Project information import and defect distribution management system and method based on intelligent matching

Abstract

According to the project information import and defect distribution management system and method based on intelligent matching provided by the invention, the one-key import module, the defect intelligent analysis module, the associated personnel matching engine, the automatic distribution and notification module and the feedback optimization module are integrated, so that automatic management of project information and intelligent distribution of defects are realized; the manual intervention cost is reduced, and the project management efficiency is remarkably improved. The system can automatically import multi-format project data, intelligently analyze defect types, severity levels and module paths, dynamically match responsible persons and push allocation suggestions, meanwhile, the matching rule weight is optimized according to manual adjustment data, the allocation accuracy and the processing efficiency are continuously improved, it is ensured that defects are allocated to the most appropriate responsible persons, the processing quality is improved, and the system is suitable for popularization and application. The closed-loop management of defect processing is realized, and subsequent tracking and optimization are facilitated. And negative sample information is continuously collected and matching rule weights are optimized, so that continuous optimization and upgrading of the system are realized to adapt to project changes and demand upgrading.

Description

Technical Field

[0001] The present application relates to the technical field of project information and defect management, and more specifically to a management system and method for project information import and defect allocation based on intelligent matching. Background Art

[0002] As software development and project management become increasingly complex, traditional approaches to project information management and defect allocation face numerous challenges. While mainstream project management tools like JIRA, Trello, and Asana provide basic project and defect tracking capabilities, they still struggle with intelligent matching and automated decision-making.

[0003] Current project information management and defect management systems separate project information from defect tracking systems, lacking dynamic linkage. Project information is typically imported manually, and traditional project management relies on manual organization of Excel documents. This approach is inefficient, error-prone, and time-consuming. Furthermore, manual determination of defect attribution is prone to omissions or improper assignments, resulting in inaccurate defect allocation and delayed repairs. Summary of the Invention

[0004] In order to solve the problem that in the above-mentioned method of using the operating status data uploaded by the refrigerator to the cloud to detect fan faults, the detected fan faults are mixed with a large amount of heater faults and refrigerator door not closed tightly data, and the accuracy rate is very low.

[0005] This application provides a project information import and defect allocation management system based on intelligent matching, including: a one-click import module with communication connection, a defect intelligent analysis module, an associated personnel matching engine, an automatic allocation and notification module, and a feedback optimization module;

[0006] The one-key import module is configured to: receive a multi-format project data source input, generate semi-structured data through the multi-format parsing unit, and output structured project data containing defect titles or description text through the data cleaning and verification unit;

[0007] The intelligent defect analysis module is configured to: in response to the input of the structured project data, process the structured project data through natural language processing, extract the defect type, severity level and module path, and generate a JSON format analysis report in combination with a rule base for dynamic matching;

[0008] The associated personnel matching engine is configured to: calculate personnel priority scores based on the analysis report through a dynamic rule matching unit, and output responsible person assignment suggestions in descending order of scores;

[0009] The automatic allocation and notification module is configured to: synchronously push the allocation suggestion to the external task system and capture the manual adjustment operation as a negative sample;

[0010] The feedback optimization module is configured to optimize the rule weights of the associated person matching engine according to the negative sample supervised learning.

[0011] In a feasible implementation, the one-key import module includes: a data source input unit, a multi-format parsing unit, and a data cleaning unit;

[0012] The data source input unit is configured to: support drag-and-drop upload, API docking, and manual path input of the multi-format project data source;

[0013] The multi-format parsing unit is configured to: parse merged cells in the Excel format in the multi-format project data source, automatically identify delimiters in the CSV format, and parse JSON key fields in the JIRA format;

[0014] The data cleaning unit is configured to perform title deduplication based on Levenshtein distance, mandatory field non-empty verification, and enumeration value compliance verification on the multi-format project data source.

[0015] In a feasible implementation, the defect intelligent analysis module includes: a key attribute extraction unit and a rule base dynamic matching unit;

[0016] The key attribute extraction unit is configured to: match predefined defect types through named entity recognition, determine severity levels in combination with sentiment analysis, and parse module paths based on a code structure tree;

[0017] The rule base dynamic matching unit is configured to map the business rules described in natural language into standardized defect types.

