Project full-process data tracking processing method and system

Abstract

The invention relates to the technical field of artificial intelligence, and particularly provides a project whole-process data tracking processing method and system, and the method comprises the steps: obtaining project planning data, extracting the relation between task information and tasks from the project planning data through a pre-training language model, and constructing a knowledge graph based on the relation between the task information and the tasks; acquiring project execution data from a plurality of data sources, and identifying task execution information from the project execution data; predicting a task progress based on a data sequence formed by all task execution information of the task, and adding state attributes for corresponding nodes of the knowledge graph based on the task progress; and analyzing an abnormal condition of a task execution state based on the knowledge graph after the state attribute is updated, analyzing an abnormal factor based on the project execution data, and generating a processing scheme for the abnormal factor by utilizing a pre-constructed knowledge base. According to the invention, the refined and intelligent level of project management is improved.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and specifically relates to a method and system for tracking and processing data throughout the entire project process. Background Art

[0002] With the increasing complexity of project management, especially in fields such as large-scale engineering, software development, scientific research, and government public projects, the need for real-time tracking of project progress and accurate problem identification has become increasingly urgent. Traditional project management tools rely primarily on manual data entry and static report analysis, but this model is insufficient when faced with the frequently changing requirements and diverse data inputs of complex projects. They struggle to effectively capture and respond to subtle changes in project dynamics, and even more so, struggle to discern potential risk factors or signs of deviation early in a project.

[0003] In particular, traditional approaches to timeline management and problem summarization reveal inadequate semantic understanding of massive amounts of project documentation and a lack of in-depth correlation analysis capabilities. This not only makes it difficult to promptly identify and correct project deviations, but also invisibly slows down overall project progress and reduces final quality. Therefore, exploring more efficient and intelligent project management methods has become a critical issue that needs to be addressed in the current project management field. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method and system for tracking and processing data throughout the entire project process to solve the above-mentioned technical problems.

[0005] In a first aspect, the present invention provides a method for tracking and processing data throughout a project, comprising: Obtaining project planning data, extracting task information and relationships between tasks from the project planning data using a pre-trained language model, wherein the task information includes planning time, and constructing a knowledge graph based on the task information and relationships between tasks; Acquire project execution data from a variety of data sources, and identify task execution information from the project execution data; Predicting task progress based on a data sequence consisting of all task execution information of the task, matching the task progress with the corresponding planned time, and adding state attributes to the corresponding nodes of the knowledge graph based on the matching results; Analyze the abnormal situation of task execution status based on the knowledge graph after updating status attributes, analyze abnormal factors based on project execution data, and use the pre-built knowledge base to generate processing solutions for the abnormal factors.

[0006] In an optional embodiment, project planning data is obtained, and task information and relationships between tasks are extracted from the project planning data using a pre-trained language model, wherein the task information includes planning time, and a knowledge graph is constructed based on the task information and relationships between tasks, including: Collect project planning data from project management platforms, document storage and email systems, and project logs; Preprocessing the project planning data, wherein the preprocessing includes cleaning the data and formatting the data; Inputting the pre-processed project planning data into a pre-trained language model to obtain task information and task relationships, wherein the task information includes resource information and planning time, and the task relationships include dependency relationships, resource allocation relationships, and time sequence relationships between tasks; Use tasks, resources, and time as nodes, and task relationships as edges to build a knowledge graph using a graph database. Generating importance labels for tasks in the knowledge graph; Set up a query interface for the knowledge graph.

[0007] In an optional embodiment, obtaining project execution data from multiple data sources and identifying task execution information from the project execution data include: Identify file names using keyword technology and collect project execution data from project management platforms, document storage and email systems, and project logs; A pre-trained language model is used to extract task execution information from the project execution data, wherein the task execution information includes executed information and approval information, wherein the executed information includes executed task information and execution time, and the approval information includes pending task information and execution time.

[0008] In an optional embodiment, the method further comprises: Based on the task information to be executed, query the corresponding target task information and the status of the predecessor tasks that the target task depends on from the knowledge graph; Extracting person-in-charge information and planning time from the target task information; Match the approval information with the person in charge information and the planned time in the target task information. If they do not completely match, an abnormal prompt message is generated; if they completely match, a prompt message indicating that the approval information is normal is generated; Confirm that the approval information is normal. If the status of the predecessor tasks that the target task depends on are all completed, then the approval information is determined to meet the approval conditions and the approval information is sent to the review terminal; if there is an unfinished status in the predecessor tasks that the target task depends on, then the approval information is determined to not meet the approval conditions and a prompt message of approval failure is generated.

[0009] In an optional embodiment, task progress is predicted based on a data sequence consisting of all task execution information of the task, the task progress is matched with the corresponding planned time, and a state attribute is added to the corresponding node of the knowledge graph based on the matching result, including: Save the executed information of the same task into the same list; Decompose the task into multiple execution stages and allocate execution time to the multiple execution stages based on the planned time of the task; Set the corresponding progress value for the execution phase based on the execution time and execution order of the execution phase; Obtain the completed execution stage corresponding to the executed task information, and based on the correspondence between the completed execution stage and the execution stage and the progress value, retrieve the progress value corresponding to the executed information; Converting the executed information in the list into a data set consisting of a progress value and an execution time; Arranging the data groups into a time data sequence according to execution time; Inputting the time data sequence and the current system time into a pre-trained long short-term memory neural network model to predict the completion time of the task; Calculating a difference between the completion time and the completion time defined by the planned time of the task, and generating a status attribute based on the difference, the status attribute including ahead of schedule completion, on track progress, and overdue completion; Add the state attribute to the corresponding task node in the knowledge graph.

[0010] In an optional embodiment, abnormalities in task execution status are analyzed based on the knowledge graph after updating status attributes, abnormal factors are analyzed based on project execution data, and processing solutions are generated for the abnormal factors using a pre-built knowledge base, including: Setting anomaly detection indicators, including the proportion of overdue tasks, task progress deviation, resource utilization, and task dependency anomalies; Traverse the task nodes in the knowledge graph and obtain the status attributes of each task node; Calculate the abnormal index value of each task; Mark tasks where the abnormal indicator value exceeds the threshold as abnormal tasks and mark them specially in the knowledge graph; By analyzing the task relationships, resource allocation, execution time, and task execution information of abnormal tasks, and using thought chain reasoning technology to locate the abnormal causes of task abnormalities based on the analysis results; Utilize the pre-built knowledge base to generate processing solutions based on the abnormal causes of abnormal tasks.

[0011] In an optional embodiment, a pre-built knowledge base is used to generate a processing solution based on the abnormal cause of the abnormal task, including: Inputting the abnormal cause of the abnormal task into the knowledge base to obtain multiple historical processing solutions; Use thought chain reasoning technology to conduct causal analysis on multiple historical treatment plans, and deduce multiple implementation steps and effects from the problem-solution-result chain; Based on the results of the implementation steps, generating multiple implementation plans with the best results, wherein the implementation plans include multiple implementation steps; Conduct a feasibility assessment on the implementation plan and output the implementation plan that passes the feasibility assessment as a treatment plan.

