To-do generation method and system, storage medium and program product

Abstract

The invention relates to the technical field of information processing, and discloses a to-do list generation method and system, a storage medium and a program product.The to-do list can be automatically recognized and generated by conducting recognition and fuzzy processing on an initial text data set through a preset large language model, so that the burden of a user is relieved, and meanwhile the user experience is improved. Blurred tasks can be processed, so that the accuracy of generation of backlogs is improved. Furthermore, importance judgment is carried out through the preset large language model, and the to-do task relevance data set of the to-do tasks is determined, so that the importance and urgency of the tasks can be automatically judged, the user is helped to preferentially process the important and urgent tasks, meanwhile, the relevance between the tasks can be understood, and the to-do tasks can be better managed. And finally, in combination with the first to-do task data set and the to-do task relevance data set, a plurality of target to-do items corresponding to the plurality of final to-do tasks can be directly generated.

Description

Technical Field

[0001] The present invention relates to the field of information processing technology, and in particular to a to-do list generation method, system, storage medium and program product. Background Art

[0002] In office communication software, users often need to record the content of to-do tasks, and traditional to-do list generation methods have the following problems: they rely on manual input and management by users, which increases the workload of users; they are usually unable to automatically identify the importance and urgency of tasks. Users need to manually set priorities, which may cause some important tasks to be neglected; they can usually only handle single tasks, but cannot understand the relationship between tasks. For example, if one task is a prerequisite for another task, traditional tools may not be able to recognize this relationship; they cannot correctly understand and process the ambiguous language used by users when describing to-do items. Summary of the invention

[0003] In view of this, the present invention provides a to-do list generation method, system, storage medium and program product to solve the problems of traditional to-do list generation methods such as inability to automatically identify and generate to-do lists, difficulty in identifying importance, lack of understanding of task relevance, and difficulty in handling ambiguous tasks.

[0004] In a first aspect, the present invention provides a to-do list generation method, which is used in a to-do list generation system, wherein the system is connected to a user terminal; the method comprises:

[0005] An initial text data set of multiple to-do tasks input by a user end is obtained; the initial text data set is recognized and fuzzy processed using a preset large language model to obtain a first to-do task data set; based on the first to-do task data set, an importance judgment is made using a preset large language model and a to-do task correlation data set of multiple to-do tasks is determined; based on the first to-do task data set and the to-do task correlation data set, multiple target to-do items of the multiple to-do tasks are generated.

[0006] The to-do list generation method provided by the present invention can automatically identify and generate to-do lists by presetting a large language model to identify and fuzzify the initial text data set, thereby reducing the burden on users. At the same time, it can handle ambiguous tasks, thereby improving the accuracy of to-do list generation. Further, by presetting a large language model to judge the importance and determine the to-do task association data set of multiple to-do tasks, the importance and urgency of the tasks can be automatically judged, thereby helping users to prioritize important and urgent tasks. At the same time, the association between tasks can be understood, thereby better managing to-do lists. Finally, combining the first to-do task data set and the to-do task association data set, multiple target to-do lists corresponding to the final multiple to-do tasks can be directly generated. Therefore, by implementing the present invention, the problems of the traditional to-do list generation method that cannot automatically identify and generate to-do items, it is difficult to identify importance, lacks task association understanding, and is difficult to handle ambiguous tasks are solved.

[0007] In an optional implementation, the initial text data set is identified and fuzzy processed using a preset large language model to obtain a first to-do task data set, including:

[0008] The initial text data set is subjected to task recognition using a preset large language model to obtain a second to-be-done task data set; based on the second to-be-done task data set, fuzzy task processing is performed using a preset large language model to obtain a first to-be-done task data set.

[0009] The to-do list generation method provided by the present invention can automatically and accurately extract relevant content that truly belongs to the to-do tasks from the initial text data set input by the user by using a preset large language model for task recognition, thereby providing support for the automatic generation of subsequent to-do lists. Furthermore, the preset large language model is used again for fuzzy task processing, which can process ambiguous tasks and provide support for improving the accuracy of subsequent to-do list generation.

[0010] In an optional implementation, the initial text data set is subjected to task identification using a preset large language model to obtain a second to-be-done task data set, including:

[0011] Preprocess the initial text data set to obtain a target text data set; based on the first preset Prompt template, use the preset large language model to perform task identification on the target text data set to obtain a third to-do task data set; input the third to-do task data set into the preset large language model for processing to obtain a fourth to-do task data set; parse the fourth to-do task data set to obtain a fifth to-do task data set; and determine the second to-do task data set based on the fifth to-do task data set and a user identity.

[0012] The to-do item generation method provided by the present invention can ensure the consistency and processability of the text, reduce the noise and irregularity of the text, and thus improve the quality of the target text data set by preprocessing the initial text data set. Furthermore, with the help of the first preset Prompt template, the preset large language model can be guided to perform task recognition on the target text data set, thereby improving recognition efficiency and accuracy. Furthermore, the preset large language model is continued to be used to process the third to-do task data set obtained by recognition, and the processed fourth to-do task data set is parsed and processed, so that the data format of the fifth to-do task data set can be guaranteed to meet the requirements. Finally, the final second to-do task data set can be obtained in combination with the user identity, which realizes the high personalization of the to-do tasks and improves the user experience and task execution efficiency. Therefore, by implementing the present invention, to-do items can be automatically identified and generated, thereby reducing the burden of users recording to-do work tasks in office communication software.

[0013] In an optional implementation, based on the second to-do task data set, fuzzy task processing is performed using a preset large language model to obtain a first to-do task data set, including:

[0014] Using the preset large language model, determine whether there is a fuzzy time description in the second to-do task data set; when there is a fuzzy time description, obtain the real-time time data and the fuzzy time description data set; based on the second preset Prompt template, input the real-time time data and the fuzzy time description data set into the preset large language model to obtain the task deadline time data set; parse and process the task deadline time data set to obtain the first to-do task data set.

[0015] The to-do list generation method provided by the present invention can quickly and accurately screen out the unclear time description parts in the second to-do task data set that has been initially extracted by judging whether there is a fuzzy time description. Furthermore, if there is a fuzzy time description, then based on the acquired real-time time data, with the help of a specific Prompt template, the large language model can be guided to convert the fuzzy task time into a clear and accurate task deadline according to the real-time time and the specific fuzzy time description. Finally, by parsing the task deadline data set, the first to-do task data set after the fuzzy task processing can be obtained. Therefore, by implementing the present invention, ambiguous tasks can be processed, thereby improving the accuracy of to-do list generation.

