Method and device for generating backlog, electronic equipment and computer storage medium

Abstract

The invention provides a to-do list generation method and device, electronic equipment and a computer storage medium, time information in a long text input by a user can be extracted to obtain a time extraction result, a scene type of the long text is identified, then a task title is generated based on the long text and the scene type, and finally, the to-do list is generated. And generating a to-do list based on the task title and the time extraction result, thereby effectively improving the generation efficiency and accuracy of the to-do list.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, electronic device, and computer storage medium for generating to-do lists. Background Technology

[0002] In modern enterprise management, employees need to process a large amount of text information every day, such as emails, meeting minutes, and project documents. These text information often contain a large number of tasks and to-do items that need to be performed.

[0003] Because documents in different scenarios have different textual focuses, manually extracting this task information and compiling task to-do lists is not only time-consuming and laborious, but also prone to missing important information, affecting work efficiency and accuracy. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, electronic device and computer storage medium for generating to-do items, in order to solve the problem of low efficiency and accuracy in the prior art of manually extracting task information and compiling to-do items.

[0005] The first aspect of this invention provides a method for generating to-do items, comprising:

[0006] Receive long text input from the user;

[0007] The time information in the long text is extracted to obtain the time extraction result, and the long text is input into a classification model to obtain the scene classification result; wherein, the classification model is obtained by training the BERT model on a text classification dataset; the text classification dataset includes historical data of different scenes;

[0008] The long text and the scene classification results are input into the office scene summarization model to obtain the task title; wherein, the office scene summarization model is trained by the text-to-text transfer model on the summary title dataset; the summary title dataset includes the weights of different scenes and the correspondence between the training sample long texts and the training sample task titles of different scenes;

[0009] Based on the task title and the extracted time, a to-do list is generated.

[0010] Optionally, the step of extracting time information from the long text to obtain time extraction results includes:

[0011] The long text is parsed using a preset parsing method to obtain the text parsing result;

[0012] The system default rules are used to perform regular expression matching on the text parsing results to obtain the first-time text matching results.

[0013] If the first time text matching result indicates that the time text corresponding to the system default rule is matched, then semantic analysis is performed on the time text corresponding to the system default rule to obtain the first semantic analysis result, and context analysis is performed on the time text corresponding to the system default rule to obtain the first context analysis result.

[0014] Based on the first semantic analysis result and the first context analysis result, a time extraction result is generated.

[0015] Optionally, after parsing the long text using a preset parsing method to obtain the text parsing result, the method further includes:

[0016] If the user sets personalized rules, then regular expression matching is performed on the personalized rules in the text parsing results to obtain the second time text matching results;

[0017] If the second time text matching result shows that no time text corresponding to the personalized rule is matched, then the step of performing regular expression matching on the system default rule in the text parsing result to obtain the first time text matching result is started.

[0018] If the second time text matching result indicates that the time text corresponding to the personalized rule is matched, then semantic analysis is performed on the time text corresponding to the personalized rule to obtain the second semantic analysis result, and context analysis is performed on the time text corresponding to the personalized rule to obtain the second context analysis result.

[0019] Based on the second semantic analysis result and the second context analysis result, a time extraction result is generated.

[0020] Optionally, generating to-do items based on the task title and the time extraction result includes:

[0021] The task time is determined based on the number of times in the extracted time results;

[0022] Based on the task title and the task time, generate a to-do list.

[0023] A second aspect of the present invention provides an apparatus for generating to-do lists, comprising:

[0024] The receiving unit is used to receive long text input by the user;

[0025] The time extraction unit is used to extract time information from the long text and obtain the time extraction result;

[0026] A scene classification unit is used to input the long text into a classification model to obtain scene classification results; wherein, the classification model is obtained by training a BERT model on a text classification dataset; the text classification dataset includes historical data of different scenes;

[0027] The task title determination unit is used to input the long text and the scene classification result into the office scene summary model to obtain the task title; wherein, the office scene summary model is trained by the summary title dataset on the text-to-text transfer model; the summary title dataset includes the weights of different scenes and the correspondence between the training sample long text and the training sample task title of different scenes;

[0028] The to-do item generation unit is used to generate to-do items based on the task title and the time extraction result.

