CN117495512B - Order data management method, device, equipment and storage medium - Google Patents
Order data management method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN117495512B CN117495512B CN202311850739.7A CN202311850739A CN117495512B CN 117495512 B CN117495512 B CN 117495512B CN 202311850739 A CN202311850739 A CN 202311850739A CN 117495512 B CN117495512 B CN 117495512B
- Authority
- CN
- China
- Prior art keywords
- order
- classification
- data
- node
- target
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 89
- 238000003860 storage Methods 0.000 title claims abstract description 15
- 238000013523 data management Methods 0.000 title claims description 20
- 238000012545 processing Methods 0.000 claims abstract description 244
- 239000013598 vector Substances 0.000 claims abstract description 78
- 238000004458 analytical method Methods 0.000 claims abstract description 47
- 238000011156 evaluation Methods 0.000 claims abstract description 36
- 238000013145 classification model Methods 0.000 claims abstract description 18
- 230000006870 function Effects 0.000 claims description 18
- 230000001364 causal effect Effects 0.000 claims description 16
- 238000005457 optimization Methods 0.000 claims description 8
- 241000288105 Grus Species 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 7
- 125000004122 cyclic group Chemical group 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000013211 curve analysis Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 description 44
- 238000012384 transportation and delivery Methods 0.000 description 27
- 230000000875 corresponding effect Effects 0.000 description 26
- 238000007726 management method Methods 0.000 description 15
- 230000006399 behavior Effects 0.000 description 11
- 238000002360 preparation method Methods 0.000 description 8
- 238000009826 distribution Methods 0.000 description 7
- 238000000605 extraction Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000003058 natural language processing Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000010921 in-depth analysis Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011867 re-evaluation Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000010200 validation analysis 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
- G06F18/24323—Tree-organised classifiers
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to the technical field of data processing, and discloses a method, a device, equipment and a storage medium for managing order data. The method comprises the following steps: acquiring personal information of a user and inquiring historical orders to acquire historical order data of the user; acquiring a plurality of order processing state nodes and constructing a target order data classification tree; performing data classification to obtain a plurality of order classification data; performing order feature analysis to obtain an order feature set, and performing feature weight analysis and vector coding to obtain a historical order evaluation vector; the user classification is carried out through the user classification model, the target user type is obtained, and an order processing result is generated.
Description
Technical Field
The present invention relates to the field of data processing, and in particular, to a method, an apparatus, a device, and a storage medium for managing order data.
Background
With the dramatic increase in online transaction volumes, merchants are faced with a large amount of complex and diverse order data that includes not only basic transaction information, but also multiple dimensions related to user behavior, preferences, and transaction status.
The traditional order management method often cannot effectively process the complex data, so that the order processing efficiency is low, and the rapidly-developed market demand cannot be met. Therefore, how to classify users by using the deep learning model can better understand the demands of clients and forecast market trends, thereby optimizing order processing strategies and improving client satisfaction.
Disclosure of Invention
The application provides a management method, device, equipment and storage medium of order data, which are used for improving the classification efficiency and accuracy of historical order data and further improving the accuracy of order processing.
In a first aspect, the present application provides a method for managing order data, where the method for managing order data includes:
receiving an order processing request sent by a client, and carrying out request analysis on the order processing request to obtain a target order placing user and a to-be-processed order corresponding to the target order placing user;
acquiring user personal information of the target ordering user, and inquiring historical order of the user personal information through an order database to obtain user historical order data;
acquiring a plurality of order processing state nodes, and constructing a corresponding target order data classification tree according to the plurality of order processing state nodes;
Based on the target order data classification tree, carrying out data classification on the historical order data of the user to obtain a plurality of order classification data;
performing order feature analysis on the plurality of order classification data respectively to obtain an order feature set of each order classification data, and performing feature weight analysis and vector coding on the order feature set of each order classification data respectively to obtain a historical order evaluation vector;
and inputting the historical order evaluation vector into a preset user classification model for user classification to obtain the target user type of the target order placing user, and generating an order processing result of the to-be-processed order according to the target user type.
In a second aspect, the present application provides an order data management apparatus, including:
the receiving module is used for receiving an order processing request sent by a client and carrying out request analysis on the order processing request to obtain a target order placing user and a to-be-processed order corresponding to the target order placing user;
the inquiry module is used for acquiring the user personal information of the target ordering user, and carrying out historical order inquiry on the user personal information through an order database to obtain user historical order data;
The construction module is used for acquiring a plurality of order processing state nodes and constructing a corresponding target order data classification tree according to the plurality of order processing state nodes;
the classification module is used for carrying out data classification on the historical order data of the user based on the target order data classification tree to obtain a plurality of order classification data;
the coding module is used for respectively carrying out order feature analysis on the plurality of order classification data to obtain an order feature set of each order classification data, and respectively carrying out feature weight analysis and vector coding on the order feature set of each order classification data to obtain a historical order evaluation vector;
the generation module is used for inputting the historical order evaluation vector into a preset user classification model to classify the users, obtaining the target user type of the target ordering user, and generating an order processing result of the to-be-processed order according to the target user type.
A third aspect of the present application provides an order data management apparatus, including: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the order data management device to perform the order data management method described above.
A fourth aspect of the present application provides a computer-readable storage medium having instructions stored therein, which when run on a computer, cause the computer to perform the above-described order data management method.
According to the technical scheme, the system can more accurately identify the intention and the requirement of the user by receiving and analyzing the order processing request sent by the client, so that the processing accuracy is improved. Meanwhile, the order data is classified and analyzed by using the Bayesian network, so that the data processing efficiency can be remarkably improved. The personal information of the user is acquired and analyzed in combination with the historical order data, so that the purchasing preference and the behavior mode of the user can be better understood. This helps to provide more personalized services such as recommendation systems and personalized order processing strategies, thereby improving user satisfaction. The constructed order data classification tree provides a dynamic and flexible method of processing order data. The method can be dynamically adjusted according to different order processing state nodes, and improves adaptability to various order types and states. The user classification model constructed by the GRU and sigmoid functions can effectively process and analyze complex order data. The deep learning method can reveal a deep mode behind the user behavior, and provides more accurate support for order processing and user service. Through the generation of the historical order evaluation vector and the user classification, the management method can provide more comprehensive and deep data support for a decision maker. The method is beneficial to optimizing decisions in aspects of inventory management, marketing strategies, customer relationship management and the like, improves the classification efficiency and accuracy of historical order data, and further improves the accuracy of order processing.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained based on these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of an embodiment of a method for managing order data according to an embodiment of the present application;
fig. 2 is a schematic diagram of an embodiment of an order data management apparatus according to an embodiment of the present application.
Detailed Description
The embodiment of the application provides a management method, device and equipment of order data and a storage medium. The terms "first," "second," "third," "fourth" and the like in the description and in the claims of this application and in the above-described figures, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For ease of understanding, the following describes a specific flow of an embodiment of the present application, referring to fig. 1, and one embodiment of a method for managing order data in an embodiment of the present application includes:
step S101, receiving an order processing request sent by a client, and carrying out request analysis on the order processing request to obtain a target order placing user and a to-be-processed order corresponding to the target order placing user;
it is to be understood that the execution body of the present application may be an order data management device, and may also be a terminal or a server, which is not limited herein. The embodiment of the present application will be described by taking a server as an execution body.
