period obtained in a background management database;
the related key order extraction module 302 is configured to extract order key data of each promotion period according to the key information and the order transaction data of each promotion period through order related degree calculation, and aggregate corresponding order key data of each promotion period to obtain order key data of a large promotion node; the key information comprises a receiving address, a user ID, user consumption habits and preferences, a trade commodity price and commodity trade quantity;
the abnormal order detection module 303 is configured to input order key data of the large promotion node into an order abnormality detection model, so as to obtain an order abnormality detection result; the order anomaly detection model is obtained through neural network training through a historical order data set;
the abnormal user confirmation module 304 is configured to extract all abnormal order users according to the abnormal detection result of the order, count historical order information of all abnormal order users to obtain commodity purchase data of all abnormal order users, and screen all abnormal order users according to the commodity purchase data of all abnormal order users to obtain final abnormal order users;
the abnormal order early warning module 305 is configured to perform abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user.
The abnormal order detection and early warning system can implement the abnormal order detection and early warning method in the method embodiment. The options in the method embodiments described above are also applicable to this embodiment and will not be described in detail here. The rest of the embodiments of the present application may refer to the content of the above method embodiments, and in this embodiment, no further description is given.
In addition, the embodiment of the application also provides a computer device, which comprises a processor and a memory, wherein the memory is used for storing a computer program, and the computer program is executed by the processor to realize the steps in any of the method embodiments.
The embodiment of the application also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the steps in any of the method embodiments when being executed by a processor.
The foregoing embodiments have been provided for the purpose of illustrating the general principles of the present application, and are not to be construed as limiting the scope of the application. It should be noted that any modifications, equivalent substitutions, improvements, etc. made by those skilled in the art without departing from the spirit and principles of the present application are intended to be included in the scope of the present application.
Claims (10)
1. The abnormal order detection and early warning method is characterized by comprising the following steps of:
dividing promotion time of a large promotion node in time intervals to obtain a plurality of promotion time intervals, screening and removing order lists of the promotion time intervals to obtain order transaction data of the promotion time intervals; the order list of each promotion period is obtained by arranging the orders of each promotion period in a list form according to the orders of each promotion period obtained in a background management database;
according to the key information and the order transaction data of each promotion period, extracting order key data of each promotion period through order association degree calculation, and summarizing the corresponding order key data of each promotion period to obtain order key data of a large promotion node; the key information comprises a receiving address, a user ID, user consumption habits and preferences, transaction commodity prices and commodity transaction quantity;
inputting order key data of the large promotion node into an order anomaly detection model to obtain an order anomaly detection result; the order anomaly detection model is obtained through neural network training through a historical order data set;
extracting all abnormal order users according to the order abnormality detection result, counting the historical order information of all abnormal order users to obtain commodity purchase data of all abnormal order users, and screening all abnormal order users according to the commodity purchase data of all abnormal order users to obtain final abnormal order users;
and carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user.
2. The method for detecting and warning abnormal orders according to claim 1, wherein the step of screening and removing the order list of each promotion period to obtain the order transaction data of each promotion period comprises the following steps:
taking order basic transaction information as a header, arranging order lists in a current promotion period to obtain each row of the current order list, and assigning indexes of each row of the current order list to obtain the row index of the current order list;
searching the data meeting the preset screening conditions in the current order list to obtain reject data; the preset screening conditions comprise that the list element value is null, the order is unpaid, the invalid order is invalid, and the order is cancelled;
searching a row index corresponding to the reject data through a preset searching function, and obtaining data to be deleted according to the row index corresponding to the reject data and the current order list;
and deleting the data to be deleted from the current order list through a preset deleting function, and obtaining the order transaction data of the current promotion period according to the deleted current order list.
3. The method for detecting and warning abnormal orders according to claim 2, wherein the step of extracting the order key data of each promotion period by calculating the order association degree according to the key information and the order transaction data of each promotion period is specifically as follows:
acquiring user information data and commodity transaction data in the current order transaction data according to the order transaction data of the current promotion number segment;
performing iterative operation processing on the user information data and the commodity transaction data to generate an order association probability map;
determining a correlation degree value between orders in the current order transaction data according to the order correlation probability map, and extracting orders meeting a preset probability threshold to obtain correlation orders;
carrying out relevance screening on the order transaction data of the current promotion period according to the relevance order to obtain relevance order transaction data;
and carrying out data extraction on the related order transaction data according to the key information to obtain the order key data of the current promotion period.
4. The method for detecting and warning abnormal orders according to claim 3, wherein the iterative operation processing is performed on the user information data and the commodity transaction data to generate an order association probability map, specifically:
adding edges between the user information data and the commodity transaction data, and establishing an initial mapping relation to form an initial association relation diagram;
judging whether an association relationship taking the commodity transaction data as a public end exists between the user information data or not based on the initial association relationship graph, and if so, establishing an adjacent relationship between the user information data and the commodity transaction data;
according to the iterative adjacency relation, acquiring the association probability among the user information data, the association probability among the commodity transaction data and the association probability among the user information data and the commodity transaction data;
and according to the adjacent relations and the association probabilities, adjusting the initial association relation graph to form the order association probability graph.
