CN115391080A - Process consistency checking method and system based on attribute filtering - Google Patents

Abstract

The invention discloses a flow consistency inspection method and a system based on attribute filtering, belonging to the field of quantity flow inspection, wherein the inspection method comprises the following specific steps: (1) constructing an attribute filter and debugging; (2) selecting a data flow and then carrying out classification recording; (3) carrying out attribute filtering on the data flow; (4) interrupting and repairing the abnormal data flow; (5) feeding back operation parameters of each data flow; according to the invention, by generating the performance curve graph of the attribute screener, the operation performance can be more intuitively fed back to the working personnel for checking, the analysis and optimization of the working personnel are facilitated, the analysis steps of the working personnel are simplified, the working efficiency of the working personnel is improved, the attribute screener is optimized by constructing the test model, the data transmission channel of the attribute screener is prevented from being occupied, the screening efficiency of the data screener is ensured, the real-time optimization of the attribute screener can be realized, and the screening accuracy is improved.

Description

Process consistency checking method and system based on attribute filtering

Technical Field

The invention relates to the field of quantitative process inspection, in particular to a process consistency inspection method and system based on attribute filtering.

Background

The process mining is a digital tool emerging in recent years, and the working principle of the process mining is to extract the time and the associated information of each process activity from an event log recorded by an information system so as to restore the actual working condition of the process, wherein the process consistency refers to evaluating whether the existing process path and the standard process path are consistent or not, quantitatively evaluating the consistency degree between the two paths, and finding out the inconsistent place between the two paths. By examining the deviation between the two and weighing the severity of the deviation, the process can be improved to improve the efficiency of the process model, so that the process model can better serve businesses. The attribute filter is one of important tools for detecting the consistency of the process, can filter according to some 'attributes' of data, is more flexible than a common filter, and can obtain a more intelligent effect than the common filter;

the existing method and system for checking the consistency of the process based on attribute filtering; in addition, when the existing method and system for checking the consistency of the process based on the attribute filtering optimize the attribute filter, the transmission channel of the attribute filter is easily occupied, the screening efficiency is reduced, and the real-time optimization of the attribute filter cannot be realized; therefore, a flow consistency checking method and system based on attribute filtering are provided.

Disclosure of Invention

The invention aims to solve the defects in the prior art, and provides a method and a system for checking process consistency based on attribute filtering.

In order to achieve the purpose, the invention adopts the following technical scheme:

a flow consistency inspection method based on attribute filtering comprises the following specific steps:

(1) Constructing an attribute filter and debugging;

(2) Selecting a data flow and then carrying out classification recording;

(3) Performing attribute filtering on the data flow;

(4) Interrupting and repairing an abnormal data flow;

(5) And feeding back the operation parameters of each data flow.

As a further scheme of the invention, the specific steps of debugging the attribute filter in the step (1) are as follows:

the method comprises the following steps: the method comprises the following steps that a worker issues selected attribute information through a management platform, then data containing the attribute information are extracted, and an attribute filter is constructed according to each extracted group of data;

step two: then, the test optimization module acquires various data information from the external Internet, automatically generates a plurality of groups of test data with different attributes according to the acquired information, loads the various test data into an attribute filter for filtering simulation, and collects each group of simulation results for verification and optimization;

step three: and manually checking and analyzing whether each group of data screening is accurate, receiving a manual analysis result by the test optimization module, calculating a screening loss value of the attribute screener through a focus loss function, evaluating the accuracy, the detection rate and the false alarm rate of the attribute screener according to the loss value and the analysis result, and generating a corresponding evaluation curve chart for a worker to check.

As a further scheme of the invention, the verification optimization in the step one specifically comprises the following steps:

the first step is as follows: selecting one simulation result from the N groups of simulation results as verification data, fitting the rest simulation results into a test model, verifying the precision of the test model by using the simulation result which is excluded firstly, calculating the prediction capability of the prediction model through the root mean square error, repeating the steps for N times, and then performing parameter optimization processing on the generated precision parameters;

the second step: initializing a parameter range, enabling a learning rate eta = [0.0001,0.1], enabling a step length to be 0.0001, listing all possible data results, selecting any subset as a test set and the rest subsets as training sets for each group of data, predicting the test set after training a model, and counting the root mean square error of the test result;

the third step: and solving an optimal parameter combination, simultaneously replacing the test set with another subset, taking the residual subset as a training set, counting the root mean square error again until each group of data is subjected to primary prediction, selecting the corresponding combination parameter when the RMSE is minimum as the optimal parameter in the data interval, then selecting the optimal parameter, and training the attribute filter by adopting a long-term iteration method.