[0018] In a feasible implementation, the dynamic rule matching unit of the associated personnel matching engine includes: a role responsibility rule library, a historical processing record analyzer, and a load statistics analyzer;

[0019] The role responsibility rule base is used to match defect modules with personnel skill tags;

[0020] The historical processing record analyzer is used to calculate the average time and success rate of personnel in resolving similar defects;

[0021] The load statistics device is used to dynamically adjust the weight factor based on the number of unclosed-loop tasks.

[0022] In a feasible implementation, the automatic allocation and notification module is further configured to:

[0023] Push the allocation results to DingTalk / WeChat Enterprise via the Webhook interface;

[0024] Generate an email template with a link to the defect details;

[0025] The task panel status is updated in real time and the manual adjustment operation is recorded.

[0026] In a feasible implementation, the feedback optimization module includes: a parameter collection unit, a supervised learning unit, and an A / B testing unit;

[0027] The parameter collection unit is used to capture the skill label mismatch and overload labeling information during manual adjustment;

[0028] The supervised learning unit is used to update the rule weights using a logistic regression algorithm;

[0029] The A / B testing unit is used to trigger an administrator alarm when the new policy allocation accuracy is lower than 85%.

[0030] In a feasible implementation, the feedback optimization module periodically triggers full model retraining with a cycle of one week.

[0031] Another aspect of the present application provides a method for managing project information import and defect allocation based on intelligent matching, which is applied to any of the aforementioned management systems for project information import and defect allocation based on intelligent matching, and includes the following steps:

[0032] Parsing multi-format project data through the one-key import module and outputting verified structured project data;

[0033] Extract the defect type, severity level and module path from the structured project data through the defect intelligent analysis module and generate a JSON analysis report;

[0034] Calculate the priority score of the responsible person through the associated personnel matching engine and output the allocation suggestion;

[0035] Pushing allocation suggestions to external systems through the automatic allocation and notification module and capturing manual adjustment data;

[0036] The feedback optimization module optimizes the matching rule weights based on the manual adjustment data.

[0037] In a feasible implementation, the step of calculating the priority score of the responsible person includes:

[0038] Matching the defect type, severity level, and module path with the personnel skill tag;

[0039] The current load factor is dynamically weighted by the number of unclosed-loop tasks;

[0040] The responsible persons are sorted in descending order by priority score, and the priority score is calculated as follows:

[0041] Priority score = (defect urgency × professional matching) / (current load + 1)).

[0042] In a feasible implementation, the step of optimizing the matching rule weights includes:

[0043] The mapping relationship between the original responsible person A and the new responsible person B in the manual adjustment record is converted into a negative sample;

[0044] Use the random forest algorithm to improve the association weight between specific skill labels and defect types;

[0045] Validate the allocation accuracy threshold of the new strategy in A / B testing.

[0046] From the above content, it can be seen that the present application provides a project information import and defect allocation management system and method based on intelligent matching. By integrating a one-click import module, a defect intelligent analysis module, an associated personnel matching engine, an automatic allocation and notification module, and a feedback optimization module, it realizes the automated management of project information and the intelligent allocation of defects, reduces the cost of manual intervention, and significantly improves project management efficiency. The system can automatically import multi-format project data, intelligently analyze defect types, severity levels, and module paths, dynamically match responsible persons, and push allocation suggestions. At the same time, it optimizes the matching rule weights based on manually adjusted data, continuously improves allocation accuracy and processing efficiency, ensures that defects are assigned to the most appropriate responsible persons, improves processing quality, realizes closed-loop management of defect processing, and facilitates subsequent tracking and optimization. It also continuously collects negative sample information and optimizes matching rule weights to achieve continuous optimization and upgrading of the system to adapt to project changes and demand upgrades. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are incorporated into and constitute a part of this specification, illustrate embodiments consistent with the implementation of the present invention, and together with the description, serve to explain the principles of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the implementation of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0048] Figure 1 This is a structural diagram of a project information import and defect allocation management system based on intelligent matching shown in an embodiment of the present application;

[0049] Figure 2 It is a flowchart of the project information import and defect allocation management method based on intelligent matching shown in an embodiment of the present application. DETAILED DESCRIPTION

[0050] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the present invention will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to provide a thorough understanding of the implementation of the example embodiments of the present invention.