[0012] In an optional embodiment, the method further comprises: Collect corresponding processing solutions for historical abnormal tasks of multiple projects and save historical data of multiple projects as corresponding multiple data sets; The knowledge base is trained using the federated learning mechanism.

[0013] In a second aspect, the present invention provides a project full-process data tracking and processing system, comprising: A graph construction module is used to obtain project planning data, extract task information and relationships between tasks from the project planning data using a pre-trained language model, wherein the task information includes planning time, and construct a knowledge graph based on the task information and relationships between tasks; An execution monitoring module, configured to obtain project execution data from various data sources and identify task execution information from the project execution data; A state setting module is used to predict the task progress based on the data sequence composed of all task execution information of the task, match the task progress with the corresponding planned time, and add state attributes to the corresponding node of the knowledge graph based on the matching result; The exception handling module is used to analyze the abnormal situation of the task execution status based on the knowledge graph after updating the status attributes, analyze the abnormal factors based on the project execution data, and use the pre-built knowledge base to generate a processing plan for the abnormal factors.

[0014] The beneficial effects of the present invention lie in the fact that the method and system for tracking and processing project data throughout the entire process provided by the present invention achieve intelligent processing of project planning and execution data by integrating pre-trained language models, knowledge graphs, progress prediction, and anomaly analysis technologies. The system extracts tasks and relationships from multi-source data and constructs a knowledge graph to enhance planning visualization and decision support; it monitors execution status in real time, predicts progress and matches plans, and dynamically updates the status of graph nodes; it automatically detects anomalies and generates processing solutions, optimizes the approval process, and improves response efficiency. This solution significantly enhances the refinement, intelligence, and overall effectiveness of project management, and has achieved good economic and social benefits in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0016] Figure 1 is a schematic flow chart of a method according to an embodiment of the present invention.

[0017] Figure 2 It is a schematic flowchart of a method for constructing a knowledge graph based on project planning data according to an embodiment of the present invention.

[0018] Figure 3 The figure is a schematic flow chart of a method for extracting task execution information according to an embodiment of the present invention.

[0019] Figure 4 The figure is a schematic flow chart of a method for setting status attributes for a task node according to an embodiment of the present invention.

[0020] Figure 5 The figure is a schematic flow chart of abnormal task screening and processing according to an embodiment of the present invention.

[0021] Figure 6 FIG. 4 is a schematic block diagram of a system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0024] The key terms appearing in the present invention are explained below.

[0025] Pre-trained language models (PLMs) are an important tool in the field of natural language processing (NLP). By pre-training on large-scale text data, they learn rich language knowledge and patterns, which can then be applied to various downstream tasks such as text classification, named entity recognition, sentiment analysis, question-answering systems, and machine translation.

[0026] Key Features: Large-scale data training: Pre-trained language models are typically trained using hundreds of millions of text data sets covering a variety of fields and topics, enabling the model to learn rich language knowledge and contextual information.

[0027] Unsupervised learning: The pre-training process usually adopts unsupervised learning, that is, the model does not require manually annotated labels, but automatically learns the statistical laws and patterns of language from text data.

[0028] Transfer learning: Through pre-training, the knowledge learned by the model can be transferred to different downstream tasks. Downstream tasks only need to fine-tune the pre-trained model using a small amount of labeled data to achieve good performance.

[0029] Common models include: BERT (Bidirectional Encoder Representations from Transformers): BERT is a pre-trained language model based on the Transformer architecture that learns text representations through bidirectional encoding. BERT's pre-training tasks include Masked Language Model (MLM) and Next Sentence Prediction (NSP).

[0030] GPT (Generative Pre-trained Transformer): The GPT series of models are autoregressive pre-trained language models based on the Transformer architecture. They learn text representations generatively and can generate coherent text sequences. The pre-training task for GPT is language modeling, which involves predicting the next word based on the previous text.

[0031] T5 (Text-to-Text Transfer Transformer): T5 is a pre-trained language model that unifies various NLP tasks into a text-to-text format. It uses the same model architecture and training process to handle different tasks by concatenating the task description and input text as the model input.

[0032] RoBERTa (Robustly optimized BERT pretraining approach): RoBERTa is an improvement and optimization of BERT, which improves the performance of the model by increasing training data, number of training steps, batch size, and removing NSP tasks.

[0033] ERNIE (Enhanced Representation through kNowledge IntEgration): ERNIE is a pre-trained language model developed by Baidu. It emphasizes knowledge enhancement and enhances the model's representation capabilities by introducing semantic information such as entities and relations.

[0034] By using pre-trained language models to process project-related data, we can accurately identify task information in the project and sort out the relationships between tasks.

[0035] The LSTM (Long Short-Term Memory) model, also known as the Long Short-Term Memory (LSTM) model, is a specialized recurrent neural network (RNN) architecture designed to address the vanishing and exploding gradient problems that plague traditional RNNs when processing long sequences of data. Proposed by Hochreiter and Schmidhuber in 1997, the LSTM model's core features three unique "gate" structures: an input gate, a forget gate, and an output gate, along with a memory cell. These structures work together to enable the LSTM to maintain and update memories over extended timeframes, effectively capturing and preserving long-term dependencies.

[0036] Specifically, the input gate controls how much new information is stored in the cell state; the forget gate controls how much information in the current cell state is retained; and the output gate determines how much information is output from the cell state. This mechanism enables LSTM to flexibly decide which information should be retained, updated, or forgotten, thereby avoiding the vanishing gradient problem and being able to exclude irrelevant or noisy information while maintaining information over long time series.

[0037] Advantages of LSTM in predicting task progress: Long-term dependency capture: LSTM, through its unique memory unit and gating mechanism, can effectively capture and retain long-term dependencies. This is particularly important for predicting task progress, as task progress is often affected by multiple time steps in the past.

[0038] Information filtering capability: The gating mechanism allows LSTM to flexibly decide which information should be retained, updated, or forgotten. This capability enables LSTM to exclude irrelevant or noisy information when processing complex task progress data, thereby improving prediction accuracy.

[0039] Nonlinear modeling: LSTM networks are capable of learning nonlinear functions, which is crucial for most real-world task progress data. Because task progress often exhibits complex nonlinear patterns and trends, LSTM is able to capture these patterns and make accurate predictions.

[0040] End-to-end learning: LSTM can learn directly from raw data, eliminating the need for manual feature engineering. This simplifies the model building process and improves the automation and efficiency of prediction. For task progress prediction, this means learning progress patterns directly from historical task data without the need for tedious feature extraction and selection.

[0041] Flexibility and scalability: LSTM can be easily combined with other network structures (such as convolutional neural networks (CNNs) and attention mechanisms), or by stacking multiple layers to enhance the model's expressiveness. This flexibility and scalability enables LSTM to adapt to more complex and diverse task progress prediction tasks.