[0016] In an optional implementation, based on the first to-do task dataset, a preset large language model is used to perform importance judgment and determine a to-do task correlation dataset of multiple to-do tasks, including:

[0017] Based on the first to-do task data set, importance judgment is performed using a preset large language model to obtain multiple target score values ​​for multiple to-do tasks, and the target score values ​​reflect the importance of the to-do tasks; based on the multiple target score values ​​and the first to-do task data set, the preset large language model is used to understand the correlation between the multiple to-do tasks to obtain a to-do task correlation data set.

[0018] The to-do list generation method provided by the present invention can automatically determine the importance and urgency of a task by using a preset large language model. Furthermore, by combining multiple target score values ​​of multiple to-do tasks, the preset large language model can be continued to be used to automatically identify the association between tasks, thereby better managing the to-do list.

[0019] In an optional implementation, based on the first to-do task data set, a preset large language model is used to perform importance judgment to obtain multiple target score values ​​of multiple to-do tasks, including:

[0020] Key information is extracted from the first to-do task data set to obtain a key information data set; based on the key information data set, multiple task deadline urgency score values, multiple priority score values, and multiple keyword score values ​​of the multiple to-do tasks are calculated respectively; based on the multiple task deadline urgency score values, multiple priority score values, and multiple keyword score values, multiple target score values ​​of the multiple to-do tasks are calculated.

[0021] The to-do list generation method provided by the present invention comprehensively considers the urgency of the deadline, priority and keyword score of the task, can comprehensively judge the importance of the task, and facilitates task management and sorting.

[0022] In an optional implementation, based on multiple target score values ​​and the first to-do task data set, a preset large language model is used to understand the relevance of multiple to-do tasks to obtain a to-do task relevance data set, including:

[0023] Obtain a preset association relationship set; input the preset association relationship set, multiple target score values ​​and the first to-do task data set into a preset large language model to obtain a to-do task association data set.

[0024] The to-do list generation method provided by the present invention establishes a standard reference framework for subsequent determination of the relevance between to-do tasks by obtaining a preset association relationship set. Furthermore, by combining multiple target score values ​​and the first to-do task data set and using a preset large language model for association understanding, the association relationship between multiple to-do tasks can be accurately determined in multiple dimensions.

[0025] In a second aspect, the present invention provides a to-do list generation system, connected to a user terminal, and used to execute the to-do list generation method of the first aspect or any corresponding embodiment thereof; the system comprises: an input interface, an output interface, a large model processing unit and a to-do list generation unit, the large model processing unit is connected to the user terminal via the input interface, the large model processing unit is also connected to the to-do list generation unit, and the to-do list generation unit is also connected to the user terminal via the output interface.

[0026] The to-do list generation system provided by the present invention solves the problems existing in traditional to-do list generation tools, such as the inability to automatically identify and generate to-do lists, difficulty in identifying importance, lack of understanding of task relevance, and difficulty in handling ambiguous tasks, by executing the to-do list generation method of the above-mentioned first aspect or any corresponding embodiment.

[0027] In a third aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the to-do list generation method of the first aspect or any corresponding embodiment thereof.

[0028] In a fourth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the to-do list generating method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0030] Figure 1 is a structural block diagram of a to-do list generation system according to an embodiment of the present invention;

[0031] Figure 2 is a flowchart of a method for generating a to-do list according to an embodiment of the present invention;

[0032] Figure 3 is a flowchart of another to-do list generation method according to an embodiment of the present invention;

[0033] Figure 4 is a schematic diagram of generating structured output content by LLM+Langchain according to an embodiment of the present invention;

[0034] Figure 5 is a schematic diagram of a fuzzy task processing flow according to an embodiment of the present invention;

[0035] Figure 6 is a flowchart of another to-do list generation method according to an embodiment of the present invention;

[0036] Figure 7 is a schematic diagram of understanding task relevance according to an embodiment of the present invention;

[0037] Figure 8 is an example diagram of task relevance relationships according to an embodiment of the present invention;

[0038] Fig. 9 is a flowchart of a to-do item generation system based on a large model according to an embodiment of the present invention;

[0039] Fig.10 is a schematic diagram of the module structure of a large model processing unit according to an embodiment of the present invention;

[0040] Fig.11 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0042] The embodiment of the present invention provides a to-do list generation method, which achieves the effects of automatically identifying and generating to-do lists, automatically judging the importance and urgency of tasks, understanding the relevance between tasks, and handling ambiguous tasks by presetting a large language model.

[0043] According to an embodiment of the present invention, an embodiment of a to-do list generating method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0044] In this embodiment, a method for generating a to-do list is provided, which can be used for example Figure 1 The to-do list generating system 1 is shown, and the to-do list generating system 1 is connected to the user terminal 2 .

[0045] Figure 2 is a flowchart of a method for generating a to-do list according to an embodiment of the present invention. Figure 2As shown, the process includes the following steps:

[0046] Step S201, obtaining an initial text data set of a plurality of to-do tasks input by a user.

[0047] The user terminal 2 provides a user-oriented user interface, and the user terminal allows the user to chat or send notifications on the office communication software.

[0048] The initial text data set may include data such as description content of to-do tasks, time-related description content, location information-related content, task priority-related content, etc.

[0049] Specifically, the user can input the descriptions of the tasks to be completed one by one in the user interface provided by the user terminal 2 according to the set input method (for example, some are in the form of text boxes, and the user directly types; some may support voice input, and the user orally dictates the tasks to be done, and the system then converts the voice into text, etc.). For example, the user may input multiple task contents such as "go to the supermarket to buy daily necessities in the morning", "organize work documents in the afternoon", and "send project progress emails to customers in the evening".

[0050] Furthermore, the user terminal 2 may integrate the input to-do tasks in a certain order or format and form an initial text data set of the corresponding multiple to-do tasks.

[0051] Furthermore, the user terminal 2 may transmit the obtained initial text data set to the to-do item generation system 1 .

[0052] Step S202: using a preset large language model to perform recognition and fuzzy processing on the initial text data set to obtain a first to-do task data set.