Specifically, first, the server establishes a stable communication connection with the client, which is typically implemented through the internet, and may employ a common communication protocol such as HTTP or WebSocket. These protocols not only provide a standardized data transmission scheme, but also ensure the security of the data transmission process by techniques such as SSL/TLS encryption. The server then receives an order processing request from the client. Upon receiving the request, the server performs request parsing, and the server is able to recognize and process input data in different formats, such as JSON or XML. The server will then perform content parsing, which typically involves Natural Language Processing (NLP) techniques, particularly when the order information is provided in natural language form. NLP technology can help servers understand and extract key information such as user identity, order content and quantity, etc. In addition, the server also needs to verify the user identity and rights to ensure the legitimacy and security of the request. And finally, the server extracts the identification information of the target order subscriber and the detailed content of the order to be processed according to the analysis result. This typically involves querying a back-end database or order management system to obtain the user's detailed information and related order data. Database queries need to be efficient and accurate, which requires reasonable database design, optimization of query algorithms, and good data indexing and caching strategies.
Step S102, acquiring user personal information of a target ordering user, and inquiring historical order of the user personal information through an order database to obtain historical order data of the user;
specifically, first, the server needs to obtain personal information of the target subscriber, which usually involves subscriber identity recognition and information retrieval. When a user submits an order, their identity is typically verified by login credentials, such as a user name and password. Once verified, the server may access personal information associated with the user account, including the user's name, address, contact, and historical purchasing habits. Next, the server uses the user's identity information to make a historical order query in the order database. Typically, the order database will be designed as a correlation table containing user information and order information, both of which are correlated by user ID. In order to optimize the query efficiency, the database typically indexes key fields such as user ID, order date, etc., thereby increasing the search speed. When a server issues a query request, a database management system (DBMS) searches all order records associated with the user based on the provided user ID.
Step S103, acquiring a plurality of order processing state nodes, and constructing a corresponding target order data classification tree according to the plurality of order processing state nodes;
specifically, first, the server identifies and obtains a plurality of order processing status nodes representing different stages of order processing, such as order submission, payment, distribution, and the like. And after the state nodes are acquired, performing node causal relationship analysis by using a preset Bayesian network. The Bayesian network is an effective statistical method, and can reveal the dependence and influence relation among nodes in different states through probabilistic reasoning so as to obtain causal relation information among the nodes. The server then hierarchies these order processing status nodes to form two main node hierarchies: a first node hierarchy and a second node hierarchy. The first node hierarchy includes primary processing states such as order receipt and preliminary reviews, while the second node hierarchy includes higher level processing states such as order validation and distribution scheduling. This hierarchical division helps to more effectively manage and monitor the order processing flow. Next, the server will perform tree structure conversion on the state nodes in the two node hierarchy, constructing an initial order data classification tree. This tree structure will reflect the entire flow of order processing, with the final completion received from the beginning. The construction of the classification tree not only facilitates understanding and visualizing the entire order processing flow, but also provides a basis for subsequent data processing and task allocation. The server then receives pre-trained order data via the initial order data classification tree and generates a first order classification task. This task will be assigned to a first and second hierarchy of nodes in the classification tree, where each state node will handle the corresponding task according to its particular responsibilities and functions. This process involves not only data processing and analysis, but also manual auditing and decision making. The server then calculates a first order data classification status attribute for each status node in the first and second node hierarchies. These attributes include the urgency of the order, customer preferences, complexity of the order, etc. Based on these attributes, the server will perform a node traversal match and an order data classification status attribute update in order to obtain a second order data classification status attribute. This process is dynamic and aims to continually adjust and optimize the order processing flow to improve efficiency and accuracy. Next, the server configures a node update policy of the initial order data sort tree based on the second order data sort status attribute. The server will adjust the structure and node attributes of the classification tree to better accommodate the actual order processing requirements. Meanwhile, the server monitors tasks of the classification tree to judge whether a second order classification task exists. If the presence of a second order classification task is detected, the server performs a loop-through classification of the task by a status node in the first and second node hierarchy to obtain a final order classification result. This process involves complex data processing algorithms and manual decisions in order to ensure that each order is properly classified and efficiently processed. Finally, the server further optimizes the initial order data classification tree according to the order classification result to obtain a corresponding target order data classification tree. This optimization process is continuous, allowing the server to constantly learn and adapt to new order processing patterns and requirements, thereby improving the efficiency and accuracy of the overall order management process. In this way, the server is able to efficiently process and manage complex order data while ensuring the efficiency and accuracy of order processing.
First, a first order data classification status attribute is calculated for a plurality of first processing status nodes in a first node hierarchy and a plurality of second processing status nodes in a second node hierarchy. These attributes include urgency of the order, processing time, customer preferences, etc., reflecting the characteristics and requirements of each order at different processing stages. Next, the server obtains first node attribute flag information for each of the first processing state nodes in the first node hierarchy and second node attribute flag information for each of the second processing state nodes in the second node hierarchy. The marking information may be predefined, such as order type, customer category, processing priority, etc., or may be derived based on historical data analysis, such as common problems, processing efficiency, etc. Next, the server calculates an order data sort status attribute for each first processing status node based on the first order sort task, the first order data sort status attribute, and the first node attribute marking information. By matching the specific requirements of each order and the capabilities of each processing node, it is ensured that each order is assigned to the node that is most suitable for processing it. Similarly, the server will also calculate an order data sort status attribute for each second processing status node based on the order data sort status attribute and the second node attribute marking information for each first processing status node. This process takes into account the transfer and variation of orders in the process flow, ensuring that each order is properly and efficiently processed in the second processing stage. And finally, the server generates a second order data classification status attribute according to the order data classification status attribute of each of the first processing status node and the second processing status node. These attributes are the result of in-depth analysis and optimization of the overall order processing flow, reflecting changes in the status and requirements of the order throughout the processing. By the method, the server can process orders more accurately and efficiently, and ensure that each order is properly focused and processed in the whole processing process, so that the whole order processing flow is optimized, and the processing efficiency and the customer satisfaction are improved.
First, when the server detects that a second order classification task exists, a third processing state node corresponding to the task is located, and a target node level to which the node belongs is determined. The specialized nodes required to handle the particular task are identified based on the particular characteristics of the order, such as urgency, processing complexity, or customer priority. The server then obtains a first aggregate order data category status attribute for all order processing status nodes in the hierarchy. These attributes provide the server with a comprehensive view of the current processing state of the entire node hierarchy, including key indicators of order processing speed, efficiency, error rate, etc. The server judges whether the target node level meets the first task classification rule by utilizing the information, so that a target judgment result is obtained. This determination ensures that order classification tasks are assigned to the most appropriate node hierarchy, thereby optimizing processing efficiency and quality. Based on this target determination, the server performs a loop-through classification for the second order classification task. If the target judgment result shows that the current node level does not meet the first task classification rule, the server feeds back the task to the target node level of the higher or lower level to which the task belongs. This flexible task redistribution mechanism enables the server to adjust task distribution according to real-time conditions to cope with changing order processing demands. And finally, the server acquires the second total order data classification status attribute corresponding to the new target node level, and judges whether the level meets the second task classification rule or not through the attribute again. If the judgment result is negative, namely the node level does not meet the task classification requirement, the server feeds back the second order classification task to the initial order data classification tree. On this more extensive level, the server reclassifies the tasks through the initial order data classification tree, ensuring that each order can be processed at the most appropriate node.