[0042] In summary, the LSTM model has significant advantages in predicting task progress. Its ability to capture long-term dependencies, information screening, nonlinear modeling, end-to-end learning, flexibility, and scalability all provide strong support for task progress prediction.

[0043] The project full-process data tracking and processing method provided by the embodiment of the present invention is executed by a computer device, and accordingly, the project full-process data tracking and processing system runs in the computer device.

[0044] Figure 1is a schematic flow chart of a method according to an embodiment of the present invention. Figure 1 The execution entity can be a project full process data tracking and processing system. According to different needs, the order of the steps in the flowchart can be changed, and some steps can be omitted.

[0045] like Figure 1 As shown, the method includes: Step 110. Obtain project planning data, use a pre-trained language model to extract task information and the relationship between tasks from the project planning data, the task information including planning time, and construct a knowledge graph based on the task information and the relationship between tasks.

[0046] Obtain project planning data: First, collect project planning data from project management tools, documents, or databases. This data should include an overall overview of the project, the task breakdown structure (WBS), task descriptions, planned timelines (such as start date, end date, milestones, etc.), and dependencies between tasks.

[0047] Use pre-trained language models to extract task information and relationships: With the help of pre-trained language models (such as BERT and GPT), natural language processing is performed on project planning data to extract key information such as task name, description, planning time, as well as dependencies such as the sequence and parallel relationships between tasks.

[0048] Build a knowledge graph: Based on the extracted task information and relationships between tasks, a knowledge graph is constructed. In the knowledge graph, each task is represented as a node, and relationships between tasks (such as dependencies) are represented as edges between nodes. Attributes, such as planning time, are also added to the task nodes.

[0049] Step 120: Obtain project execution data from multiple data sources, and identify task execution information from the project execution data.

[0050] Obtain project execution data from multiple data sources: Collect project execution data from multiple data sources such as project management tools, monitoring systems, and log records. This data should include task execution status, actual start and end times, resource usage, progress percentage, etc.

[0051] Identify task execution information: Process and analyze project execution data to identify the execution information of each task, including actual execution time, resource input, progress status, etc.

[0052] Step 130. Predict the task progress based on the data sequence composed of all task execution information of the task, match the task progress with the corresponding planned time, and add status attributes to the corresponding nodes of the knowledge graph based on the matching results.

[0053] Predicting task progress based on data sequences: For each task, we use the data sequence (such as time series data) consisting of all task execution information to adopt LSTM or other time series prediction models to predict the future progress of the task.

[0054] Matching planned time with predicted progress: Match the predicted task progress with the corresponding planned time to evaluate whether the task is progressing as planned.

[0055] Update the knowledge graph status attribute: Based on the matching results, the status attribute of the corresponding task node in the knowledge graph is updated. For example, if the task progress lags behind schedule, the status attribute is set to "lag"; if the task is completed ahead of schedule, it is set to "completed ahead of schedule".

[0056] Step 140. Analyze the abnormal situation of the task execution status based on the knowledge graph after updating the status attributes, analyze the abnormal factors based on the project execution data, and use the pre-built knowledge base to generate a processing plan for the abnormal factors.

[0057] Analyze task execution status anomalies based on the knowledge graph: Analyze task execution status anomalies based on the updated knowledge graph. For example, identify which tasks are behind schedule or lack resources.

[0058] Analyze abnormal factors based on project execution data: Further analyze project execution data to identify specific factors that lead to abnormal task execution status. This may include uneven resource allocation, skills shortages, external interference, etc.

[0059] Generate solutions using a knowledge base: Based on the anomaly factors, a pre-built knowledge base (including common problems and solutions) is used to generate solutions for the anomaly. These solutions may include adjusting resource allocation, providing training support, optimizing workflows, etc.

[0060] In one embodiment of the present invention, based on step 110, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner, such as Figure 2 shown.

[0061] Project planning data is collected from multiple channels, including project management platforms, document storage systems, email systems, and project logs. This data contains key information such as the overall project plan, task divisions for each phase, resource allocation, and timelines, forming the foundation for subsequent analysis and processing. Before entering the pre-trained language model, the collected project planning data undergoes a series of preprocessing operations. These operations include data cleansing to remove duplicate, erroneous, and irrelevant information, and data formatting to unify the data structure for easier processing. This preprocessed data significantly improves data quality, laying a solid foundation for subsequent information extraction. Next, the present invention uses the pre-trained language model to perform in-depth analysis of the project planning data. This model, with its powerful natural language understanding capabilities, accurately identifies and extracts task information (including resource information and planned time) as well as relationships between tasks (such as dependencies, resource allocation relationships, and time sequence relationships). This information is a key element in constructing a knowledge graph. After extracting task information and task relationships, the present invention utilizes graph database technology to construct a knowledge graph. In this graph, tasks, resources, and time are abstracted as nodes, while relationships between tasks are represented as edges. This graph-structured data representation makes the various tasks in a project and the relationships between them clear at a glance, providing strong support for subsequent task execution status analysis and exception handling. Furthermore, the present invention generates importance labels for task nodes in the knowledge graph to quickly identify key tasks. At the same time, by providing a query interface, users can easily retrieve and query information in the graph, meeting the diverse needs of the project management process.

[0062] In a specific example, the following specific processes are included: Step 1101. Obtain project planning data Data source: Export project planning data from project management software (such as Jira, Trello), which usually includes task names, descriptions, resource allocations, start and end dates, etc.

[0063] Alternatively, extract data from project documents (e.g., Word, Excel, PDF) that may contain detailed planning information for the project.

[0064] Data preprocessing: Clean the data and remove irrelevant information, such as redundant spaces and incorrect date formats.

[0065] Format data to ensure uniform data format for easy subsequent processing.

[0066] Step 1102. Use the pre-trained language model to extract task information and the relationship between tasks Choose a suitable pre-trained language model: Choose pre-trained language models such as BERT and GPT-3, which perform well in natural language processing tasks.

[0067] Task information extraction: Resource information: The model parses the task description and identifies the resources involved (such as personnel and equipment). For example, for the task description "Zhang San is responsible for designing the UI interface," the model can identify "Zhang San" as a resource.

[0068] Planning time: Extract the start and end dates from the task description or separate date fields. For example, for the task "2023-04-01 to 2023-04-10: Develop backend features", the model can extract the time range.

[0069] Task relationship extraction: Analyze the dependencies, sequencing, etc. in the task descriptions. For example, Task A, “Complete UI design,” may be a prerequisite for Task B, “Develop backend functionality.”

[0070] The model is used to semantically understand the task description and identify the logical relationship between tasks.

[0071] Step 1103. Build a knowledge graph based on task information and relationships between tasks Define the nodes and edges of the knowledge graph: Nodes: tasks, resources, time, etc.

[0072] Edges: dependencies between tasks, resource allocation relationships, time sequence relationships, etc.