[0053] Among them, the preset Large Language Model (LLM) is an artificial intelligence model that can recognize, summarize, translate, predict and generate text and other forms of content based on knowledge gained from large data sets. These models are artificial neural networks that can contain tens of millions or even billions of weights and are (pre) trained using self-supervised learning and semi-supervised learning.

[0054] Specifically, the preset large language model can be used to deeply understand the initial text data set, including word meaning, sentence meaning, contextual relationship, etc. Furthermore, after understanding the input natural language content, the corresponding to-do tasks can be automatically identified, including task name, task description, deadline, etc.

[0055] Furthermore, the preset large language model can also be used to process ambiguous tasks in multiple to-do tasks. For example, if a user uses ambiguous language when describing a to-do item, such as "I may need to complete this report sometime next week", then this part needs to be able to understand this ambiguous description and generate the corresponding to-do item.

[0056] Step S203: Based on the first to-do task data set, a preset large language model is used to perform importance judgment and determine to-do task correlation data sets of multiple to-do tasks.

[0057] Specifically, by presetting a large language model to judge the importance of to-do tasks, factors such as the urgency and impact scope of each to-do task can be obtained. Furthermore, combined with the importance judgment results, the correlation between multiple to-do tasks can be determined and a corresponding to-do task correlation data set can be formed.

[0058] Step S204: generating a plurality of target to-do items of a plurality of to-do tasks based on the first to-do task data set and the to-do task association data set.

[0059] The target to-do items may include information such as the name of the to-do task, task description, deadline, and importance of the task.

[0060] Specifically, by combining the first to-do task data set and the to-do task relevance data set, a target to-do item of each to-do task may be generated.

[0061] Furthermore, multiple to-do items can be stored in a database for subsequent display and management.

[0062] Furthermore, multiple to-do items may be sent to the corresponding user terminal 2 so that users can better manage their work and tasks.

[0063] The to-do list generation method provided in this embodiment can automatically identify and generate to-do lists by presetting a large language model to identify and fuzzify the initial text data set, thereby reducing the burden on users. At the same time, it can handle ambiguous tasks, thereby improving the accuracy of to-do list generation. Further, by presetting a large language model to judge the importance and determine the to-do task association data set of multiple to-do tasks, the importance and urgency of the tasks can be automatically judged, thereby helping users to prioritize important and urgent tasks. At the same time, the association between tasks can be understood, thereby better managing to-do lists. Finally, combining the first to-do task data set and the to-do task association data set, multiple target to-do lists corresponding to the final multiple to-do tasks can be directly generated. Therefore, by implementing the present invention, the problems of the traditional to-do list generation method that cannot automatically identify and generate to-do items, is difficult to identify importance, lacks task association understanding, and is difficult to handle ambiguous tasks are solved.

[0064] In this embodiment, a method for generating a to-do list is provided, which can be used for example Figure 1 The to-do list generating system 1 is shown, and the to-do list generating system 1 is connected to the user terminal 2 .

[0065] Figure 3 is a flowchart of a method for generating a to-do list according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:

[0066] Step S301, obtaining an initial text dataset of multiple to-do tasks input by the user. Figure 2 Step S201 of the illustrated embodiment will not be described in detail here.

[0067] Step S302: using a preset large language model to perform recognition and fuzzy processing on the initial text data set to obtain a first to-do task data set.

[0068] Specifically, the above step S302 includes:

[0069] Step S3021, using a preset large language model to perform task identification on the initial text data set to obtain a second to-be-done task data set.

[0070] Specifically, a preset large language model can be used to deeply understand the input initial text data set, identify key information such as task name, task description and deadline, and form a corresponding second to-do task data set.

[0071] In some optional implementations, the above step S3021 includes:

[0072] Step a1: preprocess the initial text data set to obtain the target text data set.

[0073] Step a2: Based on the first preset Prompt template, a preset large language model is used to perform task identification on the target text dataset to obtain a third to-do task dataset.

[0074] Step a3: input the third to-be-done task data set into a preset large language model for processing to obtain a fourth to-be-done task data set.

[0075] Step a4, parsing the fourth to-be-done task data set to obtain a fifth to-be-done task data set.

[0076] Step a5: determining a second to-do task data set based on the fifth to-do task data set and the user identity.

[0077] Specifically, the initial text dataset is usually a natural language text, which may contain multiple sentences or paragraphs. Therefore, the received text can be segmented into separate sentences or paragraphs. Further, the text is cleaned and standardized to ensure the consistency and processability of the text, which may include removing special characters, converting text to lowercase, processing abbreviations, etc., to reduce the noise and irregularity of the text.

[0078] Furthermore, in order to guide the LLM to perform task identification, it is necessary to construct a suitable prompt, namely the first preset prompt template, wherein the prompt identifies a text or question that can guide the model to perform a specific task.

[0079] In this embodiment, the prompt for task extraction is as follows:

[0080] "Given a text document, identify and extract the tasks that need to bedone.

[0081] #Given a text document, identify and extract the tasks that need to be completed;

[0082] Each task should be represented as an object with the following structure:

[0083] #Each task should be represented as an object with a specific structure:

[0084] ```json

[0085] {

[0086] "task_name":"The name of the task",

[0087] #Define a key named "task_name", whose corresponding value is a description of the task name;

[0088] "task_description":"A brief description of the task",

[0089] #Define a key named "task_description", whose corresponding value is a brief description of the task;

[0090] "due_time":"The deadline for the task"

[0091] #Define a key named "due_time", whose corresponding value is the deadline for the task;

[0092] }

[0093] ```

[0094] Please note that the'task_name'is a brief title of the task,'task_description'provides more details about what the task involves,and'due_time'indicates when the task needs to be completed by."

[0095] #Explain the meaning of each key in the previous JSON structure: "task_name" is a short title of the task, "task_description" provides more details about what the task involves, and "due_time" indicates the time when the task needs to be completed.

[0096] Furthermore, the above prompt can be used to guide the preset large language model to focus on task recognition, and output the result in a structured manner in JSON format, that is, the third to-do task data set is in JSON format.

[0097] Furthermore, since LLM is essentially a probabilistic model with ultra-large-scale parameters, its content generation process is to predict the next output token based on the current text, and the token is selected from a certain probability distribution. Therefore, the output of the model is unstable, and even using the Prompt technology cannot guarantee that the model can generate content in the specified format.