Step S104, based on the target order data classification tree, carrying out data classification on the historical order data of the user to obtain a plurality of order classification data;
specifically, first, order traversal analysis is performed on historical order data of a user based on a first node hierarchy in a target order data classification tree, so as to obtain a plurality of first classification data. The server screens and analyzes key attributes in the historical order, such as order generation time, processing state, payment state, etc., to obtain a plurality of first classification data. These first category data include subsets of orders categorized by time, status, or other criteria, which provide a basis for further analysis. The server then continues the traversal analysis of the user's historical order data based on these first classification data and the second level of nodes in the target order data classification tree. This step involves more complex analysis of data attributes such as payment means of orders, distribution, customer feedback, etc., to generate a plurality of second classification data. These second classification data provide a further subdivision of the order data, providing more rich and detailed order characteristic information. Next, data comparison and data fusion are performed on the plurality of first classification data and the plurality of second classification data. More valuable information is extracted by comparing common and unique features in different classification data and integrating and fusing related data. For example, the server may compare the order payment duration data with the unpaid order data to identify potential payment delay patterns; or combine the return order data with other classification data to analyze the cause and pattern of the return. Through this comparison and fusion, the server is able to obtain a plurality of more comprehensive and deep order classification data.
Step 105, respectively performing order feature analysis on the plurality of order classification data to obtain an order feature set of each order classification data, and respectively performing feature weight analysis and vector coding on the order feature set of each order classification data to obtain a historical order evaluation vector;
specifically, first, the server performs curve analysis on order payment duration data, unpaid order data, and return order data in the plurality of order classification data, respectively. Each type of data is plotted as a time series curve. And then, extracting and screening characteristic points of the order classification curves to obtain a characteristic set of each order classification data. Feature point extraction is focused on key turning points, extreme points or trend change points on the curve, which represent key characteristics of the data. The screening process selects the points with the most representativeness and influence on order classification from the extracted characteristic points. The server then calculates a feature mean and a feature standard deviation for each order feature set. The feature mean data provides an average representation of each category data, while the feature standard deviation data reflects the magnitude of the change in the data across different orders. The calculation of these two data provides important information for understanding the general trends and variability of each order category. And then, the server performs characteristic weight analysis on the characteristic mean value data and the characteristic standard deviation data to generate characteristic weight data of each order classification data. The server assigns a corresponding weight according to the degree of influence and importance of each feature on the order classification. These weight data help to more accurately reflect the importance of each feature in subsequent analysis. And then, the server performs weighting processing on the order feature set of each order classification data according to the feature weight data to obtain a weighted feature set. The influence of more important features in the feature set is enhanced through a weighting algorithm, so that the generated evaluation vector can more accurately reflect the key characteristics of order classification. Finally, the server vector encodes the weighted feature set, converts the weighted feature set into an initial order evaluation vector in a numerical form, and then splices the vectors to form a comprehensive historical order evaluation vector. This evaluation vector integrates the key features of the multiple order classification data, providing a comprehensive and accurate historical order performance overview.
And S106, inputting the historical order evaluation vector into a preset user classification model for user classification to obtain the target user type of the target order placing user, and generating an order processing result of the to-be-processed order according to the target user type.
Specifically, first, a history order evaluation vector is input into a preset user classification model. The model adopts a two-layer gating cycle unit (GRU) network structure and combines sigmoid functions to realize efficient classification. At the first layer of the model, a unidirectional GRU network consisting of 256 GRU units is responsible for preliminary feature extraction. This layer of GRU serves as a command layer, and aims to extract key hidden features from the input historical order evaluation vector, wherein the hidden features cover multi-dimensional information such as purchasing habits, preferences and payment behaviors of users. Each GRU unit is responsible for capturing a particular pattern in the time series data, thereby generating a first hidden feature vector that contains a high-level abstract representation of the historical order data. Next, in the second layer of the model, 256 sets of unidirectional GRU connections are included, each set corresponding to one GRU unit in the first layer, each set consisting of 16 GRU units. The design of this layer enables the model to analyze the first hidden feature vector more carefully and in depth, capturing more complex user behavior patterns and trends. The effect of this layer is to further refine and process the hidden features, generating a second hidden feature vector that is finer. Then, score calculation is performed on the second hidden feature vector by using a sigmoid function. The purpose of the Sigmoid function is here to map feature vectors to values within a fixed range, typically between 0 and 1, such mapping facilitating probability interpretation and classification thresholding. Through this step, a target score for the target subscriber is obtained, which score represents the probability that the subscriber belongs to a different subscriber type. And finally, the server performs user type matching according to the obtained target score. The calculated score is compared with a predefined user type threshold to determine which category the user belongs to. Once the user type is determined, the server generates order processing results for the corresponding pending orders based on this type.
In the embodiment of the application, the system can more accurately identify the intention and the demand of the user by receiving and analyzing the order processing request sent by the client, so that the processing accuracy is improved. Meanwhile, the order data is classified and analyzed by using the Bayesian network, so that the data processing efficiency can be remarkably improved. The personal information of the user is acquired and analyzed in combination with the historical order data, so that the purchasing preference and the behavior mode of the user can be better understood. This helps to provide more personalized services such as recommendation systems and personalized order processing strategies, thereby improving user satisfaction. The constructed order data classification tree provides a dynamic and flexible method of processing order data. The method can be dynamically adjusted according to different order processing state nodes, and improves adaptability to various order types and states. The user classification model constructed by the GRU and sigmoid functions can effectively process and analyze complex order data. The deep learning method can reveal a deep mode behind the user behavior, and provides more accurate support for order processing and user service. Through the generation of the historical order evaluation vector and the user classification, the management method can provide more comprehensive and deep data support for a decision maker. The method is beneficial to optimizing decisions in aspects of inventory management, marketing strategies, customer relationship management and the like, improves the classification efficiency and accuracy of historical order data, and further improves the accuracy of order processing.
In a specific embodiment, the process of executing step S103 may specifically include the following steps:
(1) Acquiring a plurality of order processing state nodes, and analyzing node causal relation of the plurality of order processing state nodes through a preset Bayesian network to obtain node causal relation information of the plurality of order processing state nodes;
(2) Carrying out node level division on a plurality of order processing state nodes according to the node causal relationship information to obtain two node levels, wherein the two node levels comprise a first node level and a second node level, the first node level comprises a plurality of first processing state nodes, and the second node level comprises a plurality of second processing state nodes;
(3) Performing tree structure conversion on a plurality of first processing state nodes in a first node hierarchy and a plurality of second processing state nodes in a second node hierarchy to obtain corresponding initial order data classification trees;
(4) Receiving pre-training order data through an initial order data classification tree, generating a first order classification task, and issuing the first order classification task to a plurality of first processing state nodes in a first node hierarchy and a plurality of second processing state nodes in a second node hierarchy;
(5) Calculating first order data classification state attributes of a plurality of first processing state nodes in a first node hierarchy and a plurality of second processing state nodes in a second node hierarchy, and carrying out node traversal matching and order data classification state attribute updating on a first order classification task according to the first order data classification state attributes to obtain second order data classification state attributes;
(6) Configuring a node updating strategy of an initial order data classification tree according to the second order data classification state attribute, executing the node updating strategy through the initial order data classification tree, performing task monitoring on the initial order data classification tree, and judging whether a second order classification task exists or not;
(7) If the order classification task exists, performing cyclic traversal classification on the second order classification task through a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy to obtain an order classification result;
(8) And carrying out order data classification tree optimization on the initial order data classification tree according to the order classification result to obtain a corresponding target order data classification tree.
