Data processing method and electronic equipment

Abstract

The invention discloses a data processing method and electronic device.The data processing method comprises the steps that interaction data input to a target application is obtained in response, the interaction data is input to a target processing model, and the target application is an application capable of calling the target processing model; and generating a target reminding event from the target item data according to a target output format by using the target processing model under the condition that the interaction data is identified to comprise the target item data.

Description

Technical Field

[0001] The present application relates to but is not limited to the field of computer technology, and in particular to a data processing method and electronic equipment. Background Art

[0002] With the development of artificial intelligence (AI) technology, various models for implementing AI functions have appeared in users' daily lives. For example, the Large Language Model (LLM) is a natural language processing model based on deep learning, which can realize tasks such as text creation, dialogue generation, summary generation, machine translation, and text classification.

[0003] As model capabilities improve, users' demands for models also gradually increase, but there is still a gap between model capabilities and user demands. For example, when users propose specific tasks, it is difficult for models to generate valuable executable operations for specific tasks; at the same time, due to the diversity of user input information structures, styles, and methods, it is further difficult for models to perform specific tasks. Therefore, how to make models provide valuable executable operations when performing specific tasks has become an urgent problem to be solved. Summary of the invention

[0004] In view of this, the present application at least provides a data processing method and an electronic device.

[0005] The technical solution of this application is implemented as follows:

[0006] In one aspect, the present application provides a data processing method, the method comprising:

[0007] In response to obtaining interaction data input to a target application, inputting the interaction data to a target processing model, the target application being an application capable of invoking the target processing model;

[0008] When it is identified that the interaction data includes target event data, the target event data is generated into a target reminder event according to the target output format using the target processing model.

[0009] In some embodiments, inputting the interaction data into a target processing model includes:

[0010] Processing the interaction data into corresponding prompt word data according to the target field format and inputting them into the target processing model to obtain target event data in the corresponding target field format from the interaction data;

[0011] Among them, the target item data that can be obtained based on prompt word data in different field formats are different.

[0012] In some embodiments, processing the interaction data into corresponding prompt word data according to the target field format and inputting it into the target processing model includes at least one of the following:

[0013] Processing the interactive data into first prompt word data according to a first field format and inputting the data into a target processing model to output a first output result;

[0014] Processing the interaction data and the first output result output by the target processing model for the interaction data into second prompt word data according to the second field format, and inputting the second prompt word data into the target processing model to output a second output result;

[0015] Processing the interaction data and the second output result output by the target processing model for the interaction data into third prompt word data according to the third field format, and inputting the third prompt word data into the target processing model to output a third output result;

[0016] Among them, the output result can represent whether the target item data exists in the interaction data.

[0017] In some embodiments, using the target processing model to generate the target event data into a target reminder event according to the target output format includes at least one of the following:

[0018] In the case where the third output result represents that the interaction data includes target item data that can constitute a reminder event, the interaction data and the third output result are processed according to the fifth field format into fifth prompt word data, which are input into the target processing model, and a first reminder event is generated according to the target output format;

[0019] In the case where the third output result does not include the time information required to constitute the reminder event, obtaining the second time information by using the target processing model to generate the second reminder event based on the second time information;

[0020] When the first time information of the first reminder event generated by the target processing model is unavailable, the second time information is acquired by using the target processing model to generate a second reminder event based on the second time information.

[0021] In some embodiments, obtaining the second time information by using the target processing model to generate a second reminder event based on the second time information includes:

[0022] The control target processing model outputs question information for time information required to constitute a reminder event;

[0023] Using the target processing model to obtain the second time information from the interactive feedback data for the question information;

[0024] The second time information and the third output result are input into the target processing model, and a second reminder event is generated according to the target output format.

[0025] In some embodiments, the method further comprises at least one of the following:

[0026] In the case where the third output result representation interaction data does not include the target item data, processing the third output result into fourth prompt word data in a fourth field format, so that the target processing model responds to the fourth prompt word data in a conversation mode;

[0027] When the time information of the second reminder event generated by the target processing model is unavailable, the third output result is processed into fourth prompt word data in a fourth field format so that the target processing model responds to the fourth prompt word data in a conversation mode.