[0073] Building a knowledge graph: Use graph databases (such as Neo4j) to build knowledge graphs.

[0074] Add the extracted task information (task name, resources, time) as nodes to the graph.

[0075] According to the extracted task relationships, corresponding edges are created in the graph.

[0076] Visualization and query of knowledge graph: Use visualization tools (such as Neo4j Browser and Graphileon) to display the knowledge graph so that you can intuitively view the relationships between tasks.

[0077] Provides a query interface that allows users to search and filter based on keywords, resources, time and other conditions in project planning data.

[0078] Step 1104. Set importance labels for task nodes Importance Label Definition: Importance labels are used to indicate the relative importance of a task within a project. This can be assessed based on multiple dimensions, such as the task's size, contribution to the project goal, resource commitment, and timeline. Labels can be qualitative (e.g., "high," "medium," or "low") or quantitative (e.g., a 1-5 rating).

[0079] Importance assessment method: Assessment based on project goals: Analyze the relationship between the task and the core goals of the project, as well as the contribution of the task to the success of the project.

[0080] Assessment based on resource input: Consider the amount of human, material, time and other resources required for the task.

[0081] Time urgency-based assessment: Assess the urgency of a task based on its deadline and the current time.

[0082] Assessment based on task dependencies: Analyze the position of a task in the task chain and the degree to which it depends on or is depended on other tasks.

[0083] Generate importance labels: Based on the above evaluation method, calculate an importance score or classification for each task. Add the importance score or classification as a label to the task node in the knowledge graph.

[0084] Label visualization and updating: In the knowledge graph visualization tool, the importance of tasks is represented by different colors, sizes, or icons. Users can update and adjust the importance labels based on project progress and actual conditions.

[0085] Sample Application: Assume there is a simple project planning data containing the following tasks: Task 1: Zhang San is responsible for designing the UI interface (2023-04-01 to 2023-04-05); Task 2: Li Si is responsible for developing backend functions (2023-04-06 to 2023-04-15), which depends on Task 1; Task 3: Wang Wu is responsible for testing (2023-04-16 to 2023-04-20), which depends on Task 2.

[0086] Using the above method, we can construct the following knowledge graph: Nodes: Task 1, Task 2, Task 3, Zhang San, Li Si, Wang Wu, 2023-04-01, 2023-04-05, 2023-04-06, ... Edges: Task 1 → Task 2 (dependency), Zhang San → Task 1 (resource allocation), 2023-04-01 → Task 1 (time relationship), ... Through this knowledge graph, we can clearly see the task allocation, time planning and dependencies between tasks in the project, which helps project managers better manage projects and make decisions.

[0087] In one embodiment of the present invention, based on step 120, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner. Figure 3 shown.

[0088] Step 1201. Identify file names using keyword technology and collect project execution data from the project management platform, document storage and email system, and project logs.

[0089] Keyword technology is used to identify file names and collect project execution data from project management platforms, document storage and email systems, and project logs. Keyword technology: Leverages keyword extraction algorithms from natural language processing technologies, such as TF-IDF, TextRank, or deep learning-based keyword generation models, to identify keywords related to project execution. These keywords can include task names, resource names, timestamps, and more. Project management platforms: Platforms such as Jira and Trello typically record task status updates, comments, attachments, and other information during project execution. Project execution data from these platforms can be obtained through API interfaces or data export functions. Document storage and email systems: Systems such as Google Drive, Dropbox, and Outlook store project-related documents and email communications. Documents and emails related to project execution can be collected through keyword searches or folder traversal. Project logs: Project logs record important events, decisions, and changes during project execution. By parsing log files, data related to project execution can be extracted.

[0090] Step 1202. Use a pre-trained language model to extract task execution information from the project execution data, where the task execution information includes executed information and approval information. The executed information includes executed task information and execution time, and the approval information includes pending task information and execution time.

[0091] Pre-trained language model: Choose pre-trained language models such as BERT and GPT-3. These models perform well in natural language processing tasks and can accurately understand the semantic information in the text.

[0092] Task execution information: includes execution information and approval information.

[0093] Executed information: This includes information about executed tasks and their execution time. For example, for the task description "Zhang San completed UI design on 2023-04-05," the model can identify "Zhang San" as the executor, "UI design" as the executed task, and "2023-04-05" as the execution time.

[0094] Approval information: This includes pending task information and execution time. For example, for a task description like "Li Si's backend development feature has been submitted for approval and is expected to be approved on April 7, 2023," the model can identify "Li Si" as the submitter, "backend development feature" as the pending task, and "2023-04-07" as the approval date.

[0095] Based on the above implementation method, after the approval information is identified, the approval data management based on the knowledge graph can automatically exclude the approval information that does not meet the approval conditions, thereby improving the approval efficiency.

[0096] (1) Based on the task information to be executed, query the corresponding target task information from the knowledge graph Extract key information from the task information to be executed, such as task name, task ID, etc.

[0097] Use this key information to search in the knowledge graph and find the corresponding target task node.

[0098] Read detailed task information from the target task node, including task description, person in charge, planned time, priority, etc.

[0099] (2) Query the status of the predecessor task that the target task depends on From the knowledge graph, find all predecessor task nodes that have dependencies with the target task.

[0100] Check the Status property of these predecessor task nodes to confirm whether they are completed.

[0101] If there are unfinished predecessor tasks, record the task name and the reason for incompletion (if there is a record in the knowledge graph).

[0102] (3) Extract the responsible person information and planning time from the target task information Extract the task leader and planned time from the queried target task information.

[0103] (4) Match the approval information with the responsible person information and planned time in the target task information Compare the approval information of the task to be executed (including the approver, approval time and approval content) with the person in charge and planned time in the target task information.

[0104] If the approval information does not completely match the target task information (for example, the approver is not the task owner, or the approval time is earlier than the planned time), the next step is to generate an exception prompt message.

[0105] (5) Generate abnormal prompt information When the approval information does not match the target task information, a corresponding abnormal prompt message is generated based on the specific content of the mismatch.

[0106] For example, if the approver is not the task owner, a prompt will appear: "The approver is not the task owner. Please re-appoint the approver."

[0107] (6) Generate prompt information that the approval information is normal When the approval information completely matches the target task information, a prompt message indicating that the approval information is normal is generated.

[0108] For example, "The approval information is consistent with the target task information, and the approval process is normal."

[0109] (7) Determine approval conditions and send approval information After confirming that the approval information is normal, check again the status of the predecessor tasks that the target task depends on.

[0110] If the status of all the predecessor tasks is completed, it is determined that the approval information meets the approval conditions.

[0111] Send the approval information to the review terminal (such as project manager, department head, etc.) for further review.

[0112] If the predecessor task that the target task depends on is in an unfinished state, it is determined that the approval information does not meet the approval conditions.

[0113] Generate a prompt message for approval failure and indicate the unfinished predecessor tasks and their reasons.

