An e-commerce order automatic processing method and system based on big data

Abstract

The application discloses an e-commerce order automatic processing method and system based on big data, relates to the technical field of order automatic processing, and comprises the following steps: obtaining order data and order behavior data corresponding to a target order, and generating order analysis data; combining a current processing state of the target order with all historical behavior information to identify an order processing intention, and generating order processing intention data; determining a dynamic processing window according to the order processing intention data, and performing order compatibility matching to generate a candidate order processing scheme set; evaluating the candidate order processing scheme set to determine a target order processing scheme; updating the order processing intention data, order processing window data and order compatibility data according to order processing feedback data to generate an order automatic processing result, realizing dynamic adjustment and intelligent optimization of an order processing process, improving order processing accuracy and automation level, and realizing continuous optimization of an order processing process.

Description

Technical Field

[0001] This invention relates to the field of automated order processing technology, specifically to a method and system for automated e-commerce order processing based on big data. Background Technology

[0002] With the continuous expansion of e-commerce, order processing, as a crucial link connecting user transactions, merchant fulfillment, and logistics, has a significant impact on order response efficiency, processing accuracy, and business continuity. As the number of e-commerce orders continues to grow and order types, user behaviors, and transaction scenarios become increasingly complex, traditional order processing methods based on fixed rules and manually configured processes are no longer sufficient to meet diverse order demands. This is especially true when order statuses are constantly changing, user behavior is dynamically evolving, and historical processing patterns are diverging; their ability to adjust processing strategies is limited, hindering dynamic optimization of the order processing process.

[0003] However, existing e-commerce order processing methods typically only process orders based on their current status or fixed business rules, lacking correlation analysis between order behavior data and full historical processing information. This makes it difficult to accurately identify the subsequent processing intentions of orders and to dynamically determine processing time and priority based on different order needs. Furthermore, existing methods lack a comprehensive matching mechanism for time constraints, status connections, and execution conflicts between order processing actions, making it difficult to generate processing solutions adapted to the current order status. Moreover, after order processing is completed, there is no continuous updating of processing intentions, processing windows, and compatibility relationships based on feedback results, leading to insufficient adaptability in the automated order processing process. Therefore, a big data-based automated e-commerce order processing method and system is needed to address the aforementioned problems. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides a method and system for automatic e-commerce order processing based on big data. This solution solves the problems raised in the background section regarding existing e-commerce order processing methods, which typically process orders based solely on their current status or fixed business rules. These methods lack correlation analysis between order behavior data and comprehensive historical processing information, making it difficult to accurately identify subsequent order processing intentions and dynamically determine processing timelines and priorities based on different order needs. Furthermore, existing methods lack a comprehensive matching mechanism for time constraints, state connections, and execution conflicts between order processing actions, making it difficult to generate processing solutions adapted to the current order status. Moreover, after order processing is completed, there is no continuous updating of processing intentions, processing windows, and compatibility relationships based on feedback results, leading to insufficient adaptability in the automatic order processing process.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for automatically processing e-commerce orders based on big data, comprising: Obtain the order data and order behavior data corresponding to the target order, and perform correlation and integration processing to generate order analysis data; Based on order analysis data, combined with the current processing status of the target order and the full historical behavior information of the corresponding order, the order processing intent of the target order is identified, and order processing intent data is generated. Based on order processing intent data, the processing time requirements and processing priorities of target orders are analyzed to determine the dynamic processing window corresponding to the target order and generate order processing window data. Based on the order processing window data, perform order compatibility matching to generate order compatibility data and obtain a set of candidate order processing solutions; The candidate order processing solutions are comprehensively evaluated to generate the target order processing solution. The system executes automatic order processing according to the target order processing plan, obtains order processing feedback data, and updates the order processing intent data, order processing window data, and order compatibility data based on the order processing feedback data to generate automatic order processing results.

[0006] Furthermore, a big data-based automated e-commerce order processing system is proposed to implement the aforementioned automated order processing method, including: The order analysis module is used to acquire the order data and order behavior data corresponding to the target order, and perform correlation and integration processing to generate order analysis data. The intent recognition module is used to identify the order processing intent of the target order based on the order analysis data, combined with the current processing status of the target order and the full historical behavior information corresponding to the order, and generate order processing intent data. The window generation module is used to analyze the processing time requirements and processing priorities of the target order based on the order processing intent data, determine the dynamic processing window corresponding to the target order, and generate order processing window data. The compatibility matching module is used to perform order compatibility matching processing based on the order processing window data, generate order compatibility data, and obtain a set of candidate order processing solutions. The solution evaluation module is used to comprehensively evaluate the set of candidate order processing solutions and generate a target order processing solution. The feedback update module is used to execute automatic order processing according to the target order processing plan, obtain order processing feedback data, and update the order processing intent data, order processing window data and order compatibility data according to the order processing feedback data to generate automatic order processing results.

[0007] Compared with the prior art, the beneficial effects of the present invention are: This solution proposes a big data-based automatic e-commerce order processing method and system. It acquires order data and order behavior data corresponding to the target order, and integrates and correlates order status information, product information, transaction information, logistics information, and user operation behavior to generate order analysis data. Simultaneously, it analyzes the correspondence between historical status change nodes, operation types, and processing results by combining the current processing status of the target order with all historical behavior information, identifying order processing intentions and improving the accuracy of judging subsequent order processing needs. Furthermore, by analyzing historical processing processes corresponding to the same processing intention based on order processing intention data, it extracts information on changes in processing timeliness and historical processing priorities, dynamically determining the target order processing time. The system identifies the corresponding processing window for each order and matches the processing actions based on the data within that window, considering time consumption, state transitions, and execution conflicts. This generates a set of candidate order processing solutions, enabling dynamic planning of the order processing process and improving the adaptability of automated order processing. By comprehensively evaluating the action composition, execution order, and state change processes of the candidate order processing solutions, the system determines the target order processing solution. During execution, it acquires order processing feedback data and updates the order processing intent data, order processing window data, and order compatibility data based on the actual processing results. This continuous optimization of the order processing strategy improves the accuracy and automation level in complex order scenarios. Attached Figure Description

[0008] Figure 1 This is a flowchart of an automatic e-commerce order processing method based on big data proposed in this invention; Figure 2 This is a flowchart of the order processing intent recognition process in this invention; Figure 3 This is a flowchart of the order compatibility matching and solution generation process in this invention; Figure 4 This is a system framework diagram of an e-commerce order automatic processing system based on big data proposed in this invention. Detailed Implementation

[0009] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0010] Reference Figure 1 - Figure 4 As shown in the figure, an embodiment of the present invention provides an automatic e-commerce order processing method based on big data, comprising: Obtain the order data and order behavior data corresponding to the target order, and perform correlation and integration processing to generate order analysis data; Specifically, in e-commerce order processing scenarios, the data corresponding to the target order comes from business data such as order transaction records, product management records, logistics records, and user operation records. During the acquisition process, firstly, based on the target order identifier, the corresponding order record is read from the order database, extracting order data such as order creation time, order status, product information, transaction information, and logistics information. The order status reflects the current processing stage of the order, product information describes the attributes of the corresponding products, transaction information records the payment and transaction process, and logistics information reflects the order delivery status. Simultaneously, user operation logs and order processing logs are used to retrieve behavioral records related to the target order, extracting user queries, modifications, cancellations, after-sales requests, and platform processing information, recording the occurrence time, operation type, and change process of each behavior, thus forming the order data and order behavior data corresponding to the target order.