[0028] In some embodiments, the method further comprises at least one of the following:

[0029] Output a prompt of whether to add the target reminder event to the to-do schedule, and add the target reminder event to the to-do schedule or not respond to the prompt based on the feedback information of the target user;

[0030] Sending the target reminder event to the target terminal device of the target user, so as to synchronize the target reminder event to the schedule management application of the target terminal device or reject the prompt of the target reminder event based on the user feedback information;

[0031] storing at least one inference result output by the target processing model for the interaction data into a corresponding target storage area;

[0032] The target processing model is used to evaluate whether the item data in the interaction data constitutes the risk and / or value of the reminder event, so as to determine the target item data from the interaction data based on the risk and / or value.

[0033] In some embodiments, the method further comprises:

[0034] Verifying the output format of the inference result to be outputted obtained by the target processing model for interactive data processing and / or the configuration parameters of the target processing model;

[0035] Based on the verification result, the target processing model is controlled to output the corresponding reasoning result, perform re-reasoning, or not output the reasoning result.

[0036] In some embodiments, the method further comprises at least one of the following:

[0037] Determine the relevance between the inference result to be output and the interaction data; based on the relevance, control the target processing model to output the corresponding inference result, perform re-inference, or not output the inference result;

[0038] Based on multiple sample data, multiple prediction outputs are generated using the model to be trained; based on the sample outputs and prediction outputs corresponding to the multiple sample data, loss information is calculated; based on the loss information, the model to be trained is updated to obtain a target processing model; wherein the sample output representation is obtained after processing the sample data in a specified format.

[0039] On the other hand, the present application also provides an electronic device, including at least one processor and at least one processing model capable of running on the processor, wherein the processor is capable of performing at least one of the following:

[0040] In response to obtaining interaction data input to a target application, inputting the interaction data to a target processing model, the target application being an application capable of invoking the target processing model;

[0041] When it is identified that the interaction data includes target event data, the target event data is generated into a target reminder event according to the target output format using the target processing model.

[0042] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application.

[0044] Figure 1 A schematic diagram of the implementation flow of a data processing method provided in this application;

[0045] Figure 2 A timing diagram for implementing an embodiment of a data processing method provided by the present application;

[0046] Figure 3 A schematic diagram of an implementation flow of an embodiment of performing model training according to the data processing method provided in the present application;

[0047] Figure 4 A schematic diagram of an implementation flow of an embodiment of verifying the inference result to be output according to the data processing method provided by the present application;

[0048] Figure 5 A hardware entity schematic diagram of an electronic device provided in this application;

[0049] Figure 6 A hardware entity schematic diagram of an electronic device provided in this application. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further elaborated in detail below in conjunction with the drawings and embodiments. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0051] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0052] The terms "first / second / third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understandable that "first / second / third" can be interchanged with a specific order or sequence where permitted so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing this application and are not intended to limit this application.

[0054] As mentioned above, LLM has functions such as text creation and dialogue generation. However, LLM answers are too broad, and because LLM is essentially a probabilistic model, there is a problem of unfixed output format. Therefore, LLM can give suggestions for simple things, but there are problems with automation feasibility, and it cannot handle uncertain output in engineering.

[0055] For example, in an application for automatically generating schedules and managing schedules, the user enters "I have to go to Xiaoxiang Mansion for lunch tomorrow and catch a bus back to Tianjin in the afternoon. Please help me arrange my schedule." The model responds:

[0056] Of course we can help you arrange your trip. The following is a trip plan based on your needs:

[0057] **morning:**

[0058] -8:00-9:30 Leave Tianjin and go to the railway station or long-distance bus station.

[0059] -Choose a suitable train to Changsha, Hunan Province.

[0060] **noon:**

[0061] - After arriving in Changsha at about 11:00, take the subway or taxi to Xiaoxiang Mansion (please check the specific address in advance).

[0062] -Enjoy lunch at Xiaoxiang Mansion.

[0063] **afternoon:**

[0064] -After the meal, you can take a short break and stroll around to experience the local customs and culture.

[0065] - Depending on your return train, leave enough time to get back to the train station or coach station.

[0066] **Evening / Nighttime:**

[0067] - Go to the railway station or long-distance bus station at around 15:00-16:30 to prepare for the return journey.

[0068] - Return to Tianjin by train or coach around 18:00-19:00.