As a further scheme of the invention, the specific steps of the classification record in the step (2) are as follows:

s1.1: according to the data information selected by the staff, capturing the corresponding data flow and carrying out sectional processing on the related data flow according to the sequence transmitted in each group of data flow;

s1.2: and then constructing a data record table to record the captured data flow information and the segmentation result, extracting the data attributes in each segment of flow, and simultaneously recording the extracted data attributes of each group into the corresponding data flow segment in the data record table.

As a further scheme of the present invention, the attribute filtering in step (3) specifically comprises the following steps:

s2.1: the attribute filter receives the extracted various data attributes, matches the extracted various data attributes with manually selected attributes and a certain specific value in the attribute room, and screens out various groups of data which are not successfully matched;

s2.2: then marking the data flow sections corresponding to the successfully unmatched groups of data, performing scale normalization processing on the successfully matched groups of data, identifying the data with matched topological structures or spatial modes, and forming corresponding elements of the data into a matched group for matching and screening;

s2.3: and updating the fields of the matched groups which are not confirmed by the appointed matching fields, re-matching, screening out the matched groups with field value difference in the matching fields, and marking the corresponding data flow sections.

As a further scheme of the invention, the specific steps of abnormal data flow repair are as follows:

p1: extracting corresponding process information according to the marking information, sending the feature data of each group of extracted process information into a bidirectional feature pyramid for feature fusion, classifying and regressing the fusion result, and outputting a detection frame and a category;

p2: and collecting detection frame information, acquiring detection frame coordinate information, performing expanded cutting on related flow information to acquire abnormal information, and feeding back each group of abnormal information and corresponding data flow sections to a worker for checking and repairing.

A flow consistency checking system based on attribute filtering comprises a management platform, an attribute filter, a test optimization module, a flow processing module, an exception feedback module, a repair detection module and a server;

the management platform is used for managing and regulating each submodule and feeding back each set of data fed back by each submodule to a worker for checking;

the attribute filter is used for receiving attribute information selected by a worker and screening inconsistent data flows according to the attribute information;

the test simulation module is used for collecting the operation information of the attribute filter and carrying out optimization adjustment on the operation information;

the flow processing module is used for capturing the corresponding data flow according to the information selected by the staff and carrying out classification processing on the data flow;

the abnormal feedback module is used for receiving the inconsistent data flows screened by the attribute filter and feeding the inconsistent data flows back to the management platform;

the repair detection module is used for monitoring the repair progress of the abnormal data flow in real time and prompting a worker to repair;

the server is used for storing data flow operation information and repair records.

Compared with the prior art, the invention has the beneficial effects that:

1. compared with the conventional inspection method, the flow consistency inspection method based on the attribute filtering comprises the steps that data containing attribute information selected by workers are extracted, an attribute filter is constructed according to the extracted data of each group, screening simulation is carried out on the data, then, manual inspection and analysis are carried out on whether the data of each group are accurately screened or not, a test optimization module receives a manual analysis result, then, a focus loss function is used for calculating a screening loss value of the attribute filter, the accuracy, the detection rate and the false alarm rate of the attribute filter are evaluated according to the loss value and the analysis result, corresponding evaluation graphs are generated to be checked by the workers, then, the attribute filter receives the extracted data attributes, the data attributes are matched with the manually selected attributes and a certain specific value in an attribute chamber, data flow sections corresponding to the data of each group which are not successfully matched are marked, corresponding flow information is extracted, abnormal information is obtained through expanding and cutting, and feedback is carried out, and a performance graph of the attribute filter is generated, so that the operation performance can be more visually fed back to the workers to be checked conveniently, analysis and optimization of the workers are simplified, the steps of the workers, and the work efficiency of the workers is improved;

2. the invention selects one simulation result from N groups of simulation results as verification data, uses the rest simulation results to fit into a test model, uses the simulation result which is excluded firstly to verify the precision of the test model, calculates the prediction capability of the prediction model through root mean square error, repeats the operation for N times, then performs parameter optimization processing on the generated precision parameters, initializes the parameter range, predicts the selected test set after training the model, counts the root mean square error of the test result, obtains the optimal parameter combination, after performing one prediction on each group of data, selects the corresponding combination parameter when RMSE is minimum as the optimal parameter in the data interval, then selects the optimal parameter, trains the attribute screener by adopting a long-term iteration method, optimizes the attribute screener by constructing the test model, avoids occupying the data transmission channel of the attribute screener, ensures the screening efficiency of the attribute screener, can realize the real-time optimization of the attribute screener, and improves the screening accuracy.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.

Fig. 1 is a block diagram of a process consistency checking method based on attribute filtering according to the present invention;

fig. 2 is a system block diagram of a process consistency check system based on attribute filtering according to the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.