[0029] Optionally, the time extraction unit includes:

[0030] The text parsing unit is used to parse the long text using a preset parsing method to obtain the text parsing result;

[0031] The first regular expression matching unit is used to perform regular expression matching on the system default rules in the text parsing result to obtain the first time text matching result;

[0032] The first analysis unit is configured to perform semantic analysis on the time text corresponding to the system default rule to obtain a first semantic analysis result and perform context analysis on the time text corresponding to the system default rule to obtain a first context analysis result if the first time text matching result indicates that the time text corresponding to the system default rule is matched.

[0033] The first time extraction sub-unit is used to generate time extraction results based on the first semantic analysis result and the first context analysis result.

[0034] Optionally, the device for generating to-do items further includes:

[0035] The second regular expression matching unit is used to perform regular expression matching on the personalized rules in the text parsing result if the user sets personalized rules, to obtain a second time text matching result; wherein, if the second time text matching result shows that no time text corresponding to the personalized rule is matched, the first regular expression unit starts to perform regular expression matching on the system default rules in the text parsing result to obtain a first time text matching result.

[0036] The second analysis unit is configured to perform semantic analysis on the time text corresponding to the personalized rule to obtain a second semantic analysis result and perform context analysis on the time text corresponding to the personalized rule to obtain a second context analysis result if the second time text matching result indicates that the time text corresponding to the personalized rule is matched.

[0037] The second time extraction subunit is used to generate time extraction results based on the second semantic analysis results and the second context analysis results.

[0038] Optionally, the to-do list generation unit includes:

[0039] The task time determination unit is used to determine the task time based on the number of times in the time extraction result;

[0040] The to-do item generation subunit is used to generate to-do items based on the task title and the task time.

[0041] A third aspect of the present invention provides an electronic device, comprising:

[0042] One or more processors;

[0043] A storage device on which one or more programs are stored;

[0044] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for generating to-do items as described in any one of the first aspects.

[0045] A fourth aspect of the present invention provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for generating to-do items as described in any one of the first aspects.

[0046] As can be seen from the above solutions, the present invention provides a method, apparatus, electronic device, and computer storage medium for generating to-do items. It can extract time information from long text input by users, obtain time extraction results, identify the scene type of the long text, generate a task title based on the long text and scene type, and finally generate to-do items based on the task title and the time extraction results, effectively improving the efficiency and accuracy of to-do item generation. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0048] Figure 1 A detailed flowchart of a method for generating to-do items provided in an embodiment of the present invention;

[0049] Figure 2 A flowchart of a time information extraction method provided in another embodiment of the present invention;

[0050] Figure 3 A flowchart of a time information extraction method provided in another embodiment of the present invention;

[0051] Figure 4 A flowchart illustrating a method for generating to-do items according to another embodiment of the present invention;

[0052] Figure 5 An architecture diagram of a method for generating to-do items according to another embodiment of the present invention;

[0053] Figure 6 A flowchart illustrating the construction process of a text classification module, as provided in another embodiment of the present invention;

[0054] Figure 7 A flowchart illustrating the construction process of a summary generation module, as provided in another embodiment of the present invention;

[0055] Figure 8 A schematic diagram of a to-do list generation device provided in another embodiment of the present invention;

[0056] Figure 9 This is a schematic diagram of an electronic device that implements a method for generating to-do items, according to another embodiment of the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0059] It should be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this invention are all information and data authorized by the user or fully authorized by all parties.

[0060] It should be noted that the concepts of "first" and "second" mentioned in this invention are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0061] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0062] This invention provides a method for generating to-do items, such as... Figure 1 As shown, the specific steps include:

[0063] S101, Receive long text input by the user.

[0064] Long texts include, but are not limited to, email content, meeting minutes, project reports, etc., and are not limited here.

[0065] In the practical application of this invention, in order to facilitate user operation, users can directly copy text content from emails and minutes as input, which is not limited here.

[0066] S102. Extract the time information from the long text to obtain the time extraction result.

[0067] In the practical application of this invention, time information in long texts can be extracted directly using tools such as Time-NLP, but not limited to Time-NLP, to obtain the time extraction results. No limitation is made here.

[0068] The Time-NLP time extraction tool is a regular expression-based tool focused on Chinese time semantic recognition. It aims to parse complex time expressions in natural language and convert them into a standard format that computers can process. Its core functions include: Chinese time semantic parsing, contextual association and multi-time processing, future time trend inference, customizable base time, and time normalization.

