Schedule planning method, system and device based on artificial intelligence

Abstract

The invention provides a schedule planning method, system and device based on artificial intelligence, and the method comprises the steps: obtaining a to-be-planned task description inputted by a user, and obtaining a subtask sequence of the user based on the to-be-planned task description; acquiring historical behavior data of the user, environment information of an area to which the user belongs and schedule information of the user, and determining personalized cue words of the user based on the historical behavior data of the user; and generating a schedule planning scheme corresponding to the description of the to-be-planned task based on the subtask sequence, the historical behavior data, the schedule arrangement information, the environment information and the personalized cue word, wherein the schedule planning scheme comprises execution time information, execution place information and resource configuration suggestion information. According to the invention, the schedule planning efficiency and rationality of complex tasks can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular, to a schedule planning method, system and device based on artificial intelligence. Background Art

[0002] By reasonably planning and using time, work tasks can be completed more effectively, unnecessary time waste can be reduced, and thus work efficiency and productivity can be improved.

[0003] Current time management tools require users to manually input tasks and arrange time, lacking intelligent support. For example, when facing complex tasks, users often need to break them down into executable subtasks by themselves, and then input the decomposed subtasks into the time management tool and allocate time for each subtask based on their feelings; this method not only takes time, but also easily leads to low efficiency due to improper decomposition, and the time allocated for each subtask is often unreasonable. Although some current time management tools can schedule the subtasks decomposed by users, existing tools usually rely on fixed rules in task scheduling and cannot dynamically integrate the user's historical behavior data and external environmental factors, resulting in unscientific and unreasonable time arrangements.

[0004] Currently, although there are some time management software on the market that attempt to introduce artificial intelligence technology, their functions are still limited to using speech recognition or text recognition to complete simple task entry. For example, the "Time Sequence app" provides the function of intelligently recognizing natural language and incorporating it into to-do items, but it cannot reasonably and efficiently schedule multiple subtasks in complex tasks. Therefore, how to improve the efficiency and rationality of schedule planning for complex tasks is a technical problem to be solved urgently. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a schedule planning method, system and device based on artificial intelligence to eliminate or improve one or more defects existing in the prior art.

[0006] One aspect of the present invention provides a schedule planning method based on artificial intelligence, the method comprising:

[0007] Obtaining a to-be-planned task description input by a user, and obtaining a subtask sequence of the user based on the to-be-planned task description;

[0008] Obtaining the historical behavior data of the user, the environmental information of the region where the user is located, and the schedule information of the user, and determining a personalized prompt word of the user based on the historical behavior data of the user;

[0009] Generate a schedule planning scheme corresponding to the to-be-planned task description based on the sub-task sequence, historical behavior data, schedule information, environmental information, and the personalized prompt words, where the schedule planning scheme includes execution time information, execution location information, and resource allocation suggestion information.

[0010] In some embodiments of the present invention, obtaining the sub-task sequence of the user based on the to-be-planned task description includes:

[0011] Input the to-be-planned task description into a pre-trained large language model to extract the long-distance dependency relationship between the words in the task description;

[0012] Construct a knowledge graph based on the long-distance dependency relationship and the structured domain knowledge base, where the nodes of the knowledge graph are the key entities in the task, and the edges of the knowledge graph are the relationships between the key entities;

[0013] Based on the constructed knowledge graph, obtain the sub-task sequence through a complexity evaluation algorithm and a minimization segmentation algorithm.

[0014] In some embodiments of the present invention, inputting the to-be-planned task description into a pre-trained large language model to extract the long-distance dependency relationship between the words in the task description includes:

[0015] The pre-trained large language model performs multi-layer bidirectional attention mechanism calculations, and combines the task decomposition paradigm library of the domain knowledge graph, the industry classic workflow template library, and the decomposition case library of historical similar tasks to call the domain adapter to extract the structured data corresponding to the to-be-planned task description, and extract the long-distance dependency relationship between the words in the task description based on the structured data.

[0016] In some embodiments of the present invention, obtaining the historical behavior data of the user, the environmental information of the region where the user is located, and the schedule information of the user includes:

[0017] Collect the user operation records in the system log, where the user operation records include the application type, timestamp, and operation duration;

[0018] Read the calendar to extract the schedule information of the user, where the schedule information includes the time, location, and participants of the schedule activity;

[0019] Obtain the task data of the to-do tasks from the to-do task application database, where the task data includes the creation time, deadline, and completion status;

[0020] Construct a time consumption feature model based on the user operation records, schedule information, and task data of the to-do tasks;

[0021] Obtain the environmental information sensed by the real-time environmental perception module, where the environmental information includes weather data.