[0051] In traditional project management, defect allocation often relies on manual judgment, resulting in inefficiencies, inaccurate allocations, and delayed responses. As projects grow in scale and complexity, traditional methods are no longer able to meet the demands of rapid iteration and high-quality delivery. Therefore, it is crucial to develop a system that automatically imports project information, intelligently analyzes defects, and accurately assigns responsible individuals.

[0052] To solve the above problems, refer to Figure 1 As shown, on the one hand, this application provides a project information import and defect allocation management system based on intelligent matching, which realizes the full-process intelligent management of project defects by integrating a one-click import module, a defect intelligent analysis module, an associated personnel matching engine, an automatic allocation and notification module, and a feedback optimization module.

[0053] Specifically, the one-click import module is responsible for receiving multi-format project data source input, parsing and cleaning the data, and outputting structured project data; the defect intelligent analysis module is used to perform natural language processing on structured project data, extract defect types, severity levels and module paths, and generate analysis reports; the associated personnel matching engine is used to calculate the priority score of the responsible person based on the analysis report and output allocation suggestions; the automatic allocation and notification module is used to push allocation suggestions to the external task system and capture manual adjustment operations as negative samples; the feedback optimization module is used to optimize the rule weights of the associated personnel matching engine based on negative sample supervised learning.

[0054] The one-click import module supports project data source input in various formats, including Excel, CSV, and JIRA. Users can import data by dragging and dropping, connecting to the API, or manually entering the path. The module also parses the imported data, addressing issues like merged cells and separator recognition, and performs data cleansing, such as deduplicating titles and validating required fields, to ensure data quality. Finally, it outputs structured project data, including defect titles or descriptions, to provide a foundation for subsequent analysis.

[0055] The Intelligent Defect Analysis module uses NLP technology to conduct in-depth analysis of structured project data, extracting key information such as defect type, severity, and module path. It then dynamically matches this extracted information against a predefined rule base and generates an analysis report in JSON format. This report includes detailed defect information and matching results, providing a basis for subsequent assignment.

[0056] Based on the analysis report, the associated personnel matching engine dynamically calculates the priority score of the responsible party through the role responsibility rule library, historical processing record analyzer, and load statistics unit. It outputs the recommended responsible party assignment in descending order of score to ensure that defects are assigned to the most appropriate responsible party.

[0057] The automatic assignment and notification module pushes assignment suggestions to external task systems, such as JIRA and Trello, automating task creation. It also pushes assignment results to instant messaging tools like DingTalk and WeChat Work via a webhook interface, ensuring timely notification to relevant personnel. The task panel status is updated in real time, and any manual adjustments made by responsible personnel to assignment results are captured as negative examples for subsequent optimization.

[0058] The feedback optimization module captures negative examples from manual adjustments, such as mismatched skill labels and overload annotations. A logistic regression algorithm is used to learn from these negative examples, updating the rule weights of the associated person matching engine to improve matching accuracy. Finally, A / B testing verifies the effectiveness of the new strategy to ensure continued improvement in allocation accuracy.

[0059] The embodiments of this application significantly improve project management efficiency by reducing manual intervention through automated import, analysis, and allocation processes. Intelligent analysis and dynamic rule matching ensure that defects are assigned to the most appropriate individuals, improving processing quality. Real-time updates of task status and capture of manual adjustments enable closed-loop management of defect handling, facilitating subsequent tracking and optimization. The feedback optimization module continuously collects negative sample information and optimizes matching rule weights, enabling continuous optimization and upgrading of the system.

[0060] In some embodiments of the present application, the one-click import module includes a data source input unit, a multi-format parsing unit, and a data cleaning and verification unit.

[0061] The data source input unit allows users to drag and drop uploads. For example, users can drag and drop files into the designated area within the interface, and the system automatically identifies the file format and imports it. Multiple formats are supported, including Excel, CSV, and JIRA. The system also provides a RESTful API interface, enabling data integration with third-party systems (such as JIRA and GitLab) for automated import. Users can also manually enter a file path or URL, and the system will download and import the file based on the path information.