[0069] Specifically, in the practical application of this invention, time information can be absolute time (e.g., "May 1, 2023"), relative time (e.g., "tomorrow" or "next week"), and general time (e.g., "morning" or "evening"), etc., without limitation. It can also handle logical relationships between multiple time points; for example, "3 PM to 5 PM on the 1st of next month" can be automatically recognized as a continuous time period. It can also automatically infer fuzzy times based on the current time; for example, entering "6 AM" in the afternoon will be recognized as the next morning.

[0070] Optionally, in another embodiment of the present invention, one implementation of step S102 is as follows: Figure 2 As shown, it includes:

[0071] S201. Use the preset parsing method to parse the long text and obtain the text parsing result.

[0072] In the actual application of this invention, the preset parsing method can be, but is not limited to, the parse method, and is not limited here.

[0073] The core meaning of the `parse` method is to transform or decompose a data representation (most commonly a string) into another structured format or specific data type that is easier for programs to process and use. This process aims to extract meaningful information from the raw data.

[0074] For example, when inputting natural language such as "7:20 tomorrow morning", the parse method will parse it and map it onto a standard timeline to get a precise time point like "2025-11-13 07:20:00" (assuming the current time is November 12, 2025).

[0075] Optionally, in another embodiment of the present invention, after parsing the long text using a preset parsing method to obtain the text parsing result, one implementation of the method for generating to-do items is as follows: Figure 3 As shown, it also includes:

[0076] S301. If the user has set personalized rules, then perform regular expression matching on the personalized rules in the text parsing results to obtain the second time text matching results.

[0077] Users can customize time processing rules according to their own usage habits, and store these user-defined configurations in a database to build a personalized configuration library. For example:

[0078] (1) User A sets the base time to 8:30 every day, then all subsequent time processing rules will be calculated and extracted based on this time.

[0079] (2) User B defines “going to work” as 8:30 and “leaving get off work” as 17:30. If this key text is included in the subsequent text parsing, the time extraction will be performed according to this rule.

[0080] Each user's personalized rules are stored in a database. When parsing text, the user's personalized configuration file is loaded first, and the personalized rules are applied. If no matching result is found for the user's configuration, the default system rules are used.

[0081] Specifically, if the second time text matching result shows that no time text corresponding to the personalized rule is matched, then step S202 and subsequent steps are executed.

[0082] This invention uses regular expressions to extract time information, which can efficiently identify complex date and time formats, extract accurate task times, and support user-defined time rules based on actual task scenarios, build a user-personalized configuration library, and is applicable to multiple scenarios to meet users' personalized needs.

[0083] S302. If the second time text matching result shows that the time text corresponding to the personalized rule is matched, then semantic analysis is performed on the time text corresponding to the personalized rule to obtain the second semantic analysis result, and context analysis is performed on the time text corresponding to the personalized rule to obtain the second context analysis result.

[0084] Specifically, semantic analysis of time text is performed using predefined rules (such as personalized rules and system default rules), including various types such as specific dates (e.g., "February 28, 2013"), relative times (e.g., "in ten minutes"), and ambiguous times (e.g., "morning" and "evening"). For enumerated time terms (e.g., today, tomorrow, next week), mapping can be performed using, but is not limited to, the CommonTimeEnum enumeration class, and the specific date can be calculated using a base time. In terms of semantic modeling, a time granularity framework can be adopted, but is not limited to, to uniformly map time units of different granularities (e.g., seconds, minutes, days, months) onto a standard timeline, resolving complex expressions such as "33 days, 2 minutes, and 4 seconds." No specific limitations are set here.

[0085] In the practical application of this invention, contextual analysis of time text can be performed. When dealing with continuous time ranges, such as "from 3 pm to 5 pm on the 1st of next month", the end time can be automatically inherited from the date part of the preceding text, and only the time part can be adjusted. For time references across sentences, the dialogue history is maintained to ensure that if the user first says "meeting tomorrow" and then says "revision to the day after tomorrow", the date of "the day after tomorrow" can be correctly understood relative to the original "tomorrow". This is not limited here.

[0086] S303. Based on the second semantic analysis results and the second context analysis results, generate time extraction results.

[0087] In the practical application of this invention, time tendency judgment can also be performed. For example, if you enter "deliver milk at 9 o'clock" at 5 pm, it will be automatically recognized as 9 am the next day. For past relative times, such as saying "get up at 6 am" in the evening, it will be recognized as 6 am the next day. The expression of the day of the week will also be automatically adjusted according to the current time. For example, if you say "meeting on Monday" on Tuesday, it will be recognized as next Monday.