[0114] Feedback the unapproved information to the submitter of the task to be executed so that he or she can understand the problem and take appropriate solutions.

[0115] In one embodiment of the present invention, based on step 130, a possible embodiment will be given below to illustrate its specific implementation scheme in a non-limiting manner, such as Figure 4 shown.

[0116] Step 1301: Extract all executed information of the same task, including but not limited to task name, executor, execution time, and execution result, etc. This information is sorted and saved in a special list for subsequent processing.

[0117] Step 1302: Decompose the task into multiple execution stages, and allocate execution time to the multiple execution stages based on the planned time of the task.

[0118] Based on the complexity of the task and the workflow, break it down into multiple specific execution stages.

[0119] Refer to the planned time of the task and allocate a reasonable execution time for each execution stage to ensure that the overall task can be completed within the planned time.

[0120] For example, a complex software development project can be broken down into specific execution phases for better management and monitoring. These phases may include: Requirements analysis phase: Communicate with customers to clarify the functional and performance requirements of the software.

[0121] Design phase: Based on the results of demand analysis, design the overall architecture and interface of the software.

[0122] Coding phase: Write the source code of the software according to the design documents.

[0123] Testing phase: Perform unit testing, integration testing, and system testing on the software to ensure the quality of the software.

[0124] Deployment phase: Deploy the software to the production environment and perform necessary configuration and debugging.

[0125] Maintenance phase: Provide technical support and maintenance services after the software is launched.

[0126] Refer to the planned timeline for the software development project (for example, a six-month project), and allocate reasonable execution times for each execution phase. These times may vary depending on the specific circumstances of the project and the efficiency of the team, but can generally be roughly estimated as follows: Requirements analysis phase: 1 month; Design phase: 1.5 months; Coding phase: 2 months; Testing phase: 1 month; Deployment phase: 0.3 months; Maintenance phase (ongoing after the project ends): arranged according to actual needs.

[0127] Step 1303: Based on the execution time and execution order of the execution phase, set a corresponding progress value for the execution phase.

[0128] Set a progress value for each execution phase, which can be determined based on execution time and execution order. The progress value can be a percentage, indicating the proportion of the phase in the overall task or the degree of completion.

[0129] Set a progress value for each stage. This progress value can be determined based on execution time and execution order, and can usually be expressed as the proportion of the stage in the overall task or the degree of completion. For example: Requirements analysis phase: The progress value is 1 / 6 (approximately 16.67%), which means that after the completion of this phase, the overall task progress is approximately 16.67%.

[0130] Design phase: The progress value is 1.5 / 6 (about 25%), which means that after the completion of this phase, the overall task progress is about 25%.

[0131] Coding phase: The progress value is 2 / 6 (approximately 33.33%), indicating that after the completion of this phase, the overall task progress is approximately 33.33%.

[0132] Testing phase: The progress value is 3 / 6 (about 50%), which means that after the completion of this phase, the overall task progress is about 50%.

[0133] Deployment phase: The progress value is 3.3 / 6 (approximately 55%), indicating that after the completion of this phase, the overall task progress is approximately 55%.

[0134] Maintenance phase: Since this phase continues after the project ends, no separate progress value is set. Instead, the overall task progress is updated according to actual needs.

[0135] Step 1304: Obtain the completed execution stage corresponding to the executed task information, and based on the correspondence between the completed execution stage and the execution stage and the progress value, retrieve the progress value corresponding to the executed information.

[0136] Based on the information of executed tasks, determine which execution stages have been completed.

[0137] The total progress value of the executed information is calculated through the correspondence between the completed execution stage, the execution stage and the progress value.

[0138] Step 1305: Convert the executed information in the list into a data set consisting of progress value and execution time.

[0139] The progress value and execution time of each executed information are combined into one data group.

[0140] These data sets will be used for subsequent time series construction and forecasting.

[0141] Step 1306: Arrange the data groups into a time data sequence according to execution time.

[0142] Arrange all data groups in the order of execution time to form a time data series.

[0143] This time data series will reflect the historical process and current status of task execution.

[0144] Step 1307: Input the time data sequence and the current system time into a pre-trained long short-term memory neural network model to predict the completion time of the task.

[0145] Use a pre-trained long short-term memory (LSTM) neural network model, taking a time data series and the current system time as input.

[0146] The model will predict the future completion time of tasks based on the patterns learned from historical data.

[0147] Step 1308: Calculate the difference between the completion time and the completion time defined by the planned time of the task, and generate status attributes based on the difference, wherein the status attributes include early completion, normal progress, and late completion.

[0148] Compare the predicted completion time with the planned completion time and calculate the difference between them.

[0149] According to the size of the difference, the corresponding status attributes are generated, such as "Ahead of schedule", "On track" and "Late completion".

[0150] Step 1309. Add the state attribute to the corresponding task node in the knowledge graph.

[0151] Find the node corresponding to the prediction task in the knowledge graph.

[0152] Add the generated status attribute to the attribute list of the node to quickly obtain task status information in project monitoring and decision-making.

[0153] In one embodiment of the present invention, based on step 140, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner. Figure 5 shown.

[0154] Step 1401. Set anomaly detection indicators, which include the proportion of overdue tasks, task progress deviation, resource utilization, and task dependency anomalies.

[0155] Overdue Task Ratio: Calculates the ratio of tasks that are currently behind their planned completion time to the total number of tasks. This metric can reflect the overall project progress delays.

[0156] Task progress deviation: This metric compares the actual progress of a task with the planned progress. This metric can quantify the degree to which a task is delayed or ahead of schedule.

[0157] Resource utilization: Analyzes the allocation and use of resources during project execution, including human resources, time resources, and material resources. Low resource utilization may mean wasted resources, while high resource utilization may lead to resource shortages.

[0158] Task dependency anomalies: Check whether the dependencies between tasks meet planning requirements. If a task's predecessor is not completed but the task has already begun, or if the execution order of tasks is disrupted, the overall progress and quality of the project may be affected.

[0159] Step 1402. Traverse the task nodes in the knowledge graph and obtain the status attributes of each task node.

[0160] After setting the anomaly detection indicators, it is necessary to traverse the task nodes in the knowledge graph and obtain the status attributes of each task node (such as progress, completion time, resource allocation, etc.).

[0161] Step 1403: Calculate the abnormality index value of each task.

[0162] Based on these state attributes and the calculation formula of the anomaly detection index, the anomaly index value of each task is calculated.

[0163] Step 1404. Mark the tasks whose abnormal indicator values ​​exceed the threshold as abnormal tasks and mark them specially in the knowledge graph.

[0164] The calculated anomaly indicator value is compared with a preset threshold. If a task's anomaly indicator value exceeds the threshold, the task is marked as abnormal and specially labeled in the knowledge graph (e.g., using a different color, icon, or text). This allows project managers to intuitively identify which tasks have anomalies and take timely action to address them.