[0062] The multi-format parsing unit has three modes: Excel format parsing, CSV format parsing, and JIRA format parsing. Among them, Excel format parsing can automatically merge cell recognition and processing to ensure data integrity, and supports .xls and .xlsx formats, and handles special character escape; CSV format parsing can automatically identify delimiters (such as commas, semicolons, tabs, etc.) to avoid data dislocation, support different encoding formats (such as UTF-8, GBK, etc.), and handle quoted fields; JIRA format parsing can directly parse JSON format data exported by JIRA, extract key fields such as project name, defect title, description, etc., and can also handle nested structures (such as subtasks, attachments, etc.) for flattening.

[0063] The data cleaning and verification unit uses the Levenshtein distance algorithm to calculate title similarity, removes duplicates from titles with similarity above a threshold, and retains the latest version of the data to ensure data currency. It also presets required fields (such as defect title, description, and severity level), prompts for missing fields, or automatically fills in default values. It also presets compliant values ​​for enumerated fields (such as severity level and defect type), prompts for non-compliant values, or automatically corrects them.

[0064] The one-click import module in this embodiment improves data import efficiency through a data source input unit, a multi-format parsing unit, and a data cleaning and verification unit. It supports multiple data source input methods to meet data import requirements in different scenarios. Through multi-format parsing and data cleaning and verification, the integrity and accuracy of the imported data are ensured. Common problems in the data import process (such as merging cells, separator recognition, etc.) are automatically handled, reducing the cost of manual intervention.

[0065] In some embodiments of the present application, the defect intelligent analysis module includes: a key attribute extraction unit and a rule base dynamic matching unit.

[0066] Specifically, the key attribute extraction unit has the functions of named entity recognition, sentiment analysis and code structure tree parsing. Among them, named entity recognition refers to predefined defect types (such as functional defects, performance defects, security defects, etc.), using NLP technology to perform named entity recognition and match predefined defect types; the sentiment analysis function is to combine sentiment analysis technology to analyze the sentiment tendency of the defect description and determine the severity level of the defect (such as serious, general, minor, etc.); the code structure tree parsing function is based on the code structure tree parsing technology to extract the module path where the defect is located, providing a basis for subsequent allocation.

[0067] The rule base dynamic matching unit maps business rules described in natural language to standardized defect types. For example, "All functional defects must be assigned to the functional testing team" can be mapped to "Functional Defect → Functional Testing Team." A rule base management interface is provided, supporting the addition, deletion, modification, and query of rules, allowing for dynamic adjustment of matching rules based on project characteristics.

[0068] The intelligent defect analysis module in this embodiment uses named entity recognition and sentiment analysis to accurately identify defect types and severity levels, improving analysis accuracy. It also uses code structure tree parsing to accurately identify the module path where the defect is located, ensuring accurate module paths. The dynamic adjustment of the rule base adapts to the business rule requirements of different projects, further enhancing the flexibility of rule matching.

[0069] In some embodiments of the present application, the associated personnel matching engine includes a dynamic rule matching unit, and the dynamic rule matching unit includes a role responsibility rule library, a historical processing record analyzer, and a load statistics analyzer.

[0070] The role responsibility rule base is used to match defect module paths with personnel skill tags to ensure professional matching. For example, a "user interface defect" can be assigned to a person with the "front-end development" skill tag.

[0071] The historical processing record analyzer calculates the average time and success rate for personnel resolving similar defects, assessing processing efficiency and quality. For example, Person A takes an average of 2 hours to resolve a "user interface defect" with a success rate of 95%.

[0072] The load counter is used to count the number of unclosed tasks currently assigned to a person and dynamically adjust the weight factor to avoid over-allocation. For example, if person B currently has five unclosed tasks and a high load, their allocation priority can be appropriately lowered.

[0073] This embodiment uses a role responsibility rule library and a historical processing record analyzer to ensure that defects are assigned to the most appropriate individuals. A load counter dynamically adjusts weight factors to avoid over-assignment and ensure a balanced team load. Priority scores are also calculated to prioritize urgent and important defects, improving overall processing efficiency.