Specifically, first, a plurality of order processing state nodes are obtained, and node causal relation analysis is performed on the plurality of order processing state nodes through a preset Bayesian network, so that node causal relation information of the plurality of order processing state nodes is obtained. A bayesian network is a graphical model for representing probabilistic relationships between variables, adapted to discover causal relationships between different order states. For example, the order status node includes order submission, payment, delivery preparation, in-delivery, order completion, and the like. By collecting historical data and applying a bayesian network, the server reveals dependencies between these states, such as payment states directly affecting the start of delivery preparation. Then, the server performs hierarchical division on the order processing state nodes according to the causal relationship information of the nodes to form two node hierarchies. This partitioning is based on causal relationship strength and logical order among nodes. For example, a first node hierarchy includes earlier processing states such as order submission and payment, and a second node hierarchy includes subsequent processing states such as dispatch preparation and dispatch. Such hierarchical partitioning helps the server to more organized process orders. Then, the server performs tree structure conversion on the state nodes in the two node levels, and an initial order data classification tree is constructed. In this tree, each node represents a processing state, and the connections between nodes reflect the flow of order processing. Thereafter, the server receives pre-trained order data through this classification tree and generates a first order classification task. These tasks are issued to the processing state nodes of the first and second node hierarchies. In this process, each node processes the received tasks according to its particular function. For example, the payment node checks the payment status of the order, and the delivery node updates the delivery status of the order. Next, the server calculates first order data classification status attributes for the process status nodes in the first and second node hierarchies and performs node traversal matching and attribute updating for the first order classification task based on the attributes. This step involves analyzing the status and characteristics of the order currently being processed by each node, such as the urgency of the order, customer preferences, etc., to more accurately match and process the tasks. Then, the server configures a node update strategy of the initial order data classification tree according to the obtained second order data classification status attribute. This includes adjusting the connections between nodes, or changing the priority of node processing tasks, to ensure efficient and accurate overall order processing flow. If a second order classification task exists, the server performs a loop-through classification of the task by the processing state nodes in the first and second node hierarchies. In this process, the server integrates the characteristics of each node with the status of the currently processed order to ensure that each order is properly classified and efficiently processed. And finally, optimizing the initial order data classification tree by the server according to the order classification result to form a final target order data classification tree. Such optimization is based on the results and efficiency of actual order processing, with the aim of improving the performance of the overall order processing flow. For example, suppose that the server finds, by analysis, that the processing time between the payment node and the delivery preparation node is too long. The server adds a new order verification node between the two nodes to ensure the accuracy of the payment information, thereby accelerating the processing speed of the whole order. In this way, the server constantly learns and adapts to optimize the overall order processing flow.
In a specific embodiment, the executing step calculates first order data classification status attributes of a plurality of first processing status nodes in the first node hierarchy and a plurality of second processing status nodes in the second node hierarchy, and performs node traversal matching and order data classification status attribute updating on the first order classification task according to the first order data classification status attributes, so as to obtain a second order data classification status attribute, which may specifically include the following steps:
(1) Calculating first order data classification status attributes of a plurality of first processing status nodes in a first node hierarchy and a plurality of second processing status nodes in a second node hierarchy;
(2) Acquiring first node attribute marking information of a plurality of first processing state nodes in a first node hierarchy, and simultaneously acquiring second node attribute marking information of a plurality of second processing state nodes in a second node hierarchy;
(3) Calculating the order data classification state attribute of each first processing state node according to the first order classification task, the first order data classification state attribute and the first node attribute marking information, and calculating the order data classification state attribute of each second processing state node according to the order data classification state attribute of each first processing state node and the second node attribute marking information;
(4) And generating a second order data classification status attribute according to the order data classification status attribute of each first processing status node and the order data classification status attribute of each second processing status node.
Specifically, first, the server calculates first order data classification status attributes of a plurality of first processing status nodes in a first node hierarchy and a plurality of second processing status nodes in a second node hierarchy. Thereby understanding the role of each status node in order processing and extracting key status attributes therefrom. For example, a first node hierarchy includes primary processing links such as order submission, user authentication, payment processing, etc., while a second node hierarchy involves subsequent processing links such as order distribution, distribution tracking, and order completion. In this step, the server analyzes the historical order data to determine key attributes of each processing link, such as the degree of urgency of the order of interest to the order submitting node, the account security level of interest to the user authentication node, the diversity and security of payment modes of interest to the payment processing node. Next, the server obtains first node attribute flag information for a plurality of first processing state nodes in a first node hierarchy and second node attribute flag information for a plurality of second processing state nodes in a second node hierarchy. Such attribute marking information is helpful in understanding the characteristics of each node. For example, for a first process state node, the attribute tags include order type (e.g., standard order, urgent order), user type (e.g., new user, old user), etc., while for a second process state node, the attribute tags include delivery mode (e.g., standard delivery, urgent delivery), delivery status (e.g., in-delivery, delivery completed), etc. Then, the server calculates the order data classification status attribute of each first processing status node according to the first order classification task, the first order data classification status attribute and the first node attribute marking information. The historical order data is combined with the current order task to evaluate the status of each order at each processing link. For example, if the current order task is to process an emergency order, the server may analyze the behavior of the user authentication node and the payment processing node in the emergency situation based on the historical data, so as to adjust the processing policy of these nodes, and ensure that the order passes through these links quickly and safely. Similarly, the server calculates the order data classification status attribute of each second processing status node according to the order data classification status attribute of each first processing status node and the second node attribute marking information. This involves a more advanced analysis, such as taking into account the various conditions that the order encounters during delivery, and how to adjust the delivery strategy based on those conditions. And finally, the server generates second order data classification status attributes according to the order data classification status attributes of each first processing status node and the order data classification status attributes of each second processing status node. And integrating all the collected and calculated information to form a comprehensive order processing strategy. For example, if the server finds a large number of emergency orders at a first processing state node while finding a shortage of delivery resources at a second processing state node, it generates an optimization strategy that increases the priority of the emergency orders in the delivery queue and allocates additional delivery resources to cope with this situation.
