An AI big data-based cross-department business coordination tracking method and system

Abstract

This invention discloses a cross-departmental business collaboration tracking method and system based on AI big data, belonging to the field of computer data processing technology. The method includes: collecting interface call messages between departmental business systems and parsing them to obtain the original collaborative event sequence; identifying and comparing submitted and received events to obtain the handover time interval, the number of handover rounds, and the actual flow node order; statistically obtaining the real-time collaborative load index of the receiving department; constructing and adjusting the expected collaborative path model to obtain a dynamic expected collaborative path model; comparing the handover time interval, the number of handover rounds, and the actual flow node order with the model to generate a collaborative status anomaly identifier and an abnormal department identifier; identifying business items with continuously missing interface call messages and generating a monitoring blind spot identifier; and accordingly pushing collaborative reminder signals and manual verification prompts. This invention can improve the accuracy of collaborative status anomaly judgment, promptly identify monitoring blind spots, and enhance the level of cross-departmental business collaboration management.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, and in particular to a cross-departmental business collaborative tracking method and system based on AI big data. Background Technology

[0002] As enterprises advance their IT infrastructure, business processes often require collaboration across multiple departmental systems. These systems interact and exchange data through API calls. Existing cross-departmental collaborative management methods typically rely on manual data entry, inter-departmental communication and confirmation, or setting fixed deadlines within the business systems to track progress. Some solutions attempt to collect and record collaborative process data using technical means, but there is usually a lack of specific solutions for situations where business data cannot be effectively collected for various reasons.

[0003] However, existing methods typically use fixed time thresholds or uniform judgment criteria to determine abnormal collaborative status, which requires improvement in accuracy. At the same time, existing methods often fail to identify business items for which relevant business data cannot be effectively collected in a timely manner, easily creating monitoring blind spots. This results in the inability to grasp the collaborative status of relevant business items in a timely manner, affecting the overall efficiency and accuracy of cross-departmental business collaborative management. Summary of the Invention

[0004] The main objective of this invention is to provide a cross-departmental business collaboration tracking method and system based on AI big data, aiming to solve the technical problems of existing cross-departmental business collaboration management methods, such as the easy misjudgment of collaboration status anomalies and the lack of effective identification methods for business matters where business data cannot be effectively collected, which easily leads to monitoring blind spots.

[0005] In a first aspect, the present invention provides a cross-departmental business collaborative tracking method based on AI big data, comprising: Collect interface call messages generated between business systems of the collection department during the process of handling business matters, and parse the interface call messages to obtain the original collaborative event sequence; Identify the submission and receipt events in the original collaborative event sequence, group and compare the submission and receipt events between two adjacent departments for the same business matter, and obtain the handover time interval, the number of handover rounds, and the actual flow node order; The original collaborative event sequence is statistically analyzed to obtain the real-time collaborative load index of the receiving department; A desired collaborative path model is constructed based on the historical collaborative event sequence, and the desired collaborative path model is adjusted based on the real-time collaborative load index to obtain a dynamic desired collaborative path model. The handover time interval, number of handover rounds, and actual flow node sequence are compared with the dynamic expected collaborative path model to generate collaborative status anomaly identifiers and corresponding abnormal department identifiers. Based on the dynamic expected collaborative path model, identify business items with continuously missing interface call messages and generate monitoring blind spot identifiers; Based on the abnormal status and abnormal department identifiers, a collaborative reminder signal is pushed, and based on the monitoring blind spot identifier, a manual verification prompt signal is triggered.

[0006] Optionally, parsing the interface call message to obtain the original cooperative event sequence includes: Send capability detection requests to the business systems of each department, and receive the version identifier of the interface call log generation module and the declaration information of the log output interface returned by the business systems of each department; The version identifier of the interface call log generation module and the declaration information of the log output interface are compared with the preset interface log capability feature library to obtain the coverage capability identifier. Based on the coverage capability identifier, the business systems of each department are classified and the data collection channels are configured to obtain the set of business systems of the departments to be collected. Establish message collection connections with the business systems of each department in the set of business systems to be collected, and collect interface call messages generated between the business systems of the departments to be collected during the process of handling business matters. Parse the API call messages, extract the key field information corresponding to the business items, and mark the API call messages that fail to parse. The timestamps of the business systems of each department are calibrated, and the extracted results are arranged in chronological order according to the calibrated timestamps to obtain the original collaborative event sequence.

[0007] Optionally, grouping and comparing submission and receipt events between two adjacent departments for the same business transaction includes: Determine whether there are multiple submission events contained in the event group. If there are multiple submission events in the event group, determine the type of multiple submission events and determine the handover time interval of the corresponding handover stage based on the determination result. Based on the type of submitted event, count the number of corresponding submitted events to obtain the number of handover rounds; Based on the sequence of each handover step and the correlation of the handover time intervals, the flow structure of business items is identified in order to determine the actual flow node sequence of business items.

[0008] Optionally, the process for determining real-time coordinated load indicators includes: Set up a sliding statistics window, group the original collaborative event sequence according to the receiving department identifier, and obtain the candidate event subsequence of each receiving department in the current statistics window; Based on the item identifier and operation type identifier, identify the backlog items that have generated a receive event but have not generated a corresponding submit event from the candidate event subsequence, count the number of backlog items, and obtain the backlog to be processed; The duration of the delay is calculated based on the time difference between the timestamp of the received event for each delayed item and the current statistical time. Based on the operation type identifier corresponding to the delayed item, a preset complexity weight is matched, and the delay time and the preset complexity weight are weighted and summed to obtain the backlog load value. The number of received events and the number of submitted events in the candidate event subsequence are counted, and the event reception rate and event processing rate per unit time are calculated respectively to obtain the difference between the two rates. The sign of the rate difference is determined, and the weighting coefficients of the backlog load value and the rate difference in the fusion calculation are dynamically adjusted based on the determination result. Weighted fusion is then performed based on the adjusted weighting coefficients to obtain the real-time collaborative load index of the receiving department.

[0009] Optionally, the construction process of the dynamic expected collaborative path model includes: Based on the historical flow node order in the historical collaborative event sequence, frequency statistics are performed to determine the desired flow node order; Based on the historical handover time intervals and the number of historical handover rounds in the historical collaborative event sequence, determine the expected processing time range and the expected number of handover rounds range; Based on the expected flow node sequence, expected processing time range, and expected handover round range, an initial expected collaborative path model is constructed. Based on the historical collaborative load index sequence corresponding to the historical collaborative event sequence, the preset load capacity limit of the receiving department, the expected processing time range, and the expected handover round range, determine the time sensitivity coefficient and round sensitivity coefficient of the receiving department. The duration adjustment amount and the round adjustment amount are determined based on the real-time collaborative load index, duration sensitivity coefficient and round sensitivity coefficient of the receiving department. Based on the duration adjustment amount and round adjustment amount, the upper limits of the expected processing duration range and the expected handover round range are updated and their validity is verified to obtain the dynamic expected collaborative path model.

[0010] Optionally, generating a collaborative status anomaly identifier and the corresponding anomaly department identifier includes: The actual flow node order is compared with the expected flow node order in the dynamic expected collaborative path model to obtain structural similarity, link mapping relationship and unmatched link identification. Based on the link mapping relationship, the actual handover time interval is compared with the corresponding dynamic expected processing time range to obtain the total time deviation; Based on the link mapping relationship, the actual number of handover rounds is compared with the corresponding dynamic expected handover round range to obtain the total number of round deviations; Based on structural similarity, total duration deviation, and total round deviation, anomaly indicators for collaborative status are determined. When generating an abnormal status indicator for collaborative status, the abnormal department indicator is determined based on the deviation of each handover link.

[0011] Optionally, generating monitoring blind spot identifiers includes: Based on the events that have occurred in the business transaction, and in conjunction with the dynamic expected collaborative path model, determine the next expected flow node of the business transaction. Based on the dynamic expected collaborative path model, determine the expected processing time range corresponding to the next expected transfer node; When the expected processing time expires, determine whether there is a corresponding submission event or reception event record for the next expected transfer node; if so, do not generate a monitoring blind spot identifier. If there are no submission or reception event records, and the duration of missing interface call messages exceeds the upper limit of the expected processing time range, a monitoring blind spot identifier will be generated.

[0012] Secondly, this invention provides a cross-departmental business collaborative tracking system based on AI big data, comprising: The message acquisition and parsing module is used to collect interface call messages generated between departmental business systems during the processing of business matters, and parse the interface call messages to obtain the original collaborative event sequence. The handover analysis module is used to identify submission and reception events in the original collaborative event sequence. It groups and compares submission and reception events between two adjacent departments for the same business matter to obtain the handover time interval, the number of handover rounds, and the actual flow node order. The load statistics module is used to statistically analyze the original collaborative event sequence and obtain the real-time collaborative load index of the receiving department. The dynamic path modeling module is used to construct a desired collaborative path model based on historical collaborative event sequences, and adjust the desired collaborative path model based on real-time collaborative load indicators to obtain a dynamic desired collaborative path model. The anomaly comparison module is used to compare the handover time interval, the number of handover rounds, and the actual flow node sequence with the dynamic expected collaborative path model, and generate collaborative status anomaly identifiers and corresponding abnormal department identifiers. The blind spot identification module is used to identify business items with continuously missing interface call messages based on the dynamic expected collaborative path model and generate monitoring blind spot identifiers. The signal push module is used to push collaborative reminder signals based on the collaborative status abnormality identifier and the abnormal department identifier, and to trigger a manual verification prompt signal based on the monitoring blind spot identifier.

[0013] Thirdly, the present invention provides an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute computer programs to implement the aforementioned cross-departmental business collaborative tracking method based on AI big data.

[0014] Fourthly, the present invention provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned cross-departmental business collaborative tracking method based on AI big data.

[0015] The beneficial effects of this invention are as follows: the collaborative process data originates from the actual interface call messages generated between departmental business systems, rather than relying on manual entry, reducing information lag and record omissions; the collaborative path model is expected to be dynamically adjusted in conjunction with the real-time load changes of the receiving department, avoiding misjudgments caused by load differences under fixed judgment criteria; collaborative anomalies and data loss risks can be identified separately through different judgment paths, avoiding the omission of some risks due to a single judgment mechanism; the identified anomalies and blind spots can be converted into reminder signals and verification prompts according to type and pushed to the relevant responsible parties, avoiding delays in handling problems due to lack of timely awareness. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a cross-departmental business collaborative tracking method based on AI big data, provided as an embodiment of the present invention.