[0022] In some embodiments of the present invention, determining the personalized prompt words of the user based on the historical behavior data of the user includes: inputting the historical behavior data and the schedule information into a trained prompt word extraction model to obtain the personalized prompt words of the user; and / or,

[0023] Generating a schedule planning scheme corresponding to the to-be-planned task description based on the subtask sequence, historical behavior data, schedule information, environmental information, and the personalized prompt words, including:

[0024] Generating a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm based on the subtask sequence, historical behavior data, schedule information, environmental information, task data of the to-do task, and the personalized prompt words.

[0025] In some embodiments of the present invention, generating a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm based on the subtask sequence, historical behavior data, schedule information, environmental information, task data of the to-do task, and the personalized prompt words includes:

[0026] Determining the best working time period of the user based on the biometric data of the user;

[0027] Calculating the time tolerance based on the public transportation data and weather data corresponding to each subtask;

[0028] Generating a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm by combining the best working time period information and time tolerance information.

[0029] In some embodiments of the present invention, the method further includes: obtaining the actual execution data of some subtasks already executed by the user, determining the planning deviation degree of the scheme based on the actual execution data and the schedule planning scheme, and generating a visual analysis report including improvement suggestions based on the planning deviation degree of the scheme; and / or,

[0030] Obtaining the user's own characteristics, storing the schedule planning scheme and the user's own characteristics in the user database; obtaining the custom modification values made by the user to the schedule planning scheme, and updating the schedule planning scheme and the user's own characteristics in the user database based on the custom modification values.

[0031] According to another aspect of the present invention, there is also provided an artificial intelligence-based schedule planning system, where the schedule planning system includes:

[0032] The AI task decomposition module obtains the task description to be planned input by the user, and obtains the user's subtask sequence based on the task description to be planned;

[0033] The behavior analysis module obtains the user's historical behavior data, the environmental information of the area where the user is located, and the user's schedule information, and determines the user's personalized prompt words based on the user's historical behavior data;

[0034] The AI time allocation module generates a schedule planning scheme corresponding to the task description to be planned based on the subtask sequence, historical behavior data, schedule information, environmental information, and the personalized prompt words. The schedule planning scheme includes execution time information, execution location information, and resource allocation suggestion information.

[0035] According to another aspect of the present invention, there is also provided an artificial intelligence-based schedule planning device, which includes a processor, a memory, and a computer program stored on the memory. The processor is used to execute the computer program, and when the computer program is executed, the device implements the steps of the method described in any of the above embodiments.

[0036] According to yet another aspect of the present invention, there is also provided a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the method described in any of the above embodiments.

[0037] The artificial intelligence-based schedule planning method disclosed in the above embodiments of the present invention obtains a subtask sequence based on the task description to be planned, and obtains the user's historical behavior data, the environmental information of the area where the user is located, the user's schedule information, and personalized prompt words. According to the obtained subtask sequence, historical behavior data, schedule information, environmental information, and personalized prompt words, etc., it completes the schedule planning of complex tasks, that is, obtains a schedule planning scheme. This method can realize the automatic decomposition of complex tasks, which not only improves the efficiency and accuracy of task decomposition, but also improves the efficiency, rationality, and accuracy of the schedule planning of complex tasks.

[0038] The additional advantages, objectives, and features of the present invention will be partially described below, and will become partially apparent to those of ordinary skill in the art after studying the following text, or can be learned from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the description and the drawings.

[0039] Those skilled in the art will understand that the objectives and advantages that can be achieved by the present invention are not limited to the above specifically described, and the above and other objectives that the present invention can achieve will be more clearly understood according to the following detailed description. Brief Description of the Drawings

[0040] The accompanying drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and do not limit the present invention. The components in the drawings are not drawn to scale, but are only for showing the principles of the present invention. To facilitate showing and describing some parts of the present invention, the corresponding parts in the drawings may be enlarged, that is, may become larger relative to other components in an exemplary device actually manufactured according to the present invention. In the drawings:

[0041] Figure 1 It is a schematic flow chart of an artificial intelligence-based schedule planning method according to an embodiment of this application.

[0042] Figure 2 It is a schematic structural diagram of an artificial intelligence-based schedule planning system according to an embodiment of this application.

[0043] Figure 3 It is a schematic flow chart of an artificial intelligence-based schedule planning method according to another embodiment of this application.

[0044] Figure 4 It is a schematic flow chart of an artificial intelligence-based schedule planning method according to still another embodiment of this application.