[0011] Based on order analysis data, combined with the current processing status of the target order and the full historical behavior information of the corresponding order, the order processing intent of the target order is identified, and order processing intent data is generated. Based on order processing intent data, the processing time requirements and processing priorities of target orders are analyzed to determine the dynamic processing window corresponding to the target order and generate order processing window data. Based on the order processing window data, perform order compatibility matching to generate order compatibility data and obtain a set of candidate order processing solutions; The candidate order processing solutions are comprehensively evaluated to generate the target order processing solution. The system executes automatic order processing according to the target order processing plan, obtains order processing feedback data, and updates the order processing intent data, order processing window data, and order compatibility data based on the order processing feedback data to generate automatic order processing results.

[0012] Furthermore, based on order analysis data, combined with the current processing status of the target order and the corresponding full historical behavior information, the order processing intent corresponding to the target order is identified, and order processing intent data is generated, specifically including: The order status information, product information, transaction information, and logistics information in the order data are extracted and processed to generate basic order data; Based on order behavior data, extract order operation time, operation type, and behavior change information to generate order behavior feature data; By linking and integrating basic order data with order behavior characteristic data, order analysis data is generated; Based on order analysis data, the information corresponding to the basic order data and order behavior feature data is processed in a unified manner, and features are aggregated according to the target order identifier to generate an order feature set; Based on the order feature set, the current processing status of the target order is divided into states, and the state change processes corresponding to payment status, shipping status, logistics status and after-sales status are marked to generate an order status change sequence. Specifically, in e-commerce order processing scenarios, the information corresponding to a target order is typically stored in various locations, including order transaction records, product management records, logistics records, and user operation records. The content described by data from different sources varies, therefore, it is necessary to first establish a unified data association based on the target order identifier. During processing, the basic records corresponding to the target order are first read from the order database, and information reflecting the current status of the order is extracted, such as payment completion status, shipping status, delivery status, and after-sales processing status. Simultaneously, the static attributes of the target order are organized by combining product characteristics such as the category, quantity, specifications, and inventory attributes of the corresponding goods, as well as transaction characteristics such as order amount, payment method, and transaction time.

[0013] After obtaining the above information, further analysis is conducted using node data generated on the logistics side to supplement the status changes during the order delivery process. For example, for a product order that has been paid for but not yet shipped, the transaction record can confirm that payment has been completed, the logistics record can confirm that no transportation node has been generated, and the product record can confirm that the corresponding product is still in the inventory preparation stage, thus forming a data set that reflects the current actual processing status of the order.

[0014] Simultaneously, it's necessary to organize the user actions and order processing behaviors surrounding this order. Since the order processing is influenced not only by the order's inherent attributes but also by user actions such as modification, querying, application, and confirmation, the order behavior process can be described by collecting time information, operation categories, and relationships of continuous behavioral changes from the order operation logs. For example, if a user repeatedly modifies their delivery address shortly after payment, this continuous behavior can serve as important supplementary information regarding order status changes; if a user frequently checks the delivery progress after logistics are generated, this can be recorded as delivery monitoring behavior.

[0015] To enable the processing of basic order attributes and dynamic behavior processes within the same analytical dimension, further correlation and fusion of data from different sources are necessary. During this fusion process, the target order identifier is used as the correlation basis, mapping order status, product attributes, transaction information, logistics nodes, and behavior records according to temporal relationships. This creates a continuous data chain of multi-dimensional information corresponding to the same order. Simultaneously, field names and data formats from different data sources are standardized. For example, payment status descriptions from different business interfaces are converted to the same status identifier, and different time formats are converted to a unified time expression method, avoiding the impact of data structure differences on subsequent processing.

[0016] After data fusion, features are aggregated for various information corresponding to the same target order. Specifically, the current order status, historical operation records, and behavioral changes are aggregated according to the order identifier, making each order an independent set of data features. This feature set not only records the current stage of the order but also retains information on status changes that occur during the order's creation, payment, shipment, and after-sales process. For example, an order's feature set can simultaneously include information such as "paid," "not shipped," "user modified delivery address," and "existing historical inquiry records," thus comprehensively describing the current processing environment of the order.

[0017] To identify patterns in order changes across different stages, the aforementioned characteristics need to be categorized into states according to the order processing flow. During this categorization, a correspondence between state nodes is established based on the order business process, treating payment, shipment, logistics, and after-sales service as different state dimensions, and recording the state transition process based on the actual occurrence time. For example, if an order changes from "pending payment" to "paid," and then from "paid" to "pending shipment," this continuous change process is recorded as a set of state change nodes. If a refund request subsequently arises, the transition from "pending shipment" to "after-sales processing" is then recorded.

[0018] By sorting and connecting each state node over time, a state change sequence reflecting the entire process of a target order can be formed, enabling subsequent analysis of historical order processing to determine the possible processing direction for the current order.

[0019] Obtain the full historical behavior information corresponding to the target order, extract the historical processing status change nodes and corresponding operation types, and generate a historical behavior sequence; Based on the historical behavior sequence, the state change nodes and corresponding operation types in different historical processing processes are associated with the final processing results, and the distribution of processing results corresponding to different state change nodes and corresponding operation types is statistically analyzed to generate a set of historical behavior association features. Based on the historical behavior association feature set, the order status change sequence is associated and matched with the corresponding status change node in the historical behavior association feature set to generate order status association feature data. Based on the order status association feature data, the status change nodes and operation types corresponding to different historical processing results are classified and processed, and the common processing features of each historical processing process are extracted to generate a set of candidate order processing intentions. Based on the candidate order processing intent set, the order status change sequence is matched with the historical status change nodes corresponding to each candidate order processing intent node by node, the number of consecutive matching nodes and the node matching order are counted, and the order processing intent matching result is generated. Based on the order processing intent matching results, the candidate order processing intent with the most consecutive matching nodes and the same node matching order is determined as the target order processing intent, and order processing intent data is generated.

[0020] Specifically, in automated order processing scenarios, the subsequent processing direction of an order is often related to its current stage and the previously generated operational trajectory. Since different orders may have the same state but different results during processing, it is necessary to backtrack the processing of completed orders and extract the patterns of change from state changes to the final processing result of historical orders to determine the possible processing direction for the current order.

[0021] Historical order processing data typically includes order status records, user operation records, platform processing records, and final completion result records. The full historical behavior information is filtered according to the business type corresponding to the target order, including complete operation records of orders of similar products, orders with the same processing flow, and orders with the same anomaly type generated within the historical period. During processing, each historical order is treated as an independent analysis object. Related records are sorted according to order processing time, and the locations where status changes occur within the order lifecycle are extracted as key change nodes. The operation that triggered the status change, the status information before and after the change, and the corresponding occurrence time are recorded simultaneously. In this way, the originally discrete operations scattered throughout the order records are transformed into a chronological trajectory of status changes. For example, when a historical order transitions from a payment-completed state to a shipping-awaiting state, the payment completion node and warehouse processing operation are recorded; if subsequently, due to a user submitting a refund request, the status change node corresponding to the after-sales request is recorded, thus forming a complete order change process.

[0022] After reconstructing the historical order processing flow, multiple state change nodes within a single order are continuously connected to form a complete change path. This path not only records which states the order went through but also preserves the transition relationships between states, describing the evolution of the order from its initial stage to its final processing result. For example, orders that are both in the payment completion stage may either directly enter the shipping process or enter different processing flows due to operations such as modifying delivery information or canceling transactions. Therefore, it is necessary to distinguish between them by considering continuous state change relationships, rather than relying solely on a single state node.