[0074] In some embodiments of the present application, the automatic allocation and notification module includes an allocation push unit, a notification generation unit, and a status update unit.

[0075] The allocation push unit is used to connect to external task systems. It provides a RESTful API interface, supporting integration with external task systems such as JIRA and Trello, enabling automated task creation and synchronization. It also pushes allocation results to instant messaging tools like DingTalk and WeChat for Business via a Webhook interface. Customizable push templates are supported, including key information such as the defect title, description, and severity level.

[0076] The notification generation unit generates email templates with links to defect details, allowing responsible individuals to quickly view defect details and address them. Email content can be customized to include additional information such as action suggestions and deadlines. Batch emails can be sent to improve notification efficiency. Email delivery status can be tracked to ensure timely delivery of notifications.

[0077] The status update unit reflects the progress of defect handling by updating the task panel in real time. Task status (such as pending, in progress, closed, etc.) is displayed visually. Manual adjustments to the assignment results by the responsible person, such as reallocation and deadline modification, are captured as negative samples for subsequent optimization.

[0078] In this implementation, instant messaging push notifications and email notifications ensure that the responsible party is informed of the assignment results promptly. Email templates with links to defect details facilitate quick review and resolution of defect details. Real-time updates of task status and recording of manual adjustments further enable closed-loop management of defect handling.

[0079] In some embodiments of the present application, the feedback optimization module includes detailed operations of a parameter collection unit, a supervised learning unit, and an A / B testing unit.

[0080] The parameter collection unit captures negative examples such as mismatched skill labels or excessive workloads in assigned results. For example, a responsible person might reassign a "User Interface Defect" task to someone with the "UI Design" skill label. This captured negative example information is stored in a database for data cleaning and preprocessing to ensure data quality.

[0081] The supervised learning unit extracts features from negative sample information (such as skill tag matching and load conditions), trains a model using a logistic regression algorithm, and updates the rule weights of the associated personnel matching engine. Based on the model training results, it adjusts the weighting factors of the role responsibility rule library, historical processing record analyzer, and load statistics to improve matching accuracy.

[0082] The A / B testing unit deploys the optimized matching rule weights to the test environment and calculates the allocation accuracy under the new policy. A threshold for allocation accuracy (e.g., 85%) is set. When the allocation accuracy of the new policy falls below the threshold, an administrator is alerted and the policy can be rolled back or optimized further.

[0083] This embodiment optimizes matching rule weights through supervised learning, enabling the system to more accurately match defects with responsible individuals. A / B testing verifies the effectiveness of the new strategy, ensuring continued improvement in allocation accuracy. Furthermore, by continuously collecting negative sample information and optimizing matching rule weights, the system can be continuously optimized and upgraded.

[0084] In some embodiments of the present application, the feedback optimization module periodically triggers full model retraining on a weekly basis.

[0085] On the other hand, this application provides a method for intelligent matching project information import and defect allocation management, referring to Figure 2 As shown, the steps include:

[0086] S100: Parse multi-format project data through a one-click import module and output verified structured project data.

[0087] Specifically, users import project data by dragging and dropping, API docking, or manual path input. The one-click import module parses the imported data, handling issues such as Excel cell merging, CSV delimiter recognition, and JIRA JSON parsing. The parsed data is then cleaned and verified, such as deduplication of titles and non-empty verification of required fields, to ensure data quality. Finally, structured project data containing defect titles or description text is output.

[0088] S200: The defect intelligent analysis module extracts the defect type, severity level, and module path from the structured project data and generates a JSON analysis report.

[0089] In this step, the Intelligent Defect Analysis module extracts the defect type, severity, and module path from the structured project data. It then uses NLP technology to perform named entity recognition on the defect description, matching it to predefined defect types. It then combines sentiment analysis to determine the severity of the defect. The Intelligent Defect Analysis module uses code structure parsing technology to extract the module path where the defect is located and generates a JSON-formatted analysis report containing detailed defect information and matching results.

[0090] S300: Calculate the priority score of the responsible person through the associated personnel matching engine and output the allocation suggestion.

[0091] This step uses the associated personnel matching engine to match defect module paths with personnel skill tags, and calculates the average time and success rate for personnel to resolve similar defects. It also counts the number of unclosed tasks currently performed by personnel, dynamically adjusts the weight factor, and outputs responsible person assignment suggestions in descending order of priority scores to ensure that defects are assigned to the most appropriate responsible person.

[0092] S400: Push allocation suggestions to external systems through the automatic allocation and notification module and capture manual adjustment data.

[0093] In this step, the automatic assignment and notification module pushes the assignment recommendations to external task systems, such as JIRA and Trello. The assignment results are sent to the responsible party via instant messaging. An email template with a link to the defect details is generated and sent to the responsible party. The automatic assignment and notification module updates the task panel status in real time and captures any manual adjustments made by the responsible party to the assignment results, which serve as negative examples for subsequent optimization.

[0094] S500: Optimizing the matching rule weights according to the manual adjustment data through the feedback optimization module.

[0095] This step uses the feedback optimization module to capture negative examples of staff members' responses to assignment results, such as mismatched skill labels and excessive workloads. A logistic regression algorithm is then used to learn from these negative examples and update the rule weights of the associated personnel matching engine. A / B testing is performed to verify the effectiveness of the new strategy and ensure continued improvement in assignment accuracy.

[0096] This application provides an intelligent matching project information import and defect allocation management method that automates the import, analysis, and allocation processes, reducing manual intervention costs and significantly improving project management efficiency. Through intelligent analysis and dynamic rule matching, defects are assigned to the most appropriate individuals, improving handling quality and achieving closed-loop defect handling management, facilitating subsequent tracking and optimization, while also enabling continuous system optimization and upgrades.

[0097] In some embodiments of the present application, the step of calculating the responsible person priority score includes:

[0098] S310: Match the defect type, severity level, module path, and personnel skill tags.

[0099] For example, matching a "user interface defect" with a person with the "front-end development" skill tag. Algorithms such as cosine similarity or Jaccard similarity can also be used to calculate professional matching. A higher matching score indicates a more suitable person for handling the defect. Professional matching scores are normalized to ensure they fall between 0 and 1, facilitating subsequent calculations and comparisons.

[0100] S320: The current load factor is dynamically weighted by the number of unclosed-loop tasks.

[0101] Specifically, the number of unclosed tasks currently performed by the personnel is counted. The greater the number of tasks, the higher the personnel load. The weight factor is dynamically adjusted based on the number of unclosed tasks. For example, using a linear weighting method, the load factor is matched based on the defective module path and the personnel skill label = the number of unclosed tasks matched based on the defective module path and the personnel skill label / the maximum load threshold matched based on the defective module path and the personnel skill label. The current load factor is normalized to ensure that its value range is between 0 and 1, facilitating subsequent calculations and comparisons.

[0102] S330: Sort the responsible persons in descending order according to their priority scores.

[0103] The priority score is calculated based on the formula: priority score = (defect urgency × professional matching) / (current load + 1).

[0104] The professional matching degree in the formula refers to the professional matching degree calculated based on the similarity between the defect module path and the personnel skill label; the current load factor refers to the current load factor calculated by dynamic weighting through the number of unclosed-loop tasks; the defect urgency is mapped according to the severity level, such as the urgency of serious defects is 3, general defects is 2, and minor defects is 1.

[0105] In this step, the priority scores can be normalized to ensure that they are between 0 and 1, which facilitates subsequent sorting and comparison. The responsible parties are sorted in descending order by priority score to ensure that the responsible parties with the highest scores are assigned defects first.

[0106] This implementation ensures that defects are assigned to the most appropriate individuals by comprehensively considering factors such as professional compatibility, current load factor, and defect urgency. Dynamically weighting the current load factor prevents over-assignment to overburdened individuals, ensuring a balanced team load. Furthermore, priority scores are calculated to prioritize urgent and important defects, improving overall processing efficiency.

[0107] In some embodiments of the present application, the step of optimizing the matching rule weights includes:

[0108] S510: Convert the mapping relationship between the original responsible person A and the new responsible person B in the manual adjustment record into a negative sample;

[0109] For example, the original responsible person A reassigns "user interface defects" to the new responsible person B. This mapping relationship is marked as a negative sample, indicating that the original assignment result is inaccurate or unreasonable. Furthermore, the negative sample data is cleaned and preprocessed, such as removing duplicate samples and handling missing values, to ensure data quality.