[0088] When time text contains multiple time points, it is also possible to analyze the temporal relationships and contextual dependencies between them. For example, the relationship between events and time can be formally represented by time ontology models (such as OWL-Time, TimeML, etc.), supporting complex semantic reasoning.

[0089] S202. Perform regular expression matching on the system default rules in the text parsing results to obtain the first-time text matching results.

[0090] It should be noted that the specific implementation of step S202 can be referred to step S301, and will not be repeated here.

[0091] S203. If the first time text matching result shows that the time text corresponding to the system default rule is matched, then semantic analysis is performed on the time text corresponding to the system default rule to obtain the first semantic analysis result, and context analysis is performed on the time text corresponding to the system default rule to obtain the first context analysis result.

[0092] It should be noted that the specific implementation method of step S203 can be referred to step S302, and will not be repeated here.

[0093] S204. Based on the first semantic analysis result and the first context analysis result, generate the time extraction result.

[0094] It should be noted that the specific implementation of step S204 can be found in step S303, and will not be repeated here.

[0095] S103. Input the long text into the classification model to obtain the scene classification result.

[0096] The classification model is obtained by training the BERT model on a text classification dataset; the text classification dataset includes historical data from different scenarios.

[0097] In the practical application of this invention, the text classification dataset can be constructed, but is not limited to, by collecting historical data from the bank's office environment, such as historical emails, historical minutes, and historical reports, labeling different categories of data with scene categories, and cleaning the data to create an internal office text classification dataset. Specifically, the text classification dataset can also be divided into a training set, a test set, and a validation set.

[0098] The BERT model was trained and fine-tuned using the text classification dataset obtained above, enabling scene classification of different types of text in real-world office scenarios. After a user inputs long text, the classification model can classify the text scene based on the characteristics of the long text, thus obtaining the scene classification result.

[0099] In the practical application of this invention, the BERT model can be pre-trained using the Masked Language Model (MLM) and Next Sentence Prediction (NSP) tasks, and can be transferred to downstream tasks for fine-tuning, without limitation here.

[0100] In the practical application of this invention, the module that implements this step can be introduced into a web development tool, API interfaces can be written, and the model can be deployed on the cloud for other modules to call. No limitation is made here.

[0101] S104. Input the long text and scene classification results into the office scene summary model to obtain the task title.

[0102] The office scenario summarization model is obtained by training the text-to-text transfer transformer (T5) on the summary title dataset. The summary title dataset includes the weights of different scenarios and the correspondence between the long text of the training samples and the task titles of the training samples in different scenarios.

[0103] The T5 model unifies all natural language processing (NLP) tasks (such as classification, summarization, and translation) into a text-to-text format, with both input and output being text sequences. It is pre-trained on massive amounts of text through a denoising autoencoder task.

[0104] The T5 model is a text-to-text model based on the Transformer architecture and employing an encoder-decoder structure.

[0105] (1) The encoder of the T5 model consists of multiple layers of Transformer blocks that encode the input text into hidden states. Each Transformer block contains a self-attention mechanism and a feedforward neural network (FFN).

[0106] (2) The decoder of the T5 model also consists of multiple layers of Transformer blocks, which generate the target text based on the output of the encoder. Each decoder layer contains self-attention, encoder-decoder attention, and a feedforward network.

[0107] This invention can be pre-trained on the Chinese corpus dataset LCSTS based on the mT5 summarization model (a multilingual version of the T5 model that supports Chinese). This dataset is a text summarization dataset with more than 2 million data points created from news summaries published by news media on social platforms, making it better suited for Chinese summarization generation tasks.

[0108] To enable the model to adaptively generate summaries based on real-world usage scenarios, a dataset was constructed using data generated in actual office environments, and the model was fine-tuned. The dataset includes office emails, meeting minutes, project reports, etc. The data was cleaned, and the characteristics of summaries generated in each scenario were preserved. A standard "summary-original text" key-value pair summary title dataset was constructed, divided into training, testing, and validation sets. A pre-trained mT5 model was loaded for text summarization training and fine-tuning. The model parameters trained on different datasets were saved, enabling adaptive summarization of various text types in real-world office scenarios.