[0165] Step 1405: Analyze the task relationship, resource allocation, execution time, and task execution information of the abnormal task, and use the thought chain reasoning technology to locate the abnormal cause of the task abnormality based on the analysis results.

[0166] First, you need to collect and analyze relevant information about the "user interface design" task, including: Task relationships: Check the dependencies between this task and other tasks to confirm whether any predecessor tasks are not completed or successor tasks have started early.

[0167] Resource allocation: Analyze whether the human resources (such as designers, developers, etc.) and time resources (such as the planned design cycle) allocated to the task are sufficient.

[0168] Execution time: Compare the actual execution time of the task with the planned execution time to determine whether there are any delays.

[0169] Task execution information: View logs, meeting records, communication emails, etc. during the task execution process to understand the details of the task execution and any problems encountered.

[0170] After collecting enough information, use the chain of thought reasoning technique to deduce the reasons that lead to the abnormality of the "User Interface Design" task. Possible abnormal reasons include: Prerequisite tasks are not completed: For example, there are ambiguities or omissions in the requirements analysis phase, resulting in unclear design direction.

[0171] Insufficient resources: Insufficient number of designers or unreasonable time allocation leads to slow design progress.

[0172] Technical difficulties: Difficult technical problems were encountered during the design process, such as the high learning cost of new design tools.

[0173] Poor communication: There are problems in communication with other team members, resulting in changes in requirements not being communicated to designers in a timely manner.

[0174] Step 1406: Input the abnormal cause of the abnormal task into the knowledge base to obtain multiple historical processing solutions.

[0175] Input the derived abnormal cause into the pre-built knowledge base and search for related historical processing solutions. Possible historical processing solutions include: Increase resources: Increase the number of designers or adjust time allocation to speed up the design process.

[0176] Technical training: Organize training courses for designers to learn new design tools or solve technical problems.

[0177] Strengthen communication: Establish a more effective communication mechanism to ensure that changes in requirements can be conveyed to designers in a timely manner.

[0178] Step 1407. Use the thought chain reasoning technology to conduct a causal analysis of multiple historical processing solutions, and deduce multiple implementation steps and effects from the problem-solution-result chain.

[0179] Conduct causal analysis on the historical solutions found, and deduce the implementation steps and expected effects of each solution from the problem-solution-result chain. For example: Add resource plan: Implementation steps: Assess the workload of existing designers and determine the number of additional designers needed; coordinate with the human resources department to recruit or deploy designers; reallocate design tasks to ensure that each designer has a reasonable workload.

[0180] Expected results: accelerated design progress and improved resource utilization.

[0181] Technical training program: Implementation steps: Determine the technical content that requires training; organize training courses or workshops; evaluate the effectiveness of the training to ensure that designers have mastered the required skills.

[0182] Expected effect: Designers’ technical capabilities will be improved, enabling them to solve technical problems in the design process more efficiently.

[0183] Step 1408. Based on the effects of the implementation steps, generate multiple implementation plans with the best effects, wherein the implementation plans include multiple implementation steps.

[0184] Randomly combine the most effective implementation steps to form multiple implementation plans.

[0185] Step 1409. Conduct a feasibility assessment on the implementation plan, and output the implementation plan that passes the feasibility assessment as a processing plan.

[0186] Conduct feasibility assessments on these options, consider the actual situation of the project, resource constraints, priorities and other factors, and select the best implementation plan.

[0187] On the basis of the above embodiments, in order to further improve the adaptability and usability of the knowledge base provided by the above embodiments, an implementable method is to collect corresponding processing solutions for historical abnormal tasks of multiple projects, and save the historical data of multiple projects as corresponding multiple data sets; and use the federated learning mechanism to train the knowledge base.

[0188] In project management practice, every project may encounter various abnormal tasks. The handling process and results of these tasks are valuable experience. In order to build a more comprehensive and effective knowledge base, it is necessary to collect the corresponding handling solutions of historical abnormal tasks from multiple projects.

[0189] Data collection channels include: Project documentation: Review the documentation of each project, including project plans, progress reports, meeting minutes, issue tracking records, etc., and extract information about abnormal tasks and their solutions.

[0190] Team communication: Conduct interviews or questionnaires with project team members to understand their experiences and lessons learned when handling unusual tasks.

[0191] Knowledge management system: If the company has a knowledge management system or project management platform, historical data can be directly extracted from it.

[0192] Data organization and cleaning: Organize the collected data to ensure that the handling plan for each abnormal task contains key information such as task description, abnormal cause, handling steps, and handling results.

[0193] Clean the data to remove duplicate, incomplete or invalid information to ensure data quality and accuracy.

[0194] Dataset construction: The organized data is categorized by project or anomaly type to construct multiple datasets. Each dataset contains a series of anomaly tasks and their solutions for subsequent analysis and training.

[0195] Use federated learning mechanism to train the knowledge base After collecting historical data from multiple projects, the federated learning mechanism can be used to train the knowledge base so that it can more accurately identify and handle abnormal tasks.

[0196] Overview of Federated Learning: Federated learning is a distributed machine learning framework that allows multiple participants to jointly train models without sharing the original data. This mechanism protects data privacy while leveraging the data resources of multiple participants to improve model performance.

[0197] Knowledge base training process: Model initialization: Select a suitable machine learning model as the basis of the knowledge base, such as a decision tree, neural network, or deep learning model.

[0198] Data distribution: Distribute multiple data sets to different participants (such as different project teams or departments) according to the requirements of federated learning.

[0199] Model training: On each participant, the model is trained using the local dataset to obtain local model parameters.

[0200] Parameter aggregation: Aggregate the local model parameters of each participant to obtain the global model parameters. This process can be repeated multiple times until the model performance reaches stability or the preset stopping condition is met.

[0201] Model evaluation and optimization: Use the validation dataset to evaluate the global model and optimize and adjust the model based on the evaluation results.

[0202] Knowledge base application: After training is complete, the knowledge base can be used to identify new abnormal tasks and provide corresponding handling suggestions. When a new abnormal task appears, the knowledge base can retrieve similar cases from the training data based on the task description and the cause of the abnormality, and provide suggestions for handling steps and expected results.

[0203] Continuous updates and maintenance: As the project progresses and new abnormal tasks emerge, the knowledge base needs to be continuously updated and maintained. This includes adding new abnormal task processing solutions, optimizing model parameters, adjusting data distribution, etc.

[0204] By collecting historical abnormal task handling solutions from multiple projects and using the federated learning mechanism to train the knowledge base, a more comprehensive, accurate, and intelligent knowledge base can be built to provide stronger support for project management.