[0110] S520: Use the random forest algorithm to improve the association weight between specific skill labels and defect types;

[0111] Specifically, features (such as skill label matching, load conditions, defect urgency, etc.) are extracted from negative sample information. For example, features such as the skill label matching and load conditions of the original responsible person A and the new responsible person B are extracted. The model is then trained using the random forest algorithm to analyze the relationship between features and allocation results. The model learns patterns in negative samples to optimize the matching rule weights. The rule weights of the associated personnel matching engine are updated based on the model training results. For example, increasing the association weight between specific skill labels and defect types makes the system more inclined to assign specific types of defects to personnel with corresponding skill labels.

[0112] S530: Verify the allocation accuracy threshold of the new strategy in an A / B test.

[0113] Deploy the optimized matching rule weights to a test environment and compare them with the old policy. Calculate the allocation accuracy of the new policy in the test environment and compare it with the old policy. Compare the allocation accuracy of the new policy to a preset threshold (e.g., 85%). If it falls below the threshold, an administrator is alerted and the policy is considered for rollback. If it exceeds the threshold, the policy is gradually rolled out to the production environment.

[0114] This embodiment optimizes matching rule weights through supervised learning, enabling the system to more accurately match defects with responsible individuals and improve assignment accuracy. A / B testing verifies the effectiveness of new strategies, ensuring continued improvement in assignment accuracy and avoiding ineffective or negative optimization. This embodiment continuously collects negative sample information and optimizes matching rule weights, enabling continuous system optimization and upgrades to adapt to project changes and evolving requirements.

[0115] Based on the above embodiments, the present application provides a project information import and defect allocation management system and method based on intelligent matching. The overall usage process is as follows:

[0116] Users import project data into the system by dragging and dropping, API docking, or manual path input; the system parses, cleans, and verifies the imported data, extracts defect types, severity levels, and module paths, and generates an analysis report in JSON format; the system calculates the priority score of the responsible person based on the analysis report and outputs allocation suggestions in descending order of the score; the system simultaneously pushes the allocation suggestions to the external task system and pushes the allocation results to the responsible person via instant messaging tools; the system updates the task panel status in real time and captures the responsible person's manual adjustments to the allocation results as negative samples; the system optimizes the matching rule weights based on the negative sample information and verifies the effectiveness of the new strategy through A / B testing.

[0117] In summary, the present application provides a project information import and defect allocation management system and method based on intelligent matching. By integrating a one-click import module, a defect intelligent analysis module, an associated personnel matching engine, an automatic allocation and notification module, and a feedback optimization module, it realizes the automated management of project information and the intelligent allocation of defects, reduces the cost of manual intervention, and significantly improves project management efficiency. The system can automatically import multi-format project data, intelligently analyze defect types, severity levels, and module paths, dynamically match responsible persons, and push allocation suggestions. At the same time, it optimizes the matching rule weights based on manually adjusted data, continuously improves allocation accuracy and processing efficiency, ensures that defects are assigned to the most appropriate responsible persons, improves processing quality, realizes closed-loop management of defect processing, and facilitates subsequent tracking and optimization. It also continuously collects negative sample information and optimizes matching rule weights to achieve continuous optimization and upgrading of the system to adapt to project changes and demand upgrades.

[0118] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.

Claims

1. A system for project information import and defect allocation management based on intelligent matching, characterized in that: include: One-click import module for communication connections, intelligent defect analysis module, associated personnel matching engine, automatic allocation and notification module, and feedback optimization module; The one-key import module is configured to: receive a multi-format project data source input, generate semi-structured data through the multi-format parsing unit, and output structured project data containing defect titles or description text through the data cleaning and verification unit; The intelligent defect analysis module is configured to: in response to the input of the structured project data, process the structured project data through natural language processing, extract the defect type, severity level and module path, and generate a JSON format analysis report in combination with a rule base for dynamic matching; The associated personnel matching engine is configured to: calculate personnel priority scores based on the analysis report through a dynamic rule matching unit, and output responsible person assignment suggestions in descending order of scores; The automatic allocation and notification module is configured to: synchronously push the allocation suggestion to the external task system and capture the manual adjustment operation as a negative sample; The feedback optimization module is configured to optimize the rule weights of the associated person matching engine according to the negative sample supervised learning.