[0023] After generating a large number of historical processing paths, the final processing results corresponding to different paths are statistically correlated. During processing, state change nodes, triggering operations, and final processing results are used as correlation analysis objects. Historical paths with the same state transition relationship are aggregated, and the frequency of occurrence of different processing results corresponding to each type of change path is counted. For cases where the same change path corresponds to multiple processing results, the proportion of different results under that path is calculated to determine the degree of correlation between the change path and different processing directions, thereby forming a set of correlation features that can reflect historical processing patterns. Specifically, the number of historical orders corresponding to each state change path is used as the total number of statistics, and the ratio of the number of orders corresponding to different final processing results to the total number is calculated to obtain the probability of occurrence of each processing result.

[0024] This set of associated features is used to store the correspondence between historical order change processes and final processing results. It includes the connections between state change nodes, combinations of operational behaviors, and the distribution of corresponding processing results. In the actual matching process, it not only determines whether the current order has reached a certain state node, but also searches for similar change paths in the history of continuous state change processes. For the current state change trajectory of the target order, the state nodes, node transition relationships, and corresponding operational behaviors are compared with historical associated features. During the matching process, the consistency between the current order change nodes and historical processing path nodes is checked sequentially according to the order of state occurrence. Simultaneously, it determines whether the transition directions between corresponding nodes are the same and records the relationships between nodes that can be continuously matched. For example, if the current order has experienced continuous changes from "payment completed" to "modify delivery information," and there are similar state transition paths in historical processing, with most entering the delivery adjustment process, then this historical path can serve as an important reference for determining the processing direction of the current order.

[0025] After obtaining the correlation results, the change paths with the same final processing result in historical processing are categorized and organized. During the categorization process, repeated combinations of state changes, operational behaviors, and node transition sequences in different orders are extracted to form common change features that represent a certain type of processing direction. For example, if multiple historical orders all involve address modification after payment and ultimately enter delivery adjustment processing, this type of state change pattern can be considered a common feature of the corresponding processing direction. By organizing the common change features corresponding to different processing results, a set of candidate order processing intentions is formed. When matching the current order's change trajectory with the historical change features corresponding to each candidate processing direction node by node, continuous comparisons are performed according to the node arrangement order. During the matching process, each node correspondence is recorded, the number of consecutively successfully matched nodes is counted, and it is determined whether the original change order is maintained between the matched nodes. When the historical change path corresponding to a certain candidate processing direction can cover a relatively long continuous change process of the current order, and the node order remains consistent with the current order, it indicates that the candidate processing direction has a high degree of correlation with the current order processing process.

[0026] After matching each candidate processing direction, the number of consecutive matching nodes and the consistency of node order corresponding to different candidate results are compared. The processing direction with the highest matching continuity and consistent change order is selected as the current order processing intention. Subsequently, the corresponding historical change features, associated status nodes, and matching results are integrated to form order processing intention data, providing a basis for subsequent order processing timeliness judgment, processing window determination, and automatic processing scheme generation.

[0027] Furthermore, based on order processing intent data, the processing time requirements and priorities of target orders are analyzed to determine the dynamic processing window corresponding to the target order and generate order processing window data, specifically including: Based on order processing intent data, extract the order processing intent, historical processing process and corresponding status change information of the target order to generate order processing demand feature data. Based on the order processing demand characteristic data, the historical processing processes corresponding to the same order processing intent are sorted out, the processing time change information corresponding to each historical processing process is extracted, and a set of processing timeliness characteristics is generated. Based on the processing timeliness feature set, the historical processing process corresponding to the target order is matched with the historical processing time change information in the processing timeliness feature set to determine the timeliness requirement corresponding to the target order and generate the timeliness requirement data of the target order. Based on the timeliness requirements of the target orders, filter historical processing records that meet the timeliness requirements of the target orders and generate a target processing record set. Based on the target processing record set, the historical processing sequence of the same order processing intent is statistically analyzed, and the order is sorted according to the frequency of occurrence of the historical processing sequence to generate order processing priority data; Based on order processing priority data, extract historical processing time range information under the corresponding priority to generate processing time range feature data; Based on the processing time range feature data, the start and end times of the corresponding historical processing time range are extracted, and time mapping processing is performed in combination with the target order generation time to generate dynamic processing window data. Based on the dynamically processed window data, the executable processing time corresponding to the current processing status of the target order is marked, and order processing window data is generated.

[0028] Specifically, during automated order processing, once the order processing intent is determined, different orders will have different processing times due to variations in their current processing stage, historical execution paths, and business requirements. Therefore, it is necessary to analyze the time requirements corresponding to the current state of the target order by combining historical order execution processes under the same processing direction, so that subsequent processing can be executed within a timeframe that conforms to historical patterns.

[0029] In the timeliness analysis process, historical orders with determined processing directions are first screened, and their corresponding processing intent identifiers, status change processes, and actual processing times are linked and organized. During the screening process, the order processing intent is used as the primary matching condition, while similarity judgments are made based on the number of status change nodes, the order of status transitions, and the current processing stage. In the judgment process, the number of completed status nodes, the node arrangement order, and the current status of the target order are compared with the corresponding information of historical orders. When the number of status nodes is close, the transition direction is consistent, and the current processing stage is the same, the corresponding historical order is used as a matching sample, retaining historical order samples with processing processes similar to the target order. Subsequently, time change information after the order enters the current processing stage is extracted from each historical sample, including the dwell time of each status node, the interval between adjacent status transitions, and the overall completion cycle, forming a set of timeliness features that reflects the time patterns of different processing paths.

[0030] After obtaining the timeliness feature set, a corresponding analysis is performed on the current status change process of the target order and its historical time characteristics. During processing, the completed status nodes of the target order are used as the starting point for matching. Processing records with the same or similar status transition paths are searched in historical samples, and historical time consumption data for the corresponding stages are extracted. For multiple matching results, the historical processing times are statistically sorted, and the distribution of historical samples within different time intervals is calculated. The time range that covers a large proportion of historical processing is selected as the timeliness requirement corresponding to the target order. This proportion is determined based on the stability of the processing results of similar historical orders. By statistically analyzing the proportion of completed orders in different time intervals to the total number of matched orders, the time intervals with a higher proportion and smaller fluctuations are selected as the effective time range. For example, when a large number of orders with the same processing intent are completed within 1 to 3 hours after entering a certain status, this time interval can be used as a reference for the time requirement of the current order's corresponding stage.

[0031] After determining the timeliness requirements of the target orders, the historical samples are screened a second time, retaining only records whose processing time matches the timeliness requirements of the target orders, forming a target processing record set. Subsequently, the historical processing in this set is broken down, and the time sequence of each processing action, the connection between actions, and the final completion status are statistically analyzed. During the statistical process, the data is sorted according to the frequency of different combinations of processing actions to obtain the historical execution frequency corresponding to different processing sequences, and the corresponding priority relationships are determined based on the processing completion status. For example, in the processing of adjustments for the same type of order, if most orders in the historical records complete information verification before executing delivery adjustments, then this execution sequence corresponds to a higher processing priority.

[0032] After determining the processing priorities, the historical processing time ranges corresponding to different priorities are extracted. During processing, the start time, end time, and stage transition time of the corresponding processing action in each historical sample are recorded, and the time information is converted into an offset relative to the order creation time, thereby eliminating the impact of differences in order creation times. By aggregating multiple historical time ranges under the same priority, the frequency and coverage ratio of each time range are statistically analyzed to form time range features that can describe the execution cycle of different processing priorities.

[0033] After obtaining the time range characteristics, the historical time offset range is mapped to the actual timeline of the target order. During the mapping process, the target order creation time is used as the time base, and the start and end times of the historical processing are converted into the actual time nodes corresponding to the current order. The executable processing time range for the current order is adjusted by considering the completed status of the target order, the currently pending processing actions, and the remaining executable processes. When the order status changes, the time range corresponding to the remaining processing stages is recalculated, ensuring that the processing time window changes with the actual progress of the order.