5. The abnormal order detection and early warning method as set forth in claim 1, wherein the order abnormal detection model is obtained by training a neural network through a historical order data set, and specifically comprises:
acquiring order key data of a plurality of historical large promotion nodes, labeling abnormal orders of the order key data of each historical large promotion node, and constructing a historical order data set according to the labeled order key data of each historical large promotion node;
and inputting the historical order data set into an initial order abnormality detection model for training of the neural network, and stopping training of the neural network when a preset training ending condition is met, so as to obtain the order abnormality detection model.
6. The method for detecting and warning abnormal orders according to claim 1, wherein the screening of all abnormal order users according to the commodity purchase data of all abnormal order users, comprises the following specific steps:
extracting the commodity quantity and commodity names of each abnormal order user according to the commodity purchase data of all abnormal order users;
sorting the commodity names of the current abnormal order users in a reverse order according to the commodity numbers of the current abnormal order users to obtain a current commodity purchase sorting table, and removing data of the current commodity purchase sorting table according to preset number conditions to obtain abnormal commodity data corresponding to the current abnormal order users; wherein the commodity names and the commodity numbers in the current commodity purchase ordering table are used as the header;
counting the abnormal commodity data corresponding to each abnormal order user, and clustering commodity categories of commodity names in the abnormal commodity data corresponding to each abnormal order user to obtain commodity categories of each abnormal order user;
counting the total number of commodities corresponding to commodity categories of each abnormal order user, screening out abnormal order users meeting preset abnormal user conditions from all the abnormal order users according to the commodity category of each abnormal order user and the total number of commodities corresponding to the commodity category, and obtaining a final abnormal order user; the preset abnormal user condition is that the commodity category is larger than a first preset value and the total number of commodities corresponding to the commodity category is larger than a second preset value.
7. The abnormal order detection and early warning method according to claim 1, wherein the abnormal order early warning is performed according to the order key data of the large promotion node and the final abnormal order user, specifically:
matching the order key data of the large promotion node with the final abnormal order user to obtain abnormal order information;
marking an abnormal order according to the abnormal order information, generating abnormal early warning information according to the abnormal order, and sending the abnormal early warning information to a corresponding background management account so that the marked abnormal order is displayed in the corresponding background management account.
8. An abnormal order detection and early warning system is characterized by comprising: the system comprises an order data screening module, a related key order extraction module, an abnormal order detection module, an abnormal user confirmation module and an abnormal order early warning module;
the order data screening module is used for dividing the promotion time of the large promotion node in time intervals to obtain a plurality of promotion time intervals, screening and removing the order list of each promotion time interval to obtain order transaction data of each promotion time interval; the order list of each promotion period is obtained by arranging the orders of each promotion period in a list form according to the orders of each promotion period obtained in a background management database;
the related key order extraction module is used for extracting the order key data of each promotion period through order related degree calculation according to the key information and the order transaction data of each promotion period, and summarizing the corresponding order key data of each promotion period to obtain the order key data of the large promotion node; the key information comprises a receiving address, a user ID, user consumption habits and preferences, transaction commodity prices and commodity transaction quantity;
the abnormal order detection module is used for inputting order key data of the large promotion node into an order abnormal detection model to obtain an order abnormal detection result; the order anomaly detection model is obtained through neural network training through a historical order data set;
the abnormal user confirmation module is used for extracting all abnormal order users according to the order abnormal detection result, counting the historical order information of all abnormal order users to obtain commodity purchase data of all abnormal order users, and screening all abnormal order users according to the commodity purchase data of all abnormal order users to obtain final abnormal order users;
and the abnormal order early warning module is used for carrying out abnormal order early warning according to the order key data of the large promotion node and the final abnormal order user.