Example 1

Referring to fig. 1, the present embodiment discloses a method for checking process consistency based on attribute filtering, which includes the following specific steps:

and constructing an attribute filter and debugging.

Specifically, a worker issues selected attribute information through a management platform, extracts data containing the attribute information, constructs an attribute filter according to the extracted data of each group, acquires various data information from the external internet through a test optimization module, automatically generates a plurality of groups of test data with different attributes according to the acquired information, loads the test data into an attribute filter for filtering simulation, collects simulation results of each group for verification and optimization, manually checks and analyzes whether the screening of each group of data is accurate or not, receives the manual analysis result through a focus loss function, calculates a screening loss value of the attribute filter, evaluates the accuracy, the detection rate and the false alarm rate of the attribute filter according to the loss value and the analysis result, and generates a corresponding evaluation curve graph for the worker to check.

It should be further explained that one simulation result is selected from N sets of simulation results as verification data, the remaining simulation results are fitted into a test model, the simulation result which is excluded first is used to verify the precision of the test model, the prediction capability of the prediction model is calculated through root mean square error, the above steps are repeated for N times, then parameter optimization processing is performed on the generated precision parameters, the parameter range is initialized, the learning rate eta = [0.0001,0.1] is made, the step length is 0.0001, all possible data results are listed, for each set of data, any subset is selected as a test set, the rest subsets are selected as a training set, the test set is predicted after the training model is performed, the root mean square error of the test results is counted, the optimal parameter combination is obtained, the test set is replaced by another subset, the rest subsets are selected as the training set, the root mean square error is counted again until each set of data is predicted once, the corresponding combination parameter when the RMSE is the minimum is selected as the optimal parameter in the data interval, the optimal parameter is selected, and long-term iterative training attributes are adopted for screening.

And selecting a data flow and then performing classification recording.

Specifically, the process processing module captures corresponding data processes according to data information selected by a worker, and performs segmentation processing on the relevant data processes according to the sequence transmitted in each group of data processes, then constructs a data record table to record each captured data process information and a segmentation result, extracts data attributes in each section of process, and records each extracted group of data attributes into the corresponding data process in the data record table.

And performing attribute filtering on the data flow.

Specifically, the attribute filter receives various extracted data attributes, matches the extracted data attributes with manually selected attributes and a specific value in the attribute chamber, screens out various groups of data which are not successfully matched, marks data process sections corresponding to the various groups of data which are not successfully matched, performs scale normalization processing on the various groups of data which are successfully matched, identifies data with matched topological structures or spatial modes, forms corresponding elements of the data into a matching group for matching and screening, performs field updating on the matching group which is not confirmed by the specified matching field, performs re-matching, screens out the matching group with field value difference in the matching field, and marks the corresponding data process sections.

And interrupting and repairing the abnormal data flow.

Specifically, corresponding process information is extracted according to the mark information, the feature data of each group of extracted process information is sent to a bidirectional feature pyramid for feature fusion, the fusion result is classified and regressed, a detection frame and a category are output, then detection frame information is collected, coordinate information of the detection frame is obtained, related process information is expanded and cut to obtain abnormal information, and each group of abnormal information and corresponding data process sections are fed back to a worker for checking and repairing.

And feeding back the operation parameters of each data flow.

Example 2

Referring to fig. 2, the embodiment discloses a process consistency checking system based on attribute filtering, which includes a management platform, an attribute filter, a test optimization module, a process processing module, an exception feedback module, a repair detection module, and a server.

The management platform is used for managing and regulating each submodule and feeding back each set of data fed back by each submodule to a worker for checking.

The attribute filter is used for receiving attribute information selected by a worker and screening inconsistent data flows according to the attribute information; and the test simulation module is used for collecting the operation information of the attribute filter and carrying out optimization adjustment.

And the flow processing module is used for capturing the corresponding data flow according to the information selected by the staff and classifying the data flow.

And the exception feedback module is used for receiving the inconsistent data flows screened out by the attribute filter and feeding back the inconsistent data flows to the management platform.

The repair detection module is used for monitoring the repair progress of the abnormal data flow in real time and prompting a worker to repair; the server is used for storing data flow operation information and repair records.

Claims (7)

1. A flow consistency inspection method based on attribute filtering is characterized by comprising the following specific steps:

(1) Constructing an attribute filter and debugging;

(2) Selecting a data flow and then carrying out classification recording;

(3) Performing attribute filtering on the data flow;

(4) Interrupting and repairing the abnormal data flow;

(5) And feeding back the operation parameters of each data flow.