[0034] Finally, the executable time ranges, after time mapping and status adjustment, are integrated to form order processing window data, providing time constraints for subsequent order compatibility matching. The dynamic processing window is not a fixed time period, but rather a dynamically changing range determined by historical similar order processing patterns, current order status, and processing priorities, thereby improving the matching degree between the automated processing and actual order requirements.

[0035] Furthermore, order compatibility matching is performed based on the order processing window data to generate order compatibility data and obtain a set of candidate order processing solutions, specifically including: Based on the order processing window data, extract the executable processing time range and current order processing status information corresponding to the target order to generate order processing constraint data; Based on the order processing constraint data and the order processing intent data, the processing intent content corresponding to the target order is decomposed into actions to generate a set of candidate processing actions; Based on the candidate processing action set, the execution sequence, state change information and time occupation information of each processing action are extracted to generate processing action association feature data. Based on the processing action association feature data, candidate processing actions are combined according to their execution order to generate a set of candidate processing paths. Based on the candidate processing path set, the time occupancy information corresponding to each candidate processing path is matched with the executable processing time range in the order processing constraint data, and candidate processing paths that meet the time requirements are filtered to generate a time-adaptive processing path set. Specifically, during the automated order processing, after determining the order processing window, it is necessary to filter and combine executable processing actions based on the current executable time range, the current order processing status, and the order processing intent. Since the same order processing intent may correspond to multiple processing actions, and different actions have differences in execution order, status changes, and time consumption, it is necessary to analyze the relationships between processing actions to form a processing path that satisfies the current order constraints.

[0036] During processing, the first step is to organize the currently executable processing time range and the order's status for the target order. The executable processing time range, derived from dynamic processing window data, represents the time interval within which subsequent processing actions are allowed to be performed for the current order. The order processing status indicates the completed processing stages and the pending stages of the current order. By associating the time range information with the status information, the boundaries of actions that can be performed for the current order are determined. For example, if an order has been paid for but has not yet entered the delivery stage, the executable actions typically focus on subsequent stages such as shipment confirmation and logistics arrangement, without performing payment-related operations again. After integrating the above constraint information, order processing constraint data is formed.

[0037] After completing the order processing constraints, the processing intent corresponding to the current order needs to be converted into specific executable actions. During processing, based on the processing goal corresponding to the order processing intent, the business operations required to achieve the goal are broken down, dividing the overall processing into multiple action nodes with clearly defined processing content. The action type, associated order status, and resulting state change relationships are recorded for each action. Specifically, the action type describes the specific processing content, the associated order status indicates the order stage to which the action applies, and the state change relationships indicate the possible transitions in the order status after the action is completed.

[0038] During the action decomposition process, the correspondence between each action and the current order stage is determined by combining the current order processing status. Completed processing content is marked, and action information that still needs to be executed in the current state is retained to form a set of candidate processing actions.

[0039] For example, the intention to adjust order delivery can be broken down into multiple consecutive actions such as address information verification, delivery information update, and logistics task regeneration. These different actions need to be executed sequentially according to business logic.

[0040] After extracting candidate processing actions, the relationships between these actions are analyzed. During processing, the execution sequence, state change information, and time consumption information for each action are extracted. The execution sequence describes the dependency order between actions, the state change information indicates the potential changes in order status after action execution, and the time consumption information indicates the processing time required for action execution. By organizing historical execution records under the same processing intent, the sequential connections between each processing action are extracted according to the actual execution order during order processing. Simultaneously, the order status changes before and after each processing action and the corresponding time consumption are recorded. Furthermore, recurring action relationships under the same processing intent are grouped together to form processing action association feature data.

[0041] Based on the processing of action-related feature data, candidate processing actions are combined into paths. During the combination process, the execution order of actions is used as an arrangement constraint, connecting actions that satisfy the dependency relationship to form multiple possible processing execution paths. Action combinations with conflicting relationships, such as two actions that both modify the same order field but with different results, are not included in the same path combination; action combinations with continuous state transition relationships are connected according to their actual execution order. Through this method, discrete processing actions are transformed into a set of candidate processing paths with an execution order.

[0042] After forming a set of candidate processing paths, the time consumption of each processing path is calculated. During the calculation, the execution time of each processing action in the path is accumulated, and combined with the waiting time between actions and the state transition time, the overall time consumption of each processing path is obtained. Subsequently, this time consumption is matched with the executable processing time range in the order processing constraint data to determine whether the corresponding processing path can be completed within the current dynamic processing window.

[0043] Candidate processing paths that meet the time constraints are retained and further filtered based on the corresponding state change process, forming a set of time-adapted processing paths. When the estimated execution time of a processing path falls within the current order's executable time range, and all actions within the path conform to the current order's state constraints, the path is considered to meet the current order's processing conditions. By filtering candidate processing paths based on time adaptation, execution plans that cannot be completed within the current processing window are reduced, providing effective candidate paths for subsequent state adaptation analysis and order processing plan determination.

[0044] Based on the time-adapted processing path set, extract the state change process corresponding to each candidate processing path, determine the connection relationship between the state change process and the current order processing status, and generate a state-adapted processing path set. Based on the set of state-adaptive processing paths, the information of the processing actions corresponding to each processing path is extracted to generate path action feature data. Based on the path action feature data, the repetition of processing actions between different processing paths is compared to generate path action association feature data; Based on path action association feature data, the execution order relationship and state change process of different processing paths are analyzed to determine the execution conflict and state connection between processing paths and generate path compatibility feature data. Based on path compatibility feature data, processing paths that meet time adaptation requirements, state adaptation requirements, and do not have execution conflicts are filtered to generate a target processing path set. Based on the target processing path set, the processing actions corresponding to each processing path are combined and encapsulated to generate a candidate order processing solution set.

[0045] Specifically, in the order processing path generation process, merely meeting the time requirements does not guarantee that the processing can be executed directly. It is also necessary to determine the connection between each processing path and the current order status, as well as whether there are issues such as duplicate actions, state conflicts, or inconsistent execution order between different paths. Therefore, it is necessary to further analyze the time-adapted processing paths, select processing paths that can be continuously connected with the current order status and do not have execution conflicts, and combine multiple feasible paths to form candidate order processing solutions.

[0046] During state adaptation analysis, the state change process corresponding to each processing action in the time-adapted processing path is extracted. The order state changes before and after each action are recorded according to the action execution order and matched with the current actual state of the target order. During the matching process, it is determined whether the initial state of the processing path is consistent with the current order state, and it is also analyzed whether subsequent state changes can be completed continuously in a predetermined order. If the first action to be executed in the processing path can affect the current order state, and the state changes generated by subsequent actions can be progressively connected, then the processing path is considered to meet the state adaptation condition. For example, if the current order has completed payment confirmation and needs to enter the delivery adjustment process, a path containing only subsequent actions such as address verification and delivery information updates can connect with the current state, while a path that re-executes payment confirmation does not meet the state adaptation requirement.

[0047] After completing the state adaptation judgment, the action composition relationships in the retained processing paths are organized. During processing, the types, number, execution order, and corresponding state changes of each action are extracted and grouped according to path identifiers to form path action feature data. This data describes the specific processing content contained in different processing paths, enabling subsequent comparison of action relationships between different paths. For example, if two processing paths both contain information verification actions, but one path subsequently enters delivery adjustment while the other subsequently enters refund processing, the action composition features can distinguish the corresponding processing directions of the two paths.