[0205] The following is another embodiment of the present invention. In project management, real-time progress display and problem feedback mechanisms can help management understand project status in a timely manner and respond to problems quickly. The S7 module provides an interactive visual interface, presenting project progress, anomalies, and risk information in intuitive charts, and supports user feedback on the model's identification and recommendation results. User feedback further refines and optimizes the model output, thereby continuously improving the accuracy and adaptability of the system. The specific steps are as follows: Real-time data visualization: Provides an interactive user interface with real-time data visualization to display project progress, anomalies, and risks; integrates user feedback to refine model outputs: By real-time monitoring and graphically displaying key project data, it provides a global view of project management, allowing users to quickly understand the current status and potential risks of the project.

[0206] Project Progress Dashboard: Provides an overall project progress dashboard, showcasing each task's current status, completion percentage, and deadlines through visual formats like Gantt charts and timelines. Color-coded delays and early completions help management quickly identify progress deviations.

[0207] Risk and exception charts: Display risk points and exceptions in the project. Through heat maps, radar charts and other charts, they intuitively display task risk levels, resource bottlenecks and abnormal tasks in the critical path, supporting management's risk assessment and decision-making.

[0208] Resource Allocation and Usage: Visualizes the current allocation and usage of resources (such as manpower, budget, and equipment), presenting resource consumption distribution through bar charts and pie charts, helping users rationally plan resource allocation and quickly identify resource shortages or over-allocations.

[0209] User feedback integration and management: Integrate the feedback information provided by users during actual use, and further improve the output effect and personalized recommendations of the model by recording user preferences and feedback.

[0210] Feedback collection mechanism: Feedback buttons are embedded in each data visualization page and recommendation result, allowing users to provide feedback on progress display, risk identification, solution recommendation and other results (such as "accurate", "partially accurate", "inaccurate", etc.), forming a user feedback data set.

[0211] Feedback data classification and labeling: Classify and label feedback data, and categorize them by feedback content type (such as progress tracking, risk identification, and solution recommendation) to provide directional data for subsequent model optimization.

[0212] Personalized preference records: Record user preference information in feedback (such as prioritizing certain types of risks, focusing on specific resource allocation, etc.), providing personalized customization support for the model, so that subsequent recommendation results are more in line with user needs.

[0213] Model output refinement and optimization: Incorporate user feedback information into the model training process, optimize and fine-tune model output through feedback data, and make the model respond more accurately to project status and user needs.

[0214] Feedback-driven model fine-tuning: Feedback data is used to fine-tune and retrain the model to correct identification errors and improve recommendation accuracy. For example, risk identification items marked as "inaccurate" by users can be trained and corrected to reduce errors.

[0215] Personalized optimization of recommendation algorithms: Based on user feedback, we optimize the weights and strategies of the recommendation algorithms, making solution recommendations and risk warnings more tailored to user needs. Frequently selected solution types or high-profile risk levels will be prioritized for subsequent recommendations and reminders.

[0216] Dynamic update and feedback adaptation: After receiving multiple similar feedbacks, the model automatically generates adaptive rules and algorithm adjustment strategies to adapt to user habits and project-specific needs, and updates them in real time in the system background.

[0217] Multi-dimensional project dynamic display: Through multi-dimensional data display and perspective switching, users can view project status from different angles and fully grasp project dynamics.

[0218] Multi-level view switching: Provides multi-level views at the task, milestone, and overall project levels, allowing users to switch from detailed to global perspectives. For example, the milestone view displays a summary of progress at each stage, while the task-level view focuses on the specific status and risks of each task.

[0219] Historical Data Comparison View: By comparing current data with historical project data, management can understand project trends and efficiency, and identify discrepancies. For critical tasks, historical delay rates and resource consumption comparisons are provided to identify potential areas for improvement.

[0220] Future trend forecast view: Predict future progress and risk trends based on time series models, and display future task completion rates and risk changes through line charts or trend charts, providing decision-making references and early warnings.

[0221] Real-time notifications and interactive reminders: By establishing a real-time notification and interactive reminder system, users can be promptly notified of key developments and risk events, facilitating rapid response.

[0222] Real-time push notifications: When the system detects new major anomalies, risks, or progress deviations, it automatically pushes notifications to users, ensuring that management can be aware of key project dynamics and respond immediately.

[0223] Customized reminder settings: Allow users to set reminder rules according to their own needs (such as delay warnings for specific tasks and the usage status of important resources), thereby achieving personalized dynamic reminders and ensuring that users obtain the most relevant information.

[0224] Interactive task reminders: When a task is about to expire or resources are about to be exceeded, the system will trigger interactive reminders and suggest users to take corresponding adjustment measures (such as task rescheduling and resource scheduling) to improve the response speed and execution efficiency of project management.

[0225] In some embodiments, the project full process data tracking and processing system may include multiple functional modules composed of computer program segments. The computer program of each program segment in the project full process data tracking and processing system may be stored in the memory of a computer device and executed by at least one processor to perform (see Figure 1 Description) The function of tracking and processing data throughout the project process.

[0226] In this embodiment, the project full process data tracking and processing system can be divided into multiple functional modules according to the functions it performs, such as Figure 6 As shown. The functional modules of system 600 may include: a graph construction module 610, an execution monitoring module 620, a state setting module 630, and an exception handling module 640. A module as referred to in the present invention refers to a series of computer program segments that can be executed by at least one processor and can perform fixed functions, and is stored in a memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0227] A graph construction module is used to obtain project planning data, extract task information and relationships between tasks from the project planning data using a pre-trained language model, wherein the task information includes planning time, and construct a knowledge graph based on the task information and relationships between tasks; An execution monitoring module, configured to obtain project execution data from various data sources and identify task execution information from the project execution data; A state setting module is used to predict the task progress based on the data sequence composed of all task execution information of the task, match the task progress with the corresponding planned time, and add state attributes to the corresponding node of the knowledge graph based on the matching result; The exception handling module is used to analyze the abnormal situation of the task execution status based on the knowledge graph after updating the status attributes, analyze the abnormal factors based on the project execution data, and use the pre-built knowledge base to generate a processing plan for the abnormal factors.

[0228] In an optional embodiment, the map construction module includes: Data collection unit, used to collect project planning data from the project management platform, document storage email system and project log; A preprocessing unit, configured to preprocess the project planning data, wherein the preprocessing includes cleaning the data and formatting the data; A model processing unit, configured to input the pre-processed project planning data into a pre-trained language model to obtain task information and task relationships, wherein the task information includes resource information and planning time, and the task relationships include dependency relationships, resource allocation relationships, and time sequence relationships between tasks; A graph construction unit is used to construct a knowledge graph using a graph database, taking tasks, resources, and time as nodes and task relationships as edges; a label generation unit, configured to generate importance labels for tasks in the knowledge graph; An interface setting unit is used to set a query interface for the knowledge graph.

[0229] Although the present invention has been described in detail with reference to the accompanying drawings and in conjunction with preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, persons of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and such modifications or substitutions shall be within the scope of the present invention. Any changes or substitutions that can be easily conceived by persons skilled in the art within the technical scope disclosed in the present invention shall be within the scope of protection of the present invention.