[0109] The fine-tuned summarization model can generate corresponding summaries based on the input text and use them as task titles. Combined with the scene classification results, training weights for different scenes are loaded, and the parameters of the office scene summarization model are adaptively configured, including model generation quality, input and output token word count, etc.

[0110] For example, for short text scenarios such as office emails, load email data training parameters, shorten the number of words output by the model, and appropriately reduce the generation quality to ensure generation efficiency; for record texts such as meeting minutes, load the corresponding training parameters and focus on extracting key themes and key matters; for long text scenarios such as project reports, increase the word limit of the model output, load the corresponding parameters, and extract the key information items in the report.

[0111] In the practical application of this invention, the module that implements this step can be introduced into a web development tool, API interfaces can be written, and the model can be deployed on the cloud for other modules to call. No limitation is made here.

[0112] S105. Based on the task title and time extraction results, generate to-do items.

[0113] In the practical application of this invention, the generated to-do list can also be submitted to the user for secondary confirmation and modification to obtain the final to-do list.

[0114] Optionally, in another embodiment of the present invention, one implementation of step S105 is as follows: Figure 4 As shown, it includes:

[0115] S401. Determine the task time based on the number of times in the time extraction results.

[0116] For example: If the text does not contain time information: the task time is set to the base time, such as 8:30 today; if the text contains only one piece of time information: the task time is set to that extraction time; if the text contains more than one piece of time information: the task time is set to the minimum extraction time including hours, minutes, and seconds, without any limitation here.

[0117] S402. Generate to-do items based on task title and task time.

[0118] Specifically, the task title and task time are combined to obtain the to-do list.

[0119] like Figure 5 The diagram shown illustrates the architecture of a method for generating to-do items according to an embodiment of the present invention. It includes a text classification module, a summary generation module, a time extraction module, and a task generation module. The text classification module receives long text input by the user, which can originate from email content, meeting minutes, project reports, or other forms of text. The long text is input into the BERT text classification model, outputting a scene classification result. The summary generation module generates a task title based on the long text and the scene classification result using the T5 summary model. The time extraction module extracts the task time using Time-NLP based on user-personalized rules and system default rules. Finally, the task generation module generates to-do items based on the task time and task title, and obtains the final to-do items after secondary confirmation by the user.

[0120] like Figure 6 The diagram shows a flowchart of a text classification module construction process provided by an embodiment of the present invention. It uses a text classification dataset of office scenarios to fine-tune the basic BERT to obtain a classification model for office scenarios. Then, it develops a model inference interface through a web interface, uses the interface to complete the model deployment, and finally obtains the text classification application.

[0121] like Figure 7The diagram shows a flowchart of the construction process of a summary generation module provided in an embodiment of the present invention. After pre-training mT5 with LCSTS to obtain a Chinese summary model, the Chinese summary model is fine-tuned with an office summary dataset to obtain an office scenario fine-tuned model. Then, a model inference interface is developed through a web interface, and the model is deployed using the interface to finally obtain an automatic summary application.

[0122] As shown in the above solutions, this invention provides a method for generating to-do items. It can extract time information from long text input by the user, obtain the time extraction result, identify the scenario type of the long text, then generate a task title based on the long text and scenario type, and finally generate to-do items based on the task title and the time extraction result. This effectively improves the efficiency and accuracy of to-do item generation. This invention is applicable to task management, information entry, and other functions in various office management systems, simplifying manual entry into automatic parsing and generation, streamlining workflows, and improving work efficiency.

[0123] Another embodiment of the present invention provides a to-do list generation device, such as... Figure 8 As shown, it specifically includes:

[0124] The receiving unit 801 is used to receive long text input by the user.

[0125] The time extraction unit 802 is used to extract time information from long texts and obtain time extraction results.

[0126] Optionally, in another embodiment of the present invention, one implementation of the time extraction unit 802 includes:

[0127] The text parsing unit is used to parse long texts using a preset parsing method to obtain the text parsing results.

[0128] The first regular expression matching unit is used to perform regular expression matching on the system default rules in the text parsing results to obtain the first-time text matching results.

[0129] The first analysis unit is used to perform semantic analysis on the time text corresponding to the system default rule if the first time text matching result shows that the time text corresponding to the system default rule is matched, and to obtain the first semantic analysis result, and to perform context analysis on the time text corresponding to the system default rule, and to obtain the first context analysis result.

[0130] The first time-based sub-unit is extracted and used to generate time-based extraction results based on the first semantic analysis results and the first context analysis results.