[0048] After obtaining the path action feature data, the action composition of different processing paths is compared. During the comparison, the processing actions contained in each path are represented in a set format to determine whether there are identical actions, duplicated actions, or mutually substitutable actions between different paths, and the relationships between actions are recorded. When multiple paths contain the same processing action, the duplicate action is marked to avoid executing the same processing content multiple times in subsequent scheme combination processes; when different paths contain mutually exclusive actions, the corresponding conflict relationship is recorded, thus forming path action association feature data.

[0049] Based on path action association feature data, a comprehensive analysis is performed on the execution order relationship and state change process between different processing paths. During the analysis, the order of actions in different paths is first compared to determine if there is a dependency relationship, and then it is determined whether the state changes generated after the execution of each path can be continuously connected. When two paths contain the same actions but with different execution orders, or when the state result of one path affects the execution conditions of subsequent actions in another path, it is determined that there is an execution conflict or abnormal state connection between the two paths, and the conflict location and scope of influence are recorded. Through this method, path compatibility feature data that can describe the degree of compatibility between paths is formed.

[0050] After completing the path compatibility analysis, candidate processing paths are filtered based on path compatibility feature data. During the filtering process, time adaptation results, state adaptation results, and path conflicts are considered simultaneously. Only processing paths that meet the requirements of the current order processing window, can connect to the current order state, and have no execution conflicts are retained. When multiple processing paths meet the above conditions, the differences between each path are retained, allowing for the selection of appropriate combinations based on different processing needs, thus forming a target set of processing paths.

[0051] After obtaining the set of target processing paths, the action relationships within each path are combined and encapsulated. During processing, based on the action relationships between paths, actions that can be executed consecutively and without conflict are integrated, while retaining the execution order, state changes, and time consumption information for each action, ensuring that the combined processing solution still possesses complete execution logic. By combining different target processing paths, multiple candidate order processing solutions are formed, providing candidate objects for subsequent comprehensive evaluation of solutions and selection of the optimal processing solution.

[0052] Furthermore, a comprehensive evaluation of the candidate order processing solutions is performed to generate the target order processing solution, specifically including: Based on the set of candidate order processing solutions, extract the number of processing actions, the order of action execution, and the state change process corresponding to each candidate order processing solution to generate solution feature data; Based on the scheme feature data, the number of processing actions and the repetition of actions corresponding to each candidate order processing scheme are analyzed to generate scheme action evaluation data; Based on the scheme feature data, the execution order of actions corresponding to each candidate order processing scheme is analyzed to determine the continuous execution of processing actions and generate scheme execution evaluation data. Based on the scheme feature data, the state change process corresponding to each candidate order processing scheme is analyzed, the consistency between the processing state transition and the order processing intention is determined, and scheme state evaluation data is generated. Based on the action evaluation data, execution evaluation data, and status evaluation data of the scheme, the action evaluation results, execution evaluation results, and status evaluation results corresponding to each candidate order processing scheme are correlated and integrated to generate scheme evaluation result data. Based on the evaluation results data, the evaluation levels of different candidate order processing schemes are compared to determine the priority order among the candidate order processing schemes and generate a scheme ranking sequence. Based on the ranking sequence of the solutions, the candidate order processing solution with the highest ranking is selected to determine the target order processing solution. Specifically, during the automated order processing, the set of candidate order processing solutions may contain multiple solutions that meet time constraints, state constraints, and compatibility requirements. However, different solutions differ in execution complexity, action continuity, and degree of matching with the order processing intent. Therefore, it is necessary to comprehensively evaluate the candidate solutions from three dimensions: processing action composition, execution process, and state changes, and select the processing solution that best matches the current order requirements from multiple executable solutions.

[0053] During the scheme evaluation process, the processing procedures corresponding to each candidate order processing scheme are first structured and organized. For each candidate scheme, the types of processing actions, the number of actions, the order of action execution, and the state change information generated during execution are extracted and grouped according to the scheme identifier. Among them, the number of processing actions reflects the scale of operations that need to be completed during the scheme execution, the order of action execution describes the sequential dependencies between actions, and the state change process represents the trajectory of the order as it transitions from the current state to the target processing state. The above information forms the scheme feature data.

[0054] During the action evaluation process, the composition of processing actions in each candidate solution is analyzed. During processing, the actions included in each candidate solution are broken down, and the frequency of different actions within the same solution is counted. Simultaneously, duplicate actions are compared between different solutions. Solutions containing a large number of duplicate actions indicate potentially redundant processing procedures; solutions with fewer actions that effectively cover the target processing intent have higher execution efficiency. For example, in order information adjustment processing, if one solution includes multiple identical information verification actions, while another solution only retains necessary verification steps, the latter's action composition is more reasonable. By comprehensively analyzing the number of actions, the degree of action repetition, and action coverage, solution action evaluation data is generated.

[0055] After completing the action evaluation, the execution order of the candidate solutions is analyzed. During the analysis, based on the action arrangement relationships in the solution feature data, it is determined whether adjacent processing actions satisfy execution dependencies, and the number of consecutively executed actions and the maintenance of their order are counted. When multiple actions can be executed consecutively according to the order processing flow, the solution is considered to have good execution continuity; when there are situations where actions require waiting for the completion of a preceding state or the execution order is unreasonable, the corresponding execution evaluation result is reduced. By comparing the action connection relationships of different candidate solutions, solution execution evaluation data is formed.

[0056] During the status evaluation process, the order status changes before and after the execution of each candidate solution are analyzed. In processing, the status transition paths generated during solution execution are compared with the target status change direction corresponding to the target order processing intent to determine whether the solution can make the order status change in the expected direction. The judgment process focuses on analyzing whether the status transitions are continuous, whether there are invalid state jumps, and whether the final state conforms to the order processing intent. For example, if the order processing intent is to complete the delivery information adjustment, if a solution can change the order status from "pending adjustment" to "delivery rescheduling," then the status change process corresponding to that solution is more consistent with the processing intent. The above analysis forms the solution status evaluation data.

[0057] After obtaining the action evaluation data, execution evaluation data, and status evaluation data, the evaluation results corresponding to different evaluation dimensions are correlated and organized. During processing, the action composition, execution process, and status changes corresponding to each candidate order processing solution are grouped according to the solution identifier, forming a complete evaluation description from multiple evaluation results for the same candidate order processing solution. Specifically, the action evaluation data reflects the number of processing actions, the repetition of actions, and the rationality of action composition; the execution evaluation data reflects the continuous execution of processing actions and the matching of execution order; and the status evaluation data reflects the correspondence between the order status transition process and the target order processing intent. During the correlation and organization process, the results corresponding to different evaluation dimensions are uniformly expressed and combined and stored according to the candidate order processing solutions, so that each candidate solution corresponds to solution evaluation result data containing action evaluation results, execution evaluation results, and status evaluation results. For example, if a candidate order processing solution contains few repetitive actions, and the actions can be executed continuously according to the order processing flow, and the order status can be transitioned to the target processing status after execution, then the solution evaluation result data for this solution records the matching situation under the corresponding evaluation dimension, providing a basis for subsequent solution ranking.

[0058] After obtaining the evaluation results data, the comprehensive evaluation results of different candidate order processing solutions are compared and ranked. During the ranking process, solutions are arranged according to their evaluation level or comprehensive score, with those having higher evaluation results, better matching execution conditions, and more compliant state transitions placed first. After ranking, a priority sequence of candidate order processing solutions is formed.

[0059] Based on the ranking sequence of the proposed solutions, the candidate order processing solution with the highest ranking is selected as the target order processing solution. During this process, not only are the comprehensive evaluation results of the solutions considered, but also the action relationships, execution order, and state change paths of the corresponding solutions are preserved, ensuring that subsequent execution phases can automatically process orders according to the established processing logic. The final determined target order processing solution serves as the basis for the automatic processing of the current order.