In a specific embodiment, if the executing step exists, the process of performing the cyclic traversal classification on the second order classification task through the plurality of first processing state nodes in the first node hierarchy and the plurality of second processing state nodes in the second node hierarchy to obtain the order classification result may specifically include the following steps:
(1) If the first order classification task exists, a third processing state node corresponding to the second order classification task is acquired and positioned, and a target node level to which the third processing state node belongs is determined;
(2) Acquiring first total order data classification state attributes of all order processing state nodes in the target node hierarchy, judging whether the target node hierarchy meets a first task classification rule according to the first total order data classification state attributes, and obtaining a target judgment result;
(3) Performing cyclic traversal classification on the initial order data classification tree according to the target judgment result and the second order classification task to obtain an order classification result;
(4) When the target judgment result is not met, determining that the target node level does not meet the first task classification rule, and feeding back a second order classification task to the target node level to which the target node level belongs;
(5) Acquiring a second total order data classification state attribute corresponding to the target node level, and judging whether the target node level meets a second task classification rule or not according to the second total order data classification state attribute;
(6) If the target node level does not meet the second task classification rule, feeding the second order classification task back to the initial order data classification tree, and performing task order classification on the second order classification task through the initial order data classification tree to obtain an order classification result.
Specifically, first, the server acquires and locates a node in a third processing state corresponding to the second order classification task when detecting that the task exists, and determines a target node level to which the node belongs. The server obtains first total order data classification status attributes of all order processing status nodes in the target node hierarchy, and judges whether the node hierarchy meets a first task classification rule based on the attributes. Each state node in the hierarchy of nodes is evaluated for its current processing power and state by a comprehensive data analysis. For example, if the target node hierarchy is a node hierarchy that handles logistics, the server may analyze the current delivery volume, delivery efficiency, delivery schedule rate, etc. of each logistics node to determine if this hierarchy is capable of efficiently handling the newly added second order classification task. If the judgment result of the server is that the target node level does not meet the first task classification rule, the current logistics node level cannot process the newly added task due to overload or other reasons. In this case, the server feeds back the second order classification task to the node hierarchy of the higher or lower level to which the target node hierarchy belongs, to find a more appropriate processing node. For example, if the logistics node hierarchy is unable to process the newly added task, the server moves the task up to the order review hierarchy to re-evaluate the order's priority and processing policy. And then, the server acquires a second total order data classification status attribute corresponding to the target node level, and further judges whether the node level meets a second task classification rule or not according to the attribute. This process involves analyzing more detailed data, such as taking into account processing time, customer satisfaction, etc. for a particular type of order. If the final judging result is still that the target node level does not meet the task classification rule, the server feeds back the second order classification task to the initial order data classification tree. Through the initial order data classification tree, the server performs more aggressive task order classification on the second order classification task to find the most appropriate processing strategy. This involves a re-evaluation of the entire order processing flow, such as adjusting the processing priority of different order types, or reallocating processing resources. This process ensures that each order is processed at the most appropriate time and at the most appropriate node.
In a specific embodiment, the process of executing step S104 may specifically include the following steps:
(1) Performing order traversal analysis on the historical order data of the user based on a first node hierarchy in the target order data classification tree to obtain a plurality of first classification data;
(2) Performing order traversal analysis on the historical order data of the user based on the first classification data and the second node level in the target order data classification tree to obtain second classification data;
(3) Performing data comparison and data fusion on the first classification data and the second classification data to obtain order classification data, wherein the order classification data comprises: order payment duration data, unpaid order data, and return order data.
Specifically, first, based on a first node hierarchy in a target order data classification tree, detailed order traversal analysis is performed on user historical order data to obtain a plurality of first classification data. The first node hierarchy typically involves early stages of order processing, such as order submission, user authentication, payment processing, and the like. For each such node, the server may analyze the relevant historical order data, such as looking at the number, type, time distribution, etc. of orders in the order submission node. These analyses help reveal features and trends of different types of orders early in the process, e.g., the server finds that the number of orders of a particular type increases abnormally over a certain period of time, or that certain types of orders have a higher failure rate in the user authentication link. Next, based on the first classification data and a second node hierarchy in the target order data classification tree, further order traversal analysis is performed on the user historical order data to obtain a plurality of second classification data. The second node hierarchy involves subsequent links to order processing, such as delivery preparation, delivery in progress, order completion, etc. In this stage of analysis, the server may explore, for example, the time efficiency of orders during delivery, the frequency of problems during delivery, etc. For example, the server analyzes the average preparation time of the order in the delivery preparation stage, or the damage rate and delay rate of the order during delivery. Then, these first classification data and second classification data are subjected to careful data comparison and fusion. Similar or related portions of the first and second classification data are combined to form a more comprehensive view angle. For example, the server may combine information regarding payment processing efficiency in the first category data with information regarding delivery preparation time in the second category data to evaluate the effect of payment efficiency on overall order processing time. In addition, data fusion also involves integrating data of different dimensions, such as combining the length of payment for an order, unpaid order, and return order data. In this way, the server obtains a plurality of order classification data including order payment duration data, unpaid order data, and return order data.
In a specific embodiment, the process of executing step S105 may specifically include the following steps:
(1) Respectively carrying out curve analysis on order payment duration data, unpaid order data and return order data in the order classification data to obtain a plurality of order classification curves, wherein the order classification curves comprise order payment duration curves, unpaid order change curves and return order curves;
(2) Extracting curve characteristic points and screening the curve characteristic points of the plurality of order classification curves respectively to obtain an order characteristic set of each order classification data;
(3) Respectively carrying out characteristic mean value and characteristic standard deviation calculation on each order characteristic set to obtain characteristic mean value data and characteristic standard deviation data of each order classification data;
(4) Carrying out characteristic weight analysis on the characteristic mean value data and the characteristic standard deviation data to obtain characteristic weight data of each order classification data;
(5) Carrying out feature weighting on the order feature set of each order classification data according to the feature weight data to obtain a weighted feature set of each order classification data;
(6) And carrying out vector coding on the weighted feature set to obtain an initial order evaluation vector of each order classification data, and carrying out vector splicing on the initial order evaluation vector to obtain a historical order evaluation vector.
Specifically, first, the server performs curve analysis on a plurality of order classification data, including order payment duration data, unpaid order data and return order data, respectively. These data types reflect different aspects of order processing, and through curve analysis, the server is able to visualize the time series trend of these data. For example, an order payment duration curve shows the average time from creation of an order to completion of payment over time; the unpaid order change curve shows the change trend of the number of unpaid orders in a period of time; the return order curve reflects the change in the quantity of the return orders over time. Through these curves, the server intuitively observes dynamic changes in various order status over time. The server then performs feature point extraction and screening on these order classification curves. The server identifies key feature points on each curve, such as peaks, valleys, inflection points, etc., that reveal key changes and trends in the data. For example, one peak of the order payment duration curve represents a significant increase in order payment duration over a period of time due to delays or other problems with the payment system. By identifying these key feature points, the server is able to more accurately understand the key changes and potential causes of the data. Then, the server calculates a feature mean and a feature standard deviation for each order feature set. The feature mean provides a measure of the average performance of a feature, while the feature standard deviation provides information about how well the features vary from order to order. Then, the server performs feature weight analysis on the feature mean data and the feature standard deviation data to determine the relative importance of each feature to the overall order processing flow and customer satisfaction. For example, if the server finds a strong correlation between the order payment duration and customer satisfaction, this feature will be given a higher weight. Conversely, if a feature is not highly correlated with a key traffic indicator, its weight will be low. Based on these feature weight data, the server then performs a weighting process on the order feature set for each order category data. The server adjusts its impact in the final analysis according to the importance of each feature. This weighted feature set more accurately reflects the actual impact of each feature on order processing flow and customer satisfaction. Finally, the server vector encodes the weighted feature sets, converting them into initial order assessment vectors in numerical form. The complex multidimensional data is simplified into a standardized numerical form, so that further analysis and processing are facilitated. For example, the server may convert the weighted characteristics of the order payment duration, the number of outstanding orders, and the number of return orders into a series of numerical values, each numerical value representing a weighted score for the corresponding characteristic. The server then concatenates the initial order assessment vectors to form a comprehensive historical order assessment vector. This vector integrates all key order features, providing a comprehensive assessment of historical order performance.
In a specific embodiment, the process of executing step S106 may specifically include the following steps:
(1) Inputting a historical order evaluation vector into a preset user classification model, wherein the user classification model comprises two layers of GRUs and a sigmoid function, the first layer of GRUs are formed by connecting one unidirectional GRU for a command layer, 256 GRU units are totally contained in the second layer of GRUs, 256 groups of unidirectional GRU connections are contained in the second layer of GRUs, each group of GRU in the second layer of GRUs corresponds to each GRU unit in the first layer of GRU, and each group of unidirectional GRU connections contain 16 GRU units;
(2) Extracting hidden features of the historical order evaluation vector through the first layer GRU to obtain a first hidden feature vector;
(3) Extracting hidden features of the first hidden feature vector through the second layer GRU to obtain a second hidden feature vector;
(4) Performing score calculation on the second hidden feature vector through a sigmoid function to obtain a target score of the target ordering user;
(5) And performing user type matching on the target score to obtain a target user type, and generating an order processing result of the to-be-processed order according to the target user type.
Specifically, first, the server inputs the historical order rating vector into a preset user classification model. This model is designed to contain two layers of GRU networks and a sigmoid function. The first layer of GRU is used as a command layer and consists of 256 unidirectional GRU units. A GRU (gate loop unit) is a special type of neural network unit adapted to process time series data, such as order history data. They can effectively capture time-dependent relationships in the data, which helps understand user behavior and order pattern aspects. In this layer, each GRU unit is responsible for capturing different aspects and features in the historical order assessment vector, such as the time pattern of the order, the purchasing habits of the user, and so forth. Then, the server performs hidden feature extraction on the historical order evaluation vector through the first layer GRU of the 256 units, and generates a first hidden feature vector. This process is analogous to high-level encoding of raw data, converting it into a form that is more abstract and more representative of the underlying schema. For example, the server may identify that certain users tend to make purchases at a particular time, or that certain types of products have a higher return rate during a particular season. The second layer GRU then further processes this first hidden feature vector. This layer contains 256 sets of unidirectional GRU connections, each set corresponding to one GRU unit in the first layer, each set containing 16 GRU units. This design enables the second layer to more carefully analyze the features extracted by the first layer, further refining and enhancing the important information in the data. For example, if a first tier identifies high returns over a period of time, a second tier may further analyze the relationship between these returns and a particular product category, user group, or purchase channel. The server then calculates a score for a second hidden feature vector obtained from the second tier GRU via a sigmoid function. The Sigmoid function here functions to convert the feature vector into a probability score representing the probability that the target subscriber belongs to a particular subscriber type. Finally, the server performs user type matching on the target score. The calculated probability scores are compared with predefined user type criteria to determine which class the user most belongs to. The server then generates an order processing result for the corresponding pending order according to the determined target user type. The server customizes the order processing policy according to the classification of the user.
The method for managing order data in the embodiment of the present application is described above, and the following describes an apparatus for managing order data in the embodiment of the present application, referring to fig. 2, one embodiment of the apparatus for managing order data in the embodiment of the present application includes:
the receiving module 201 is configured to receive an order processing request sent by a client, and perform request analysis on the order processing request to obtain a target order placing user and a to-be-processed order corresponding to the target order placing user;
the query module 202 is configured to obtain user personal information of the target ordering user, and perform historical order query on the user personal information through an order database to obtain user historical order data;
the construction module 203 is configured to obtain a plurality of order processing status nodes, and construct a corresponding target order data classification tree according to the plurality of order processing status nodes;
the classification module 204 is configured to perform data classification on the historical order data of the user based on the target order data classification tree to obtain a plurality of order classification data;
the encoding module 205 is configured to perform order feature analysis on the plurality of order classification data respectively to obtain an order feature set of each order classification data, and perform feature weight analysis and vector encoding on the order feature set of each order classification data respectively to obtain a historical order evaluation vector;
The generating module 206 is configured to input the historical order evaluation vector into a preset user classification model for user classification, obtain a target user type of the target user, and generate an order processing result of the to-be-processed order according to the target user type.
Through the cooperation of the components, the system can more accurately identify the intention and the requirement of the user by receiving and analyzing the order processing request sent by the client, thereby improving the processing accuracy. Meanwhile, the order data is classified and analyzed by using the Bayesian network, so that the data processing efficiency can be remarkably improved. The personal information of the user is acquired and analyzed in combination with the historical order data, so that the purchasing preference and the behavior mode of the user can be better understood. This helps to provide more personalized services such as recommendation systems and personalized order processing strategies, thereby improving user satisfaction. The constructed order data classification tree provides a dynamic and flexible method of processing order data. The method can be dynamically adjusted according to different order processing state nodes, and improves adaptability to various order types and states. The user classification model constructed by the GRU and sigmoid functions can effectively process and analyze complex order data. The deep learning method can reveal a deep mode behind the user behavior, and provides more accurate support for order processing and user service. Through the generation of the historical order evaluation vector and the user classification, the management method can provide more comprehensive and deep data support for a decision maker. The method is beneficial to optimizing decisions in aspects of inventory management, marketing strategies, customer relationship management and the like, improves the classification efficiency and accuracy of historical order data, and further improves the accuracy of order processing.
The present application also provides an order data management apparatus, where the order data management apparatus includes a memory and a processor, where the memory stores computer readable instructions that, when executed by the processor, cause the processor to execute the steps of the order data management method in the foregoing embodiments.
The present application also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, or may be a volatile computer readable storage medium, where instructions are stored in the computer readable storage medium, when the instructions are executed on a computer, cause the computer to perform the steps of the order data management method.
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, systems and units may refer to the corresponding processes in the foregoing method embodiments, which are not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, including several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random acceS memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The above embodiments are merely for illustrating the technical solution of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.
Claims (8)
1. A method for managing order data, the method comprising:
receiving an order processing request sent by a client, and carrying out request analysis on the order processing request to obtain a target order placing user and a to-be-processed order corresponding to the target order placing user;
acquiring user personal information of the target ordering user, and inquiring historical order of the user personal information through an order database to obtain user historical order data;
acquiring a plurality of order processing state nodes, and constructing a corresponding target order data classification tree according to the plurality of order processing state nodes; the method specifically comprises the following steps: acquiring a plurality of order processing state nodes, and analyzing node causal relation of the plurality of order processing state nodes through a preset Bayesian network to obtain node causal relation information of the plurality of order processing state nodes; performing node hierarchy division on the plurality of order processing state nodes according to the node causal relationship information to obtain two node hierarchies, wherein the two node hierarchies comprise a first node hierarchy and a second node hierarchy, the first node hierarchy comprises a plurality of first processing state nodes, and the second node hierarchy comprises a plurality of second processing state nodes; performing tree structure conversion on a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy to obtain corresponding initial order data classification trees; receiving pre-training order data and generating a first order classification task through the initial order data classification tree, and issuing the first order classification task to a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy; calculating first order data classification state attributes of a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy, and carrying out node traversal matching and order data classification state attribute updating on the first order classification task according to the first order data classification state attributes to obtain second order data classification state attributes; configuring a node updating strategy of the initial order data classification tree according to the second order data classification status attribute, executing the node updating strategy through the initial order data classification tree, performing task monitoring on the initial order data classification tree, and judging whether a second order classification task exists or not; if so, performing cyclic traversal classification on the second order classification task through a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy to obtain an order classification result; performing order data classification tree optimization on the initial order data classification tree according to the order classification result to obtain a corresponding target order data classification tree;
Based on the target order data classification tree, carrying out data classification on the historical order data of the user to obtain a plurality of order classification data;
performing order feature analysis on the plurality of order classification data respectively to obtain an order feature set of each order classification data, and performing feature weight analysis and vector coding on the order feature set of each order classification data respectively to obtain a historical order evaluation vector;
inputting the historical order evaluation vector into a preset user classification model for user classification to obtain a target user type of the target ordering user, and generating an order processing result of the to-be-processed order according to the target user type; the method specifically comprises the following steps: inputting the historical order evaluation vector into a preset user classification model, wherein the user classification model comprises two layers of GRUs and a sigmoid function, the first layer of GRU is formed by connecting one unidirectional GRU for a command layer, 256 GRU units are totally arranged in the first layer of GRU, 256 unidirectional GRU units are connected in the second layer of GRU, each GRU unit in the second layer of GRU corresponds to each GRU unit in the first layer of GRU, and 16 GRU units are connected in each unidirectional GRU unit; extracting hidden features of the historical order evaluation vector through the first layer GRU to obtain a first hidden feature vector; extracting hidden features of the first hidden feature vector through the second layer GRU to obtain a second hidden feature vector; performing score calculation on the second hidden feature vector through the sigmoid function to obtain a target score of the target ordering user; and performing user type matching on the target score to obtain a target user type, and generating an order processing result of the to-be-processed order according to the target user type.
2. The method according to claim 1, wherein calculating first order data classification status attributes of a plurality of first processing status nodes in the first node hierarchy and a plurality of second processing status nodes in the second node hierarchy, and performing node sequential traversal matching and order data classification status attribute updating on the first order classification task according to the first order data classification status attributes, to obtain second order data classification status attributes, includes:
calculating first order data classification status attributes of a plurality of first processing status nodes in the first node hierarchy and a plurality of second processing status nodes in the second node hierarchy;
acquiring first node attribute marking information of a plurality of first processing state nodes in the first node hierarchy, and simultaneously acquiring second node attribute marking information of a plurality of second processing state nodes in a second node hierarchy;
calculating the order data classification state attribute of each first processing state node according to the first order classification task, the first order data classification state attribute and the first node attribute marking information, and calculating the order data classification state attribute of each second processing state node according to the order data classification state attribute of each first processing state node and the second node attribute marking information;
And generating a second order data classification status attribute according to the order data classification status attribute of each first processing status node and the order data classification status attribute of each second processing status node.
3. The method according to claim 1, wherein the performing, if any, the cyclic traversal classification on the second order classification task by the plurality of first processing state nodes in the first node hierarchy and the plurality of second processing state nodes in the second node hierarchy to obtain an order classification result includes:
if yes, a third processing state node corresponding to the second order classification task is obtained and positioned, and a target node level to which the third processing state node belongs is determined;
acquiring first total order data classification state attributes of all order processing state nodes in the target node hierarchy, and judging whether the target node hierarchy meets a first task classification rule according to the first total order data classification state attributes to obtain a target judgment result;
performing cyclic traversal classification on the initial order data classification tree according to the target judgment result and the second order classification task to obtain an order classification result;
When the target judgment result is not met, determining that the target node level does not meet a first task classification rule, and feeding back the second order classification task to the target node level to which the target node level belongs;
acquiring a second total order data classification status attribute corresponding to the target node level, and judging whether the target node level meets a second task classification rule according to the second total order data classification status attribute;
and if the target node level does not meet a second task classification rule, feeding the second order classification task back to the initial order data classification tree, and performing task order classification on the second order classification task through the initial order data classification tree to obtain an order classification result.
4. A method of managing order data according to claim 3, wherein said data classifying said user history order data based on said target order data classification tree to obtain a plurality of order classification data comprises:
performing order traversal analysis on the historical order data of the user based on a first node hierarchy in the target order data classification tree to obtain a plurality of first classification data;
Performing order traversal analysis on the user history order data based on the plurality of first classification data and a second node hierarchy in the target order data classification tree to obtain a plurality of second classification data;
performing data comparison and data fusion on the plurality of first classified data and the plurality of second classified data to obtain a plurality of order classified data, wherein the order classified data comprises: order payment duration data, unpaid order data, and return order data.
5. The method of claim 4, wherein the performing order feature analysis on the plurality of order classification data to obtain an order feature set of each order classification data, and performing feature weight analysis and vector encoding on the order feature set of each order classification data to obtain a historical order evaluation vector, respectively, comprises:
respectively carrying out curve analysis on order payment duration data, unpaid order data and return order data in the plurality of order classification data to obtain a plurality of order classification curves, wherein the order classification curves comprise an order payment duration curve, an unpaid order change curve and a return order curve;
Extracting curve characteristic points and screening the curve characteristic points of the plurality of order classification curves respectively to obtain an order characteristic set of each order classification data;
respectively carrying out characteristic mean value and characteristic standard deviation calculation on each order characteristic set to obtain characteristic mean value data and characteristic standard deviation data of each order classification data;
carrying out characteristic weight analysis on the characteristic mean value data and the characteristic standard deviation data to obtain characteristic weight data of each order classification data;
carrying out feature weighting on the order feature set of each order classification data according to the feature weight data to obtain a weighted feature set of each order classification data;
and carrying out vector coding on the weighted feature set to obtain an initial order evaluation vector of each order classification data, and carrying out vector splicing on the initial order evaluation vector to obtain a historical order evaluation vector.