[0131] The specific working process of the units disclosed in the above embodiments of the present invention can be found in the corresponding method embodiments, and will not be repeated here.

[0132] Optionally, in another embodiment of the present invention, one implementation of the to-do list generation device further includes:

[0133] The second regular expression matching unit is used to perform regular expression matching on the personalized rules in the text parsing results if the user has set personalized rules, so as to obtain the second time text matching result.

[0134] If the second time text matching result shows that no time text corresponding to the personalized rule is matched, then the first regular expression unit starts to perform regular expression matching on the system default rule in the text parsing result to obtain the first time text matching result.

[0135] The second analysis unit is used to perform semantic analysis on the time text corresponding to the personalized rule if the second time text matching result shows that the time text corresponding to the personalized rule is matched, to obtain the second semantic analysis result, and to perform context analysis on the time text corresponding to the personalized rule, to obtain the second context analysis result.

[0136] The second time extraction subunit is used to generate time extraction results based on the second semantic analysis results and the second context analysis results.

[0137] The specific working process of the units disclosed in the above embodiments of the present invention can be found in the corresponding method embodiments, and will not be repeated here.

[0138] Scene classification unit 803 is used to input long text into the classification model to obtain scene classification results.

[0139] The classification model is obtained by training the BERT model on a text classification dataset; the text classification dataset includes historical data from different scenarios.

[0140] The task title determination unit 804 is used to input long text and scene classification results into the office scene summary model to obtain the task title.

[0141] The office scenario summary model is obtained by training a text-to-text transfer model using a summary title dataset. The summary title dataset includes the weights of different scenarios and the correspondence between the long text of the training samples and the task titles of the training samples in different scenarios.

[0142] The to-do item generation unit 805 is used to generate to-do items based on the task title and time extraction results.

[0143] For details on the specific operation of the units disclosed in the above embodiments of the present invention, please refer to the corresponding method embodiments, such as... Figure 1 As shown, it will not be elaborated further here.

[0144] Optionally, in another embodiment of the present invention, one implementation of the to-do list generation unit 805 includes:

[0145] The task time determination unit is used to determine the task time based on the amount of time extracted from the time extraction results.

[0146] The to-do list generation sub-unit is used to generate to-do items based on task titles and task times.

[0147] The specific working process of the units disclosed in the above embodiments of the present invention can be found in the corresponding method embodiments, and will not be repeated here.

[0148] As can be seen from the above solution, the present invention provides a to-do list generation device that can extract time information from long text input by the user, obtain time extraction results, identify the scene type of the long text, then generate a task title based on the long text and scene type, and finally generate to-do list based on the task title and time extraction results, effectively improving the generation efficiency and accuracy of to-do list.

[0149] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0150] Another embodiment of the present invention provides an electronic device, such as... Figure 9 As shown, it includes:

[0151] One or more processors 901.

[0152] Storage device 902, on which one or more programs are stored.

[0153] When the one or more programs are executed by the one or more processors 901, the one or more processors 901 implement the to-do list generation method as described in the above embodiments.

[0154] Another embodiment of the present invention provides a computer storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the method for generating to-do items as described in the above embodiments.

[0155] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0156] It should be noted that the computer-readable medium described above in this invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0157] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0158] Another embodiment of the present invention provides a computer program product, which, when executed, is used to perform the above-described method for generating to-do items.

[0159] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments of the present invention.

[0160] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in this invention is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely exemplary forms for implementing the invention.

[0161] While several specific implementation details are included in the foregoing discussion, these should not be construed as limiting the scope of the invention. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0162] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to the specific combination of the above-described technical features, but also includes other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with technical features of the present invention (but not limited to) that have similar functions.

Claims

1. A method of generating a to-do list, characterized by, The method comprises the following steps: receiving a long text input by a user; extracting time information in the long text to obtain a time extraction result, and inputting the long text into a classification model to obtain a scene classification result; wherein the classification model is obtained by training a BERT model based on a text classification data set; and the text classification data set comprises historical data of different scenes; inputting the long text and the scene classification result into an office scene summarization model to obtain a task title; wherein the office scene summarization model is obtained by training a text-to-text transfer model based on a summary title data set; and the summary title data set comprises a correspondence between a training sample long text and a training sample task title in different scenes and weights of different scenes; generating a to-do list based on the task title and the time extraction result.