[0045] Reference numerals:

[0046] AI task decomposition module 10, behavior analysis module 20, AI time allocation module 30, AI review and analysis module 40, domain knowledge graph construction unit 11, complexity evaluation unit 12, user portrait modeling unit 21, environmental sensitivity analysis unit 22, dynamic weight adjustment unit 23, conflict detection subunit 31, resource optimization subunit 32, elastic buffer setting unit 33, data structure 41, multi-dimensional comparison unit 42, root cause analysis unit 43 Detailed implementation manners

[0047] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the implementation manners and the accompanying drawings. Herein, the illustrative implementation manners of the present invention and their descriptions are used to explain the present invention, but do not limit the present invention.

[0048] Herein, it should also be noted that in order to avoid obscuring the present invention due to unnecessary details, only the structures and / or processing steps closely related to the solution according to the present invention are shown in the drawings, while other details less related to the present invention are omitted.

[0049] It should be emphasized that the term "including / comprising" when used herein refers to the presence of features, elements, steps, or components, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0050] Here, it should also be noted that, unless otherwise specified, the term "connection" in this text can not only refer to direct connection, but also indirect connection with intermediaries, not only represent wired connection, but also wireless connection, and can be specifically changed based on the actual application scenario.

[0051] In recent years, with the rapid development of artificial intelligence technology, large language models have made remarkable progress in the field of natural language processing (NLP). These models are based on the Transformer architecture and can capture long-distance dependencies in language through the attention mechanism, thus performing well in tasks such as text generation and semantic understanding. The application scope of large language models has been continuously expanding, from simple text generation to the processing of complex tasks, such as intelligent customer service, content creation, knowledge Q&A, etc. However, although these models show powerful capabilities in language understanding and generation, their application in the field of time management is still in its infancy. The inventors found in their research that by leveraging the natural language processing ability and intelligent decision-making ability of large language models, it is possible to help users efficiently manage time, that is, to achieve task decomposition, task arrangement, and time review of complex tasks, etc., which can significantly improve time management efficiency; based on this, this application proposes an artificial intelligence-based schedule planning method, system, and device, and this schedule planning method fills the gap in intelligent time management of large language models.

[0052] In the following text, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0053] Figure 1 It is a schematic flowchart of an artificial intelligence-based schedule planning method according to an embodiment of this application. As Figure 1 shown, this artificial intelligence-based schedule planning method at least includes steps S10 to S30.

[0054] Step S10: Obtain the task description to be planned input by the user, and obtain the sub-task sequence of the user based on the task description to be planned.

[0055] This step can be implemented based on the AI task decomposition module of the schedule planning system, that is, the AI task decomposition module receives the task description to be planned by the user, decomposes the task description to be planned, and finally obtains the sub-task sequence.

[0056] Exemplarily, obtaining the user's subtask sequence based on the to-be-planned task description may include the following steps: inputting the to-be-planned task description into a pre-trained large language model to extract the long-distance dependency relationship between the words in the task description; constructing a knowledge graph based on the long-distance dependency relationship and a structured domain knowledge base, where the nodes of the knowledge graph are the key entities in the task and the edges of the knowledge graph are the relationships between the key entities; and obtaining the subtask sequence based on the constructed knowledge graph through a complexity evaluation algorithm and a minimization segmentation algorithm.

[0057] In the above embodiment, the AI task decomposition module can receive the to-be-planned task description input by the user through the natural language interaction interface, perform semantic parsing and knowledge extraction using a pre-trained large language model, and generate an executable subtask sequence including a hierarchical structure. The pre-trained large language model can be a large language model with a Transformer architecture. Specifically, the pre-trained large language model can perform multi-layer bidirectional attention mechanism calculations and call a domain adapter based on the task decomposition paradigm library of the domain knowledge graph, the industry classic workflow template library, and the decomposition case library of historical similar tasks to extract the structured data corresponding to the to-be-planned task description, and extract the long-distance dependency relationship between the words in the task description based on the structured data.

[0058] In the above embodiment, using a large language model with a Transformer architecture to perform multi-layer bidirectional attention mechanism calculations on the task description input by the user through the natural language interaction interface, calling a domain adapter from the pre-trained large language model based on the task decomposition paradigm library of the domain knowledge graph, the industry classic workflow template library, and the decomposition case library of historical similar tasks to extract structured data, capturing the long-distance dependency relationship between the words in the task description, and accurately understanding the overall semantics of the task. In addition, constructing the dependency relationship and the domain knowledge base into a knowledge graph and providing it for the large language model to call and analyze to obtain the classification preferences for task classification and user personalization, using the key entities and relationships in the task as nodes and edges to form a structured knowledge network, and obtaining the classification preferences for task classification and user personalization. Moreover, based on the knowledge graph obtained by the large language model, using a domain knowledge graph construction unit, a complexity evaluation unit, and a minimization segmentation algorithm, performing semantic parsing and knowledge extraction using a pre-trained large language model, generating an executable subtask sequence including a hierarchical structure, and obtaining multiple subtasks and execution suggestions. This method combines the pre-trained large language model and the domain knowledge graph, realizes the automatic decomposition of complex tasks, generates a logical subtask sequence, and significantly improves the efficiency and accuracy of task decomposition.