6. An order data management apparatus, characterized in that the order data management apparatus includes:
the receiving module is used for receiving an order processing request sent by a client and carrying out request analysis on the order processing request to obtain a target order placing user and a to-be-processed order corresponding to the target order placing user;
The inquiry module is used for acquiring the user personal information of the target ordering user, and carrying out historical order inquiry on the user personal information through an order database to obtain user historical order data;
the construction module is used for acquiring a plurality of order processing state nodes and constructing a corresponding target order data classification tree according to the plurality of order processing state nodes; the method specifically comprises the following steps: acquiring a plurality of order processing state nodes, and analyzing node causal relation of the plurality of order processing state nodes through a preset Bayesian network to obtain node causal relation information of the plurality of order processing state nodes; performing node hierarchy division on the plurality of order processing state nodes according to the node causal relationship information to obtain two node hierarchies, wherein the two node hierarchies comprise a first node hierarchy and a second node hierarchy, the first node hierarchy comprises a plurality of first processing state nodes, and the second node hierarchy comprises a plurality of second processing state nodes; performing tree structure conversion on a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy to obtain corresponding initial order data classification trees; receiving pre-training order data and generating a first order classification task through the initial order data classification tree, and issuing the first order classification task to a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy; calculating first order data classification state attributes of a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy, and carrying out node traversal matching and order data classification state attribute updating on the first order classification task according to the first order data classification state attributes to obtain second order data classification state attributes; configuring a node updating strategy of the initial order data classification tree according to the second order data classification status attribute, executing the node updating strategy through the initial order data classification tree, performing task monitoring on the initial order data classification tree, and judging whether a second order classification task exists or not; if so, performing cyclic traversal classification on the second order classification task through a plurality of first processing state nodes in the first node hierarchy and a plurality of second processing state nodes in the second node hierarchy to obtain an order classification result; performing order data classification tree optimization on the initial order data classification tree according to the order classification result to obtain a corresponding target order data classification tree;
The classification module is used for carrying out data classification on the historical order data of the user based on the target order data classification tree to obtain a plurality of order classification data;
the coding module is used for respectively carrying out order feature analysis on the plurality of order classification data to obtain an order feature set of each order classification data, and respectively carrying out feature weight analysis and vector coding on the order feature set of each order classification data to obtain a historical order evaluation vector;
the generation module is used for inputting the historical order evaluation vector into a preset user classification model to classify the users, obtaining the target user type of the target ordering user, and generating an order processing result of the to-be-processed order according to the target user type; the method specifically comprises the following steps: inputting the historical order evaluation vector into a preset user classification model, wherein the user classification model comprises two layers of GRUs and a sigmoid function, the first layer of GRU is formed by connecting one unidirectional GRU for a command layer, 256 GRU units are totally arranged in the first layer of GRU, 256 unidirectional GRU units are connected in the second layer of GRU, each GRU unit in the second layer of GRU corresponds to each GRU unit in the first layer of GRU, and 16 GRU units are connected in each unidirectional GRU unit; extracting hidden features of the historical order evaluation vector through the first layer GRU to obtain a first hidden feature vector; extracting hidden features of the first hidden feature vector through the second layer GRU to obtain a second hidden feature vector; performing score calculation on the second hidden feature vector through the sigmoid function to obtain a target score of the target ordering user; and performing user type matching on the target score to obtain a target user type, and generating an order processing result of the to-be-processed order according to the target user type.
7. An order data management apparatus, characterized in that the order data management apparatus comprises: a memory and at least one processor, the memory having instructions stored therein;
the at least one processor invoking the instructions in the memory to cause the order data management apparatus to perform the order data management method of any of claims 1-5.
8. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the order data management method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311850739.7A CN117495512B (en) | 2023-12-29 | 2023-12-29 | Order data management method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311850739.7A CN117495512B (en) | 2023-12-29 | 2023-12-29 | Order data management method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117495512A CN117495512A (en) | 2024-02-02 |
| CN117495512B true CN117495512B (en) | 2024-04-16 |
Family
ID=89669394
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311850739.7A Active CN117495512B (en) | 2023-12-29 | 2023-12-29 | Order data management method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117495512B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107203866A (en) * | 2017-06-26 | 2017-09-26 | 北京京东尚科信息技术有限公司 | The processing method and device of order |
| CN107203789A (en) * | 2017-06-23 | 2017-09-26 | 北京小度信息科技有限公司 | Distribution model method for building up, distribution method and relevant apparatus |
| CN110533502A (en) * | 2019-08-12 | 2019-12-03 | 北京三快在线科技有限公司 | The recognition methods of goal behavior and device and electronic equipment |
| CN116433339A (en) * | 2023-06-15 | 2023-07-14 | 全屋优品科技(深圳)有限公司 | Order data processing method, device, equipment and storage medium |
-
2023
- 2023-12-29 CN CN202311850739.7A patent/CN117495512B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107203789A (en) * | 2017-06-23 | 2017-09-26 | 北京小度信息科技有限公司 | Distribution model method for building up, distribution method and relevant apparatus |
| CN107203866A (en) * | 2017-06-26 | 2017-09-26 | 北京京东尚科信息技术有限公司 | The processing method and device of order |
| CN110533502A (en) * | 2019-08-12 | 2019-12-03 | 北京三快在线科技有限公司 | The recognition methods of goal behavior and device and electronic equipment |
| CN116433339A (en) * | 2023-06-15 | 2023-07-14 | 全屋优品科技(深圳)有限公司 | Order data processing method, device, equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 基于分类算法的农产品电商推荐系统的设计与实现;李亚峰;刘志丽;;农家参谋;20200518(15);全文 * |
| 李亚峰 ; 刘志丽 ; .基于分类算法的农产品电商推荐系统的设计与实现.农家参谋.2020,(15),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117495512A (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111324642A (en) | Model algorithm type selection and evaluation method for power grid big data analysis | |
| CN109685635A (en) | Methods of risk assessment, air control server-side and the storage medium of financial business | |
| CN113742492B (en) | Insurance scheme generation method and device, electronic equipment and storage medium | |
| CN110147389B (en) | Account processing method and device, storage medium and electronic device | |
| CN103838857B (en) | Automatic service combination system and method based on semantics | |
| US20230229975A1 (en) | Offline security value determination system and method | |
| CN110310114A (en) | Object classification method, device, server and storage medium | |
| CN117151870B (en) | Portrait behavior analysis method and system based on guest group | |
| CN111612519A (en) | Method, device and storage medium for identifying potential customers of financial product | |
| CN117670066B (en) | Questor management method, system, equipment and storage medium based on intelligent decision | |
| CN114399367A (en) | Insurance product recommendation method, device, equipment and storage medium | |
| CN116739794B (en) | User personalized scheme recommendation method and system based on big data and machine learning | |
| CN111612583B (en) | Personalized shopping guide system based on clustering | |
| CN117495512B (en) | Order data management method, device, equipment and storage medium | |
| CN115169508B (en) | Data analysis and management method based on supply chain platform | |
| CN116595262A (en) | Travel scheme recommendation method and device, electronic equipment and computer storage medium | |
| WO2011084238A1 (en) | Method and apparatus of adaptive categorization technique and solution for services selection based on pattern recognition | |
| CN113628748A (en) | Method, device and equipment for evaluating risk bearing tendency of user and storage medium | |
| Hanif | Applications of data mining techniques for churn prediction and cross-selling in the telecommunications industry | |
| CN113627997A (en) | Data processing method and device, electronic equipment and storage medium | |
| CN115222177A (en) | Service data processing method and device, computer equipment and storage medium | |
| CN115689128B (en) | Customer data analysis method and system based on CRM | |
| CN117668236B (en) | Analysis method, system and storage medium of patent standard fusion system | |
| CN116795357B (en) | Processing method, device, equipment and storage medium of business rule | |
| CN117668205B (en) | Smart logistics customer service processing method, system, equipment and storage medium |
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 |