[0098] Therefore, in order to solve the above problems, this embodiment combines the open source LLM framework Langchain to build a structured content extraction pipeline, such as Figure 4 As shown, first, the third to-do task data set is input into the preset large language model for processing to obtain a preliminary fourth to-do task data set. Further, the JsonOutputParser (Json output parser) module is used to extract structured JSON data from the fourth to-do task data set to form a fifth to-do task data set.

[0099] Among them, the JsonOutputParser module can be implemented based on specific rules and regular expressions; Figure 4 The Prompt Engineering module is used to perform step a2 above.

[0100] Furthermore, after extracting relevant information of the to-do tasks from the fifth to-do task dataset and concatenating the user id to generate task_id, a corresponding second to-do task dataset containing user id, task id, task name, task description and deadline can be generated.

[0101] In an optional implementation, the output example is as follows:

[0102] {

[0103] "user_id":1,

[0104] "task_id":1,

[0105] "task_name":"Complete Report",

[0106] "task_description":"Complete next week's report and submit",

[0107] "due_time":"2023-09-30"

[0108] }

[0109] Among them, user_id represents the user id; task_id represents the task id; task_name represents the task name; task_description represents the task description.

[0110] Step S3022: Based on the second to-do task data set, fuzzy task processing is performed using a preset large language model to obtain a first to-do task data set.

[0111] Specifically, fuzzy task processing can be performed by presetting a large language model, and the ambiguous task descriptions in the second to-do task data set can be processed. In this embodiment, the task deadline is mainly targeted. If the user uses ambiguous language in the notification, such as "complete the report sometime next week", this part will be understood and converted into a specific deadline or time range.

[0112] In some optional implementations, the above step S3022 includes:

[0113] Step b1, using a preset large language model, determining whether there is a fuzzy time description in the second to-do task data set.

[0114] Step b2: when there is fuzzy time description, obtain real-time time data and fuzzy time description data set.

[0115] Step b3: Based on the second preset Prompt template, the real-time data and the fuzzy time description data set are input into the preset large language model to obtain the task deadline data set.

[0116] Step b4: parse and process the task deadline data set to obtain a first to-be-done task data set.

[0117] Specifically, Figure 5 As shown, firstly, the second to-do task data set is input into the preset large language model, and it is determined whether there is a fuzzy time description in the second to-do task data set: if not, the deadline is directly converted into the standard %Y-%m-%d %H:%M:%S time format and the fuzzy task processing flow is jumped out; if it exists, the corresponding fuzzy time description data set is extracted, which can be achieved by the following prompt:

[0118] "Given a task with a'due_time'field,determine whether the'due_time'contains a vague time description.If a vague time description exists,extractand output it.Avague time description could be phrases like'end of day','nextweek','in a few days',etc."

[0119] #Given a task with a 'due_time' field, determine whether 'due_time' contains a fuzzy time description. If a fuzzy time description exists, extract and output it. The fuzzy time description may be a phrase such as 'at the end of the day', 'next week', 'a few days later', etc.;

[0120] Please note that the task should be represented as an object with the following structure:

[0121] #Please note that tasks should be represented by the following structure;

[0122] ```json

[0123] {

[0124] "user_id":1,

[0125] "task_id":1,

[0126] "task_name":"The name of the task",

[0127] "task_description":"A brief description of the task",

[0128] "due_time":"The deadline for the task"

[0129] }

[0130] ```

[0131] The'task_name'is a brief title of the task,'task_description'provides more details about what the task involves,and'due_time'indicates when thetask needs to be completed by.

[0132] Furthermore, the current time, that is, real-time time data, can be obtained by calling a Python tool script and the real-time time data can be used as a basis for calculating the task deadline.

[0133] Furthermore, the real-time time data and the fuzzy time description data set are input into the preset large language model, and under the guidance of the second preset Prompt template, the specific task deadline can be calculated.

[0134] Among them, the second preset Prompt template is as follows:

[0135] "Given a vague time description and the current time, calculate the precise deadline for the task and output it in the format '%Y-%m-%d%H:%M:%S'.

[0136] #Given a fuzzy time description and the current time, calculate the precise deadline of the task and output the result in the format of 'year-month-day hour:minute:second' (i.e. '%Y-%m-%d %H:%M:%S');

[0137] vague_time_description:{{vague_time_description}}

[0138] Current_time:{{Current_time}}

[0139] "

[0140] Among them, {vague_time_description} indicates that this should be replaced by the actual fuzzy time description content, which is used to indicate to LLM where the corresponding fuzzy time information needs to be filled in. It is a convenient way to embed corresponding data in specific use.

[0141] {Current_time} represents a placeholder setting for filling in the actual current time data, which can be used to accurately pass the actual current time information to LLM, and together with the fuzzy time description, allow LLM to perform corresponding time calculation operations.

[0142] Furthermore, by parsing the task deadline dataset output by the preset large language model, the fuzzy time description can be replaced with the calculated accurate task deadline.

[0143] In an alternative embodiment, the output example is as follows:

[0144] {

[0145] "user_id":1,

[0146] "task_id":1,

[0147] "task_name":"Complete Report",

[0148] "task_description":"Complete next week's report and submit",

[0149] "due_time":"2023-09-30 00:00:00"

[0150] }

[0151] Furthermore, through the above processing, ambiguous tasks can be handled, thereby improving the accuracy of to-do list generation.

[0152] Step S303: Based on the first to-do task dataset, a preset large language model is used to perform importance judgment and determine the to-do task correlation datasets of multiple to-do tasks. Figure 2 Step S203 of the illustrated embodiment will not be described in detail here.

[0153] Step S304: Generate multiple target to-do items of multiple to-do tasks based on the first to-do task dataset and the to-do task association dataset. Figure 2 Step S204 of the illustrated embodiment will not be described in detail here.