[0060] Furthermore, based on the target order processing plan, automatic order processing is executed, order processing feedback data is obtained, and the order processing intent data, order processing window data, and order compatibility data are updated according to the order processing feedback data to generate automatic order processing results, specifically including: Based on the target order processing plan, extract the corresponding processing action composition information, execution sequence information, and processing status change information to generate order execution control data; Based on the order execution control data, the processing actions corresponding to the target order are executed sequentially, and the status change information during the execution of each processing action is obtained to generate order execution status data; Based on the order execution status data, extract the processing completion status, abnormal status and processing result change information of the target order, and generate order processing feedback data. Based on order processing feedback data, a correlation analysis is performed on the actual processing process of the target order and the intent content corresponding to the order processing intent data to generate order processing intent adjustment feature data. Based on order processing feedback data and order processing intent adjustment feature data, the corresponding processing intent content in the order processing intent data is updated to generate updated order processing intent data. Based on the order processing feedback data, extract the time execution information corresponding to the actual order processing process, update the processing time range in the order processing window data, and generate updated order processing window data; Based on order processing feedback data, extract information on action connection changes, status changes, and execution conflicts during the execution of the target order processing solution, and update the order compatibility data accordingly. Based on the updated order processing intent data, order processing window data, and order compatibility data, generate automatic order processing results.

[0061] Specifically, during automated order processing, after the target order processing plan is executed, the actual processing may be affected by changes in user operations, business status, and execution environment, leading to differences between the actual execution result and the initial plan. Therefore, it is necessary to continuously collect data on status changes, anomalies, and processing results during order execution, and dynamically adjust the order processing intent, processing time window, and path compatibility based on feedback information, so that subsequent order processing can be optimized according to the actual execution situation.

[0062] When executing a target order processing plan, the relationships between the processing actions in the plan are first analyzed, and the execution order, triggering conditions, and state change relationships of each action are organized. During processing, execution control information is generated according to the action sequence determined by the plan, and the execution conditions, associated order status, and expected change results for each action are bound together. For example, if an order processing plan includes three consecutive actions: information verification, delivery adjustment, and status confirmation, the execution control information records the sequential relationship between the three actions and the state conditions that trigger the next action after the previous action is completed, thus forming order execution control data.

[0063] During order execution, corresponding processing actions are triggered sequentially based on order execution control data, and status change information is collected in real time. The collected information includes whether the action is completed, the current order status, the time of status change, and intermediate results generated during execution. During processing, the actual execution status is compared with the expected status changes of the corresponding actions in the plan. Normal transitions, delayed changes, or abnormal states that occur during action execution are recorded and correlated according to the action execution order to form order execution status data.

[0064] After collecting the execution status data, the order execution status data is organized, extracting the final completion status, execution exception status, and changes in processing results. During processing, the actual execution result is compared with the target status in the target order processing plan to determine whether the order was processed as expected. For cases where the target status was not achieved, the location, type, and scope of the exception are further analyzed. For example, during the execution of delivery adjustments, if the delivery information is successfully updated but the logistics task regeneration fails, the corresponding exception status and changes in processing results are recorded, thus forming order processing feedback data.

[0065] After obtaining order processing feedback data, the correspondence between the actual processing process and the initial order processing intent is analyzed. During the analysis, the state change paths, anomalies, and final processing results generated during actual execution are compared with the historical change patterns corresponding to the order processing intent to determine whether the current actual order processing direction is consistent with the original identification result. When frequent state transitions different from the initial intent occur during actual processing, or when the final processing result deviates from the original processing intent, the corresponding change features are extracted to form order processing intent adjustment feature data.

[0066] Based on order processing feedback data and order processing intent adjustment feature data, the original order processing intent is updated. During the update process, newly added status change nodes, exception handling records, and final completion results during actual execution are integrated with the original intent features to adjust the correlation between corresponding processing intents. For example, in historical data, a certain type of order typically shows delivery adjustment needs, but in actual execution, a large number of orders ultimately enter the refund process. Therefore, this status change pattern is added to the processing intent judgment criteria, enabling subsequent similar orders to obtain a more accurate processing direction.

[0067] Simultaneously, the original order processing window is dynamically adjusted based on the actual execution time information in the order processing feedback data. During processing, the actual start time, completion time, and stage transition time of each processing action are extracted and compared with the corresponding time range of the original processing window. When the actual execution time deviates from the original time range for a long period, the time variation pattern of historical orders under the same processing intent is re-analyzed, and the time range of the corresponding processing stage is adjusted. For example, if the review time required for processing a certain type of order gradually increases from 1 hour in historical statistics to more than 2 hours, the corresponding processing window is updated to make the subsequent order processing time arrangement more consistent with the actual execution situation.

[0068] During the update of path compatibility relationships, information on action transitions, state changes, and execution conflicts generated during the execution of the target order processing plan is extracted. During processing, the action relationships in the actual execution process are compared with the original order compatibility data to determine whether there are execution deviations, abnormal state transitions, or new associations between the original action combinations. When certain action combinations frequently experience execution conflicts or abnormal state transitions during actual execution, the compatibility relationships between the corresponding action combinations are adjusted; when certain action combinations can consistently complete state transitions according to the predetermined execution order, the association characteristics of the corresponding action combinations are retained or enhanced, forming updated order compatibility data.

[0069] After completing the above updates, the adjusted order processing intent data, order processing window data, and order compatibility data will be integrated to form an automatic order processing result. This result not only includes the actual processing status and completion status of the current order, but also the processing patterns after feedback correction. This allows subsequent orders of the same type to be automatically processed based on the updated processing intent recognition, time window determination, and path matching rules, improving the adaptability of the order processing process.

[0070] Furthermore, a big data-based automated e-commerce order processing system is proposed to implement any of the above-mentioned automated order processing methods, including: The order analysis module is used to obtain the order data and order behavior data corresponding to the target order, and perform correlation and integration processing to generate order analysis data. The intent recognition module is used to identify the order processing intent of the target order based on the order analysis data, combined with the current processing status of the target order and the full historical behavior information of the order, and generate order processing intent data. The window generation module is used to analyze the processing time requirements and processing priorities of target orders based on order processing intent data, determine the dynamic processing window corresponding to the target order, and generate order processing window data. The compatibility matching module is used to perform order compatibility matching based on the order processing window data, generate order compatibility data, and obtain a set of candidate order processing solutions. The solution evaluation module is used to comprehensively evaluate the set of candidate order processing solutions and generate the target order processing solution. The feedback update module is used to automatically process orders according to the target order processing plan, obtain order processing feedback data, and update the order processing intent data, order processing window data, and order compatibility data based on the order processing feedback data to generate automatic order processing results.

[0071] Furthermore, the intent recognition module includes: The basic extraction unit is used to extract and process order status information, product information, transaction information and logistics information from order data to generate basic order data. The behavioral feature unit is used to extract order operation time, operation type, and behavioral change information based on order behavior data, and generate order behavior feature data. The state sequence generation unit is used to divide the current processing state of the target order according to the order feature set, and to mark the state change process corresponding to the payment state, delivery state, logistics state and after-sales state to generate an order state change sequence. The historical association unit is used to obtain the full historical behavior information corresponding to the target order, extract the historical processing status change nodes and corresponding operation types, generate historical behavior sequences, and generate a set of historical behavior association features based on the historical behavior sequences. The intent matching unit is used to perform association matching processing between the order status change sequence and the corresponding status change node based on the historical behavior association feature set, and generate the order processing intent matching result. The intent determination unit is used to determine the target order processing intent based on the order processing intent matching result and generate order processing intent data.