[0059] Step S20: Obtain the historical behavior data of the user, the environmental information of the region where the user is located, and the schedule information of the user, and determine the personalized prompt words of the user based on the historical behavior data of the user.

[0060] This step can be implemented based on the behavior analysis module of the schedule planning system. When scheduling the subtask sequence, it combines the user's historical data and real-time environment perception data to dynamically generate the optimal time planning scheme, adapt to the needs of different users and the changes in the external environment, and improve the rationality and flexibility of time arrangement.

[0061] Exemplarily, obtaining the historical behavior data of the user, the environmental information of the region where the user is located, and the schedule information of the user may specifically include: collecting the user operation records in the system log, where the user operation records include the application type, timestamp, and operation duration; reading the calendar to extract the schedule information of the user, where the schedule information includes the time, location, and participants of the schedule activity; obtaining the task data of the to-do tasks from the to-do task application database, where the task data includes the creation time, deadline, and completion status; constructing a time consumption feature model based on the user operation records, schedule information, and task data of the to-do tasks; obtaining the environmental information sensed by the real-time environment perception module, where the environmental information includes weather data. Among them, when reading the calendar to obtain the schedule information of the user, the API interface of the external calendar application can be called, or the system's own calendar can be directly read.

[0062] In the above embodiment, the behavior analysis module can establish a time consumption feature model based on the user historical behavior database, and combine multi-dimensional information such as weather data and location data obtained by the real-time environment perception module to construct a parameter system for dynamic time planning or behavior prediction.

[0063] Specifically, a lightweight data collection agent can be deployed to obtain detailed information such as the timestamp and operation duration of the user's operation of various applications from the system log; use the API interface opened by the calendar application to extract information such as the time, location, and participants of the user's scheduled schedule activities; read data such as the creation time, deadline, and completion status of the tasks from the database of the to-do task application to form a user historical behavior data set and establish a time consumption feature model. In the above steps, a pre-trained large language model is used to combine multi-dimensional information such as weather data and location data obtained by the real-time environment perception module to construct a dynamic time planning parameter system and user portrait modeling unit, environmental sensitivity analysis unit, and dynamic weight adjustment unit.

[0064] In some embodiments of the present invention, determining the personalized prompt words of the user based on the historical behavior data of the user specifically includes: inputting the historical behavior data and the schedule information into a trained prompt word extraction model to obtain the personalized prompt words of the user. The input information of the prompt word extraction model is historical behavior data and schedule information, and the output data is the personalized prompt words of the user.

[0065] Step S30: Generate a schedule planning scheme corresponding to the to-be-planned task description based on the subtask sequence, historical behavior data, schedule information, environmental information, and the personalized prompt words. The schedule planning scheme includes execution time information, execution location information, and resource allocation suggestion information.

[0066] This step can be implemented based on the AI time allocation module of the schedule planning system. The AI time allocation module can perform spatio-temporal resource matching on the subtask sequence and generate a schedule planning scheme including specific time periods, execution locations, and resource allocation suggestions.

[0067] Exemplarily, a schedule planning scheme corresponding to the to-be-planned task can be generated based on a constraint satisfaction algorithm. That is, a schedule planning scheme corresponding to the to-be-planned task is generated through a constraint satisfaction algorithm based on the subtask sequence, historical behavior data, schedule information, environmental information, task data of the to-do task, and the personalized prompt words.

[0068] In some embodiments of the present invention, generating a schedule planning scheme corresponding to the to-be-planned task through a constraint satisfaction algorithm based on the subtask sequence, historical behavior data, schedule information, environmental information, task data of the to-do task, and the personalized prompt words includes: determining the best working time period of the user based on the biometric data of the user; calculating the time tolerance based on the public transportation data and weather data corresponding to each subtask; generating a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm in combination with the best working time period information and time tolerance information.

[0069] In the above embodiments, the constraint satisfaction algorithm is used to perform spatio-temporal resource matching on the subtask sequence, that is, a schedule planning scheme including specific time periods, execution locations, and resource allocation suggestions is generated. At the same time, the AI time allocation module also has the ability to analyze the user's biometric data to judge the best working time period and the ability to integrate real-time data such as public transportation and weather to calculate the time tolerance, so as to reasonably group related tasks and improve the task execution efficiency.