2. The intelligent matching project information import and defect allocation management system according to claim 1 is characterized in that: The one-key import module includes: a data source input unit, a multi-format parsing unit and a data cleaning unit; The data source input unit is configured to: support drag-and-drop upload, API docking, and manual path input of the multi-format project data source; The multi-format parsing unit is configured to: parse merged cells in the Excel format in the multi-format project data source, automatically identify delimiters in the CSV format, and parse JSON key fields in the JIRA format; The data cleaning unit is configured to perform title deduplication based on Levenshtein distance, mandatory field non-empty verification, and enumeration value compliance verification on the multi-format project data source.

3. The intelligent matching project information import and defect allocation management system according to claim 1 is characterized in that: The defect intelligent analysis module includes: a key attribute extraction unit and a rule base dynamic matching unit; The key attribute extraction unit is configured to: match predefined defect types through named entity recognition, determine severity levels in combination with sentiment analysis, and parse module paths based on a code structure tree; The rule base dynamic matching unit is configured to map the business rules described in natural language into standardized defect types.

4. The intelligent matching project information import and defect allocation management system according to claim 1 is characterized in that: The dynamic rule matching unit of the associated personnel matching engine includes: a role responsibility rule library, a historical processing record analyzer and a load statistics recorder; The role responsibility rule base is used to match defect modules with personnel skill tags; The historical processing record analyzer is used to calculate the average time and success rate of personnel in resolving similar defects; The load statistics device is used to dynamically adjust the weight factor based on the number of unclosed-loop tasks.

5. The system for importing project information and allocating defects based on intelligent matching according to claim 1 is characterized in that: The automatic allocation and notification module is further configured to: Push the allocation results to DingTalk / WeChat Enterprise via the Webhook interface; Generate an email template with a link to the defect details; The task panel status is updated in real time and the manual adjustment operation is recorded.

6. The system for importing project information and allocating defects based on intelligent matching according to claim 1 is characterized in that: The feedback optimization module includes: a parameter collection unit, a supervised learning unit and an A / B testing unit; The parameter collection unit is used to capture the skill label mismatch and overload labeling information during manual adjustment; The supervised learning unit is used to update the rule weights using a logistic regression algorithm; The A / B testing unit is used to trigger an administrator alarm when the new policy allocation accuracy is lower than 85%.

7. The intelligent matching project information import and defect allocation management system according to claim 6 is characterized in that: The feedback optimization module periodically triggers full model retraining on a weekly basis.

8. A method for managing project information import and defect allocation based on intelligent matching, applied to the management system for project information import and defect allocation based on intelligent matching according to any one of claims 1 to 7, characterized in that the steps include: Parsing multi-format project data through the one-key import module and outputting verified structured project data; Extract the defect type, severity level and module path from the structured project data through the defect intelligent analysis module and generate a JSON analysis report; Calculate the priority score of the responsible person through the associated personnel matching engine and output the allocation suggestion; Pushing allocation suggestions to external systems through the automatic allocation and notification module and capturing manual adjustment data; The feedback optimization module optimizes the matching rule weights based on the manual adjustment data.

9. The method for project information import and defect allocation management based on intelligent matching according to claim 8 is characterized in that: The steps of calculating the priority score of the responsible person include: Matching the defect type, severity level, and module path with the personnel skill tag; The current load factor is dynamically weighted by the number of unclosed-loop tasks; The responsible persons are sorted in descending order by priority score, and the priority score is calculated as follows: Priority score = (defect urgency × professional matching) / (current load + 1)).

10. The method for project information import and defect allocation management based on intelligent matching according to claim 8, characterized in that: The step of optimizing the matching rule weights includes: The mapping relationship between the original responsible person A and the new responsible person B in the manual adjustment record is converted into a negative sample; Use the random forest algorithm to improve the association weight between specific skill labels and defect types; Validate the allocation accuracy threshold of the new strategy in A / B testing.

Images