[0072] Furthermore, the compatibility matching module includes: The constraint generation unit is used to extract the executable processing time range and current order processing status information corresponding to the target order based on the order processing window data, and generate order processing constraint data. The action decomposition unit is used to decompose the processing intent content corresponding to the target order into action based on the order processing constraint data and the order processing intent data, and generate a set of candidate processing actions. The path generation unit is used to extract the execution sequence, state change information and time occupation information of each processing action based on the candidate processing action set, and generate a candidate processing path set according to the execution sequence. The time matching unit is used to match the time occupancy information corresponding to the candidate processing path set with the executable processing time range in the order processing constraint data to generate a time-adapted processing path set. The state matching unit is used to generate a state adaptation processing path set based on the connection relationship between the state change process corresponding to the time adaptation processing path set and the current order processing status. The solution generation unit is used to combine and encapsulate the processing actions corresponding to the target processing path set to generate a set of candidate order processing solutions.

[0073] Furthermore, the feedback update module includes: The execution control unit is used to extract the corresponding processing action composition information, execution sequence information, and processing status change information based on the target order processing scheme, and generate order execution control data. The status acquisition unit is used to sequentially execute the processing actions corresponding to the target order based on the order execution control data, and to acquire the status change information during the execution of each processing action to generate order execution status data. The feedback generation unit is used to extract the processing completion status, abnormal status and processing result change information of the target order based on the order execution status data, and generate order processing feedback data. The intent update unit is used to perform correlation analysis between the actual processing process of the target order and the intent content corresponding to the order processing intent data based on the order processing feedback data, and update the order processing intent data. The window update unit is used to extract the time execution information corresponding to the actual order processing process based on the order processing feedback data, and update the order processing window data. The compatibility update unit is used to extract action connection change information, status change information, and execution conflict information during the execution of the target order processing scheme based on the order processing feedback data, and update the order compatibility data.

[0074] In summary, the advantages of this invention are as follows: By associating and integrating order data, order behavior data, and all historical behavior information, the correspondence between order status changes and historical processing results is analyzed, enabling accurate identification of order processing intent; simultaneously, a dynamic processing window is determined based on the historical processing process corresponding to the order processing intent, and order compatibility matching is performed by combining the time constraints, status connections, and execution conflicts between processing actions to generate a processing scheme adapted to the current order status; furthermore, the order processing intent data, order processing window data, and order compatibility data are updated based on order processing feedback data, enabling the automatic order processing process to be dynamically adjusted according to the actual execution situation, thereby improving the accuracy, adaptability, and automation level of the order processing process.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for automatically processing e-commerce orders based on big data, characterized in that, include: Obtain the order data and order behavior data corresponding to the target order, and perform correlation and integration processing to generate order analysis data; Based on order analysis data, combined with the current processing status of the target order and the full historical behavior information of the corresponding order, the order processing intent of the target order is identified, and order processing intent data is generated. Based on order processing intent data, the processing time requirements and processing priorities of target orders are analyzed to determine the dynamic processing window corresponding to the target order and generate order processing window data. Based on the order processing window data, perform order compatibility matching to generate order compatibility data and obtain a set of candidate order processing solutions; The candidate order processing solutions are comprehensively evaluated to generate the target order processing solution. The system executes automatic order processing according to the target order processing plan, obtains order processing feedback data, and updates the order processing intent data, order processing window data, and order compatibility data based on the order processing feedback data to generate automatic order processing results.

2. The method for automatically processing e-commerce orders based on big data according to claim 1, characterized in that, The step of identifying the order processing intent corresponding to the target order based on order analysis data, combined with the current processing status of the target order and the corresponding full historical behavior information, and generating order processing intent data specifically includes: The order status information, product information, transaction information, and logistics information in the order data are extracted and processed to generate basic order data; Based on order behavior data, extract order operation time, operation type, and behavior change information to generate order behavior feature data; By linking and integrating basic order data with order behavior characteristic data, order analysis data is generated; Based on order analysis data, the information corresponding to the basic order data and order behavior feature data is processed in a unified manner, and features are aggregated according to the target order identifier to generate an order feature set; Based on the order feature set, the current processing status of the target order is divided into states, and the state change processes corresponding to payment status, shipping status, logistics status and after-sales status are marked to generate an order status change sequence. Obtain the full historical behavior information corresponding to the target order, extract the historical processing status change nodes and corresponding operation types, and generate a historical behavior sequence; Based on the historical behavior sequence, the state change nodes and corresponding operation types in different historical processing processes are associated with the final processing results, and the distribution of processing results corresponding to different state change nodes and corresponding operation types is statistically analyzed to generate a set of historical behavior association features. Based on the historical behavior association feature set, the order status change sequence is associated and matched with the corresponding status change node in the historical behavior association feature set to generate order status association feature data. Based on the order status association feature data, the status change nodes and operation types corresponding to different historical processing results are classified and processed, and the common processing features of each historical processing process are extracted to generate a set of candidate order processing intentions. Based on the candidate order processing intent set, the order status change sequence is matched with the historical status change nodes corresponding to each candidate order processing intent node by node, the number of consecutive matching nodes and the node matching order are counted, and the order processing intent matching result is generated. Based on the order processing intent matching results, the candidate order processing intent with the most consecutive matching nodes and the same node matching order is determined as the target order processing intent, and order processing intent data is generated.

3. The method for automatically processing e-commerce orders based on big data according to claim 1, characterized in that, The process, based on order processing intent data, analyzes the processing time requirements and priorities of target orders to determine the corresponding dynamic processing window for each target order and generates order processing window data. Specifically, this includes: Based on order processing intent data, extract the order processing intent, historical processing process and corresponding status change information of the target order to generate order processing demand feature data. Based on the order processing demand characteristic data, the historical processing processes corresponding to the same order processing intent are sorted out, the processing time change information corresponding to each historical processing process is extracted, and a set of processing timeliness characteristics is generated. Based on the processing timeliness feature set, the historical processing process corresponding to the target order is matched with the historical processing time change information in the processing timeliness feature set to determine the timeliness requirement corresponding to the target order and generate the timeliness requirement data of the target order. Based on the timeliness requirements of the target orders, filter historical processing records that meet the timeliness requirements of the target orders and generate a target processing record set. Based on the target processing record set, the historical processing sequence of the same order processing intent is statistically analyzed, and the order is sorted according to the frequency of occurrence of the historical processing sequence to generate order processing priority data; Based on order processing priority data, extract historical processing time range information under the corresponding priority to generate processing time range feature data; Based on the processing time range feature data, the start and end times of the corresponding historical processing time range are extracted, and time mapping processing is performed in combination with the target order generation time to generate dynamic processing window data. Based on the dynamically processed window data, the executable processing time corresponding to the current processing status of the target order is marked, and order processing window data is generated.