2. The to-do item generation method according to claim 1, wherein, The step of extracting time information in the long text to obtain a time extraction result comprises the following steps: parsing the long text based on a preset parsing method to obtain a text parsing result; performing regular matching on a system default rule in the text parsing result to obtain a first time text matching result; if the first time text matching result indicates that a time text corresponding to the system default rule is matched, performing semantic analysis on the time text corresponding to the system default rule to obtain a first semantic analysis result, and performing context analysis on the time text corresponding to the system default rule to obtain a first context analysis result; generating a time extraction result based on the first semantic analysis result and the first context analysis result.

3. The to-do item generation method according to claim 2, wherein, After the step of parsing the long text based on a preset parsing method to obtain a text parsing result, the following steps are further included: if a user has set a personalized rule, performing regular matching on the personalized rule in the text parsing result to obtain a second time text matching result; if the second time text matching result indicates that a time text corresponding to the personalized rule is not matched, starting to perform the step of performing regular matching on a system default rule in the text parsing result to obtain a first time text matching result; if the second time text matching result indicates that a time text corresponding to the personalized rule is matched, performing semantic analysis on the time text corresponding to the personalized rule to obtain a second semantic analysis result, and performing context analysis on the time text corresponding to the personalized rule to obtain a second context analysis result; generating a time extraction result based on the second semantic analysis result and the second context analysis result.

4. The to-do item generation method according to claim 1, wherein, The step of generating a to-do list based on the task title and the time extraction result comprises the following steps: determining a task time based on a number of times in the time extraction result; generating a to-do list based on the task title and the task time.

5. A to-do list generating apparatus characterized by comprising: The method comprises the following steps: a receiving unit configured to receive a long text input by a user; a time extraction unit configured to extract time information in the long text to obtain a time extraction result; a scene classification unit, configured to input the long text into a classification model to obtain a scene classification result, wherein the classification model is obtained by training a BERT model based on a text classification data set, and the text classification data set comprises historical data of different scenes; a task title determination unit, configured to input the long text and the scene classification result into an office scene summarization model to obtain a task title, wherein the office scene summarization model is obtained by training a text-to-text transfer model based on a summary title data set, and the summary title data set comprises a correspondence between a training sample long text and a training sample task title of different scenes and a weight of different scenes; a to-do list generation unit, configured to generate a to-do list based on the task title and the time extraction result.

6. The to-do item generating apparatus according to claim 5, wherein The time extraction unit comprises: a text analysis unit, configured to analyze the long text based on a preset analysis method to obtain a text analysis result; a first regular matching unit, configured to perform regular matching on a system default rule in the text analysis result to obtain a first time text matching result; a first analysis unit, configured to, if the first time text matching result indicates that the time text corresponding to the system default rule is matched, perform semantic analysis on the time text corresponding to the system default rule to obtain a first semantic analysis result, and perform context analysis on the time text corresponding to the system default rule to obtain a first context analysis result; a first time extraction subunit, configured to generate a time extraction result based on the first semantic analysis result and the first context analysis result.

7. The to-do item generating apparatus according to claim 6, wherein Further comprising: a second regular matching unit, configured to, if a user has set a personalized rule, perform regular matching on the personalized rule in the text analysis result to obtain a second time text matching result; wherein, if the second time text matching result indicates that the time text corresponding to the personalized rule is not matched, the first regular unit starts to perform the regular matching on the system default rule in the text analysis result to obtain the first time text matching result; a second analysis unit, configured to, if the second time text matching result indicates that the time text corresponding to the personalized rule is matched, perform semantic analysis on the time text corresponding to the personalized rule to obtain a second semantic analysis result, and perform context analysis on the time text corresponding to the personalized rule to obtain a second context analysis result; a second time extraction subunit, configured to generate a time extraction result based on the second semantic analysis result and the second context analysis result.

8. The to-do item generating apparatus according to claim 5, wherein The to-do list generation unit comprises: a task time determination unit, configured to determine a task time based on the number of times in the time extraction result; a to-do list generation subunit, configured to generate a to-do list based on the task title and the task time.

9. An electronic device, comprising: comprise: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the to-do item generation method as claimed in any one of claims 1 to 4.

10. A computer storage medium, characterized in that, A computer program product, having stored thereon a computer program, wherein the computer program is executed by a processor to implement the to-do item generation method as claimed in any one of claims 1 to 4.

Images