[0070] As an interaction medium between users and models, prompts directly guide the model to generate output results that meet user expectations. The behavior and output quality of the model largely depend on the design and input of prompts. This is because although large language models have mastered the statistical patterns and structures of language through training with a vast amount of text data, they essentially do not possess the ability to deeply understand the semantics of language. Therefore, the output of the model is completely driven by the input prompts, and the selection and design of prompts directly affect the accuracy and effectiveness of the model output. The present invention can generate different personalized prompts based on the historical behavior data of different users.

[0071] In addition, when making a schedule plan, in order to handle possible task conflicts or unexpected situations, a conflict detection sub-module, a resource optimization sub-module, and an elastic buffer setting unit can also be used to dynamically allocate spare time periods.

[0072] In addition, in terms of task review in the prior art, enterprise-level users usually need to rely on professionals to analyze time logs and generate efficiency reports. Although some traditional tools can count the distribution of time consumption, their functions are limited to data statistics and cannot achieve semantic-level analysis and causal inference. Existing tools lack the ability of adaptive optimization and cannot dynamically adjust the time allocation strategy based on historical data, resulting in the difficulty of continuously improving time management efficiency. Therefore, in order to further improve time management efficiency, this application also compares the deviation between the actual execution and the plan, generates a visual improvement report through causal reasoning, and feeds it back to the knowledge base.

[0073] That is, in some embodiments of the present invention, the schedule planning method based on artificial intelligence may further include the following steps: obtaining the actual execution data of some subtasks executed by the user; determining the plan deviation degree based on the actual execution data and the schedule planning scheme; generating a visual analysis report containing improvement suggestions based on the plan deviation degree. In this embodiment, by comparing the deviation degree between the actual execution data and the planning scheme, a multi-dimensional evaluation index system is established, and a visual analysis report containing improvement suggestions is automatically generated.

[0074] In some other embodiments of the present invention, the artificial intelligence-based schedule planning method may further include the following steps: obtaining the user's own characteristics, and storing the schedule planning scheme and the user's own characteristics in the user database; obtaining the custom modification values made by the user to the schedule planning scheme, and updating the schedule planning scheme and the user's own characteristics in the user database based on the custom modification values. For example, the user's own characteristics may include the user's personal information, the user's task processing efficiency, and the user's physical condition, etc. In this embodiment, the user database stores the schedule planning scheme and the user's own characteristics, and the user can customize and modify the schedule planning scheme based on his own status; further, the original schedule planning scheme stored in the user database can be updated based on the schedule planning scheme customized by the user, and at this time, the original user's own characteristics stored in the user database can be updated based on the difference between the schedule planning scheme customized by the user and the original schedule planning scheme.

[0075] Figure 4 It is a schematic flowchart of the artificial intelligence-based schedule planning method according to another embodiment of the present application, as Figure 4 shown, this method can further update the pre-trained language model and the user's own characteristics based on the custom modification results of the schedule planning schemes corresponding to multiple groups of users; exemplarily, a reference model can be determined based on the custom modification results of multiple groups of users to the schedule planning scheme, the model parameters of the pre-trained language model can be updated based on this reference model, and the user's own characteristics can be updated or optimized based on the residual results.

[0076] Correspondingly, the present invention also provides an artificial intelligence-based schedule planning system, as Figure 2 shown, the schedule planning system at least includes: an AI task decomposition module 10, a behavior analysis module 20, and an AI time allocation module 30.

[0077] Among them, the AI task decomposition module 10 obtains the to-be-planned task description input by the user, and obtains the user's subtask sequence based on the to-be-planned task description; the behavior analysis module 20 obtains the user's historical behavior data, the environmental information of the area where the user is located, and the user's schedule information, and determines the user's personalized prompt words based on the user's historical behavior data; the AI time allocation module 30 generates a schedule planning scheme corresponding to the to-be-planned task description based on the subtask sequence, historical behavior data, schedule information, environmental information, and the personalized prompt words, and the schedule planning scheme includes execution time information, execution location information, and resource allocation suggestion information.

[0078] In addition to the above, the schedule planning system may further include an AI review and analysis module 40. The AI review and analysis module 40 establishes a multi-dimensional evaluation index system by comparing the deviation between the actual execution data and the planning scheme, and combines it with the knowledge graph to automatically generate a visual analysis report containing improvement suggestions, and feeds back the review conclusions to the domain knowledge bases of the AI task decomposition module and the behavior analysis module 20.