[0069] **Kind tips:**

[0070] 1. Check and book tickets in advance to ensure a smooth trip.

[0071] 2. Please be sure to carry your personal documents and luggage to ensure driving safety.

[0072] 3. Adjust the itinerary according to the actual situation and extend the stay if necessary.

[0073] Have a nice trip! ”

[0074] It can be seen that the LLM output format is not fixed, which makes it difficult for the automation program to generate processing logic for fixed tasks.

[0075] The present application provides a data processing method, which can be executed by a processor of an electronic device. The electronic device may refer to a server, a laptop, a tablet computer, a desktop computer, a smart TV, a set-top box, a mobile device (such as a mobile phone, a portable video player, a personal digital assistant, a dedicated messaging device), or other device with data processing capabilities.

[0076] Figure 1 A schematic diagram of the implementation flow of a data processing method provided in this application, such as Figure 1 As shown, the method includes the following steps S10 to S11:

[0077] Step S10, in response to obtaining the interaction data input to the target application, inputting the interaction data to the target processing model, wherein the target application is an application capable of calling the target processing model.

[0078] Here, the target processing model refers to a model with AI capabilities.

[0079] In some embodiments, the target processing model may be an AI model having an event prompt function, a content generation function, a content optimization function, or other data processing functions. For example, the target processing model may be an LLM.

[0080] In some embodiments, the target processing model may be a processing model deployed locally, or a processing model deployed in the cloud or on an edge device.

[0081] In some embodiments, the target processing model may be a model for processing interaction data determined from at least one processing model. For example, the target processing model may be determined from at least one processing model based on the AI ​​capabilities required to process the interaction data.

[0082] Here, the target application refers to an application that can receive user interaction information and call a target processing model to process the interaction data.

[0083] In some embodiments, the target application may be an intelligent assistant. For example, the target application may be a personal intelligent agent such as Lenovo Xiaotian or AInow.

[0084] In some embodiments, the target application may also be a social application or an instant messaging application. For example, the target application may be WeChat, TEAMS, QQ, etc.

[0085] Thus, in some embodiments, the interactive data input into the target application may be question and answer data, chat data, planned content input by the target user, etc., input into an intelligent assistant, a social application, or an instant messaging application.

[0086] In some embodiments, the interaction data input to the target application may be continuous input with an interval time shorter than a preset time threshold. For example, input with an interval time shorter than 10 seconds may be considered continuous input, and the corresponding input data is one interaction data.

[0087] In some embodiments, when the target user inputs a large amount of data to the target application, the target user's input data may be segmented into multiple segments, and the multiple segments are sequentially sent to the target processing model as interaction data for model processing. For example, when the target user inputs a large segment of voice information to the target application, the voice information input by the target user may be converted into text information and segmented in units of 500 words; or, the voice information input by the target user may be segmented in units of 10 seconds.

[0088] In some embodiments, the interactive data input to the target application may also be processing data returned by the target processing model during the process of executing model reasoning.

[0089] In this way, in response to obtaining the interaction data, the interaction data is input into the target processing model to call the target processing model to provide the corresponding AI function or service.

[0090] In some embodiments, the AI ​​functions or services that the target processing model can provide may include AI summary function, hand-drawn redrawing function, AI interpretation function, AI translation function, AI search function, AI raw picture function, AI price comparison function, AI writing function, AI video generation function, AI setting function, AI editing function, AI dialogue function, and the like.

[0091] Step S11, when it is identified that the interaction data includes target event data, the target event data is generated into a target reminder event according to a target output format by using the target processing model.

[0092] Here, the target item data refers to data related to the items to be processed by the user. For example, the target item data may include data related to the item type, content, time, person, etc.

[0093] In some embodiments, the target item data may be determined based on interaction data input by the target user once, or may be determined based on multiple interaction data input by the target user multiple times.

[0094] For example, when the interaction data input by the user at one time contains all the data related to the target matter, the target matter data can be determined based on the interaction data input at one time; when the interaction data input by the user at one time contains only part of the data related to the target matter, the user can be prompted to continue entering the interaction data to supplement the remaining data.

[0095] The target output format refers to the data output format preset for the target matter.

[0096] In some embodiments, different data output formats are set for different target matter data.