[0154] The to-do list generation method provided in this embodiment can ensure the consistency and processability of the text, reduce the noise and irregularity of the text, and thus improve the quality of the target text data set by preprocessing the initial text data set. Further, the preset large language model can be guided to perform task recognition on the target text data set with the help of the first preset Prompt template, thereby improving the recognition efficiency and accuracy. Further, the preset large language model is continued to be used to process the third to-do task data set obtained by recognition, and the processed fourth to-do task data set is parsed and processed, so that the data format of the fifth to-do task data set can be guaranteed to meet the requirements. Finally, the final second to-do task data set can be obtained in combination with the user identity. Further, by judging whether there is a fuzzy time description, the part about the unclear time description can be quickly and accurately screened out in the second to-do task data set that has been initially extracted. Further, if there is a fuzzy time description, based on the acquired real-time time data, with the help of a specific Prompt template, the large language model can be guided to convert the fuzzy task time into a clear and accurate task deadline according to the real-time time and the specific fuzzy time description. Finally, by parsing the task deadline data set, the first to-do task data set after fuzzy task processing can be obtained. Therefore, by implementing the present invention, to-do items can be automatically identified and generated, thereby reducing the burden of users recording to-do tasks in office communication software. Further, ambiguous tasks can be processed, thereby improving the accuracy of to-do item generation.

[0155] In this embodiment, a method for generating a to-do list is provided, which can be used for example Figure 1 The to-do list generating system 1 is shown, and the to-do list generating system 1 is connected to the user terminal 2 .

[0156] Figure 6 is a flowchart of a method for generating a to-do list according to an embodiment of the present invention. Figure 6 As shown, the process includes the following steps:

[0157] Step S601, obtaining an initial text dataset of multiple to-do tasks input by the user. Figure 2 Step S201 of the illustrated embodiment will not be described in detail here.

[0158] Step S602: Use a preset large language model to identify and fuzzy process the initial text data set to obtain a first to-do task data set. Figure 3 Step S302 of the illustrated embodiment will not be described in detail here.

[0159] Step S603: Based on the first to-do task data set, a preset large language model is used to perform importance judgment and determine to-do task correlation data sets of multiple to-do tasks.

[0160] Specifically, the above step S603 includes:

[0161] Step S6031: Based on the first to-do task data set, importance judgment is performed using a preset large language model to obtain multiple target score values ​​for multiple to-do tasks.

[0162] Among them, the target score value reflects the importance of the task to be done.

[0163] Specifically, a preset large language model may be used to analyze factors such as content, urgency, and context of multiple to-do tasks corresponding to the first to-do task data set and determine the importance score of each to-do task.

[0164] Among them, the urgency of the task may be judged based on the deadline, while the importance of the task may be related to the keywords in the task description.

[0165] In some optional implementations, the above step S6031 includes:

[0166] Step c1, extracting key information from the first to-do task data set to obtain a key information data set.

[0167] Step c2, based on the key information data set, respectively calculating a plurality of task deadline urgency score values, a plurality of priority score values, and a plurality of keyword score values ​​of a plurality of to-do tasks.

[0168] Step c3, calculating multiple target score values ​​of the multiple to-do tasks based on the multiple task deadline urgency score values, the multiple priority score values ​​and the multiple keyword score values.

[0169] Specifically, by extracting key information from the first to-do task dataset, a key information dataset including task deadlines, task names, keywords and phrases in task descriptions, and other information can be obtained.

[0170] Furthermore, the urgency score of the task deadline can be calculated based on the time interval between the deadline and the current time. The closer the deadline is, the higher the score is. This can be done by the following steps:

[0171] (1) Obtain the deadline for the task;

[0172] (2) Calculate the time interval between the current time and the deadline;

[0173] (3) Use a function to map the time interval to the task deadline urgency score value: Task deadline urgency score value = max(0, 1-(time interval / maximum allowed time interval))

[0174] The maximum allowed time interval represents a time limit or deadline threshold set in task management or time management, and can be used to indicate how long a task is considered urgent or needs to be completed as soon as possible.

[0175] Furthermore, if the user has assigned a priority to the task, the priority score value may be determined directly according to the priority specified by the user.

[0176] Further, the task description may be analyzed for keywords or phrases, such as "urgent", "important", and corresponding keyword score values ​​may be determined: if these keywords are present, a higher score is assigned.

[0177] Furthermore, the above-calculated scores can be comprehensively calculated to obtain a comprehensive importance score by comprehensively evaluating the importance of the task. In this embodiment, this is achieved by weighted summation:

[0178] Target score value = w1*deadline urgency score + w2*user-specified priority + w3*keyword score

[0179] Among them, w1, w2, and w3 represent weight factors, which can be adjusted according to specific needs to adjust the importance of different factors.

[0180] Furthermore, each to-do task can be marked as high, medium, low importance or other corresponding levels according to the target score value, so that the task management system can sort the tasks and remind the user accordingly.

[0181] In an optional embodiment, the output example is as follows:

[0182] {

[0183] "user_id":1,

[0184] "task_id":1,

[0185] "task_name":"Complete Report",

[0186] "task_description":"Complete next week's report and submit",

[0187] "timestamp":"2023-09-30",

[0188] "task_importance":"high"

[0189] }

[0190] Furthermore, by comprehensively considering the urgency of the task deadline, priority, and keyword score, the importance of the task can be comprehensively judged, which facilitates task management and sorting.

[0191] Step S6032: Based on the multiple target score values ​​and the first to-do task data set, the preset large language model is used to understand the relevance of the multiple to-do tasks to obtain a to-do task relevance data set.

[0192] Specifically, the preset large language model can be used to analyze the keywords, dependencies and other information between the to-do tasks and understand the relevance between the tasks. For example, if one task is a prerequisite for another task, the preset large language model can be used to identify and establish the relationship between the two tasks.

[0193] In some optional implementations, the above step S6032 includes:

[0194] Step d1, obtaining a preset association relationship set.

[0195] Step d2: inputting a preset association relationship set, a plurality of target score values ​​and a first to-be-done task data set into a preset large language model to obtain a to-be-done task association data set.

[0196] The preset association relationship set may include the following three association relationships:

[0197] (1) Dependency: Identify whether one task is a prerequisite for another task. For example, if task A is a prerequisite for task B, the system needs to understand this dependency.

[0198] (2) Parallel: Identifies the parallel relationship between two or more tasks, indicating that they can be performed simultaneously without depending on each other.

[0199] (3) Sequential: Identifies the sequential relationship between tasks, indicating that one task must be executed after another task.

[0200] Furthermore, the input of relevance understanding is the output of importance judgment, and the output is a to-do task association graph, where the directed edges of the graph represent the dependency relationships between tasks.

[0201] Specifically, Figure 7 As shown, the data output by the importance judgment, i.e., multiple target score values, can be converted into nodes of a relationship graph, and the above preset association relationship set and the generated nodes are input into the preset large language model together. The Prompt used is as follows:

[0202] ```

[0203] system:

[0204] Your task is to determine the relationship between given task nodes and add the "inputs" field to each task node. The relationships can be one of the following: "Dependency", "Parallel", "Sequential", or "None".

[0205] #Your task is to determine the relationship between given task nodes and add an "inputs" field to each task node. The relationship can be one of the following: "Dependency", "Parallel", "Sequential" or "None". The "inputs" field should list the task nodes that the current node depends on, is parallel to it, follows sequentially, or has no relationship with it;

[0206] user:

[0207] The possible values ​​for the "relationship" field include "Dependency", "Parallel", "Sequential", and "None". Please provide the correct relationship for each task node and specify the associated task nodes in the "inputs" fieldaccordingly.

[0208] #Specific content of user request:

[0209] The possible values ​​of the "relationship" field include "Dependency", "Parallel", "Sequential" and "None";

[0210] Please provide the correct relationship for each task node and specify the related task nodes accordingly in the "inputs" field;

[0211] Here is an example for your reference:

[0212] #This is a hint, indicating that an example will be given for reference next.

[0213] Task Nodes:

[0214] # List task nodes

[0215] 1.Task Name:Task A

[0216] 2.Task Name:Task B

[0217] 3.Task Name:Task C

[0218] 4.Task Name:Task D

[0219] Relationships:

[0220] #List the relationships between task nodes

[0221] 1.Relationship:Dependency

[0222] Task:Task A

[0223] Inputs:[Task B,Task C]

[0224] 2.Relationship:Parallel

[0225] Task:Task B

[0226] Inputs:[Task C]

[0227] 3.Relationship:Sequential

[0228] Task:Task C

[0229] Inputs:[Task D]

[0230] 4.Relationship:None

[0231] Task:Task D

[0232] Inputs:[]

[0233] Please classify the relationships between the given task nodes andadd the "inputs"field to each task node accordingly.

[0234] # To clarify the task again, it is required to classify the relationship between the given task nodes and add the "inputs" field to each task node accordingly

[0235] ```

[0236] Furthermore, the dependency relationship of each task node can be obtained according to the output of the preset large language model, and a task correlation relationship graph can be constructed.

[0237] In an optional embodiment, the task correlation diagram is as follows: Figure 8 shown.

[0238] Furthermore, the corresponding to-do task correlation data set can be obtained by combining the task correlation relationship diagram.

[0239] Step S604: Generate multiple target to-do items of multiple to-do tasks based on the first to-do task dataset and the to-do task association dataset. Figure 2 Step S204 of the illustrated embodiment will not be described in detail here.

[0240] The to-do list generation method provided in this embodiment comprehensively considers the urgency of the deadline, priority and keyword score of the task, and can comprehensively judge the importance of the task, which is convenient for task management and sorting. Furthermore, by obtaining a preset association relationship set, a standard reference framework is established for the subsequent judgment of the association between the to-do tasks. Furthermore, by combining multiple target score values ​​and the first to-do task data set, the preset large language model is used to understand the association, and the association between multiple to-do tasks can be accurately determined in multiple dimensions.

[0241] In this embodiment, a to-do list generation system is provided. Figure 1 As shown, the to-do list generating system 1 is connected to the user terminal 2 and is used to execute the to-do list generating method provided by the above embodiment of the present invention.

[0242] Specifically, the to-do list generating system 1 includes an input interface 11 , an output interface 12 , a large model processing unit 13 and a to-do list generating unit 14 .

[0243] The large model processing unit 13 is connected to the user terminal 2 via the input interface 11 , and is also connected to the to-do list generating unit 14 , and the to-do list generating unit 14 is also connected to the user terminal 2 via the output interface 12 .

[0244] Specifically, the user terminal 2 can input the initial text data set to the large model processing unit 13 through the input interface 11 .

[0245] Furthermore, the large model processing unit 13 can perform task identification, fuzzy task processing, importance judgment and relevance understanding on the received large model processing unit 13 in sequence, and output the processing results to the to-do list generation unit 14.

[0246] Furthermore, the to-do item generating unit 14 may generate corresponding target to-do items according to the output content received from the large model processing unit 13 .

[0247] Furthermore, the to-do item generating unit 14 may send the generated target to-do item to the user terminal 2 through the output interface 12 to be displayed to the corresponding user.

[0248] The specific processing process may refer to the process description of the to-do list generation method provided in the above embodiment of the present invention, which will not be repeated here.

[0249] The to-do list generation system provided in this embodiment solves the problems existing in traditional to-do list generation tools, such as the inability to automatically identify and generate to-do lists, difficulty in identifying importance, lack of understanding of task relevance, and difficulty in handling ambiguous tasks, by executing the to-do list generation method of the above-mentioned first aspect or any corresponding implementation manner.

[0250] In one example, a to-do item generation system based on a large model is provided, comprising:

[0251] (1) Input interface: The input interface receives chat or notification content containing agent information.

[0252] (2) Large model processing unit (main invention point): mainly includes four sub-steps: task identification, fuzzy task processing, importance judgment and relevance understanding.

[0253] ① Task identification: The natural language understanding part of the large model processing unit receives the natural language content from the input interface. This part uses LLM to deeply understand the input text, including word meaning, sentence meaning, contextual relationship, etc. After understanding the input natural language content, the task identification part needs to identify the to-do tasks. This involves extracting key information from the text, such as task name, task description, deadline, etc.

[0254] ②Fuzzy task processing: The fuzzy task processing part needs to handle ambiguous tasks. For example, if the user uses ambiguous language when describing a to-do item, such as "I may need to complete this report sometime next week", then this part needs to be able to understand this ambiguous description and generate the corresponding to-do item.

[0255] ③ Importance judgment: The importance judgment part needs to judge the importance of the identified tasks. This may involve analyzing factors such as the urgency of the task and the scope of impact.

[0256] ④ Relevance understanding: The relevance understanding part requires understanding the relevance between the tasks to be done. For example, if one task is a prerequisite for another task, then the two tasks are relevance.

[0257] (3) To-do item creation: After the large model processing unit processes the input content, the to-do item generation unit generates to-do items based on the output of the large model processing unit.

[0258] (4) Output interface: Finally, the output interface displays the generated to-do items to the user.

[0259] Further, if Fig. 9 As shown, the process of the to-do item generation system based on the large model provided in the above example is explained.

[0260] 1. Input interface: In this embodiment, a user interface is first provided, which allows users to chat or send notifications on the office communication software. The user clicks "Generate To-Do Item" on the interface to pass the content to the system through the input interface.

[0261] 2. Large model processing unit (such as Fig.10 As shown in the figure): The large model processing unit receives text data from the input interface and converts the text data into structured to-do item data through four steps: task identification, fuzzy task processing, importance judgment and relevance understanding. The specific processing flow is as follows:

[0262] 2.1. Task Identification: The task identification part receives the natural language content from the input interface, uses a large language model to deeply understand the input text, and identifies key information such as the task name, task description, and deadline. The details are as follows:

[0263] 2.1.1 Data preprocessing: The text data from the input interface is usually a piece of natural language text, which may contain multiple sentences or paragraphs. The first step of data preprocessing is to split the received text into separate sentences or paragraphs. Next, the text is cleaned and standardized to ensure the consistency and processability of the text. This includes operations such as removing special characters, converting text to lowercase, and processing abbreviations to reduce the noise and irregularity of the text.

[0264] 2.1.2 Prompt Engineering: In order to guide LLM to perform task identification, a suitable prompt needs to be constructed. Prompt is a text or question that guides the model to perform a specific task. In task identification, the design of prompt is very important. The prompt extracted by the task in this embodiment is shown in step S3021 above. This prompt will guide LLM to focus on task identification and output the result in a structured JSON format.

[0265] 2.1.3. Calling the LLM model: Since LLM is essentially a probability model with super-large-scale parameters, its content generation process is to predict the next output token based on the current text, and the token is selected from a certain probability distribution. Therefore, the output of the model is unstable. Even if the Prompt Engineering technology is used, it cannot guarantee that the model can generate content in the specified format. In order to solve this problem, this embodiment combines the open source LLM framework Langchain to build a structured content extraction pipeline, such as Figure 4 As shown. First, the prompt words obtained by Prompt Engineering are input into LLM to obtain preliminary output results, and then input into the JsonOutputParser module to extract structured JSON data from the output of LLM. JsonOutputParser is implemented based on specific rules and regular expressions. The modules for extracting structured output below are all implemented using similar technologies and will not be repeated here.

[0266] 2.1.4. Generate to-do items: Based on the task-related information extracted from the LLM output, after concatenating the user id and generating task_id, the output example of this step is shown in the above step S3021, including user id, task id, task name, task description and deadline.

[0267] 2.2. Fuzzy task handling: The fuzzy task handling part can handle ambiguous task descriptions, mainly for task deadlines. If the user uses vague language in the notification, such as "complete the report sometime next week", this part will be able to understand and convert it into a specific deadline or time range.

[0268] Specific steps are as follows Figure 5 As shown:

[0269] 2.2.1. Extract fuzzy time description: Input the JSON structured data obtained by task identification in step 2.1 into LLM to determine whether there is a fuzzy time description. If not, directly convert the deadline into the standard %Y-%m-%d %H:%M:%S time format and jump out of the fuzzy task processing module. Otherwise, extract the fuzzy time description. The prompt design of this step is shown in the above step S3022.

[0270] 2.2.2. Get the current time. Call the Python tool script to get the current time. The purpose of this step is to serve as a benchmark for calculating the task deadline.

[0271] 2.2.3. Specific time calculation: The current time and the fuzzy time description are input into LLM to calculate the specific task deadline. The prompt design of this step is as shown in the above step S3022. This step will guide LLM to convert the fuzzy time description into an accurate task deadline.

[0272] 2.2.4. Parse the output response of LLM and replace the fuzzy time description with the calculated accurate task deadline. The output example of this step is shown in the above step S3022.

[0273] 2.3. Importance judgment: The importance judgment part analyzes factors such as the content, urgency, and context of the task to determine the importance of the task. The urgency of the task may be judged based on the deadline, while the importance of the task may be related to the keywords in the task description. The specific steps are as follows:

[0274] 2.3.1. Key information extraction: Obtain JSON structured task information from the fuzzy task processing step 2.2, including task deadline, task name, keywords and phrases in the task description, etc.

[0275] 2.3.2 Calculation of the urgency score of the task deadline: Calculate the urgency score of the task deadline. The score is calculated based on the time interval between the deadline and the current time. The closer the deadline is, the higher the score. This can be achieved by following the steps below:

[0276] (1) Get the deadline of the task.

[0277] (2) Calculate the time interval between the current time and the deadline.

[0278] (3) Use the function to map the time interval to the urgency score: urgency score = max(0, 1-(time interval / maximum allowed time interval))

[0279] The maximum allowed time interval is a time limit or deadline threshold set in task management or time management. It indicates how long a task is considered urgent or needs to be completed as soon as possible.

[0280] 2.3.3. User-specified priority: If the user has assigned a priority to the task, the user-specified priority is directly included in the comprehensive evaluation.

[0281] 2.3.4. Keyword score calculation: Analyze whether the task description contains keywords or phrases, such as "urgent" and "important". If these keywords are present, assign a higher score.

[0282] 2.3.5. Comprehensive evaluation calculation: Using the formula for comprehensive evaluation of task importance, the above-calculated scores are comprehensively calculated to obtain a comprehensive importance score. This embodiment is implemented by weighted summation:

[0283] Composite Importance Score = w1*Deadline Urgency Score + w2*User-Specified Priority + w3*Keyword Score

[0284] Among them, w1, w2, and w3 are weight factors used to adjust the importance of different factors. These weight factors can be adjusted according to specific needs.

[0285] 2.3.8 Output results:

[0286] According to the comprehensive importance score, the task is marked as high, medium, low importance or other corresponding levels so that the task management system can sort the tasks and remind the user accordingly. The output example of this step is shown in the above step S6031.

[0287] 2.4. Relevance Understanding: The relevance understanding part understands the relevance between tasks by analyzing the keywords, dependencies, and other information between tasks. For example, if one task is a prerequisite for another task, the system will identify and establish the relationship between the two tasks.

[0288] The relationships in this embodiment include the following three types:

[0289] Dependency: Identify whether one task is a prerequisite for another task. For example, if task A is a prerequisite for task B, the system needs to understand this dependency.

[0290] Parallel: Identifies the parallel relationship between two or more tasks, indicating that they can be performed simultaneously without relying on each other.

[0291] Sequential: Identifies the sequential relationship between tasks, indicating that one task must be executed after another task.