[0079] As Figure 2 shown, the AI task decomposition module 10 may specifically include a domain knowledge graph construction unit 11 and a complexity evaluation unit 12. The domain knowledge graph construction unit 11 automatically retrieves technical literature and case libraries in related fields according to task keywords; the complexity evaluation unit 12 evaluates task complexity, required time and other attributes based on the knowledge graph, the retrieved results of technical literature in related fields, and the results of the user behavior analysis knowledge base. And the AI task decomposition module adopts a minimum segmentation algorithm to identify logical breakpoints in the task description based on the attention mechanism, and generates a logical subtask sequence.

[0080] The behavior analysis module 20 may specifically include a user portrait modeling unit 21, an environment sensitivity analysis unit 22, and a dynamic weight adjustment unit 23; the user portrait modeling unit 21 continuously records and analyzes the historical task execution efficiency data of users; the environment sensitivity analysis unit 22 establishes an association matrix between weather factors, geographical locations and task execution success rates; the dynamic weight adjustment unit 23 uses a reinforcement learning algorithm to update the priority of time planning parameters in real time.

[0081] The AI time allocation module may include a conflict detection subunit 31, a resource optimization subunit 32, and an elastic buffer setting unit 33; the conflict detection subunit 31 verifies the schedule feasibility based on the calculation result of the spatio-temporal overlap degree; the resource optimization subunit 32 automatically matches a preset resource configuration template according to information such as task type and the user behavior analysis knowledge base; the elastic buffer setting unit 33 dynamically allocates spare time periods based on task complexity.

[0082] The AI review and analysis module may include a three-layer data structure 41, a multi-dimensional comparison unit 42, a root cause analysis unit 43, and a knowledge update loop; the three-layer data structure 41 covers a data layer, a feature layer, and a knowledge layer. The data layer aggregates the original records of the user execution log database, the feature layer extracts multiple KPI indicators such as the user's time utilization efficiency, and the knowledge layer covers the knowledge graph that conforms to the user characteristics; the multi-dimensional comparison unit 42 provides various visualization schemes such as pie charts, bar charts, and Gantt chart comparison views of the planning scheme and the actual execution; the root cause analysis unit 43 uses a causal reasoning algorithm to identify the key influencing factors of execution deviation; the knowledge update loop automatically feeds back the review conclusions to the knowledge bases of the AI task decomposition module and the behavior analysis module 20.

[0083] In addition to the above, the pre-trained large language model in this application adopts a multi-task joint training architecture, which also has the following capabilities: a domain term recognition function that supports semantic parsing for professional tasks; a fuzzy requirement clarification function that clarifies task boundaries through interaction.

[0084] In addition, in order to further improve the efficiency and accuracy of schedule planning, the schedule planning method and system of this application can be optimized based on the following steps:

[0085] Based on the user historical data set, utilize the text classification and feature extraction capabilities of the large language model to construct a hybrid feature extraction model; for the historical behavior data of different users, design different prompt templates, and continuously optimize these prompt templates through the large language model.

[0086] Establish an adaptive learning closed-loop, a knowledge update mechanism including the following elements: store successful cases in the best practice library for reference in subsequent tasks; add failure modes to the risk mode recognition library for early warning in future tasks; adjust the time management ability evaluation value in the user profile to more accurately reflect the user's time management level, and feedback the review conclusion to the domain knowledge bases of the AI task decomposition module 10 and the behavior analysis module 20 to achieve continuous optimization of the system.

[0087] Figure 3 It is a schematic flowchart of an AI-based schedule planning method according to another embodiment of this application, as Figure 3 shown. This schedule planning method mainly includes the following steps: S1: The system receives the user's original task description; S2: The AI automatically generates an executable subtask sequence; S3: The AI dynamically generates a time planning scheme; S4: Real-time monitor the task execution status; S5: The AI generates a time management review report ( Figure 3 not shown in the figure).

[0088] In the step where the system receives the user's original task description, it can specifically include the following sub-steps: S1.1 (based on the multi-modal input interface), convert the user's oral task description into text data through the speech recognition unit, and at the same time receive the document attachments uploaded by the user through the graphical interface to form a multi-source input data pool; S1.2 (semantic standardization processing), use a bidirectional LSTM network to denoise the original input, including eliminating redundant information in colloquial expressions, unifying the naming norms of professional terms, and identifying and labeling the time sensitivity labels of task elements.

[0089] In the step of AI automatically generating an executable subtask sequence, the following specific subtasks can be included: S2.1 Knowledge-enhanced task decomposition, where a domain adapter is called to extract the following structured data from a pre-trained language model: a task decomposition paradigm library based on a domain knowledge graph, an industry classic workflow template library, and a decomposition case library of historical similar tasks; S2.2 Dynamic granularity regulation, where the optimal decomposition depth is determined through a mixed-integer programming algorithm, the matching degree between the user's professional background and the task domain is evaluated, and the distribution characteristics of available time resources are analyzed.

[0090] In the step of AI dynamically generating a time planning scheme, the following specific subtasks can be included: Analyze the user's biometric data to determine the best working time period; Integrate real-time data such as public transportation and weather to calculate the time tolerance; Evaluate the task correlation degree, optimize and batch process in combination with the knowledge graph.

[0091] In addition, when AI dynamically generates a time planning scheme, the following subtasks can also be included: S3.1 Multidimensional constraint modeling, S3.2 Rolling horizon optimization.

[0092] In S3.1 Multidimensional constraint modeling, a decision space containing but not limited to the following variables can be constructed:

[0093] Variable type Data source Weight coefficient User's work and rest pattern Historical data of multiple devices Adaptive adjustment Environmental factors Real-time data such as meteorology and traffic Dynamic weight Task coupling degree Sub-task dependency graph Determination of topological sorting

[0094] In S3.2 Rolling horizon optimization, a model predictive control (MPC) framework is used for iterative optimization, which is executed once every 6 hours: Obtain the latest task execution progress data; Recalculate the time cost of the remaining subtasks; Generate an N+1 version of the planning scheme.

[0095] In the step of real-time monitoring of the task execution status, the following specific subtasks can be included: S4.1 Multi-source perception data fusion, establishing a real-time monitoring matrix containing the following data streams: application usage log analysis, intelligent device sensor data collection, and progress updates actively fed back by users.

[0096] In the step of AI generating a time management review report, the following specific subtasks can be included: S5.1 Knowledge distillation-based report generation, where the review conclusions are refined through the following three-layer architecture: (1) Data layer: Aggregate the original records of the execution log database, (2) Feature layer: Extract multiple KPI indicators of time utilization efficiency, (3) Knowledge layer: Generate improvement suggestions that conform to the principles included in the knowledge base; S5.2 Adaptive learning closed-loop, establishing a knowledge update mechanism containing the following elements: Store successful cases in the best practice library, add failure modes to the risk mode recognition library, and adjust the time management ability evaluation value in the user profile.

[0097] The schedule planning method and schedule planning system of artificial intelligence in the above embodiments achieve the automatic decomposition of complex tasks through a pre-trained large language model and a domain knowledge graph, generate a logical sub-task sequence, and significantly improve the efficiency and accuracy of task decomposition. By combining user historical behavior data and real-time environment perception data, a dynamic optimal time planning scheme is generated to adapt to the needs of different users and changes in the external environment, improving the rationality and flexibility of time arrangement. Through a multi-dimensional evaluation index system and a causal reasoning algorithm, a visual analysis report containing improvement suggestions is generated to help users accurately identify problems in time management and provide optimization strategies. Through a knowledge update loop and an adaptive learning closed-loop, the task decomposition, scheduling, and review modules are continuously optimized to continuously improve the intelligent level of the system and the efficiency of user time management. That is, through the semantic understanding ability of the large model and the adaptive learning closed-loop, the present invention significantly improves the efficiency of complex task decomposition, the scientific nature of dynamic planning, and the continuous optimization ability of time management, and solves the technical defects of traditional tools relying on manual work and lacking intelligent analysis and adaptive adjustment.

[0098] Correspondingly, the present invention also provides a schedule planning device based on artificial intelligence. The device includes a processor, a memory, and a computer program stored on the memory. The processor is used to execute the computer program, and when the computer program is executed, the device implements the steps of the method described in any one of the above embodiments.

[0099] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the method described in any one of the above embodiments. The computer-readable storage medium may be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium well-known in the technical field.

[0100] Those of ordinary skill in the art should understand that the various exemplary components, systems, and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Specifically, whether to implement in hardware or software depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present invention are programs or code segments used to execute the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave on a transmission medium or a communication link.

[0101] It should be clear that the present invention is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present invention is not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present invention.

[0102] In the present invention, features described and / or exemplified for one embodiment can be used in the same way or in a similar way in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0103] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and modifications can be made to the embodiments of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An artificial intelligence-based schedule planning method, characterized in that, The method includes: Obtaining a to-be-planned task description input by the user, and obtaining a sub-task sequence of the user based on the to-be-planned task description; Obtaining the historical behavior data of the user, the environmental information of the region where the user is located, and the schedule information of the user, and determining a personalized prompt word of the user based on the historical behavior data of the user; Generating a schedule planning scheme corresponding to the to-be-planned task description based on the sub-task sequence, historical behavior data, schedule information, environmental information, and the personalized prompt word, where the schedule planning scheme includes execution time information, execution location information, and resource allocation recommendation information.