[0097] For example, when the target item is a to-do reminder, the target output format includes at least the reminder time, reminder content and reminder method; for another example, when the target item is a meeting reminder, the target output format includes at least the reminder time, meeting time, meeting location and meeting subject; for another example, when the target item is an appointment reminder, the target output format includes at least the reminder time, appointment time, related persons and appointment location; and so on.

[0098] In this way, the target processing model can extract target event data from the interaction data input by the target user once or multiple times, and generate target reminder events according to the preset target output format.

[0099] In some embodiments, the target reminder event may be an event output to the target user to prompt the target user whether to establish a corresponding reminder task. For example, after extracting data related to the scheduled meeting from the interaction data, a target reminder event is generated according to the output format of the meeting time, meeting subject, and meeting location, and the target reminder event is displayed and output to the target user, so that the target user can determine whether to establish a corresponding meeting reminder task based on the target reminder event.

[0100] In some embodiments, the target reminder event may also be a reminder task established on the terminal device of the target user. For example, after extracting the meeting reservation related data from the interaction data, a target reminder event is generated according to the output format of the meeting time, meeting subject, and meeting location, and the target reminder event is automatically recorded on the terminal device of the target user.

[0101] In the data processing method provided by the present application, first, in response to obtaining the interactive data input to the target application, the interactive data is input to the target processing model; then, when it is identified that the interactive data includes the target event data, the target event data is generated as a target reminder event according to the target output format using the target processing model. In this way, during the data processing process, by specifying the target output format for the target processing model, the output result of the target processing model can be interpreted based on the target output format, and the data processing flow of the target processing model can be standardized based on the output result, so that the target processing model can provide valuable executable operations when performing specific tasks.

[0102] In some embodiments, the interaction data is input into the target processing model, that is, the above step S10, can be implemented as the following step S101:

[0103] Step S101, processing the interaction data into corresponding prompt word data according to the target field format and inputting the corresponding prompt word data into the target processing model, so as to obtain target item data corresponding to the target field format from the interaction data;

[0104] Among them, the target item data that can be obtained based on prompt word data in different field formats are different.

[0105] Here, the interaction data may be interaction data directly input by the target user to the target application, or may be intermediate processing result data obtained after the target processing model performs model processing based on the interaction data directly input by the target user.

[0106] The target field format refers to the field format set according to the preset model processing flow to extract the event data corresponding to the specified model processing stage from the interaction data.

[0107] In some embodiments, the target field format may include a human setting field of the target processing model at the current processing stage, a model processing target field, a model output format field, and / or reference data of the current processing stage.

[0108] In some embodiments, the human settings of the target processing model at the current processing stage may include information extraction experts, information recognition experts, intelligent assistants, and the like.

[0109] In some embodiments, the model processing target refers to the model processing task to be completed by the model in the current model processing stage. In some embodiments, the model processing target may include extracting all event information related to life events from the reference data, identifying future events that need to be processed by the user from the reference data, judging whether there are to-do items from the reference data, generating to-do items including reminder time based on the reference data, generating query information related to the reference data, completing information on to-do items based on the reference data, and so on.

[0110] In some embodiments, the model output format field is used to specify that the model outputs the event data extracted from the reference data in a preset output format. For example, when the person who processes the target model is an information extraction expert, the output format field can specify that the data is output in the format of "event type", "event content" and "event title". In some embodiments, the model output format field can also specify the data format of the model output data, for example, the data format of the model output data can be specified as JSON (JavaScript Object Notation) format.

[0111] Thus, in some embodiments, prompt word data including different personal settings, model processing targets, model output formats and reference data can be generated according to the target field format and interaction data; after the prompt word data is input into the target processing model, the target processing model can process the reference data according to the prompt word data, that is, extract the target matter data corresponding to the target field format from the reference data.

[0112] At the same time, since at different stages of model processing, at least one of the human settings, model processing goals, model output formats and reference data contained in different field formats is different, the target event data that can be obtained by the prompt word data in different field formats is different.

[0113] For example, when the referenced data is the same in different field formats, but the human settings, model processing targets, or model output formats are different, the types of target event data output by the target processing model are different.

[0114] For another example, when the human setting, model processing target and model output format are the same in different field formats, but the referenced data are different, the content of the target event data output by the target processing model is different.