[0292] The input of relevance understanding is the output of importance judgment, and the output is the to-do task association graph, where the directed edges represent the dependency between tasks. The specific steps are as follows: Figure 7 As shown:

[0293] 2.4.1. Graph node generation: Convert the output data of importance judgment into nodes of the relationship graph.

[0294] 2.4.2. The association relationship conversion example and the nodes generated in step 2.4.1 are input into the large model together. The Prompt used in this embodiment is as shown in the above step S6032.

[0295] 2.4.3. Obtain the dependency of task nodes based on the output of LLM and construct a task correlation diagram. The output of this step is as follows: Figure 7 shown.

[0296] 3. To-do item generation unit: The to-do item generation unit generates to-do items based on the output of the large model processing unit. This includes information such as task name, task description, deadline, and task importance. The generated to-do items will be stored in the database for subsequent display and management.

[0297] 4. Output interface: Finally, the generated to-do items will be displayed to users through the output interface. Users can access these to-do items in the Yilian APP to better manage their work and tasks.

[0298] The AI ​​to-do item generation system on the office communication software provided in this example has the following beneficial effects:

[0299] (1) Task identification: Prompt Engineering is used to fully utilize the content extraction and instruction execution capabilities of LLM. The Langchain framework is combined to intelligently extract task names, descriptions, time and other information from the text, automatically identify and generate to-do items, thereby reducing the burden of users recording to-do tasks in office communication software.

[0300] (2) Fuzzy task processing: Combine Python tool scripts and Prompt Engineering to build an LLM workflow pipeline, using the accuracy of tool scripts and the reasoning ability of LLM to convert fuzzy task times into clear and accurate task deadlines.

[0301] (3) Importance judgment: The importance of the task is judged by comprehensively considering the urgency of the deadline, the priority specified by the user, and the keyword score, which facilitates task management and sorting.

[0302] (4) Understanding of associations: Design a task association diagram and use Prompt Engineering to build an LLM workflow pipeline to automatically identify the associations between tasks, thereby better managing to-do items.

[0303] The embodiment of the present invention also provides a computer device for executing the above Figure 2 , Figure 3 and Figure 6 The to-do item generation method shown.

[0304] See also Fig.11 , Fig.11 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Fig.11 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Fig.11 A processor 10 is taken as an example.

[0305] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0306] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0307] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0308] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0309] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0310] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0311] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0312] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for generating a to-do list, characterized in that: Used in a to-do list generation system, the system is connected to a user terminal; the method comprises: Acquire an initial text data set of a plurality of to-do tasks input by the user terminal; Using a preset large language model to perform recognition and fuzzy processing on the initial text data set to obtain a first to-do task data set; Based on the first to-do task dataset, using the preset large language model to perform importance judgment and determine the to-do task relevance dataset of the plurality of to-do tasks; Based on the first to-do task data set and the to-do task association data set, a plurality of target to-do items of the plurality of to-do tasks are generated.

2. The method according to claim 1, characterized in that The initial text data set is identified and fuzzy processed using a preset large language model to obtain a first to-do task data set, including: Using the preset large language model to perform task identification on the initial text data set to obtain a second to-be-done task data set; Based on the second to-do task data set, fuzzy task processing is performed using the preset large language model to obtain the first to-do task data set.

3. The method according to claim 2, characterized in that Using the preset large language model to perform task identification on the initial text data set to obtain a second to-be-done task data set, including: Preprocessing the initial text data set to obtain a target text data set; Based on the first preset Prompt template, using the preset large language model to perform task recognition on the target text dataset to obtain a third to-do task dataset; Inputting the third to-be-done task data set into the preset large language model for processing to obtain a fourth to-be-done task data set; Parsing the fourth to-be-done task data set to obtain a fifth to-be-done task data set; Based on the fifth set of to-do task data and the user identity, the second set of to-do task data is determined.

4. The method according to claim 2, characterized in that: Based on the second to-do task data set, fuzzy task processing is performed using the preset large language model to obtain the first to-do task data set, including: Using the preset large language model, determining whether there is a fuzzy time description in the second to-do task data set; When there is fuzzy time description, obtain real-time time data and fuzzy time description data set; Based on a second preset Prompt template, the real-time data and the fuzzy time description data set are input into the preset large language model to obtain a task deadline data set; The task deadline data set is parsed to obtain the first to-do task data set.

5. The method according to claim 1, characterized in that Based on the first to-do task dataset, using the preset large language model to perform importance judgment and determine the to-do task relevance dataset of the plurality of to-do tasks, including: Based on the first to-do task data set, using the preset large language model to perform importance judgment, and obtain multiple target score values ​​of multiple to-do tasks, wherein the target score values ​​reflect the importance of the to-do tasks; Based on the multiple target score values ​​and the first to-do task data set, the preset large language model is used to understand the relevance of the multiple to-do tasks to obtain the to-do task relevance data set.

6. The method according to claim 5, characterized in that Based on the first to-do task dataset, the preset large language model is used to perform importance judgment to obtain multiple target score values ​​of the multiple to-do tasks, including: Extracting key information from the first to-do task data set to obtain a key information data set; Based on the key information data set, respectively calculating a plurality of task deadline urgency score values, a plurality of priority score values, and a plurality of keyword score values ​​of the plurality of to-do tasks; The plurality of target score values ​​of the plurality of to-do tasks are calculated based on the plurality of task deadline urgency score values, the plurality of priority score values, and the plurality of keyword score values.

7. The method according to claim 5, characterized in that Based on the multiple target score values ​​and the first to-do task data set, the preset large language model is used to understand the relevance of the multiple to-do tasks to obtain the to-do task relevance data set, including: Get the preset association relationship set; The preset association relationship set, the multiple target score values ​​and the first to-do task data set are input into the preset large language model to obtain the to-do task association data set.

8. A to-do list generation system, connected to a user terminal; characterized in that: Used to execute the to-do list generation method described in any one of claims 1 to 7; the system includes: an input interface, an output interface, a large model processing unit and a to-do list generation unit, the large model processing unit is connected to the user terminal through the input interface, the large model processing unit is also connected to the to-do list generation unit, and the to-do list generation unit is also connected to the user terminal through the output interface.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the to-do list generation method according to any one of claims 1 to 7.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the to-do list generating method according to any one of claims 1 to 7.

Images