9. A computer device comprising a processor and a memory for storing a computer program which when executed by the processor implements the method of detecting and pre-warning an abnormal order according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements the abnormal order detection and early warning method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310715693.1A CN116934418B (en) | 2023-06-15 | 2023-06-15 | Abnormal order detection and early warning method, system, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310715693.1A CN116934418B (en) | 2023-06-15 | 2023-06-15 | Abnormal order detection and early warning method, system, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116934418A true CN116934418A (en) | 2023-10-24 |
| CN116934418B CN116934418B (en) | 2024-03-19 |
Family
ID=88390358
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310715693.1A Active CN116934418B (en) | 2023-06-15 | 2023-06-15 | Abnormal order detection and early warning method, system, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116934418B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017116311A1 (en) * | 2015-12-30 | 2017-07-06 | Agency For Science, Technology And Research | Method of detecting fraud in procurement and system thereof |
| CN107798571A (en) * | 2016-08-31 | 2018-03-13 | 阿里巴巴集团控股有限公司 | Identifying system, the method and device of malice address/malice order |
| CN109711955A (en) * | 2019-02-18 | 2019-05-03 | 杭州跨境邦信息技术有限公司 | Difference based on current order comments method for early warning, system, blacklist library method for building up |
| CN110648195A (en) * | 2019-08-28 | 2020-01-03 | 苏宁云计算有限公司 | User identification method and device and computer equipment |
| CN110874778A (en) * | 2018-08-31 | 2020-03-10 | 阿里巴巴集团控股有限公司 | Abnormal order detection method and device |
| CN112308352A (en) * | 2019-07-30 | 2021-02-02 | 阿里巴巴集团控股有限公司 | Order identification method and device and electronic equipment |
| CN112825175A (en) * | 2019-11-20 | 2021-05-21 | 顺丰科技有限公司 | Client abnormity early warning method, device and equipment |
| CN113469789A (en) * | 2021-07-01 | 2021-10-01 | 易纳购科技(北京)有限公司 | Abnormal order detection method and device and computer equipment |
| CN113516529A (en) * | 2021-04-26 | 2021-10-19 | 新东方教育科技集团有限公司 | Abnormal order determining method and device, storage medium and electronic equipment |
| CN114331593A (en) * | 2021-12-14 | 2022-04-12 | 深圳壹账通智能科技有限公司 | Malicious order identification and processing method, device, equipment and storage medium |
| CN114742477A (en) * | 2022-06-09 | 2022-07-12 | 未来地图(深圳)智能科技有限公司 | Enterprise order data processing method, device, equipment and storage medium |
-
2023
- 2023-06-15 CN CN202310715693.1A patent/CN116934418B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017116311A1 (en) * | 2015-12-30 | 2017-07-06 | Agency For Science, Technology And Research | Method of detecting fraud in procurement and system thereof |
| CN107798571A (en) * | 2016-08-31 | 2018-03-13 | 阿里巴巴集团控股有限公司 | Identifying system, the method and device of malice address/malice order |
| CN110874778A (en) * | 2018-08-31 | 2020-03-10 | 阿里巴巴集团控股有限公司 | Abnormal order detection method and device |
| CN109711955A (en) * | 2019-02-18 | 2019-05-03 | 杭州跨境邦信息技术有限公司 | Difference based on current order comments method for early warning, system, blacklist library method for building up |
| CN112308352A (en) * | 2019-07-30 | 2021-02-02 | 阿里巴巴集团控股有限公司 | Order identification method and device and electronic equipment |
| CN110648195A (en) * | 2019-08-28 | 2020-01-03 | 苏宁云计算有限公司 | User identification method and device and computer equipment |
| CN112825175A (en) * | 2019-11-20 | 2021-05-21 | 顺丰科技有限公司 | Client abnormity early warning method, device and equipment |
| CN113516529A (en) * | 2021-04-26 | 2021-10-19 | 新东方教育科技集团有限公司 | Abnormal order determining method and device, storage medium and electronic equipment |
| CN113469789A (en) * | 2021-07-01 | 2021-10-01 | 易纳购科技(北京)有限公司 | Abnormal order detection method and device and computer equipment |
| CN114331593A (en) * | 2021-12-14 | 2022-04-12 | 深圳壹账通智能科技有限公司 | Malicious order identification and processing method, device, equipment and storage medium |
| CN114742477A (en) * | 2022-06-09 | 2022-07-12 | 未来地图(深圳)智能科技有限公司 | Enterprise order data processing method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116934418B (en) | 2024-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222272B (en) | Potential customer mining and recommending method | |
| CN111523976B (en) | Commodity recommendation method and device, electronic equipment and storage medium | |
| CN107341716B (en) | Malicious order identification method and device and electronic equipment | |
| CN109711955B (en) | Poor evaluation early warning method and system based on current order and blacklist base establishment method | |
| US20180300748A1 (en) | Technologies for granular attribution of value to conversion events in multistage conversion processes | |
| CN107153656B (en) | Information searching method and device | |
| CN111325609A (en) | Commodity recommendation list determining method and device, electronic equipment and storage medium | |
| US8799193B2 (en) | Method for training and using a classification model with association rule models | |
| CN111666275B (en) | Data processing method and device, electronic equipment and storage medium | |
| CN111241367A (en) | Method and system for supervising network catering platform based on custom rule | |
| CN112001754A (en) | User portrait generation method, device, equipment and computer readable medium | |
| CN110766428A (en) | Data value evaluation system and method | |
| JP2019215778A (en) | Service provision system, advertisement related service provision system, user side facility and user side advertisement facility | |
| CN112396433A (en) | Method and system for identifying false commodity comments based on behavior of person to be evaluated | |
| CN111861679A (en) | Commodity recommendation method based on artificial intelligence | |
| CN114860941A (en) | Industry data management method and system based on data brain | |
| CN113128218A (en) | Key field extraction method and device for bidding information | |
| TWI747290B (en) | Customer grouping service customized marketing method and system | |
| CN111222923B (en) | Method and device for judging potential clients, electronic equipment and storage medium | |
| CN116934418B (en) | Abnormal order detection and early warning method, system, equipment and storage medium | |
| CN111311381A (en) | Commodity recommendation method and system | |
| CN111091416A (en) | Method and device for predicting probability of hotel purchase robot | |
| CN113837843A (en) | Product recommendation method, device, medium and electronic equipment | |
| CN113420018A (en) | User behavior data analysis method, device, equipment and storage medium | |
| JP4231130B2 (en) | Promising customer extraction device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |