Timeline anomaly detection methods, systems, devices, and products for flexible workforce

Abstract

This application discloses a method, system, device, and product for detecting timeline anomalies in flexible employment. The method includes: receiving a target personnel identifier and a verification time range; based on a flexible employment timeline intelligent agent, performing the following processing: obtaining a task time series from the flexible employment platform and task execution time records from the customer system according to the target personnel identifier and verification time range; performing multi-dimensional verification on the task time series to determine abnormal time nodes and anomaly types; multi-dimensional verification includes basic temporal logic verification, time interval rationality verification, and multi-source consistency verification; based on an event temporal logic reasoning model, and combined with task execution time records, business scenario rule bases, and anomaly types, performing source tracing analysis on abnormal time nodes to obtain the cause of the anomaly; and generating a timeline risk analysis report. This application can improve the efficiency and accuracy of identifying timeline fraud in flexible employment scenarios.

Description

Technical Field

[0001] This application relates to the fields of flexible employment and artificial intelligence technology, and in particular to a method, system, electronic device, computer-readable storage medium, and computer program product for timeline anomaly detection in flexible employment. Background Technology

[0002] Flexible employment, with its advantages of flexible staffing, controllable costs, and high adaptability, has become an important way for various enterprises to optimize human resource allocation. Flexible employment platforms have also developed rapidly, becoming the core carrier connecting employers and flexible workers. In the entire process of flexible employment, from task assignment and execution to acceptance and settlement, each stage generates corresponding time node data, forming a complete task time sequence. This task time sequence is the core evidence reflecting the authenticity and compliance of flexible employment operations.

[0003] However, the current flexible employment industry faces numerous challenges in terms of complex business scenarios, diverse participants, and fragmented task execution processes. Furthermore, the business data of employers and flexible employment platforms are often stored across different systems, leading to significant difficulties in the collection, verification, and anomaly identification of task time series. Existing technologies typically employ manual verification or simple, single-dimensional checks for time-related risks in flexible employment tasks, only allowing for basic checks on individual time points and failing to achieve comprehensive, systematic verification of the entire task time series. Furthermore, violations such as fraudulent order acceptance, task delegation, and process falsification in flexible employment scenarios are often concealed by altering or forging task time point data. Existing detection methods, lacking professional temporal logic reasoning capabilities and business scenario adaptability, struggle to penetrate such falsified data and accurately pinpoint the root cause of timeline anomalies. This results in difficulties for flexible employment platforms and employers in verifying business authenticity, lack of basis for risk control decisions, and untimely handling of violations, ultimately impacting the compliant operation of flexible employment businesses and potentially leading to financial losses and legal disputes. Summary of the Invention

[0004] In view of the above, embodiments of this application provide a method, system, electronic device, computer-readable storage medium, and computer program product for detecting timeline anomalies in flexible employment, to solve at least one of the above-mentioned technical problems.

[0005] In a first aspect, embodiments of this application provide a timeline anomaly detection method for flexible employment, comprising: receiving a target personnel identifier and a verification time range input by a user; performing the following processing based on a flexible employment timeline intelligent agent: obtaining a task time series from a flexible employment platform and obtaining task execution time records from a customer system based on the target personnel identifier and the verification time range, wherein the task time series includes multiple time nodes of the target personnel in the flexible employment task cycle; performing multi-dimensional verification on the task time series according to preset verification rules and task execution time records, and determining abnormal time nodes and anomaly types in the task time series based on the verification results of the multi-dimensional verification; the multi-dimensional verification includes basic temporal logic verification, time interval rationality verification, and multi-source consistency verification; performing source tracing analysis on abnormal time nodes based on a pre-trained event temporal logic reasoning model, combined with task execution time records, a preset business scenario rule base, and anomaly types, to obtain the anomaly causes that led to the occurrence of abnormal time nodes; and generating a timeline risk analysis report for the target personnel based on the abnormal time nodes, anomaly types, and anomaly causes.

[0006] According to some embodiments of this application, optionally, multiple time nodes include at least some of the following time nodes: registration time, signing time, contract effective time, task dispatch time, task start time, task completion time, acceptance time, settlement time, and invoice issuance time. The multiple time nodes are sorted in chronological order. Based on preset verification rules and task execution time records, the task time sequence is verified in multiple dimensions. Based on the verification results of the multi-dimensional verification, abnormal time nodes and abnormal types in the task time sequence are determined, including: verifying whether the task time sequence corresponding to the target personnel conforms to the preset flexible employment business logic sequence; determining the time nodes in the task time sequence corresponding to the target personnel that do not conform to the flexible employment business logic sequence as abnormal time nodes, and determining the abnormal type as a basic sequence abnormality.

[0007] According to some embodiments of this application, optionally, the task time series is subjected to multi-dimensional verification based on preset verification rules and task execution time records, and abnormal time nodes and abnormal types in the task time series are determined based on the verification results of the multi-dimensional verification, including: verifying whether the time interval between adjacent key time nodes in the task time series is within a preset time range; wherein, adjacent key time nodes include at least one of the following pairs: task dispatch time and task start time, task completion time and task acceptance time, task acceptance time and settlement time, settlement time and invoice issuance time; adjacent key time nodes whose time interval exceeds the preset time range are determined as abnormal time nodes, and the abnormal type is determined to be time interval abnormal.

[0008] According to some embodiments of this application, optionally, the task time series is subjected to multi-dimensional verification based on preset verification rules and task execution time records, and abnormal time nodes and abnormal types in the task time series are determined based on the verification results of the multi-dimensional verification, including: comparing a first key time node in the task time series with a second key time node of the same action in the task execution time record; the first key time node includes at least one of task dispatch time, task completion time, task acceptance time and settlement time, and the second key time node includes at least one of task start time in the customer system, actual task completion time, task result confirmation time and payment instruction issuance time; if the time deviation between the first key time node and the second key time node exceeds a preset threshold, the first key time node and / or the second key time node are determined to be abnormal time nodes, and the abnormal type is determined to be multi-source time inconsistency abnormality.

[0009] According to some embodiments of this application, optionally, the task time series includes at least the task completion time; based on preset verification rules and task execution time records, the task time series is subjected to multi-dimensional verification, and based on the verification results of the multi-dimensional verification, abnormal time nodes and abnormal types in the task time series are determined, further including: obtaining metadata of task completion vouchers associated with the task completion time, the task completion vouchers including captured images or videos, and the metadata including the capture time and upload time; verifying whether the metadata and the task completion time conform to preset business logic; if the metadata and the task completion time do not conform to the business logic, the task completion time is determined as an abnormal time node, and the abnormal type is determined to be an abnormality of the result authenticity.

[0010] According to some embodiments of this application, optionally, the task time series includes task dispatch time, task completion time, contract effective time, and contract termination time; based on preset verification rules and task execution time records, the task time series is subjected to multi-dimensional verification, and based on the verification results of the multi-dimensional verification, abnormal time nodes and abnormal types in the task time series are determined, further including: verifying whether the task dispatch time and task completion time are within the contract validity period between the contract effective time and the contract termination time; if the task dispatch time and / or task completion time are not within the contract validity period, then the task dispatch time and / or task completion time are determined as abnormal time nodes, and the abnormal type is determined to be an abnormal contract validity period time series.

[0011] According to some embodiments of this application, optionally, based on a pre-trained event-sequence logic reasoning model, and combined with task execution time records, a preset business scenario rule base, and exception types, a source analysis is performed on abnormal time nodes to obtain the abnormal causes that led to the abnormal time nodes. This includes: providing the event-sequence logic reasoning model with the task time sequence, task execution time records, multi-dimensional verification results, and exception types marked with abnormal time nodes as combined inputs; the event-sequence logic reasoning model, based on a preset business scenario rule base, matches and analyzes the combined inputs with the rules in the business scenario rule base and outputs the abnormal causes; wherein, the abnormal causes include customer operation errors, system synchronization delays, false task records, or compliance exception operations.

[0012] According to some embodiments of this application, optionally, the event-time logic reasoning model, based on a preset business scenario rule base, matches and analyzes the combined input with the rules in the business scenario rule base, and outputs the cause of the anomaly, including: when the anomaly type is a multi-source time inconsistency anomaly, and the time node in the task execution time record is earlier than the corresponding time node in the task time sequence, and there is a system synchronization delay feature, the cause of the anomaly is determined to be a system synchronization delay; when the anomaly type is a basic time sequence anomaly or a multi-source time inconsistency anomaly, and there is a supplementary record or explanation from the customer system for the anomaly time node, the cause of the anomaly is determined to be a customer operation error; when the anomaly type is a basic time sequence anomaly, a time interval anomaly, or an anomaly of questionable authenticity of the result, and the anomaly features in the combined input match the false task feature rule in the business scenario rule base, the cause of the anomaly is determined to be a false task record; when the anomaly type is a basic time sequence anomaly or a time interval anomaly, and the personnel identifier, task identifier, or project identifier in the combined input matches the applicable subject identifier of the compliance special process rule in the business scenario rule base, the cause of the anomaly is determined to be a compliance special operation.

[0013] According to some embodiments of this application, optionally, the timeline risk analysis report includes a timeline anomaly details list; the timeline anomaly details list includes multiple fields such as personnel identifier, task identifier, anomaly type, anomaly time node identifier, actual time of the anomaly time node, anomaly cause, evidence index, and evidence link.

[0014] According to some embodiments of this application, optionally, the flexible employment timeline agent is also used to perform the following processes: calculating the risk base score of the abnormal time node according to the abnormality type and a preset first scoring rule; calculating the risk additional score of the abnormal time node according to the abnormality cause and a preset second scoring rule; obtaining the risk score of the abnormal time node based on the risk base score and the risk additional score; determining the risk level of the abnormal time node according to the risk score and multiple preset score ranges; wherein, the timeline abnormality details list also includes at least one field of the risk score and risk level of the abnormal time node.

[0015] According to some embodiments of this application, optionally, the timeline risk analysis report also includes a timeline panorama of the target personnel. The timeline panorama displays multiple time nodes in the task time sequence in chronological order. The multiple time nodes are distinguished and marked using a first visual identifier and a second visual identifier. The first visual identifier is used to mark time nodes that have been verified as normal, and the second visual identifier is used to mark abnormal time nodes. The timeline panorama also includes target connecting lines, which are used to mark the temporal relationship between the multiple time nodes.

[0016] Secondly, this application provides a timeline anomaly detection system for flexible employment, characterized by comprising: an interaction module for receiving a target personnel identifier and a verification time range input by a user; and a flexible employment timeline intelligent agent for performing the following processes: obtaining a task time sequence from a flexible employment platform and a task execution time record from a customer system based on the target personnel identifier and the verification time range, wherein the task time sequence includes multiple time nodes of the target personnel in the flexible employment task cycle; performing multi-dimensional verification on the task time sequence based on preset verification rules and task execution time records, and determining abnormal time nodes and anomaly types in the task time sequence based on the verification results of the multi-dimensional verification; the multi-dimensional verification includes basic temporal logic verification, time interval rationality verification, and multi-source consistency verification; performing source tracing analysis on abnormal time nodes based on a pre-trained event temporal logic reasoning model, combined with task execution time records, a preset business scenario rule base, and anomaly types, to obtain the anomaly causes that led to the occurrence of abnormal time nodes; and generating a timeline risk analysis report for the target personnel based on the abnormal time nodes, anomaly types, and anomaly causes.

[0017] Thirdly, embodiments of this application provide an electronic device, which includes a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, it implements the steps of the timeline anomaly detection method for flexible employment as described above.

[0018] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps of the timeline anomaly detection method for flexible employment as described above.

[0019] Fifthly, embodiments of this application provide a computer program product, which includes computer program instructions that, when executed by a processor, implement the steps of the timeline anomaly detection method for flexible employment as described above.

[0020] The timeline anomaly detection method, system, electronic device, computer-readable storage medium, and computer program product for flexible employment provided by the embodiments of this application, on the one hand, collects task time series and execution time records across platforms through a flexible employment timeline intelligent agent. Combined with three types of verification—basic temporal logic, time interval rationality, and multi-source consistency—it can not only identify explicit time node deviations but also recognize implicit anomalies such as temporal logic contradictions and cross-system data inconsistencies. This allows it to identify instances where individuals in flexible employment scenarios are using falsified or altered time nodes to conceal fraudulent order acceptance, subcontracting, and task process manipulation, thereby reducing the risk of financial losses and legal disputes. On the other hand, the flexible employment timeline intelligent agent, based on a pre-trained event temporal logic reasoning model and business scenario rule base, performs full-link logical tracing of abnormal time nodes. It can not only detect alterations at a single time point but also identify the chain of time contradictions caused by the alteration. Furthermore, by combining anomaly types to pinpoint the nature of the fraud, it accurately locates traces of timeline manipulation, helping platforms and employers curb process manipulation and data forgery at the source. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings of the embodiments of this application will be briefly described below.

[0022] Figure 1 This is a flowchart illustrating a timeline anomaly detection method for flexible employment provided in an embodiment of this application.

[0023] Figure 2 This is a schematic flowchart of step S1022 in the timeline anomaly detection method for flexible employment provided in the embodiments of this application.

[0024] Figure 3 Another schematic diagram of S1022 in the timeline anomaly detection method for flexible employment provided in the embodiments of this application.

[0025] Figure 4 This is a flowchart illustrating step S1023 of the timeline anomaly detection method for flexible employment provided in an embodiment of this application.

[0026] Figure 5 This is a structural block diagram of a timeline anomaly detection system for flexible employment provided in an embodiment of this application.

[0027] Figure 6 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0028] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0029] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0030] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0031] Various modifications and variations can be made to this application without departing from its spirit or scope, which will be apparent to those skilled in the art. Therefore, this application is intended to cover modifications and variations falling within the scope of the corresponding claims (the claimed technical solutions) and their equivalents. It should be noted that the implementation methods provided in the embodiments of this application can be combined with each other without contradiction.

[0032] Before describing the technical solutions provided in the embodiments of this application, in order to facilitate understanding of the embodiments of this application, this application first specifically explains the problems existing in the prior art: Flexible employment, with its advantages of flexible staffing, controllable costs, and high adaptability, has become an important way for various enterprises to optimize human resource allocation. Flexible employment platforms have also developed rapidly, becoming the core carrier connecting employers and flexible workers. In the entire process of flexible employment, from task assignment and execution to acceptance and settlement, each stage generates corresponding time node data, forming a complete task time sequence. This task time sequence is the core evidence reflecting the authenticity and compliance of flexible employment operations.

[0033] However, the current flexible employment industry faces numerous challenges in terms of complex business scenarios, diverse participants, and fragmented task execution processes. Furthermore, the business data of employers and flexible employment platforms are often stored across different systems, leading to significant difficulties in the collection, verification, and anomaly identification of task time series. Existing technologies typically employ manual verification or simple, single-dimensional checks for time-related risks in flexible employment tasks, only allowing for basic checks on individual time points and failing to achieve comprehensive, systematic verification of the entire task time series. Furthermore, violations such as fraudulent order acceptance, task delegation, and process falsification in flexible employment scenarios are often concealed by altering or forging task time point data. Existing detection methods, lacking professional temporal logic reasoning capabilities and business scenario adaptability, struggle to penetrate such falsified data and accurately pinpoint the root cause of timeline anomalies. This results in difficulties for flexible employment platforms and employers in verifying business authenticity, lack of basis for risk control decisions, and untimely handling of violations, ultimately impacting the compliant operation of flexible employment businesses and potentially leading to financial losses and legal disputes.

[0034] To address at least one of the aforementioned technical problems, this application provides a method, system, electronic device, computer-readable storage medium, and computer program product for detecting timeline anomalies in flexible employment.

[0035] The following section first introduces the timeline anomaly detection method for flexible employment provided in this application.

[0036] The timeline anomaly detection method for flexible employment provided in this application embodiment can be mainly implemented through a flexible employment timeline intelligent agent. The flexible employment timeline intelligent agent is an AI intelligent agent used for timeline anomaly detection in flexible employment. It is configured with a workflow for cross-system data collection, multi-dimensional time-series verification, anomaly tracing analysis, and risk report generation. It is adapted to the cross-platform, multi-node, and highly time-series characteristics of flexible employment businesses. It is used to identify violations in flexible employment scenarios, such as individuals tampering with or forging false time nodes to cover up false order acceptance, proxy execution, and task process falsification. It can independently complete the entire automated process from data retrieval to risk analysis, improving the efficiency, comprehensiveness, and accuracy of flexible employment timeline anomaly detection.

[0037] Figure 1 This is a flowchart illustrating a timeline anomaly detection method for flexible employment provided in an embodiment of this application. Figure 1 As shown, the timeline anomaly detection method for flexible employment provided in this application embodiment may include the following steps S101, S1021 to S1024.

[0038] S101: Receive the target personnel identifier and verification time range input by the user.

[0039] Users can input the target personnel identifier of the person to be verified, such as platform account, ID number, or employee number, on the operation page of the flexible employment timeline intelligent agent, according to their flexible employment risk control needs. They can also input the verification time range. The verification time range can be any period within the entire cycle of the flexible employment project, covering the entire project cycle, or selecting a single day, week, month, quarter, or custom period within the project. The specific time range can be flexibly adjusted according to actual circumstances, and this application does not impose any limitations on it.

[0040] Based on the flexible employment timeline agent, perform the following steps S1021 to S1024.

[0041] S1021: Based on the target personnel identification and the verification time range, obtain the task time series from the flexible employment platform and the task execution time record from the customer system. The task time series includes multiple time nodes of the target personnel in the flexible employment task cycle.

[0042] The flexible staffing timeline agent can interface with both the flexible staffing platform and the client system. In S1021, the flexible staffing timeline agent can obtain the task time sequence of the target personnel within the verification time range from the flexible staffing platform based on the target personnel identifier and the verification time range. This task time sequence is full-process time-series data recorded on the platform side, which may include some or all time nodes within the flexible staffing task cycle, such as task assignment, order acceptance, start of execution, completion and submission, acceptance, and settlement and payment.

[0043] In S1021, the flexible employment timeline agent can retrieve the task execution time records of the target personnel within the verification time range from the customer system based on the target personnel identification and verification time range. These task execution time records are time-series data recorded on the customer side (employer side), and may include task start time, actual task execution records (such as clock-in time, location upload time, etc.), and some or all of the key time nodes in the four-stream data (such as contract signing time, payment time, invoice issuance time, etc.).

[0044] In some embodiments, the flexible employment timeline agent can also perform format standardization processing on task time series and task execution time records, such as unifying timestamps and field naming, to reduce comparison errors caused by format differences.

[0045] S1022: Based on the preset verification rules and task execution time records, perform multi-dimensional verification on the task time series, and determine the abnormal time nodes and abnormal types in the task time series based on the verification results of the multi-dimensional verification; the multi-dimensional verification includes basic timing logic verification, time interval rationality verification and multi-source consistency verification.

[0046] The flexible employment timeline agent can perform multi-dimensional verification of task time sequences based on preset verification rules and task execution time records. Multi-dimensional verification can include basic time sequence logic verification, time interval reasonableness verification, and multi-source consistency verification. Basic time sequence logic verification checks the temporal sequence of tasks themselves to determine if there are logical contradictions, such as task completion time earlier than order acceptance time, or acceptance time earlier than submission time. Time interval reasonableness verification checks for anomalies in the time intervals between different time nodes in the task time sequence, such as completing highly complex tasks in a short time, or the interval between task execution and acceptance exceeding the normal cycle.

[0047] To conceal violations, fraudulent order acceptance or outsourced execution often involves setting time points that contradict actual business processes, such as having completion times without execution times, or accepting consecutive orders for cross-regional tasks without scheduled travel times. This application's embodiments, through basic timing logic verification and time interval reasonableness verification, can directly identify these illogical time anomalies.

[0048] Multi-source consistency verification can be used to cross-compare the task time series on the platform side with the task execution time records on the customer side to determine whether there is a discrepancy in the time data at the same node, such as a large difference between the task completion time recorded by the platform and the actual acceptance time by the customer, or no execution record on the customer side after the platform shows that the order has been accepted.

[0049] Task time series and task execution time records are independent data sources across entities and systems. Those who engage in fraudulent order taking or proxy execution can typically only tamper with the time data of one party (such as the task completion time on the flexible employment platform), but cannot simultaneously tamper with the actual execution records of the customer system (such as the task acceptance time or attendance records on the customer's side). This application's embodiment cross-compares the data from both parties through multi-source consistency verification, directly identifying inconsistencies in the time data and accurately demonstrating how individuals in flexible employment scenarios can conceal fraudulent order taking, proxy execution, and task process falsification by altering or forging time data from a single platform.

[0050] After multi-dimensional verification, the flexible employment timeline agent can mark time nodes with deviations, contradictions, or unreasonableness as abnormal time nodes, and match the corresponding abnormal type according to the verification type, such as time sequence logic contradictions, time interval exceeding limits, and inconsistencies in multi-source data.

[0051] S1023: Based on a pre-trained event sequence logic reasoning model, and combined with task execution time records, a preset business scenario rule base, and exception types, perform source analysis on abnormal time nodes to obtain the abnormal causes that led to the occurrence of abnormal time nodes.

[0052] The event-time logic reasoning model can be a time-series logic reasoning model adapted to the timeline risk control scenario of flexible employment, such as a general large model, a temporal convolutional network (TCN) model, a long short-term memory network (LSTM) model, a graph neural network (GNN) model, or a multi-model fusion architecture (such as a TCN+LSTM model), etc. This application does not limit it.

[0053] When training an event-driven temporal logic reasoning model, real business data from the flexible employment industry can be used as the training set. The training set can include a predetermined number of sample data points: task time series marked with abnormal time nodes, task execution time records, multi-dimensional verification results, anomaly types, and anomaly causes. The sample data can cover different job positions (such as remote part-time work, on-site service, project outsourcing, etc.), different task types, and different violation scenarios (such as fraudulent order taking, proxy execution, data tampering, process fraud, etc.). Simultaneously, the business specifications, temporal logic requirements, and risk control rules of the flexible employment industry can be transformed into structured feature data and integrated into the training process.

[0054] During training, the mapping relationship between abnormal features and abnormal causes can be used as the learning objective, and model parameters can be iteratively adjusted through forward and backward propagation. In some embodiments, a preset business scenario rule base can be introduced as a constraint during training to prevent the model's inference results from violating the actual business logic of the industry, until the model's source tracing accuracy and / or recall rate reach a preset threshold, thus completing model training.

[0055] The event-time logic reasoning model is primarily used for precise source analysis of abnormal time nodes in the time series of flexible employment tasks, uncovering implicit correlations between abnormal nodes, such as distinguishing between time data deviations caused by operational errors and timeline anomalies caused by subjective fraud, providing accurate root cause evidence for risk control. In S1023, the flexible employment timeline agent can input the task time series marked with abnormal time nodes, task execution time records, multi-dimensional verification results, and anomaly types into the event-time logic reasoning model. The event-time logic reasoning model extracts features from the input multi-dimensional data, retrieves a preset business scenario rule base, and matches the extracted abnormal features with the matching rules for abnormal features and causes in the business scenario rule base to determine the corresponding abnormal cause.

[0056] For example, taking an offline installation task as an example, its business scenario rules are: check-in → execution → submission → acceptance. Submission and check-in must be at least 3 hours apart, and acceptance must occur after submission with at least a 1-day interval. Assume the target employee's actual task timeline is: 9:00 AM check-in → 12:00 PM submission → 10:00 AM acceptance the next day, which conforms to the business scenario rules. However, to leave early, the target employee alters the platform submission time to 10:00 AM. The altered task timeline becomes: 9:00 AM check-in → 10:00 AM submission → 10:00 AM acceptance the next day.

[0057] The event sequence logic reasoning model, combined with the business scenario rule base, can identify the following three chain contradictions caused by the tampering: (1) The interval between the submission at 10:00 and the check-in at 9:00 is only 1 hour, which violates the execution time rule of ≥3 hours. (2) The tampered submission time (10:00) is 24 hours away from the actual acceptance time of the customer system (10:00 the next day), which seems to comply with the rule, but combined with the fact that there is no construction progress record on the customer side from 10:00 to 12:00, the logical contradiction of submission without execution is identified. (3) The target person has a check-in record in another area at 10:30, which forms a time conflict of cross-regional operation at the same time with the tampered submission time of 10:00. The event sequence logic reasoning model can identify three chain contradictions: inconsistent working hours, missing execution records and cross-regional time conflict, and completely sort out this contradiction chain, lock the traces of fraud, such as through the cross-verification of multiple node contradictions, determine that the abnormal reason is that the target person tampered with the platform submission time and did not actually complete the prescribed working hours.

[0058] S1024: Generate a timeline risk analysis report for the target personnel based on the abnormal time point, abnormal type, and abnormal cause.

[0059] The flexible employment timeline intelligent agent integrates abnormal time nodes, abnormal types, and abnormal causes to generate a timeline risk analysis report for the target personnel. In some embodiments, the timeline risk analysis report may also include supporting information such as verification process records and model reasoning basis. The timeline risk analysis report supports online preview, export, and / or data traceability, providing risk control personnel with complete and traceable decision-making basis for conducting offline verification, handling violations, or subsequent monitoring.

[0060] The timeline anomaly detection method for flexible employment provided in the embodiments of this application, on the one hand, collects task time series and execution time records across platforms through a flexible employment timeline intelligent agent. Combining this with three types of verification—basic temporal logic, reasonableness of time intervals, and multi-source consistency—it can not only identify explicit time node deviations but also recognize implicit anomalies such as temporal logic contradictions and cross-system data inconsistencies. This allows it to identify instances where individuals in flexible employment scenarios are using falsified or altered time nodes to conceal fraudulent order acceptance, subcontracting, and task process manipulation, thereby reducing the risk of financial losses and legal disputes. On the other hand, the flexible employment timeline intelligent agent, based on a pre-trained event temporal logic reasoning model and a business scenario rule base, performs full-link logical tracing of abnormal time nodes. It can not only detect alterations at a single time point but also identify the chain of time contradictions caused by the alteration. Furthermore, by combining anomaly types to pinpoint the nature of the fraud, it accurately locates traces of timeline manipulation, helping platforms and employers curb violations such as process manipulation and data forgery at the source.

[0061] According to some embodiments of this application, optionally, the multiple time nodes in the task time sequence include at least some of the time nodes such as registration time, signing time, contract effective time, task dispatch time, task start time, task completion time, acceptance time, settlement time, and invoice issuance time, and the multiple time nodes can be sorted in chronological order.

[0062] S1022: Based on the preset verification rules and task execution time records, perform multi-dimensional verification on the task time series, and determine the abnormal time nodes and abnormal types in the task time series based on the verification results of the multi-dimensional verification. This may include the following steps one and two.

[0063] Step 1: Verify whether the task time sequence corresponding to the target personnel conforms to the preset flexible employment business logic sequence.

[0064] The timing sequence of flexible employment business logic can be flexibly set according to industry or business type, and this application does not impose any limitations on it. For example, in some examples, the timing sequence of flexible employment business logic may include at least one of the following: contract effective time is later than signing time, task assignment time is later than contract effective time, task start time is later than task assignment time, task completion time is later than task start time, acceptance time is later than task completion time, settlement time is later than acceptance time, and invoice issuance time is later than settlement time. The flexible employment timeline intelligent agent can compare and verify the task time sequence corresponding to the target personnel with the timing sequence of flexible employment business logic one by one.

[0065] Step 2: Identify the time nodes in the task time series corresponding to the target personnel that do not conform to the business logic of flexible employment as abnormal time nodes, and determine the abnormality type as basic time sequence abnormality.

[0066] If the verification finds that any one or more time nodes in the task time series do not conform to the timing of the flexible employment business logic, then the one or more time nodes and their associated time nodes with timing conflicts can be marked as abnormal time nodes, and the abnormality type can be determined as a basic timing abnormality.

[0067] For example, if the task completion time in the task time sequence is earlier than the task start time, which does not conform to the flexible employment business logic sequence of task completion time being later than task start time, then the task completion time and task start time can be marked as abnormal time nodes, and the abnormality type can be determined as a basic time sequence abnormality.

[0068] In this way, through basic temporal logic verification, explicit temporal violations such as reversed node order and contradictory timelines can be quickly identified, and time node forgery without actual business support can be directly identified. At the same time, it can provide accurate type clues and anomaly indications for subsequent source tracing analysis of event temporal logic reasoning models, thereby improving the efficiency and accuracy of timeline anomaly detection as a whole.

[0069] Figure 2 This is a schematic flowchart of step S1022 in the timeline anomaly detection method for flexible employment provided in an embodiment of this application. Figure 2 As shown, according to some embodiments of this application, optionally, S1022: performing multi-dimensional verification on the task time series according to preset verification rules and task execution time records, and determining the abnormal time nodes and abnormal types in the task time series based on the verification results of the multi-dimensional verification, may include the following steps S201 and S202.

[0070] S201: Verify whether the time interval between adjacent key time nodes in the task time series is within the preset time range.

[0071] S201 identifies hidden time anomalies, such as intervals that are too short or too long, by verifying whether the intervals between adjacent key time nodes conform to actual business practices. In some examples, adjacent key time nodes may include at least one of the following pairs: task assignment time and task start time, task completion time and task acceptance time, task acceptance time and settlement time, and settlement time and invoice issuance time. These adjacent key time nodes are core time node pairs connecting various business processes in flexible employment scenarios, and are also time node pairs that are prone to fraud. For example, task assignment and start times can be easily tampered with to cover up unprepared or fraudulent order acceptance; acceptance and settlement times can be illegally adjusted to achieve early payment; and settlement and invoice issuance times can be forged to circumvent tax compliance checks. Adjacent key time nodes can be flexibly adjusted according to actual circumstances, and this application does not impose any restrictions on this.

[0072] The preset time ranges corresponding to different adjacent key time nodes can be different, and the preset time ranges corresponding to each adjacent key time node can be flexibly adjusted according to the actual situation. This application does not impose any restrictions on this. For example, taking project outsourcing tasks as an example, the preset interval between task assignment time and task start time is 1 to 2 working days. The preset interval between task completion time and task acceptance time is 1 to 3 working days. The preset interval between task acceptance time and settlement time is 3 to 7 working days. The preset interval between settlement time and invoice issuance time is 1 to 2 working days.

[0073] In S201, the flexible employment timeline agent can extract all adjacent key time node pairs in the task time series and calculate the time interval between each adjacent key time node pair. Then, it compares the time interval between each adjacent key time node with the corresponding preset time range one by one to verify whether the time interval between each adjacent key time node is within the preset time range.

[0074] S202: Identify adjacent critical time nodes whose time intervals exceed the preset time range as abnormal time nodes, and determine the abnormality type as time interval abnormality.

[0075] In S202, the flexible employment timeline agent can identify adjacent key time nodes whose time intervals exceed the preset time range as abnormal time nodes and determine the abnormality type as time interval abnormality.

[0076] In some embodiments, time interval anomalies can be further subdivided into multiple subtypes, such as excessively long interval anomalies and excessively short interval anomalies, while also associating them with corresponding adjacent key time node information, such as an excessively long time interval between task assignment and start, or an excessively short time interval between acceptance and settlement. Additionally, interval deviation values ​​can be associated, i.e., the number of days or hours that the time interval between adjacent key time nodes exceeds or falls short of a preset time range. This allows for more refined anomaly detection results, providing more accurate clues for subsequent event sequence logic reasoning models to conduct source tracing analysis, helping the model quickly locate the cause of the anomaly. For example, an excessively long interval may indicate human delays or process bottlenecks, while an excessively short interval may indicate data tampering or fraudulent processes, further improving the accuracy of timeline anomaly detection and the efficiency of risk control measures.

[0077] In this way, by using the flexible employment timeline intelligent agent to verify the reasonableness of the interval between adjacent key time nodes, it is possible to accurately identify hidden risks where the node sequence is normal but the interval is abnormal, accurately identify violations such as false order acceptance (such as starting immediately after order assignment without reasonable preparation time) and process fraud (such as settling immediately after acceptance without compliance review time), thereby reducing the false alarm rate.

[0078] Figure 3 This is another schematic flowchart of S1022 in the timeline anomaly detection method for flexible employment provided in an embodiment of this application. For example... Figure 3 As shown, according to some embodiments of this application, optionally, S1022: performing multi-dimensional verification on the task time series according to preset verification rules and task execution time records, and determining the abnormal time nodes and abnormal types in the task time series based on the verification results of the multi-dimensional verification, may include the following steps S301 and S302.

[0079] S301: Compare the first critical time node in the task time series with the second critical time node of the same action in the task execution time record.

[0080] S301 identifies anomalies caused by single-source time data tampering or forgery by cross-comparing the time nodes of the same action across the flexible staffing platform and the customer system. In some embodiments, the first key time node is a core node in the task time sequence on the flexible staffing platform side, and the first key time node may include at least one of the following: task dispatch time, task completion time, task acceptance time, and settlement time. The second key time node is a node in the task execution time record on the customer system side that corresponds to the same business action as the first key time node, and the second key time node may include at least one of the following: task start time, actual task completion time, task result confirmation time, and payment instruction issuance time in the customer system.

[0081] The first and second key time nodes are matching node pairs with the same action and across systems. The matching relationship is as follows: the task dispatch time corresponds to the task start time, the task completion time corresponds to the actual task execution completion time, the task acceptance time corresponds to the task result confirmation time, and the settlement time corresponds to the payment instruction issuance time.

[0082] In S301, the flexible employment timeline agent can extract the first key time node to be verified from the task time series on the platform side, and simultaneously extract the corresponding second key time node from the task execution time records on the client system side. Next, the time formats of the first and second key time nodes are standardized, such as unifying timestamps, time zones, and date formats, to reduce the impact of format differences on the comparison results.

[0083] Next, the flexible employment timeline agent compares each first key time node with its corresponding second key time node according to a preset matching relationship, calculating the time deviation between the two. For example, if the task completion time on the platform side is 18:00 on a certain day, and the actual task completion time on the client system side is 20:00 on the same day, the time deviation between the two is 2 hours.

[0084] S302: If the time deviation between the first critical time node and the second critical time node exceeds a preset threshold, the first critical time node and / or the second critical time node are determined to be abnormal time nodes, and the abnormality type is determined to be multi-source time inconsistency abnormality.

[0085] The flexible employment timeline agent compares the time deviation between the first and second key time nodes in each group with a preset time deviation threshold. This time deviation threshold can be set according to the actual situation of different flexible employment business scenarios. For example, the time deviation threshold for on-site service tasks is 4 hours, and the time deviation threshold for remote work tasks is 2 hours, adapting to the time error tolerance of different scenarios. This application does not limit this.

[0086] If the time deviation between the first and second key time nodes of a certain group exceeds a preset time deviation threshold, the flexible employment timeline agent will determine the abnormal time node based on the actual business scenario. For example, if the flexible employment timeline agent clearly identifies data tampering in a single system (such as a time node on the platform side or the customer system side being manually modified), then only the time node on that side (the first or second key time node) will be determined as an abnormal time node. If the flexible employment timeline agent cannot determine the tampering entity temporarily, then both the first and second key time nodes will be determined as abnormal time nodes, pending confirmation by subsequent event sequence logic reasoning model source tracing analysis.

[0087] If the time deviation between the first and second key time nodes exceeds a preset threshold, the anomaly type is determined to be a multi-source time inconsistency anomaly. The flexible employment timeline agent synchronously records the core information of the first and second key time nodes. For example, the core information may include the corresponding matched key time node pair, the time deviation value, the preset time deviation threshold, and the deviation exceeding the specified value.

[0088] In this way, by using a flexible employment timeline intelligent agent to cross-compare and verify the time nodes of the flexible employment platform and the client's system, risk control vulnerabilities such as tampering or falsification of single-source time data can be accurately identified. This makes it difficult to cover up violations such as false order acceptance, proxy execution, and process fraud by modifying a single system's time node, thus improving the penetration of timeline anomaly detection. Through cross-comparison of time nodes across systems, hidden anomalies such as tasks showing as completed on the platform but having no actual execution record on the client's side, or tasks having been accepted on the client's side but not updated synchronously on the platform's side, can be effectively identified. This ensures the authenticity and compliance of business processes and reduces the risk of financial losses and legal disputes for both the platform and employers caused by fraudulent business transactions.

[0089] According to some embodiments of this application, optionally, S1022: according to preset verification rules and task execution time records, perform multi-dimensional verification on the task time series, and determine the abnormal time nodes and abnormal types in the task time series based on the verification results of multi-dimensional verification, and may also include the following steps three to five.

[0090] Step 3: Obtain the metadata of the task completion credential associated with the task completion time.

[0091] The task completion credentials may include the captured images or videos, and the metadata may include the capture time and upload time.

[0092] Specifically, the flexible employment timeline agent can extract the task completion time of target personnel from the task time series, and then, based on the task completion time, associate and match the task completion vouchers uploaded by the target personnel when submitting task results. The task completion vouchers are the core materials proving that the target personnel have actually completed the task, and can include photos or videos of the task results.

[0093] Next, the flexible employment timeline agent can read the metadata of the task completion voucher. Metadata can include shooting time, such as the shutter speed of a mobile phone photo or the start time of a camera recording. It can also include upload time, such as the time the target person uploaded the voucher to the flexible employment platform. This metadata is basic information automatically attached when the voucher is generated, automatically recorded by the shooting device or the uploading system, and cannot be tampered with through conventional editing methods, ensuring data objectivity.

[0094] Taking on-site cleaning service tasks as an example, the flexible employment timeline agent extracts the task completion time of the target personnel from the task time sequence as 16:00 on the same day. Based on this task completion time, it matches and identifies three on-site cleaning result images uploaded by the target personnel. Parsing the metadata of the three on-site cleaning result images, it extracts the shooting times as 15:30, 15:32, and 15:35 on the same day, and the upload time as 15:50 on the same day.

[0095] Step 4: Verify whether the metadata and task completion time conform to the preset business logic.

[0096] The business logic is primarily used to constrain the chronological relationship and time interval between metadata and task completion time, ensuring that task completion credentials are generated and uploaded within a reasonable timeframe before and after task completion. In some examples, the business logic may include requirements such as the shooting time not being later than the task completion time, the upload time not being later than the task completion time, and the interval between the shooting time and the task completion time not exceeding a preset reasonable range. The business logic can be flexibly adjusted according to actual circumstances, and this application does not impose any limitations on it.

[0097] In step four, the flexible employment timeline agent compares the metadata with the task completion time, verifies whether the metadata and task completion time conform to the preset business logic, and records the verification results.

[0098] Step 5: If the metadata and the task completion time do not match the business logic, then the task completion time is identified as an abnormal time node, and the abnormality type is determined to be an abnormality of the authenticity of the result.

[0099] If the metadata and the task completion time do not match the business logic, the core indication is that the task completion time may be falsely reported or tampered with. In this case, the flexible employment timeline agent will identify the task completion time as an abnormal time node and determine the abnormality type as an abnormality in the authenticity of the result.

[0100] In addition, the flexible employment timeline intelligent agent can synchronously record core abnormal information, such as non-compliant business logic rules, metadata, and deviations from task completion time, providing clear evidence for subsequent event time-series logic reasoning models to conduct source tracing analysis (such as determining whether it is a false order, falsified results, or operational error).

[0101] Thus, by using a flexible employment timeline intelligent agent to verify the authenticity of results, a penetrating verification of the authenticity of task completion time can be achieved. This accurately identifies violations such as fraudulent order acceptance, falsified results, and time tampering, including situations like using supplementary photos to cover up incomplete tasks or altering task completion times. This improves the comprehensiveness and accuracy of timeline anomaly detection. Furthermore, results authenticity verification can complement basic time-series verification, time-interval verification, and multi-source consistency verification. It covers multiple levels of anomaly scenarios in flexible employment timelines, including time-series logic, cross-source comparison, and results corroboration. This effectively addresses the practical pain points of difficulty in verifying the authenticity of results and identifying fraudulent tasks on flexible employment platforms, reducing financial losses and business non-compliance risks faced by both platforms and employers.

[0102] According to some embodiments of this application, the task time sequence may optionally include task dispatch time, task completion time, contract effective time, and contract termination time.

[0103] Accordingly, S1022: Based on the preset verification rules and task execution time records, perform multi-dimensional verification on the task time series, and determine the abnormal time nodes and abnormal types in the task time series based on the verification results of the multi-dimensional verification. This may also include the following steps six and seven.

[0104] Step Six: Verify that the task dispatch time and task completion time are within the contract validity period, which is between the contract effective time and the contract termination time.

[0105] Specifically, the flexible employment timeline agent extracts the task assignment time, task completion time, contract effective time, and contract termination time for the same flexible employment task from the task time sequence of the target personnel. Based on the contract effective time and contract termination time, the contract validity period can be determined, that is, the time interval from the contract effective time to the contract termination time. Then, it verifies whether the task assignment time and task completion time are within the contract validity period, that is, whether the task assignment time is between the contract effective time and the contract termination time, and whether the task completion time is between the contract effective time and the contract termination time, and records the verification results.

[0106] Taking long-term labor outsourcing tasks (such as monthly cleaning outsourcing) as an example, the flexible employment timeline agent extracts the time nodes of the same task from the task time series: the contract effective date is the 1st of the month, the contract termination date is the 30th of the month, the task assignment date is the 10th of the month, and the task completion date is the 25th of the month. Based on the contract effective date and contract termination date, the contract validity period is determined to be from the 1st to the 30th of the month. Then, it is verified that the task assignment date (the 10th of the month) is between the 1st and the 30th of the month, and the task completion date (the 25th of the month) is between the 1st and the 30th of the month, and the verification result is recorded as meeting the requirements. For example, if the extracted time nodes are the contract effective date of the 1st of the month, the contract termination date of the 30th of the month, the task assignment date of the previous month, and the task completion date of the current month, the verification result is recorded as: the task assignment date is not within the contract validity period, but the task completion date is within the contract validity period.

[0107] Step 7: If the task dispatch time and / or task completion time are not within the contract validity period, the task dispatch time and / or task completion time shall be identified as abnormal time nodes, and the abnormality type shall be determined as contract validity period timing abnormality.

[0108] The flexible employment timeline agent, based on the verification results from step six, specifically marks time nodes that are not within the contract validity period. If only the task assignment time is outside the contract validity period, then only the task assignment time is marked as an abnormal time node. If only the task completion time is outside the contract validity period, then only the task completion time is marked as an abnormal time node. If both the task assignment time and the task completion time are outside the contract validity period, then both the task assignment time and the task completion time are marked as abnormal time nodes.

[0109] For abnormal time points that are not within the contract's validity period, the anomaly type is determined to be a contract validity period timing anomaly. Contract validity period timing anomalies differ from basic timing anomalies, time interval anomalies, etc., and directly point to the core risk of task execution not conforming to the contract validity period agreement.

[0110] In addition, the flexible employment timeline intelligent agent can synchronously record core abnormal information, such as the contract effective time, contract termination time, the specific time of the abnormal time node, and the deviation between the abnormal time node and the contract validity period (such as earlier than the contract effective time, later than the contract termination time, and the number of days / hours of deviation), providing a basis for subsequent event time sequence logic reasoning models to carry out source tracing analysis (such as determining whether it is a process violation, time tampering, task delay or operational error).

[0111] In this way, by using the flexible employment timeline intelligent agent to verify the contract validity period, the contract agreement is linked to the actual execution. This enables a penetrating verification of the match between task execution and contract validity, accurately identifying non-compliant behaviors such as early order assignment, overdue completion, and execution of tasks without contract validity support. This reduces the risks of legal disputes and unclear responsibilities caused by contract time discrepancies, protects the legitimate rights and interests of the flexible employment platform, employers, and flexible workers, and reduces performance disputes caused by executing tasks beyond the contract period.

[0112] Figure 4 This is a schematic flowchart of step S1023 in the timeline anomaly detection method for flexible employment provided in an embodiment of this application. Figure 4 As shown, according to some embodiments of this application, optionally, S1023: Based on a pre-trained event sequence logic reasoning model, and combined with task execution time records, a preset business scenario rule base and exception types, perform source analysis on the abnormal time nodes to obtain the abnormal cause of the abnormal time nodes, which may include the following steps S401 and S402.

[0113] S401: Provide the event time sequence logic reasoning model with the task time sequence marked with abnormal time nodes, task execution time records, multi-dimensional verification results and abnormal types as combined inputs.

[0114] Specifically, the flexible employment timeline intelligent agent incorporates the complete verification results generated during the multi-dimensional verification process into the combined input. These multi-dimensional verification results may include the judgment records of one or more verification links such as basic time sequence logic verification, time interval rationality verification, multi-source consistency verification, result authenticity verification, and contract validity period time sequence verification, as well as deviation data of abnormal time nodes, cross-system data comparison differences, and logical contradictions between metadata and time nodes.

[0115] S402: The event-sequence logic reasoning model is based on a preset business scenario rule base. It matches and analyzes the combined inputs with the rules in the business scenario rule base and outputs the cause of the anomaly.

[0116] Specifically, the event-driven temporal logic reasoning model extracts features from multi-dimensional data in the combined input. For example, it combines the temporal dependency patterns of the flexible employment industry learned during the model training phase to uncover anomalous features in the combined input. Anomalous features can include correlation features between anomalous time nodes, anomalous types, and multi-dimensional verification results. Anomalous features can also include temporal features such as temporal dependencies between task time nodes, time interval features, and multi-source data deviation features.

[0117] For example, when the anomaly type is an anomaly in the authenticity of the results, the deviation values ​​between the metadata recorded in the multi-dimensional verification results and the task completion time (time-series features), logical contradictions (such as the shooting time being later than the task completion time, correlation features), and other anomaly features are extracted simultaneously. When the anomaly type is an anomaly in the contract validity period time-series, the deviation days between the anomaly time node and the contract validity period (time-series features), and the details of the node time-series conflicts recorded in the multi-dimensional verification (correlation features), and other anomaly features are extracted.

[0118] Next, the event sequence logic reasoning model retrieves the preset business scenario rule base, which contains matching rules for different abnormal characteristics and causes under various flexible employment business scenarios.

[0119] Next, the event-time logic reasoning model matches the extracted anomaly features with rules in the business scenario rule base to determine the major categories of anomaly causes corresponding to the anomaly features, such as customer operational errors, system synchronization delays, fraudulent task records, or compliance exceptions. In some examples, it can further determine the specific sub-causes under the major categories of anomaly causes corresponding to the anomaly features. Then, combining the temporal logic of the task time series and task execution time records, it performs a contradiction check on the initially matched anomaly causes, removing reasoning results that do not conform to the temporal logic and actual business. For example, if evidence of forged documents is found, then customer operational errors and system synchronization delays are removed.

[0120] Finally, the event sequence logic reasoning model filters out abnormal causes (major categories of abnormal causes and / or specific sub-causes under major categories of abnormal causes) that match the combined input features and have no logical contradictions, completing the source analysis of abnormal time nodes and providing accurate root cause basis for subsequent risk control measures.

[0121] In some embodiments, customer operational errors may further include manual entry errors for time nodes, errors in associating cross-task or cross-contract node information, errors in matching task completion voucher uploads with time nodes, and errors in selecting the format for manually entered time information.

[0122] In some embodiments, system synchronization delay may further include data synchronization delay between the flexible employment platform and the customer system, data entry lag in business system nodes, timeout in cross-system data interface transmission, and timestamp asynchronization.

[0123] In some embodiments, false task records may further include false order acceptance without actual execution, task execution by proxy with altered time, re-photographing or stealing task completion vouchers to forge results, intentionally altering time nodes, modifying time intervals to cover up tasks not being executed according to specifications, etc.

[0124] In some embodiments, compliance exceptions may further include order assignment ahead of schedule due to urgent customer business needs, project completion exceeding the deadline due to unforeseen circumstances, settlement delays due to compliance requirements, invoicing delays due to compliance requirements, and timeline deviations caused by objective limitations of on-site operations.

[0125] According to some embodiments of this application, optionally, S402: The event sequence logic reasoning model, based on a preset business scenario rule base, matches and analyzes the combined input with the rules in the business scenario rule base and outputs the cause of the anomaly, which may include the following steps eight to eleven.

[0126] Step 8: When the exception type is a multi-source time inconsistency exception, and the time node in the task execution time record is earlier than the corresponding time node in the task time series, and system synchronization delay characteristics exist, the cause of the exception is determined to be system synchronization delay. For example, system synchronization delay characteristics may include features such as cross-system data interface response timeout and data synchronization queue backlog.

[0127] Step 9: When the anomaly type is a basic timing anomaly or a multi-source time inconsistency anomaly, and there are supplementary records or explanations from the customer's system regarding the anomaly time node, the cause of the anomaly is determined to be a customer operational error.

[0128] When the anomaly type is a basic timing anomaly or a multi-source time inconsistency anomaly, the judgment is based on the customer's operational error being a human-intentional mistake. After the error occurs, the customer will usually submit a formal supplementary record or explanation for filing and correction, thus distinguishing it from anomalies caused by subjective fraud, system synchronization delay, or other reasons.

[0129] The event sequence logic reasoning model searches for supplementary records or explanations submitted by the customer system regarding the abnormal time node, such as explanations of time entry errors by customer-side maintenance personnel, correction records of cross-task node association errors, and filing explanations for manual reporting format errors. For basic time sequence anomalies or multi-source time inconsistency anomalies, if supplementary records or explanations from the customer system exist regarding the abnormal time node, the cause of the anomaly is determined to be customer operational error.

[0130] Step 10: When the anomaly type is a basic time-series anomaly, a time interval anomaly, or an anomaly indicating doubt about the authenticity of the result, and the anomaly features in the combined input match the false task feature rules in the business scenario rule base, the cause of the anomaly is determined to be a false task record.

[0131] The essence of falsified task records is the subjective and intentional falsification of timelines, processes, or results, which usually reveals obvious traces of falsification in abnormal features. For example, rules for falsified task features may include rules for contradictory timelines, rules for falsified results documentation, rules for missing execution traces, and rules for time tampering traces.

[0132] For example, rules regarding temporal logical contradictions may include: a task completion time earlier than the task assignment time without any record-keeping; zero time interval between multiple nodes of the same task (e.g., completion immediately upon assignment); and cross-regional or cross-time period conflicts between task execution node times and personnel location or scheduling times. Rules regarding falsified achievement certificates may include: the task completion certificate being photographed later than the task completion time; inconsistencies between the metadata photographing equipment and location and the execution equipment and on-site location required by the task; the presence of certificate falsification characteristics; and uploading only blank or vague certificates without actual achievement information.

[0133] For example, rules for missing execution traces may include situations where the platform shows the task is completed, but the customer's system has no record of task execution or result confirmation; there are records of task time nodes, but no corresponding operation logs or process approval records to corroborate the execution; and multi-source data comparison shows that only a single system has task records, while other related systems have no synchronized data.

[0134] For example, time tampering trace rules can include abnormal modification records of task time nodes, such as multiple modifications within a short period of time; the deviation between the system-recorded time of a time node and the actual generation time being greater than a preset threshold; and the task dispatch time being changed from before the contract took effect to after the contract took effect.

[0135] If the abnormal features in the combined input are combined with the false task feature rules, the event-sequence logic reasoning model determines that the abnormality is caused by a false task record.

[0136] Step 11: When the exception type is a basic time-series exception or a time-interval exception, and the personnel identifier, task identifier, or project identifier in the combined input matches the applicable subject identifier of the compliance exception process rule in the business scenario rule base, the exception reason is determined to be a compliance exception operation.

[0137] In some embodiments, the business scenario rule base stores compliance exception process rules and their corresponding applicable entity identifiers, such as emergency business early dispatch, sudden project overdue completion, compliance process delays, etc., and their corresponding personnel identifiers, task identifiers, or project identifiers. When the exception type is a basic time-series exception or a time-interval exception, and the exception characteristics match the compliance exception process rules, and the personnel identifier, task identifier, or project identifier in the combined input matches the applicable entity identifier, the exception reason is determined to be a compliance exception operation. For example, if the exception type is a basic time-series exception, such as the dispatch time being earlier than the contract effective time, the exception characteristics meet the emergency business early dispatch process rules, and the project identifier to which the task belongs is consistent with the emergency project identifier registered in the rule base, then it is determined to be a compliance exception operation.

[0138] This approach effectively distinguishes between objective anomalies at the system level, unintentional human error, deliberate fraudulent irregularities, and compliant special cases, improving the accuracy of anomaly tracing and reducing misjudgments caused by relying on single conditions. Precise anomaly tracing provides clear action guidelines for subsequent risk control and process rectification, achieving a closed-loop capability for flexible employment timeline risk control—from anomaly detection to root cause identification and precise handling.

[0139] According to some embodiments of this application, the timeline risk analysis report may optionally include a list of timeline anomaly details.

[0140] The timeline anomaly details list can include multiple fields such as personnel identifier, task identifier, anomaly type, anomaly time node identifier, actual time of the anomaly time node, anomaly cause, evidence index, and evidence link.

[0141] The evidence index can be an index of evidence for anomaly determination and cause tracing, such as multi-dimensional verification results, system log records, feature matching records, customer-side supplementary explanations, and / or identification filing information. This facilitates risk control personnel in quickly retrieving and locating core evidence, improving evidence verification efficiency. The evidence link can be the access address or storage link to the original evidence files or data records for anomaly determination and cause tracing conclusions. Clicking the address or link directly retrieves the corresponding original evidence files or data records, enabling one-click tracing and visual viewing of anomaly evidence. This solves the problems of scattered evidence and difficulty in retrieval in traditional risk control, while providing objective and truthful factual basis for anomaly handling.

[0142] According to some embodiments of this application, optionally, the flexible employment timeline agent can also be used to perform steps twelve to fifteen.

[0143] Step 12: Calculate the risk baseline score for the abnormal time point based on the anomaly type and the preset first scoring rule.

[0144] The first scoring rule can be based on the risk control scenario of the flexible employment industry, combined with pre-set scoring rules for the inherent risk level, probability of violation, and scope of impact of different anomaly types. For example, in some examples, the first scoring rule can be for anomalies in the authenticity of results and contract validity period (defined as high inherent risk, high probability of violation, and impact involving funds and compliance), with a base score set at 80-100 points. For anomalies in multi-source time inconsistency and time interval, the base score is set at 50-79 points. For anomalies in basic time sequence, the base score is set at 30-49 points.

[0145] In step twelfth, the flexible employment timeline agent can calculate the basic risk score for the abnormal time node based on the anomaly type and the preset first scoring rule. For example, if the anomaly type is an anomaly in the authenticity of the result, its basic risk score is directly determined to be 90 points according to the first scoring rule. If the anomaly type is a basic time series anomaly, its basic risk score is determined to be 35 points. If the anomaly type is a multi-source time inconsistency anomaly, its basic risk score is determined to be 65 points.

[0146] Step 13: Calculate the risk bonus score for the abnormal time point based on the cause of the abnormality and the preset second scoring rule.

[0147] The second scoring rule can be a pre-defined scoring adjustment rule based on the cause of the anomaly. Specifically, it can be divided into different levels—additional points, deductions, or zero points—based on the risk attribute of the anomaly cause. This is used to fine-tune the base risk score and differentiate the risk differences caused by different factors within the same anomaly type. For example, in some examples, the second scoring rule could be: For false task records, +10 to +20 points; for customer operational errors (non-subjective, low-risk attributes), -5 to 0 points; for system synchronization delays, 0 points; and for compliance exceptions, 0 points.

[0148] In step thirteen, the flexible employment timeline agent can calculate the risk bonus score for abnormal time nodes based on the cause of the anomaly and the preset second scoring rule. For example, if the cause of the anomaly is a false task record (such as forging results by re-shooting vouchers), the risk bonus score is +15 points according to the second scoring rule. If the cause of the anomaly is a customer operational error (such as a manual entry error for a time node), the risk bonus score is -5 points. If the cause of the anomaly is a system synchronization delay (such as a timeout in cross-system interface transmission), the risk bonus score is 0 points. If the cause of the anomaly is a compliance exception operation (such as early dispatch of urgent business), the risk bonus score is 0 points.

[0149] Step Fourteen: Based on the basic risk score and the additional risk score, obtain the risk score for the abnormal time node.

[0150] The base risk score and additional risk score can be calculated according to a preset comprehensive scoring formula to obtain the comprehensive risk score for that abnormal time point. For example, in some examples, the base risk score and additional risk score can be summed to obtain the risk score for the abnormal time point.

[0151] Step 15: Determine the risk level of abnormal time points based on the risk score and multiple preset score ranges.

[0152] Specifically, the flexible employment timeline agent retrieves the preset risk level classification rules, which divide the range of risk scores into multiple continuous and non-overlapping score intervals, and uniquely matches the corresponding risk level for each score interval.

[0153] The flexible employment timeline agent can match the risk score of an abnormal time point with multiple preset score ranges, and determine the risk level corresponding to the abnormal time point based on the matching results. For example, in some examples, the preset score ranges and corresponding risk levels can be: above 80 points is high risk, 50-79 points is medium risk, 1-49 points is low risk, and below 0 points is no risk.

[0154] Correspondingly, the timeline anomaly details list can also include at least one field from the risk score and risk level of the anomaly time node. That is, the flexible employment timeline intelligent agent can simultaneously enter the calculated risk score and the determined risk level into the timeline anomaly details list, and present them in association with the original fields such as personnel identifier, task identifier, anomaly type, and anomaly cause, so as to achieve the integrated collection of anomaly information and risk quantification information.

[0155] Thus, by using a flexible employment timeline intelligent agent to execute risk scoring and level determination processes, the abstract risks of flexible employment timeline anomalies can be transformed into quantifiable scores and levels, enabling quantitative risk assessment. Furthermore, incorporating risk scores and levels into the timeline anomaly details list provides risk control personnel with clear priorities for handling these issues, helping them to conduct differentiated and targeted risk control measures. This further improves the flexible employment timeline anomaly detection system, effectively enhancing the efficiency and scientific rigor of risk control measures, and reducing operational losses and compliance risks caused by untimely or inappropriate handling of anomaly risks.

[0156] According to some embodiments of this application, the timeline risk analysis report may optionally include a timeline panorama of the target personnel. The timeline panorama can display multiple time nodes in the task time sequence in chronological order. The timeline panorama can visually present the entire process sequence of the target personnel's corresponding tasks, clearly showing the sequential relationship, correlation logic, and node status of each time node, thereby providing intuitive visual support for risk control personnel's risk verification and anomaly tracing.

[0157] In this timeline panorama, multiple time nodes can be distinguished and marked using primary and secondary visual identifiers. This enables rapid visual identification of normal and abnormal time nodes, reducing the cost for risk control personnel in differentiating time node states. Primary visual identifiers can be used to mark time nodes that have been verified as normal across multiple dimensions, while secondary visual identifiers can be used to mark abnormal time nodes. The two types of visual identifiers can employ differentiated visual designs, which can be flexibly adjusted according to visualization needs. For example, in some examples, the primary visual identifier can be a green circle to mark normal task dispatch times, task start times, and compliant task completion times. The secondary visual identifier can be a red triangle to mark abnormal time nodes, with a brief annotation next to the secondary visual identifier indicating the type of abnormality, such as abnormal result authenticity or temporal logical contradictions, further improving the efficiency of abnormal node identification.

[0158] The timeline panorama can also include target connecting lines. These lines are used to mark the temporal relationships between multiple time nodes, clearly presenting the execution order of each time node within the same task. This allows risk control personnel to intuitively grasp the entire timeline logic of a task from initiation to completion. For example, the target connecting line can be a solid black line, connecting the dispatch time, start time, execution node time, completion time, and settlement time of the same task in chronological order. If there are temporal anomalies between adjacent or related time nodes, differentiated markings can be made on the corresponding target connecting lines, such as marking with red dashed segments or adding prompts indicating temporal anomalies. This provides a secondary visual identifier for the anomalous node, allowing risk control personnel to not only locate the anomalous node but also intuitively see the temporal inconsistencies it corresponds to.

[0159] Thus, compared to time nodes in plain text or tables, a panoramic timeline can transform abstract task sequence relationships and node statuses into intuitive graphical information. Risk control personnel can quickly grasp the entire timeline of task execution for target personnel without having to check text data line by line, accurately know the sequence of each time node, the specific location and number of abnormal time nodes, reduce the cost of interpreting time sequence information, and improve the overall efficiency of risk control verification.

[0160] Based on the timeline anomaly detection method for flexible employment provided in any of the above embodiments, this application also provides a timeline anomaly detection system for flexible employment.

[0161] Figure 5 This is a structural block diagram of a timeline anomaly detection system for flexible employment provided in an embodiment of this application. Figure 5 As shown, the timeline anomaly detection system 50 for flexible employment provided in this application embodiment may include: Interaction module 501 is used to receive the target personnel identifier and verification time range input by the user; Flexible staffing timeline agent 502 is used to perform the following processes: Based on the target personnel identification and verification time range, the task time series is obtained from the flexible employment platform, and the task execution time record is obtained from the customer system. The task time series includes multiple time nodes of the target personnel in the flexible employment task cycle. Based on the preset verification rules and task execution time records, the task time series is verified in multiple dimensions, and based on the verification results of the multiple dimensions, the abnormal time nodes and abnormal types in the task time series are determined; the multiple dimensions verification includes basic time sequence logic verification, time interval rationality verification, and multi-source consistency verification. Based on a pre-trained event sequence logic reasoning model, and combined with task execution time records, a pre-set business scenario rule base and exception types, the abnormal time nodes are traced and analyzed to obtain the abnormal causes that led to the occurrence of the abnormal time nodes. Generate a timeline risk analysis report for the target personnel based on the abnormal time point, abnormal type, and abnormal cause.

[0162] The timeline anomaly detection system for flexible employment provided in the embodiments of this application, on the one hand, collects task time series and execution time records across platforms through a flexible employment timeline intelligent agent. Combining three types of verification—basic temporal logic, reasonableness of time intervals, and multi-source consistency—it can not only identify explicit time node deviations but also recognize implicit anomalies such as temporal logic contradictions and cross-system data inconsistencies. This allows it to identify instances where individuals in flexible employment scenarios are using falsified or altered time nodes to conceal fraudulent order acceptance, subcontracting, and task process manipulation, thereby reducing the risk of financial losses and legal disputes. On the other hand, the flexible employment timeline intelligent agent, based on a pre-trained event temporal logic reasoning model and a business scenario rule base, performs full-link logical tracing of abnormal time nodes. It can not only detect alterations at a single time point but also identify the chain of time contradictions caused by the alteration. Furthermore, by combining anomaly types to pinpoint the nature of the fraud, it accurately locates traces of timeline manipulation, helping platforms and employers curb violations such as process manipulation and data forgery at the source.

[0163] It should be noted that the timeline anomaly detection system 50 for flexible employment provided in this application embodiment has the same or corresponding technical features as the timeline anomaly detection method for flexible employment provided in any of the above embodiments, and produces the same technical effects. For the sake of brevity, further details are omitted here.

[0164] The electronic device in this application embodiment may be a user terminal device, a server, other computing devices, or a cloud server. Figure 6This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. The electronic device may include a processor 601 and a memory 602 storing computer program instructions. When the processor 601 executes the computer program instructions, it implements the process or function of the timeline anomaly detection method for flexible employment in any of the above embodiments.

[0165] Specifically, processor 601 may include a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. Memory 602 may include mass storage for data or instructions. For example, memory 602 may be at least one of the following: hard disk drive (HDD), read-only memory (ROM), random access memory (RAM), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, universal serial bus (USB) drive, or other physical / tangible memory storage device. Alternatively, memory 602 may include removable or non-removable (or fixed) media. Furthermore, memory 602 may be internal or external to an electronic device. Memory 602 may be non-volatile solid-state memory. In other words, typically memory 602 includes a tangible (non-transitory) computer-readable storage medium (such as a memory device) encoded with computer-executable instructions, and when the software is executed (e.g., by one or more processors), it can perform the operations described in the methods of the embodiments of this application. The processor 601 implements the process or function of any of the timeline anomaly detection methods for flexible employment in the above embodiments by reading and executing computer program instructions stored in the memory 602.

[0166] In one example Figure 6The illustrated electronic device may also include a communication interface 603 and a bus 610. The processor 601, memory 602, and communication interface 603 are connected via bus 610 and communicate with each other. Communication interface 603 is primarily used to enable communication between modules, devices, units, and / or equipment in the embodiments of this application. Bus 610 may include hardware, software, or both, and can couple components of the online data traffic billing device together. For example, the bus may include at least one of the following: Accelerated Graphics Port (AGP) or other graphics bus, Enhanced Industry Standard Architecture (EISA) bus, Front Side Bus (FSB), HyperTransport (HT) Interconnect, Industry Standard Architecture (ISA) bus, Infinite Bandwidth Interconnect, Low Pin Count (LPC) bus, memory bus, Microchannel Architecture (MCA) bus, Peripheral Component Interconnect (PCI) bus, PCI-Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association Local (VLB) bus, or other suitable buses. Bus 610 may include one or more buses. Although specific buses are described or illustrated in the embodiments of this application, any suitable bus or interconnection method may be considered in the embodiments of this application.

[0167] In conjunction with the methods in the above embodiments, this application also provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the process or function of any of the timeline anomaly detection methods for flexible employment in the above embodiments.

[0168] In addition, this application also provides a computer program product that stores computer program instructions. When the computer program instructions are executed by a processor, they implement the process or function of any of the timeline anomaly detection methods for flexible employment described above.

[0169] The flowcharts and / or block diagrams of methods, apparatuses, systems, and computer program products according to embodiments of this application have been exemplarily described above, and related aspects have been described. It should be understood that each block or combination thereof in the flowcharts and / or block diagrams may be implemented by computer program instructions, by dedicated hardware performing a specified function or action, or by a combination of dedicated hardware and computer instructions. For example, these computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to form a machine that enables the implementation of the function / action specified in each block or combination thereof in the flowcharts and / or block diagrams, executable via such processor. Such a processor may be a general-purpose processor, a dedicated processor, a special-purpose application processor, or a field-programmable logic circuit.

[0170] The functional blocks shown in the structural block diagrams of this application can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc.; when implemented in software, they are programs or code segments used to perform the required tasks. Programs or code segments can be stored in memory or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. Code segments can be downloaded via computer networks such as the Internet or intranets.

[0171] It should be noted that this application is not limited to the specific configurations and processes described above or shown in the figures. The above descriptions are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the described systems, devices, modules, or units can be referred to the corresponding processes in the method embodiments, and need not be repeated here. It should be understood that the scope of protection of this application is not limited thereto. Any person skilled in the art can conceive of various equivalent modifications or substitutions within the scope of the technology disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application.

Claims

1. A method for detecting timeline anomalies in flexible employment, characterized in that, include: Receive user input of the target personnel identifier and verification time range; Based on the flexible employment timeline agent, the following processes are performed: Based on the target personnel identification and verification time range, the task time sequence is obtained from the flexible employment platform, and the task execution time record is obtained from the customer system. The task time sequence includes multiple time nodes of the target personnel in the flexible employment task cycle. Based on preset verification rules and task execution time records, the task time series is verified in multiple dimensions, and based on the verification results of the multiple dimensions, the abnormal time nodes and abnormal types in the task time series are determined; the multiple dimensions verification includes basic time sequence logic verification, time interval rationality verification, and multi-source consistency verification. Based on a pre-trained event sequence logic reasoning model, and combined with task execution time records, a pre-set business scenario rule base and exception types, the abnormal time nodes are traced and analyzed to obtain the abnormal causes that led to the occurrence of the abnormal time nodes. Generate a timeline risk analysis report for the target personnel based on the abnormal time point, abnormal type, and abnormal cause.

2. The method according to claim 1, characterized in that, The multiple time nodes include at least some of the following time nodes: registration time, signing time, contract effective time, task assignment time, task start time, task completion time, acceptance time, settlement time, and invoice issuance time. The multiple time nodes are ordered in chronological order. Based on preset verification rules and task execution time records, the task time series is subjected to multi-dimensional verification. Based on the verification results, abnormal time nodes and abnormal types within the task time series are determined, including: Verify whether the task time sequence corresponding to the target personnel conforms to the preset flexible employment business logic sequence; The time nodes in the task time series corresponding to the target personnel that do not conform to the time sequence of flexible employment business logic are identified as abnormal time nodes, and the abnormality type is determined to be a basic time sequence abnormality.

3. The method according to claim 1, characterized in that, Based on preset verification rules and task execution time records, the task time series is subjected to multi-dimensional verification. Based on the verification results, abnormal time nodes and abnormal types within the task time series are determined, including: Verify whether the time interval between adjacent key time nodes in the task time sequence is within a preset time range; wherein, the adjacent key time nodes include at least one of the following pairs: task dispatch time and task start time, task completion time and task acceptance time, task acceptance time and settlement time, and settlement time and invoice issuance time; Adjacent critical time nodes whose time intervals exceed the preset time range are identified as abnormal time nodes, and the abnormality type is determined to be time interval abnormality.

4. The method according to claim 1, characterized in that, Based on preset verification rules and task execution time records, the task time series is subjected to multi-dimensional verification. Based on the verification results, abnormal time nodes and abnormal types within the task time series are determined, including: Compare the first key time node in the task time sequence with the second key time node of the same action in the task execution time record; the first key time node includes at least one of the task dispatch time, task completion time, task acceptance time and settlement time, and the second key time node includes at least one of the task start time in the customer system, task actual execution completion time, task result confirmation time and payment instruction issuance time. If the time deviation between the first critical time node and the second critical time node exceeds a preset threshold, the first critical time node and / or the second critical time node are determined to be abnormal time nodes, and the abnormality type is determined to be a multi-source time inconsistency abnormality.

5. The method according to claim 1, characterized in that, The task time series includes at least the task completion time; Based on preset verification rules and task execution time records, the task time series is subjected to multi-dimensional verification. Based on the verification results, abnormal time nodes and abnormal types within the task time series are determined. This also includes: Obtain metadata of a task completion credential associated with the task completion time, wherein the task completion credential includes a captured image or video, and the metadata includes the capture time and upload time; Verify whether the metadata and task completion time conform to the preset business logic; If the metadata and the task completion time do not match the business logic, the task completion time will be identified as an abnormal time node, and the abnormality type will be determined as an abnormality in the authenticity of the result.

6. The method according to claim 1, characterized in that, The task time series includes task dispatch time, task completion time, contract effective time, and contract termination time; The step of performing multi-dimensional verification on the task time series based on preset verification rules and the task execution time records, and determining the abnormal time nodes and abnormal types in the task time series based on the verification results of the multi-dimensional verification, further includes: Verify whether the task dispatch time and task completion time are within the contract validity period between the contract effective time and the contract termination time; If the task dispatch time and / or task completion time are not within the contract validity period, the task dispatch time and / or task completion time will be identified as abnormal time nodes, and the abnormality type will be determined as contract validity period timing abnormality.

7. The method according to claim 1, characterized in that, Based on a pre-trained event-sequence logic reasoning model, and combined with task execution time records, a pre-defined business scenario rule base, and exception types, the source analysis of abnormal time nodes is performed to obtain the abnormal causes that led to the occurrence of abnormal time nodes, including: The task time series marked with abnormal time nodes, task execution time records, multi-dimensional verification results, and abnormal types are provided as combined inputs to the event time sequence logic reasoning model. The event sequence logic reasoning model, based on the preset business scenario rule base, matches and analyzes the combined input with the rules in the business scenario rule base, and outputs the cause of the anomaly. The reasons for the anomalies include customer operational errors, system synchronization delays, false task records, or compliance exceptions.

8. The method according to claim 7, characterized in that, The event-sequence logic reasoning model, based on the preset business scenario rule base, matches and analyzes the combined input with the rules in the business scenario rule base, and outputs the cause of the anomaly, including: When the anomaly type is a multi-source time inconsistency anomaly, and the time node in the task execution time record is earlier than the corresponding time node in the task time sequence, and there is a system synchronization delay characteristic, the cause of the anomaly is determined to be system synchronization delay. When the anomaly type is a basic timing anomaly or a multi-source time inconsistency anomaly, and there are supplementary records or explanations from the customer system regarding the anomaly time nodes, the cause of the anomaly is determined to be a customer operational error. When the anomaly type is a basic time-series anomaly, a time interval anomaly, or an anomaly indicating doubt about the authenticity of the result, and the anomaly features in the combined input match the false task feature rules in the business scenario rule base, the cause of the anomaly is determined to be a false task record. When the anomaly type is a basic time-series anomaly or a time-interval anomaly, and the personnel identifier, task identifier, or project identifier in the combined input matches the applicable subject identifier of the compliance exception process rule in the business scenario rule base, the anomaly reason is determined to be a compliance exception operation.

9. The method according to claim 1, characterized in that, The timeline risk analysis report includes a list of timeline anomalies; The timeline anomaly details list includes multiple fields such as personnel identifier, task identifier, anomaly type, anomaly time node identifier, actual time of the anomaly time node, anomaly cause, evidence index, and evidence link.

10. The method according to claim 9, characterized in that, The flexible staffing timeline agent is also used to perform the following processes: Calculate the basic risk score for the abnormal time point based on the abnormality type and the preset first scoring rule; Calculate the risk bonus score for the abnormal time point based on the cause of the abnormality and the preset second scoring rule; Based on the basic risk score and the additional risk score, the risk score for abnormal time nodes is obtained; Based on the risk score and multiple preset score ranges, the risk level of abnormal time points is determined; The timeline anomaly details list also includes at least one field from the risk score and risk level of the anomaly time node.

11. The method according to claim 1, 9, or 10, characterized in that, The timeline risk analysis report also includes a panoramic view of the target personnel's timeline, which displays multiple time nodes in the task's time sequence in chronological order. The multiple time nodes are distinguished and marked using a first visual identifier and a second visual identifier. The first visual identifier is used to mark time nodes that have been verified as normal, and the second visual identifier is used to mark the abnormal time nodes. The timeline panorama is also provided with target connecting lines, which are used to mark the temporal relationship between multiple time nodes.

12. A timeline anomaly detection system for flexible employment, characterized in that, include: The interaction module is used to receive the target personnel identifier and verification time range input by the user; The flexible staffing timeline agent is used to perform the following processes: Based on the target personnel identification and verification time range, the task time sequence is obtained from the flexible employment platform, and the task execution time record is obtained from the customer system. The task time sequence includes multiple time nodes of the target personnel in the flexible employment task cycle. Based on preset verification rules and task execution time records, the task time series is verified in multiple dimensions, and based on the verification results of the multiple dimensions, the abnormal time nodes and abnormal types in the task time series are determined; the multiple dimensions verification includes basic time sequence logic verification, time interval rationality verification, and multi-source consistency verification. Based on a pre-trained event sequence logic reasoning model, and combined with task execution time records, a pre-set business scenario rule base and exception types, the abnormal time nodes are traced and analyzed to obtain the abnormal causes that led to the occurrence of the abnormal time nodes. Generate a timeline risk analysis report for the target personnel based on the abnormal time point, abnormal type, and abnormal cause.

13. An electronic device, characterized in that, The electronic device includes a processor and a memory storing computer program instructions; when the electronic device executes the computer program instructions, it implements the timeline anomaly detection method for flexible employment as described in any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the timeline anomaly detection method for flexible employment as described in any one of claims 1-11.

15. A computer program product, characterized in that, It includes computer program instructions that, when executed by a processor, implement the timeline anomaly detection method for flexible employment as described in any one of claims 1-11.

Images