[0018] Figure 2 This is a flowchart illustrating the event grouping and flow structure determination process for a cross-departmental business collaborative tracking method based on AI big data, as provided in an embodiment of the present invention.

[0019] Figure 3 A flowchart illustrating the construction of a dynamic expected collaborative path model for a cross-departmental business collaborative tracking method based on AI big data, provided in this embodiment of the invention.

[0020] Figure 4 This is a structural block diagram of a cross-departmental business collaborative tracking system based on AI big data, provided as an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figure 1 This invention discloses a cross-departmental business collaborative tracking method based on AI big data, which may include: S1: Collect interface call messages generated between departmental business systems during the processing of business matters, and parse the interface call messages to obtain the original collaborative event sequence.

[0023] Specifically, S1 includes the following sub-steps: S1.1: Send capability detection requests to the business systems of each department, and receive the version identifier of the interface call log generation module and the declaration information of the log output interface returned by the business systems of each department.

[0024] S1.2: Compare the version identifier of the interface call log generation module and the log output interface declaration information with the preset interface log capability feature library to obtain the coverage capability identifier.

[0025] The coverage capability identifier is used to characterize whether the corresponding department's business system has the ability to generate interface call messages that can be collected.

[0026] Specifically, the log output interface declaration information is parsed to extract interface capability feature information. This interface capability feature information is then compared with a pre-defined set of required capability items recorded in the interface log capability feature library to obtain a preliminary judgment result. The interface capability feature information includes at least one of the supported log field types, log output formats, and log invocation methods. The set of required capability items represents the minimum combination of capabilities required to generate a collectable interface invocation message, and this minimum combination includes field types, output formats, and invocation methods.

[0027] Furthermore, the version identifier of the interface call log generation module is matched with the list of known abnormal versions recorded in the preset interface log capability feature library; if a corresponding known abnormal version is matched, the initial judgment result is corrected according to the defect type associated with the known abnormal version to obtain the corrected judgment result; if no corresponding known abnormal version is matched, the initial judgment result is used as the corrected judgment result; based on the corrected judgment result, the coverage capability identifier of the corresponding department's business system is generated.

[0028] This determination method ensures that departmental business systems that claim to support data collection but actually have known defects are not mistakenly judged as having data collection capabilities, thereby reducing invalid data collection and parsing attempts caused by insufficient log capabilities or version defects before data collection occurs.

[0029] S1.3: Based on the coverage capability identifier, classify and process the business systems of each department and configure the collection channels to obtain the set of business systems of the departments to be collected.

[0030] Specifically, based on the coverage capability identifier, the business systems of each department are divided into a first set of departmental business systems with capability and a second set of departmental business systems without capability; log collection components are issued to each departmental business system in the second set of departmental business systems, or a bypass traffic mirroring connection is established with its network egress to obtain a compensation collection channel; the first set of departmental business systems and the second set of departmental business systems configured with compensation collection channels are merged to obtain the set of departmental business systems to be collected.

[0031] In this way, departmental business systems that were originally deemed unsuitable for data collection due to insufficient log capabilities can still be included in the scope of data collection through the compensation collection channel, thereby improving the coverage and completeness of the set of departmental business systems to be collected.

[0032] S1.4: Establish message collection connections with the business systems of each department in the set of business systems to be collected, and collect interface call messages generated between the business systems of the departments to be collected during the process of handling business matters.

[0033] S1.5: Parse the interface call message, extract the key field information corresponding to the business item, and mark the interface call message that is parsed abnormally.

[0034] The key fields include a unique identifier for the business transaction, an initiating department identifier, a receiving department identifier, an operation type identifier, and a timestamp.

[0035] In addition, parsing anomalies include missing key fields (i.e., at least one of the following could not be extracted: the event identifier, the initiating department identifier, the receiving department identifier, the operation type identifier, or the timestamp) and format anomalies (i.e., the extracted field content does not conform to the preset field format rules).

[0036] Accordingly, abnormal messages with traceability (extractable event identifiers) are retained for manual verification and are not lost due to parsing anomalies; abnormal messages without this basis are excluded from the main sequence to avoid interfering with subsequent grouping comparisons in the absence of valid identifiers.

[0037] S1.6: Calibrate the timestamps of the business systems of each department, and arrange the extracted results in chronological order according to the calibrated timestamps to obtain the original collaborative event sequence.

[0038] S2: Identify the submission and reception events in the original collaborative event sequence, group and compare the submission and reception events between two adjacent departments for the same business matter, and obtain the handover time interval, the number of handover rounds, and the actual flow node order.

[0039] Specifically, the flowchart for determining the event grouping and workflow structure is as follows: Figure 2 As shown, it includes the following sub-steps: S2.1: Identify submit and receive events from the original collaborative event sequence based on the operation type identifier.

[0040] S2.2: Group the submission and receipt events of the same business matter between two adjacent departments according to the event identifier, the initiating department identifier, and the receiving department identifier to obtain event groups.

[0041] S2.3: Determine whether there are multiple submission events contained in the event group. If there are multiple submission events contained in the event group, determine the type of multiple submission events and determine the handover time interval of the corresponding handover stage based on the determination result.

[0042] Specifically, if there are multiple submissions, the system determines whether the submissions are resubmitted or split in parallel based on whether there are any return or rejection flags associated with the business matter between adjacent submission events. If it is determined to be resubmitted, the system determines the handover time interval of the corresponding handover stage by using the last submission event in the event group and the nearest receiving event thereafter. If it is determined to be split in parallel, the system matches the corresponding receiving event for each submission event to obtain multiple parallel handover time intervals.

[0043] S2.4: Based on the type of the submitted event, count the number of corresponding submitted events to obtain the number of handover rounds.

[0044] In one embodiment, the number of commit events that are rolled back and resubmitted is counted to obtain the number of handover rounds.

[0045] S2.5: Based on the sequence of each handover step and the correlation of the handover time intervals, identify the flow structure of business items in order to determine the actual flow node sequence of business items.

[0046] Specifically, the handover time intervals corresponding to each handover stage of a business matter are obtained; handover stages that share the same initiating department or the same receiving department are identified as related handover stages, and the overlap of the handover time intervals between each pair of related handover stages is judged; if the handover time intervals of related handover stages overlap, the corresponding handover stages are determined to be in parallel relationship and are assigned to the same parallel branch; if the handover time intervals of related handover stages do not overlap, the corresponding handover stages are determined to be in linear connection relationship according to the order of the start time of the handover time intervals.

[0047] Furthermore, based on the determination results of parallel and linear connections, a flow structure corresponding to the business item is constructed. The flow structure can be a linear flow structure, a parallel flow structure, or a hybrid flow structure containing linear connections and parallel branches. The actual flow node order of the business item is determined based on the position and sequence of each handover link in the flow structure. This identification method ensures that handover links with objectively existing parallel progression relationships can be correctly classified into parallel branches, avoiding misjudgment as abnormal flow order due to temporal overlap.

[0048] S2.6: For incomplete groups that contain only submit events but no corresponding receive events or only receive events but no corresponding submit events, record them as groups to be closed and mark the corresponding waiting time.

[0049] By distinguishing between rollback resubmissions and parallel split submissions, statistical bias in handover rounds is reduced; by identifying the parallel flow structure based on the overlapping time intervals of related handover stages, parallel stages are prevented from being misjudged as having abnormal order; and by marking incomplete matching groups as pending closure, related matters are prevented from being missed or confused with normal groups.

[0050] S3: Statistically analyze the original collaborative event sequence to obtain the real-time collaborative load index of the receiving department.

[0051] Specifically, S3 includes the following sub-steps: S3.1: Set a sliding statistics window, group the original collaborative event sequence according to the receiving department identifier, and obtain the candidate event subsequence of each receiving department in the current statistics window.

[0052] S3.2: Based on the item identifier and operation type identifier, identify the backlog items that have generated a receive event but have not generated a corresponding submit event from the candidate event subsequence, count the number of backlog items, and obtain the backlog to be processed.

[0053] In one embodiment, the average processing time and standard deviation of the processing time of historical processed items corresponding to the operation type of the backlog item are obtained. The sum of the average processing time and the standard deviation of the processing time of the preset confidence multiple is used as the dynamic backlog judgment threshold. The preset confidence multiple is set according to the required backlog judgment confidence level. It is determined whether the elapsed time of the backlog item since the corresponding received event timestamp exceeds the dynamic backlog judgment threshold. If it exceeds the threshold, it is determined to be a backlog item, and the backlog amount to be processed is calculated.

[0054] This dynamic threshold setting method allows the judgment criteria to be adaptively adjusted according to the fluctuation characteristics of the processing time of each operation type. Compared with a uniform fixed threshold, it reduces the situation where operation types with large fluctuations in processing time are misjudged as stalled, and operation types with generally short processing times are missed due to overly lenient thresholds.

[0055] S3.3: Calculate the dwell time based on the time difference between the timestamp of the received event for each dwelling item and the current statistical time.

[0056] S3.4: Match the preset complexity weight according to the operation type identifier corresponding to the delayed item, and sum the delayed duration with the preset complexity weight to obtain the backlog load value.

[0057] In one embodiment, a preset complexity weight table is queried based on the operation type identifier of the delayed item to obtain the corresponding complexity weight coefficient; the preset complexity weight table is calibrated based on the ratio of the average processing time of historical processed items of the same operation type to the global baseline processing time.

[0058] Furthermore, the maximum dwell time of historical delayed items of the same operation type is obtained, and the ratio of the dwell time to the maximum dwell time is used as the normalized dwell time. The normalized dwell time is weighted and summed with the complexity weight coefficient to obtain the individual backlog contribution value corresponding to the delayed item. The individual backlog contribution values ​​of all delayed items in the current statistical window are accumulated to obtain the backlog load value.

[0059] Therefore, in mixed scenarios with multiple operation types, the backlog load value does not mask the actual congestion of complex tasks due to the dominance of simple tasks, making the backlog load value closer to the actual processing pressure of the receiving department.

[0060] S3.5: Count the number of received events and the number of submitted events in the candidate event subsequence, calculate the event reception rate and event processing rate per unit time, and obtain the rate difference between the two.

[0061] S3.6: Determine the sign of the rate difference and dynamically adjust the weighting coefficients of the backlog load value and the rate difference in the fusion calculation based on the determination result. Perform weighted fusion based on the adjusted weighting coefficients to obtain the real-time collaborative load index of the receiving department.

[0062] In one embodiment, when the rate difference is positive, the weighting coefficient corresponding to the backlog load value is increased; when the rate difference is negative, the weighting coefficient corresponding to the backlog load value is decreased.

[0063] S4: Construct a desired collaborative path model based on the historical collaborative event sequence, and adjust the desired collaborative path model based on the real-time collaborative load index to obtain a dynamic desired collaborative path model.

[0064] Specifically, the flowchart for constructing the dynamic expected collaborative path model is as follows: Figure 3 As shown, it includes the following sub-steps: S4.1: Based on the historical flow node order in the historical collaborative event sequence, perform frequency statistics to determine the desired flow node order.

[0065] S4.2: Determine the expected processing time range and the expected number of handover rounds based on the historical handover time interval and the number of historical handover rounds in the historical collaborative event sequence.

[0066] It should be noted that, considering that in cross-departmental business collaboration scenarios, both historical processing time and historical handover rounds may generate outliers due to unconventional factors such as holiday shutdowns or special interventions, in one optional implementation, S4.2 can use statistical methods that are insensitive to outliers to determine the expected processing time range and the expected handover round range, so that the determined expected intervals can more accurately reflect the typical processing characteristics under normal collaboration processes; specific statistical methods can be selected by those skilled in the art based on the actual historical data distribution characteristics, and this invention does not limit them.

[0067] S4.3: Construct an initial expected collaborative path model based on the expected flow node order, expected processing time range, and expected handover round range.

[0068] Specifically, based on the expected flow node sequence, the handover links between adjacent nodes are identified to obtain handover link identifiers; it is determined whether there are missing data in the expected processing time range and expected handover round range corresponding to each handover link identifier; for handover links with missing data, the expected processing time range and expected handover round range are filled in according to the expected processing time range and expected handover round range of adjacent handover links or handover links of the same type to obtain the filled expected processing time range and expected handover round range.

[0069] Furthermore, the expected flow node sequence, handover link identifier, and the range of expected processing time and expected handover rounds are associated and mapped to construct an initial expected collaborative path model.

[0070] S4.4: Based on the historical collaborative load index sequence corresponding to the historical collaborative event sequence, the preset load capacity limit of the receiving department, the expected processing time range, and the expected handover round range, determine the time sensitivity coefficient and round sensitivity coefficient of the receiving department.

[0071] Specifically, based on the historical collaborative load index sequence corresponding to the historical collaborative event sequence and the preset load capacity limit of the receiving department, the historical load utilization rate of each historical moment is calculated; based on the receiving event timestamps corresponding to each historical handover link, the historical load utilization rate of the corresponding historical moment is matched to obtain the historical load utilization rate corresponding to each historical handover link.

[0072] Furthermore, based on the actual historical processing time corresponding to each historical handover stage, the offset amount exceeding the upper limit of the expected processing time range after completion in the initial expected collaborative path model is calculated to obtain the historical processing time offset sequence; based on the actual historical handover rounds corresponding to each historical handover stage, the offset amount exceeding the upper limit of the expected handover rounds range after completion in the initial expected collaborative path model is calculated to obtain the historical handover rounds offset sequence.

[0073] Furthermore, by regressing the historical load utilization rate corresponding to each historical handover stage with the historical processing time offset sequence and the historical handover round offset sequence, the time sensitivity coefficient and round sensitivity coefficient corresponding to the receiving department are obtained.

[0074] The preset load capacity limit is determined based on the peak load index of the receiving department during the historical statistical period or the rated design value of the department's manpower and system processing capacity.

[0075] S4.5: Determine the duration adjustment amount and the round adjustment amount based on the real-time collaborative load index, duration sensitivity coefficient and round sensitivity coefficient of the receiving department.

[0076] Specifically, the current load utilization rate is calculated based on the real-time collaborative load indicators of the receiving department and the preset load capacity limit; the duration adjustment amount is determined based on the current load utilization rate and the duration sensitivity coefficient; and the round adjustment amount is determined based on the current load utilization rate and the round sensitivity coefficient.

[0077] S4.6: Based on the duration adjustment amount and round adjustment amount, update and verify the upper limit of the expected processing duration range and the expected handover round range to obtain the dynamic expected collaborative path model.

[0078] Furthermore, the duration adjustment is added to the upper limit of the expected processing duration range after completion in the initial expected collaborative path model to obtain the upper limit of the duration adjustment; the round adjustment is added to the upper limit of the expected handover round range after completion in the initial expected collaborative path model to obtain the upper limit of the round adjustment; based on the upper limit of the duration adjustment and the upper limit of the round adjustment, the expected processing duration range and the expected handover round range after completion in the initial expected collaborative path model are updated respectively, and the validity of the updated expected processing duration range and the expected handover round range is verified to obtain the dynamic expected collaborative path model.

[0079] It should be noted that the validity verification adjustment refers to avoiding the upper limit value after adjustment from falling below the lower limit value of the corresponding interval, which would cause the interval to fail, in the process of adjusting the upper limit value according to the adjustment amount, thereby ensuring that the dynamic expected collaborative path model always maintains a valid interval structure; the specific verification method can be set by those skilled in the art according to actual needs, and this invention does not limit it.

[0080] S4 determines the expected baseline through statistical methods that are insensitive to outlier historical samples, and calibrates a unique load sensitivity coefficient for each receiving department. It drives online adjustments with a uniform load utilization rate, so that the dynamic expected collaborative path model can accurately reflect the typical characteristics of normal collaborative processes and reflect the differentiated responses of different departments to load pressure. Compared with conventional methods that use uniform thresholds or uniform adjustment ranges, it is more in line with the actual scenario of cross-departmental collaboration.

[0081] S5: Compare the handover time interval, number of handover rounds, and actual flow node sequence with the dynamic expected collaborative path model to generate collaborative status anomaly identifiers and corresponding abnormal department identifiers.

[0082] Specifically, S5 includes the following sub-steps: S5.1: Compare the actual flow node order with the expected flow node order in the dynamic expected collaborative path model to obtain structural similarity, link mapping relationship and unmatched link identifier.

[0083] Specifically, the actual flow node order is compared with the expected flow node order in the dynamic expected collaborative path model to obtain the structural similarity and the link mapping relationship between the actual handover link and the expected handover link; based on the actual handover link or expected handover link that does not form a corresponding relationship during the structural similarity comparison, the unmatched link identifier is obtained.

[0084] S5.2: Based on the link mapping relationship, compare the actual handover time interval with the corresponding dynamic expected processing time range to obtain the total time deviation.

[0085] Furthermore, based on the link mapping relationship, the actual handover time interval of each corresponding handover link is compared with the dynamic expected processing time range of the corresponding expected handover link to obtain the time deviation of each handover link; the time deviation of each handover link is aggregated to obtain the total time deviation of the business matter.

[0086] S5.3: Based on the link mapping relationship, compare the actual number of handover rounds with the corresponding dynamic expected handover round range to obtain the total number of round deviations.

[0087] Accordingly, based on the link mapping relationship, the actual number of handover rounds for each corresponding handover link is compared with the dynamic expected handover round range for the corresponding expected handover link to obtain the round deviation of each handover link; the round deviations of each handover link are aggregated to obtain the total round deviation of the business matter.

[0088] S5.4: Determine the abnormal status identifier of the collaborative state based on structural similarity, total duration deviation, and total round deviation.

[0089] In one optional implementation, the three factors are normalized according to their respective historical fluctuation ranges: structural similarity, total duration deviation, and total round deviation. The normalized structural similarity, total duration deviation, and total round deviation are then weighted and fused according to preset weights to obtain a comprehensive anomaly score. It is then determined whether the comprehensive anomaly score exceeds a preset comprehensive anomaly threshold. If it does, a collaborative state anomaly identifier is generated.

[0090] S5.5: When generating an abnormal status identifier for collaborative status, determine the abnormal department identifier based on the deviation of each handover link.

[0091] Furthermore, by comparing the duration deviation and round deviation of each handover stage, as well as the handover stage corresponding to the mismatched stage identifier, the receiving department identifier corresponding to the handover stage with the largest deviation contribution is identified as the abnormal department identifier.

[0092] S5 normalizes and weights deviation information from three dimensions—actual workflow structure, processing time, and handover rounds—and further locates the specific handover link with the greatest deviation contribution after an anomaly is determined. This helps reduce false alarms caused by reasonable variations in business paths and improves the accuracy of anomaly location by relevant departments.

[0093] S6: Based on the dynamic expected collaborative path model, identify business items with continuously missing interface call messages and generate monitoring blind spot identifiers.

[0094] Specifically, S6 includes the following sub-steps: S6.1: Based on the events that have occurred in the business transaction, and in conjunction with the dynamic expected collaborative path model, determine the next expected flow node of the business transaction.

[0095] In one embodiment, the nodes and timestamps corresponding to the submission and reception events generated by the business transaction are obtained, and the submission and reception events whose corresponding interface call messages are marked as parsing errors are removed; the removed nodes are sorted by timestamp to obtain a candidate node sequence.

[0096] Furthermore, the candidate node sequence is compared with the expected flow node order in the dynamic expected collaborative path model to determine the last node in the candidate node sequence that conforms to the expected flow node order, which is taken as the last node that has generated a submission event or a reception event; according to the expected flow node order, the next node in the expected flow node order is determined as the next expected flow node; if the candidate node sequence is empty, the starting node in the dynamic expected collaborative path model is taken as the next expected flow node.

[0097] S6.2: Based on the dynamic expected collaborative path model, determine the expected processing time range corresponding to the next expected transfer node.

[0098] Furthermore, based on the last node and the next expected transfer node, the corresponding handover link is determined in the expected transfer node sequence. If the last node does not exist, the handover link corresponding to the starting node and the next node in the expected transfer node sequence is determined as the reference handover link. The expected processing time range of the reference handover link is used to approximately represent the expected processing time required for the business item to reach the starting node. The expected processing time range corresponding to the handover link or the reference handover link is extracted from the dynamic expected collaborative path model as the expected processing time range corresponding to the next expected transfer node.

[0099] This approach allows for an approximate estimate of the expected processing time range even when no valid events have yet occurred (no final node), by referring to adjacent known handover stages, thus avoiding blind spot judgments due to a lack of historical events.

[0100] S6.3: When the expected processing time range expires, determine whether there is a corresponding submission event or reception event record for the next expected transfer node; if so, do not generate a monitoring blind spot identifier.

[0101] S6.4: If there are no submission or reception event records, and the duration of missing interface call messages exceeds the upper limit of the expected processing time range, a monitoring blind spot identifier will be generated.

[0102] S7: Push collaborative reminder signals based on the abnormal status and abnormal department identifiers, and trigger manual verification prompt signals based on the monitoring blind spot identifiers.

[0103] Accordingly, refer to Figure 4 This invention also provides a cross-departmental business collaborative tracking system based on AI big data, which may include: The message acquisition and parsing module is used to collect interface call messages generated between departmental business systems during the processing of business matters, and parse the interface call messages to obtain the original collaborative event sequence. The handover analysis module is used to identify submission and reception events in the original collaborative event sequence. It groups and compares submission and reception events between two adjacent departments for the same business matter to obtain the handover time interval, the number of handover rounds, and the actual flow node order. The load statistics module is used to statistically analyze the original collaborative event sequence and obtain the real-time collaborative load index of the receiving department. The dynamic path modeling module is used to construct a desired collaborative path model based on historical collaborative event sequences, and adjust the desired collaborative path model based on real-time collaborative load indicators to obtain a dynamic desired collaborative path model. The anomaly comparison module is used to compare the handover time interval, the number of handover rounds, and the actual flow node sequence with the dynamic expected collaborative path model, and generate collaborative status anomaly identifiers and corresponding abnormal department identifiers. The blind spot identification module is used to identify business items with continuously missing interface call messages based on the dynamic expected collaborative path model and generate monitoring blind spot identifiers. The signal push module is used to push collaborative reminder signals based on the collaborative status abnormality identifier and the abnormal department identifier, and to trigger a manual verification prompt signal based on the monitoring blind spot identifier.

[0104] This invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program; when the computer program is executed by the processor, the processor performs the steps of the cross-departmental business collaborative tracking method based on AI big data as described in any of the above embodiments.

[0105] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the AI ​​big data-based cross-departmental business collaborative tracking method as described in any of the above embodiments.

[0106] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0107] For a description of the computer-readable storage medium provided by the present invention, please refer to the above method embodiments; the present invention will not be described in detail here.

[0108] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0109] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cross-departmental business collaborative tracking method based on AI big data, characterized in that, Includes the following steps: Collect interface call messages generated between business systems of the collection department during the process of handling business matters, and parse the interface call messages to obtain the original collaborative event sequence; Identify the submission and reception events in the original collaborative event sequence, group and compare the submission and reception events between two adjacent departments for the same business matter, and obtain the handover time interval, the number of handover rounds, and the actual flow node order; The real-time collaborative load index of the receiving department is obtained by statistically analyzing the original collaborative event sequence. A desired collaborative path model is constructed based on the historical collaborative event sequence, and the desired collaborative path model is adjusted based on the real-time collaborative load index to obtain a dynamic desired collaborative path model. The handover time interval, the number of handover rounds, and the actual flow node sequence are compared with the dynamic expected collaborative path model to generate a collaborative status anomaly identifier and a corresponding abnormal department identifier. Based on the dynamic expected collaborative path model, identify business items with continuously missing interface call messages and generate monitoring blind spot identifiers; Based on the abnormal collaborative status identifier and the abnormal department identifier, a collaborative reminder signal is pushed, and based on the monitoring blind spot identifier, a manual verification prompt signal is triggered.

2. The cross-departmental business collaborative tracking method based on AI big data according to claim 1, characterized in that, The original cooperative event sequence obtained from the parsing interface call message includes: Send capability detection requests to the business systems of each department, and receive the version identifier of the interface call log generation module and the declaration information of the log output interface returned by the business systems of each department; The version identifier of the interface call log generation module and the declaration information of the log output interface are compared with the preset interface log capability feature library to obtain the coverage capability identifier. Based on the coverage capability identifier, the business systems of each department are classified and the collection channels are configured to obtain a set of business systems of the departments to be collected. Establish message collection connections with the business systems of each department in the set of business systems to be collected, and collect the interface call messages generated between the business systems of the departments to be collected during the processing of business matters. Parse the interface call message, extract the key field information corresponding to the business item, and mark the interface call message that fails to be parsed; The timestamps of the business systems of each department are calibrated, and the extracted results are arranged in chronological order according to the calibrated timestamps to obtain the original collaborative event sequence.

3. The cross-departmental business collaborative tracking method based on AI big data according to claim 1, characterized in that, The grouping and comparison of submission and receipt events between two adjacent departments for the same business transaction includes: Determine whether there are multiple submission events contained in the event group. If there are multiple submission events in the event group, determine the type of multiple submission events and determine the handover time interval of the corresponding handover stage based on the determination result. Based on the type of the submission event, the number of corresponding submission events is counted to obtain the number of handover rounds; Based on the sequence of each handover step and the correlation of the handover time intervals, the flow structure of the business matter is identified to determine the actual flow node sequence of the business matter.

4. The cross-departmental business collaborative tracking method based on AI big data according to claim 1, characterized in that, The process of determining the real-time coordinated load index includes: Set up a sliding statistics window, group the original collaborative event sequence according to the receiving department identifier, and obtain the candidate event subsequence of each receiving department in the current statistics window; Based on the item identifier and operation type identifier, identify the backlog items that have generated a receive event but have not generated a corresponding submit event from the candidate event subsequence, count the number of the backlog items, and obtain the backlog to be processed; The duration of the delay is calculated based on the time difference between the timestamp of the received event for each delayed item and the current statistical time. Based on the operation type identifier corresponding to the delayed item, a preset complexity weight is matched, and the delay time and the preset complexity weight are weighted and summed to obtain the backlog load value; The number of received events and the number of submitted events in the candidate event subsequence are counted, and the event reception rate and event processing rate per unit time are calculated respectively to obtain the rate difference between the two. The sign of the rate difference is determined, and the weighting coefficients of the backlog load value and the rate difference in the fusion operation are dynamically adjusted according to the determination result. Based on the adjusted weighting coefficients, weighted fusion is performed to obtain the real-time collaborative load index of the receiving department.

5. The cross-departmental business collaborative tracking method based on AI big data according to claim 1, characterized in that, The construction process of the dynamic expected collaborative path model includes: Based on the historical flow node order in the historical collaborative event sequence, frequency statistics are performed to determine the desired flow node order; Based on the historical handover time intervals and the number of historical handover rounds in the historical collaborative event sequence, determine the expected processing time range and the expected number of handover rounds range; Based on the expected flow node order, the expected processing time range, and the expected handover round range, an initial expected collaborative path model is constructed. Based on the historical collaborative load index sequence corresponding to the historical collaborative event sequence, the preset load capacity limit of the receiving department, the expected processing time range, and the expected handover round range, determine the time sensitivity coefficient and round sensitivity coefficient of the receiving department. Based on the real-time collaborative load index of the receiving department, the duration sensitivity coefficient, and the round sensitivity coefficient, determine the duration adjustment amount and the round adjustment amount; Based on the duration adjustment amount and the round adjustment amount, the upper limits of the expected processing duration range and the expected handover round range are updated and their validity is verified to obtain a dynamic expected collaborative path model.

6. The cross-departmental business collaborative tracking method based on AI big data according to claim 1, characterized in that, The generation of collaborative status anomaly identifiers and corresponding anomaly department identifiers includes: The actual flow node order is compared with the expected flow node order in the dynamic expected collaborative path model to obtain structural similarity, link mapping relationship and unmatched link identifier; Based on the aforementioned link mapping relationship, the actual handover time interval is compared with the corresponding dynamic expected processing time range to obtain the total time deviation. Based on the aforementioned link mapping relationship, the actual number of handover rounds is compared with the corresponding dynamic expected handover round range to obtain the total number of round deviations; Based on the structural similarity, the total duration deviation, and the total round deviation, an abnormal collaborative state identifier is determined; When the aforementioned collaborative status anomaly identifier is generated, the anomaly department identifier is determined based on the deviation of each handover link.

7. The cross-departmental business collaborative tracking method based on AI big data according to claim 1, characterized in that, The generated monitoring blind spot identifier includes: Based on the events that have occurred in the business transaction, and in conjunction with the dynamic expected collaborative path model, determine the next expected flow node of the business transaction. Based on the dynamic expected collaborative path model, the expected processing time range corresponding to the next expected transfer node is determined; When the expected processing time range expires, determine whether there is a corresponding submission event or reception event record for the next expected transfer node; if so, do not generate a monitoring blind spot identifier. If no submission or reception event record exists, and the duration of missing interface call messages exceeds the upper limit of the expected processing time range, a monitoring blind spot identifier is generated.

8. A cross-departmental business collaborative tracking system based on AI big data, based on the cross-departmental business collaborative tracking method based on AI big data as described in any one of claims 1 to 7, characterized in that, include: The message acquisition and parsing module is used to collect interface call messages generated between departmental business systems during the processing of business matters, and parse the interface call messages to obtain the original collaborative event sequence. The handover analysis module is used to identify submission and reception events in the original collaborative event sequence, and to group and compare submission and reception events between two adjacent departments for the same business matter to obtain the handover time interval, the number of handover rounds, and the actual flow node order. The load statistics module is used to statistically analyze the original collaborative event sequence to obtain the real-time collaborative load index of the receiving department. The dynamic path modeling module is used to construct a desired collaborative path model based on the historical collaborative event sequence, and adjust the desired collaborative path model according to the real-time collaborative load index to obtain a dynamic desired collaborative path model. The anomaly comparison module is used to compare the handover time interval, the number of handover rounds, and the actual flow node sequence with the dynamic expected collaborative path model to generate a collaborative status anomaly identifier and a corresponding anomaly department identifier. The blind spot identification module is used to identify business items with continuously missing interface call messages based on the dynamic expected collaborative path model, and generate monitoring blind spot identifiers. The signal push module is used to push collaborative reminder signals based on the collaborative status abnormality identifier and the abnormal department identifier, and to trigger a manual verification prompt signal based on the monitoring blind spot identifier.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the cross-departmental business collaborative tracking method based on AI big data as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program, wherein the computer program, when executed by a processor, implements the cross-departmental business collaborative tracking method based on AI big data as described in any one of claims 1 to 7.

Images