2. The artificial intelligence-based schedule planning method according to claim 1, wherein Obtaining the sub-task sequence of the user based on the to-be-planned task description, including: Inputting the to-be-planned task description into a pre-trained large language model, and extracting the long-distance dependency relationship between the words in the task description; Constructing a knowledge graph based on the long-distance dependency relationship and a structured domain knowledge base, where the nodes of the knowledge graph are the key entities in the task, and the edges of the knowledge graph are the relationships between the key entities; Based on the constructed knowledge graph, obtaining the sub-task sequence through a complexity evaluation algorithm and a minimization segmentation algorithm.

3. The artificial intelligence-based schedule planning method according to claim 2, wherein Inputting the to-be-planned task description into a pre-trained large language model, and extracting the long-distance dependency relationship between the words in the task description, including: The pre-trained large language model performs multi-layer bidirectional attention mechanism calculation, and calls a domain adapter to extract the structured data corresponding to the to-be-planned task description in combination with a task decomposition paradigm library of a domain knowledge graph, an industry classic workflow template library, and a decomposition case library of historical similar tasks, and extracts the long-distance dependency relationship between the words in the task description based on the structured data.

4. The artificial intelligence-based schedule planning method according to claim 1, characterized in that, Obtaining the historical behavior data of the user, the environmental information of the region where the user is located, and the schedule information of the user, including: Collecting user operation records in the system log, where the user operation records include application type, timestamp, and operation duration; Reading a calendar to extract the schedule information of the user, where the schedule information includes the time, location, and participants of the schedule activity; Obtaining task data of to-do tasks from a to-do application database, where the task data includes creation time, deadline, and completion status; Constructing a time consumption feature model based on the user operation records, schedule information, and task data of the to-do tasks; Obtaining environmental information sensed by a real-time environment perception module, where the environmental information includes weather data.

5. The artificial intelligence-based schedule planning method according to claim 4, wherein, Determining the personalized prompt word of the user based on the historical behavior data of the user, including: inputting the historical behavior data and the schedule information into a trained prompt word extraction model to obtain the personalized prompt word of the user; and / or, Generating a schedule planning scheme corresponding to the to-be-planned task description based on the sub-task sequence, historical behavior data, schedule information, environmental information, and the personalized prompt word, including: Generate a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm based on the sub-task sequence, historical behavior data, schedule information, environmental information, task data of the to-do task, and the personalized prompt words.

6. The artificial intelligence-based schedule planning method according to claim 5, wherein, Generate a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm based on the sub-task sequence, historical behavior data, schedule information, environmental information, task data of the to-do task, and the personalized prompt words, including: Determine the user's optimal working time period based on the user's biometric data; Calculate the time tolerance based on the public transportation data and weather data corresponding to each sub-task; Generate a schedule planning scheme corresponding to the to-be-planned task description through a constraint satisfaction algorithm in combination with the optimal working time period information and time tolerance information.

7. The artificial intelligence-based schedule planning method according to claim 1, wherein The method further includes: obtaining the actual execution data of some sub-tasks executed by the user, determining the plan planning deviation degree based on the actual execution data and the schedule planning scheme, and generating a visual analysis report including improvement suggestions based on the plan planning deviation degree; and / or, Obtain the user's own characteristics, store the schedule planning scheme and the user's own characteristics in the user database; obtain the custom modification values made by the user to the schedule planning scheme, and update the schedule planning scheme and the user's own characteristics in the user database based on the custom modification values.

8. An artificial intelligence-based schedule planning system, characterized in that, The schedule planning system includes: An AI task decomposition module, which obtains the to-be-planned task description input by the user and obtains the user's sub-task sequence based on the to-be-planned task description; A behavior analysis module, which obtains the user's historical behavior data, the environmental information of the area where the user is located, and the user's schedule information, and determines the user's personalized prompt words based on the user's historical behavior data; An AI time allocation module, which generates a schedule planning scheme corresponding to the to-be-planned task description based on the sub-task sequence, historical behavior data, schedule information, environmental information, and the personalized prompt words, and the schedule planning scheme includes execution time information, execution location information, and resource allocation suggestion information.

9. An artificial intelligence-based schedule planning device, the device comprising a processor, a memory, and a computer program stored on the memory, characterized in that, The processor is used to execute the computer program, and when the computer program is executed, the device implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images