4. The method for automatically processing e-commerce orders based on big data according to claim 1, characterized in that, The step of performing order compatibility matching based on order processing window data to generate order compatibility data and obtain a set of candidate order processing solutions specifically includes: Based on the order processing window data, extract the executable processing time range and current order processing status information corresponding to the target order to generate order processing constraint data; Based on the order processing constraint data and the order processing intent data, the processing intent content corresponding to the target order is decomposed into actions to generate a set of candidate processing actions; Based on the candidate processing action set, the execution sequence, state change information and time occupation information of each processing action are extracted to generate processing action association feature data. Based on the associated feature data of the processing actions, the candidate processing actions are combined according to their execution order to generate a set of candidate processing paths. Based on the candidate processing path set, the time occupancy information corresponding to each candidate processing path is matched with the executable processing time range in the order processing constraint data, and candidate processing paths that meet the time requirements are filtered to generate a time-adaptive processing path set. Based on the time-adapted processing path set, extract the state change process corresponding to each candidate processing path, determine the connection relationship between the state change process and the current order processing status, and generate a state-adapted processing path set. Based on the set of state-adaptive processing paths, the information of the processing actions corresponding to each processing path is extracted to generate path action feature data. Based on the path action feature data, the repetition of processing actions between different processing paths is compared to generate path action association feature data; Based on path action association feature data, the execution order relationship and state change process of different processing paths are analyzed to determine the execution conflict and state connection between processing paths and generate path compatibility feature data. Based on path compatibility feature data, processing paths that meet time adaptation requirements, state adaptation requirements, and do not have execution conflicts are filtered to generate a target processing path set. Based on the target processing path set, the processing actions corresponding to each processing path are combined and encapsulated to generate a candidate order processing solution set.

5. The method for automatically processing e-commerce orders based on big data according to claim 1, characterized in that, The process of comprehensively evaluating the set of candidate order processing solutions to generate a target order processing solution specifically includes: Based on the set of candidate order processing solutions, extract the number of processing actions, the order of action execution, and the state change process corresponding to each candidate order processing solution to generate solution feature data; Based on the scheme feature data, the number of processing actions and the repetition of actions corresponding to each candidate order processing scheme are analyzed to generate scheme action evaluation data; Based on the scheme feature data, the execution order of actions corresponding to each candidate order processing scheme is analyzed to determine the continuous execution of processing actions and generate scheme execution evaluation data. Based on the scheme feature data, the state change process corresponding to each candidate order processing scheme is analyzed, the consistency between the processing state transition and the order processing intention is determined, and scheme state evaluation data is generated. Based on the action evaluation data, execution evaluation data, and status evaluation data of the scheme, the action evaluation results, execution evaluation results, and status evaluation results corresponding to each candidate order processing scheme are correlated and integrated to generate scheme evaluation result data. Based on the evaluation results data, the evaluation levels of different candidate order processing schemes are compared to determine the priority order among the candidate order processing schemes and generate a scheme ranking sequence. Based on the sorting sequence of the schemes, the candidate order processing scheme with the highest sorting result is selected to determine the target order processing scheme.

6. The method for automatically processing e-commerce orders based on big data according to claim 1, characterized in that, The process involves automatically processing orders according to the target order processing plan, obtaining order processing feedback data, and updating the order processing intent data, order processing window data, and order compatibility data based on the order processing feedback data to generate automatic order processing results. Specifically, this includes: Based on the target order processing plan, extract the corresponding processing action composition information, execution sequence information, and processing status change information to generate order execution control data; Based on the order execution control data, the processing actions corresponding to the target order are executed sequentially, and the status change information during the execution of each processing action is obtained to generate order execution status data; Based on the order execution status data, extract the processing completion status, abnormal status and processing result change information of the target order, and generate order processing feedback data. Based on order processing feedback data, a correlation analysis is performed on the actual processing process of the target order and the intent content corresponding to the order processing intent data to generate order processing intent adjustment feature data. Based on order processing feedback data and order processing intent adjustment feature data, the corresponding processing intent content in the order processing intent data is updated to generate updated order processing intent data. Based on the order processing feedback data, extract the time execution information corresponding to the actual order processing process, update the processing time range in the order processing window data, and generate updated order processing window data; Based on order processing feedback data, extract information on action connection changes, status changes, and execution conflicts during the execution of the target order processing solution, and update the order compatibility data accordingly. Based on the updated order processing intent data, order processing window data, and order compatibility data, generate automatic order processing results.

7. A big data-based e-commerce order automatic processing system, used to implement the order automatic processing method as described in any one of claims 1-6, characterized in that, include: The order analysis module is used to acquire the order data and order behavior data corresponding to the target order, and perform correlation and integration processing to generate order analysis data. The intent recognition module is used to identify the order processing intent of the target order based on the order analysis data, combined with the current processing status of the target order and the full historical behavior information corresponding to the order, and generate order processing intent data. The window generation module is used to analyze the processing time requirements and processing priorities of the target order based on the order processing intent data, determine the dynamic processing window corresponding to the target order, and generate order processing window data. The compatibility matching module is used to perform order compatibility matching processing based on the order processing window data, generate order compatibility data, and obtain a set of candidate order processing solutions. The solution evaluation module is used to comprehensively evaluate the set of candidate order processing solutions and generate a target order processing solution. The feedback update module is used to execute automatic order processing according to the target order processing plan, obtain order processing feedback data, and update the order processing intent data, order processing window data and order compatibility data according to the order processing feedback data to generate automatic order processing results.

8. The e-commerce order automatic processing system based on big data according to claim 7, characterized in that, The intent recognition module includes: The basic extraction unit is used to extract and process order status information, product information, transaction information and logistics information from order data to generate basic order data. A behavior feature unit is used to extract order operation time, operation type and behavior change information based on order behavior data to generate order behavior feature data. A state sequence generation unit is used to perform state division processing on the current processing state of the target order according to the order feature set, and to mark the state change process corresponding to the payment state, delivery state, logistics state and after-sales state to generate an order state change sequence. The historical association unit is used to obtain the full historical behavior information corresponding to the target order, extract the historical processing status change nodes and the corresponding operation types, generate a historical behavior sequence, and generate a historical behavior association feature set based on the historical behavior sequence. The intent matching unit is used to perform association matching processing on the order status change sequence and the corresponding status change node based on the historical behavior association feature set, and generate the order processing intent matching result. An intent determination unit is used to determine the target order processing intent based on the order processing intent matching result and generate order processing intent data.

9. The e-commerce order automatic processing system based on big data according to claim 7, characterized in that, The compatibility matching module includes: The constraint generation unit is used to extract the executable processing time range and current order processing status information corresponding to the target order based on the order processing window data, and generate order processing constraint data. An action decomposition unit is used to decompose the processing intent content corresponding to the target order into action based on the order processing constraint data and the order processing intent data, and generate a set of candidate processing actions. A path generation unit is used to extract the execution sequence relationship, state change information and time occupation information of each processing action based on the candidate processing action set, and generate a candidate processing path set according to the execution sequence relationship. A time matching unit is used to match the time occupancy information corresponding to the candidate processing path set with the executable processing time range in the order processing constraint data to generate a time-adapted processing path set. A state matching unit is used to generate a state adaptation processing path set based on the connection relationship between the state change process corresponding to the time adaptation processing path set and the current order processing state. The scheme generation unit is used to combine and encapsulate the processing actions corresponding to the target processing path set to generate a set of candidate order processing schemes.

10. The e-commerce order automatic processing system based on big data according to claim 7, characterized in that, The feedback update module includes: An execution control unit is used to extract corresponding processing action composition information, execution sequence information, and processing status change information based on the target order processing scheme, and generate order execution control data. A status acquisition unit is used to sequentially execute the processing actions corresponding to the target order according to the order execution control data, and acquire the status change information during the execution of each processing action to generate order execution status data. The feedback generation unit is used to extract the processing completion status, abnormal status and processing result change information of the target order based on the order execution status data, and generate order processing feedback data. The intent update unit is used to perform correlation analysis between the actual processing process of the target order and the intent content corresponding to the order processing intent data based on the order processing feedback data, and update the order processing intent data. A window update unit is used to extract the time execution information corresponding to the actual order processing process based on the order processing feedback data, and update the order processing window data. The compatibility update unit is used to extract action connection change information, status change information and execution conflict information during the execution of the target order processing scheme based on the order processing feedback data, and update the order compatibility data.

Images