2. The method for flow consistency check based on attribute filtering as claimed in claim 1, wherein the specific steps of debugging the attribute filter in step (1) are as follows:

the method comprises the following steps: the method comprises the following steps that a worker issues selected attribute information through a management platform, then data containing the attribute information are extracted, and an attribute filter is constructed according to each extracted group of data;

step two: then, the test optimization module acquires various data information from the external Internet, generates a plurality of groups of test data with different attributes according to the acquired information, loads the various test data into an attribute filter for filtering simulation, and collects the simulation results of the groups for verification and optimization;

step three: and manually checking and analyzing whether each group of data screening is accurate, receiving a manual analysis result by the test optimization module, calculating a screening loss value of the attribute screener through a focus loss function, evaluating the accuracy, the detection rate and the false alarm rate of the attribute screener according to the loss value and the analysis result, and generating a corresponding evaluation curve chart for a worker to check.

3. The method for checking process consistency based on attribute filtering as claimed in claim 2, wherein the verification optimization in the first step specifically comprises the following steps:

the first step is as follows: selecting one simulation result from the N groups of simulation results as verification data, fitting the rest simulation results into a test model, verifying the precision of the test model by using the simulation result which is excluded firstly, calculating the prediction capability of the prediction model through the root mean square error, repeating the steps for N times, and then performing parameter optimization processing on the generated precision parameters;

the second step: initializing a parameter range, enabling a learning rate eta = [0.0001,0.1], enabling a step length to be 0.0001, listing all possible data results, selecting any subset as a test set and the rest subsets as training sets for each group of data, predicting the test set after training a model, and counting the root mean square error of the test result;

the third step: and (3) solving an optimal parameter combination, simultaneously replacing the test set with another subset, then taking the residual subset as a training set, counting the root mean square error again until each group of data is predicted once, selecting the corresponding combination parameter when the RMSE is minimum as the optimal parameter in the data interval, then selecting the optimal parameter, and training the attribute filter by adopting a long-term iteration method.

4. The method for checking process consistency based on attribute filtering as claimed in claim 1, wherein the specific steps of the classification record in the step (2) are as follows:

s1.1: capturing corresponding data flows according to the data information selected by the staff, and carrying out sectional processing on the related data flows according to the sequence transmitted in each group of data flows;

s1.2: and then constructing a data record table to record the captured data flow information and the segmentation result, extracting the data attributes in each segment of flow, and simultaneously recording the extracted data attributes of each group into the corresponding data flow segment in the data record table.

5. The method for checking process consistency based on attribute filtering as claimed in claim 1, wherein the attribute filtering in step (3) specifically comprises the following steps:

s2.1: the attribute filter receives the extracted various data attributes, matches the extracted various data attributes with the manually selected attributes and a certain specific value in the attribute chamber, and screens out various groups of data which are not successfully matched;

s2.2: then marking the data flow section corresponding to each group of data which is not successfully matched, then carrying out scale normalization processing on each group of data which is successfully matched, identifying the data with a matched topological structure or spatial mode, and then forming a matched group by corresponding elements of each data for matching and screening;

s2.3: and updating the fields of the matching groups which are not confirmed by the appointed matching fields, re-matching, screening out the matching groups with field value difference in the matching fields, and marking the corresponding data flow sections.

6. The method for flow consistency check based on attribute filtering as claimed in claim 5, wherein the specific steps for abnormal data flow repair are as follows:

p1: extracting corresponding process information according to the marking information, sending the feature data of each group of extracted process information into a bidirectional feature pyramid for feature fusion, classifying and regressing the fusion result, and outputting a detection frame and a category;

p2: and collecting detection frame information, acquiring detection frame coordinate information, performing expanded cutting on related flow information to acquire abnormal information, and feeding back each group of abnormal information and corresponding data flow sections to workers for checking and repairing.

7. A flow consistency checking system based on attribute filtering is characterized by comprising a management platform, an attribute filter, a test optimization module, a flow processing module, an exception feedback module, a repair detection module and a server;

the management platform is used for managing and regulating each submodule and feeding back each set of data fed back by each submodule to a worker for checking;

the attribute filter is used for receiving attribute information selected by a worker and screening inconsistent data flows according to the attribute information;

the test simulation module is used for collecting the operation information of the attribute filter and carrying out optimization adjustment on the operation information;

the flow processing module is used for capturing the corresponding data flow according to the information selected by the staff and classifying the data flow;

the abnormal feedback module is used for receiving the inconsistent data flows screened by the attribute filter and feeding back the inconsistent data flows to the management platform;

the repair detection module is used for monitoring the repair progress of the abnormal data flow in real time and prompting a worker to repair;

the server is used for storing data flow operation information and repair records.

Images