Claims

1. A method for tracking and processing data throughout a project, characterized in that: include: Obtaining project planning data, extracting task information and relationships between tasks from the project planning data using a pre-trained language model, wherein the task information includes planning time, and constructing a knowledge graph based on the task information and relationships between tasks; Acquire project execution data from a variety of data sources, and identify task execution information from the project execution data; Predicting task progress based on a data sequence consisting of all task execution information of the task, matching the task progress with the corresponding planned time, and adding state attributes to the corresponding nodes of the knowledge graph based on the matching results; Analyze the abnormal situation of task execution status based on the knowledge graph after updating status attributes, analyze abnormal factors based on project execution data, and use the pre-built knowledge base to generate processing solutions for the abnormal factors.

2. The method according to claim 1, characterized in that Obtain project planning data, extract task information and relationships between tasks from the project planning data using a pre-trained language model, wherein the task information includes planning time, and construct a knowledge graph based on the task information and relationships between tasks, including: Collect project planning data from project management platforms, document storage and email systems, and project logs; Preprocessing the project planning data, wherein the preprocessing includes cleaning the data and formatting the data; Inputting the pre-processed project planning data into a pre-trained language model to obtain task information and task relationships, wherein the task information includes resource information and planning time, and the task relationships include dependency relationships, resource allocation relationships, and time sequence relationships between tasks; Use tasks, resources, and time as nodes, and task relationships as edges to build a knowledge graph using a graph database. Generating importance labels for tasks in the knowledge graph; Set up a query interface for the knowledge graph.

3. The method according to claim 1, characterized in that Acquiring project execution data from a variety of data sources and identifying task execution information from the project execution data include: Identify file names using keyword technology and collect project execution data from project management platforms, document storage and email systems, and project logs; A pre-trained language model is used to extract task execution information from the project execution data, wherein the task execution information includes executed information and approval information, wherein the executed information includes executed task information and execution time, and the approval information includes pending task information and execution time.

4. The method according to claim 3, characterized in that The method further comprises: Based on the task information to be executed, query the corresponding target task information and the status of the predecessor tasks that the target task depends on from the knowledge graph; Extracting person-in-charge information and planning time from the target task information; Match the approval information with the person in charge information and the planned time in the target task information. If they do not completely match, an abnormal prompt message is generated; if they completely match, a prompt message indicating that the approval information is normal is generated; Confirm that the approval information is normal. If the status of the predecessor tasks that the target task depends on are all completed, then the approval information is determined to meet the approval conditions and the approval information is sent to the review terminal; if there is an unfinished status in the predecessor tasks that the target task depends on, then the approval information is determined to not meet the approval conditions and a prompt message of approval failure is generated.

5. The method according to claim 3, characterized in that Predicting task progress based on a data sequence consisting of all task execution information of the task, matching the task progress with the corresponding planned time, and adding state attributes to the corresponding nodes of the knowledge graph based on the matching results, including: Save the executed information of the same task into the same list; Decompose the task into multiple execution stages and allocate execution time to the multiple execution stages based on the planned time of the task; Set the corresponding progress value for the execution phase based on the execution time and execution order of the execution phase; Obtain the completed execution stage corresponding to the executed task information, and based on the correspondence between the completed execution stage and the execution stage and the progress value, retrieve the progress value corresponding to the executed information; Converting the executed information in the list into a data set consisting of a progress value and an execution time; Arranging the data groups into a time data sequence according to execution time; Inputting the time data sequence and the current system time into a pre-trained long short-term memory neural network model to predict the completion time of the task; Calculating a difference between the completion time and the completion time defined by the planned time of the task, and generating a status attribute based on the difference, the status attribute including ahead of schedule completion, on track progress, and overdue completion; Add the state attribute to the corresponding task node in the knowledge graph.

6. The method according to claim 5, characterized in that Analyze abnormalities in task execution status based on the knowledge graph after updating status attributes, analyze abnormal factors based on project execution data, and use the pre-built knowledge base to generate processing solutions for the abnormal factors, including: Setting anomaly detection indicators, including the proportion of overdue tasks, task progress deviation, resource utilization, and task dependency anomalies; Traverse the task nodes in the knowledge graph and obtain the status attributes of each task node; Calculate the abnormal index value of each task; Mark tasks where the abnormal indicator value exceeds the threshold as abnormal tasks and mark them specially in the knowledge graph; By analyzing the task relationships, resource allocation, execution time, and task execution information of abnormal tasks, and using thought chain reasoning technology to locate the abnormal causes of task abnormalities based on the analysis results; Utilize the pre-built knowledge base to generate processing solutions based on the abnormal causes of abnormal tasks.

7. The method according to claim 6, characterized in that Leverage pre-built knowledge base to generate solutions based on the causes of abnormal tasks, including: Inputting the abnormal cause of the abnormal task into the knowledge base to obtain multiple historical processing solutions; Use thought chain reasoning technology to conduct causal analysis on multiple historical treatment plans, and deduce multiple implementation steps and effects from the problem-solution-result chain; Based on the results of the implementation steps, generating multiple implementation plans with the best results, wherein the implementation plans include multiple implementation steps; Conduct a feasibility assessment on the implementation plan and output the implementation plan that passes the feasibility assessment as a treatment plan.

8. The method according to claim 7, characterized in that The method further comprises: Collect corresponding processing solutions for historical abnormal tasks of multiple projects and save historical data of multiple projects as corresponding multiple data sets; The knowledge base is trained using the federated learning mechanism.

9. A project full-process data tracking and processing system, characterized by: include: A graph construction module is used to obtain project planning data, extract task information and relationships between tasks from the project planning data using a pre-trained language model, wherein the task information includes planning time, and construct a knowledge graph based on the task information and relationships between tasks; An execution monitoring module, configured to obtain project execution data from various data sources and identify task execution information from the project execution data; A state setting module is used to predict the task progress based on the data sequence composed of all task execution information of the task, match the task progress with the corresponding planned time, and add state attributes to the corresponding node of the knowledge graph based on the matching result; The exception handling module is used to analyze the abnormal situation of the task execution status based on the knowledge graph after updating the status attributes, analyze the abnormal factors based on the project execution data, and use the pre-built knowledge base to generate a processing plan for the abnormal factors.

10. The system according to claim 9, characterized in that The graph construction module includes: Data collection unit, used to collect project planning data from the project management platform, document storage email system and project log; A preprocessing unit, configured to preprocess the project planning data, wherein the preprocessing includes cleaning the data and formatting the data; A model processing unit, configured to input the pre-processed project planning data into a pre-trained language model to obtain task information and task relationships, wherein the task information includes resource information and planning time, and the task relationships include dependency relationships, resource allocation relationships, and time sequence relationships between tasks; A graph construction unit is used to construct a knowledge graph using a graph database, taking tasks, resources, and time as nodes and task relationships as edges; a label generation unit, configured to generate importance labels for tasks in the knowledge graph; An interface setting unit is used to set a query interface for the knowledge graph.

Images