[0115] In some embodiments, the interaction data is processed into corresponding prompt word data according to the target field format and input into the target processing model, that is, the above step S101 can be implemented as at least one of the following steps S1011 to S1013:

[0116] Step S1011, processing the interaction data into first prompt word data according to a first field format and inputting the first prompt word data into the target processing model to output a first output result.

[0117] Here, the interaction data refers to the data inputted by the target user into the target application.

[0118] In this way, the interactive data input by the target user into the target application is processed into the first prompt word data according to the first field format.

[0119] For example, according to the preset model processing flow, the first prompt word data is used to extract all event information related to life events from the interactive data, then the first field format may include personality settings, model processing goals, model output format and input data of the target user; wherein, the personality setting is "information extraction expert", the model processing goal is to extract all event information related to life events from the input data, the output format is "event type: """, "event content: """, "event title: """, and the reference data is the interactive data entered by the target user.

[0120] In this way, after the first prompt word data is input into the target processing model, the first output result output by the target processing model can characterize whether there is target event data in the interaction data, that is, the first output result can characterize whether there are life events in the interaction data, or whether there is content related to life events.

[0121] For example, when the target processing model determines that the interaction data does not contain any life events, the fields such as "Event Type:"", "Event Content:"", and "Event Title:"" in the first output result output by the target processing model are all empty;

[0122] For another example, when the target processing model determines that the interaction data contains at least one life event, in the first output result output by the target processing model, target matter data corresponding to fields such as "Event type:"", "Event content:"", and "Event title:"" are generated for each life event.

[0123] Step S1012, processing the interaction data and the first output result output by the target processing model for the interaction data into second prompt word data according to a second field format, and inputting the second prompt word data into the target processing model to output a second output result.

[0124] Here, the interaction data refers to the data input by the target user to the target application, and the interaction data input by the target user and the first output result output by the target processing model for the interaction data are used as reference data in the second prompt word. It can be seen that the second prompt word data is the prompt information for controlling the target processing model to perform the next stage of data processing based on the data processing result of the target processing model in the previous stage.

[0125] For example, according to the preset model processing flow, the second prompt word data is used to determine whether there are events that future users need to handle from the interaction data and the first output result. The second field format may include personality settings, model processing goals, model output formats, and reference data; wherein the personality setting is "information identification expert", the model processing goal is to identify events that future users need to do from the input data and the first output result, the output format is "event content: """, "things that future users need to do: """, and the reference data is the interaction data and the first output result.

[0126] In this way, after the second prompt word data is input into the target processing model, the second output result output by the target processing model can represent whether there is an event that the user needs to do in the future in the reference data.

[0127] For example, the target processing model can identify each "event content" in the first output result based on the knowledge base related to the target user, and determine whether there is an event that the future user needs to handle. For example, if the identified "event content" is not an event that the future user needs to handle, in the second output result, the label value of "things that the future user needs to do:"" corresponding to the "event content" is set to "0"; otherwise, if the identified "event content" is an event that the future user needs to handle, in the second output result, the label value of "things that the future user needs to do:"" corresponding to the "event content" is set to "1".

[0128] It should be noted that the label value of “Things that users need to do in the future:”” can be set according to actual application needs and is not limited to “0” and “1”.

[0129] In this way, through the “Things that users need to do in the future:”” field in the second output result, it can be represented whether there is target item data in the referenced interaction data.

[0130] Step S1013, processing the interaction data and the second output result output by the target processing model for the interaction data into third prompt word data according to a third field format, and inputting the third prompt word data into the target processing model to output a third output result;

[0131] Here, the interaction data refers to the data input by the target user to the target application, and the interaction data and the second output result output by the target processing model for the interaction data and the first output result are used as reference data in the third prompt word. It can be seen that the third prompt word data is the prompt information for controlling the target processing model to perform the next stage of data processing based on the data processing result of the target processing model in the previous stage.

[0132] For example, according to the preset model processing flow, the third prompt word data is used to determine the to-do items from the interaction data and the second output results. Therefore, the third field format may include a model processing target, a model output format, and reference data; wherein the model processing target is to identify the to-do items from the interaction data and the second output results, the output format is “Whether to-do items are included: “”,” To-do content: “”,” and the reference data is the interaction data and the second output result.

[0133] In this way, after the third prompt word data is input into the target processing model, the second output result output by the target processing model can represent whether there is a to-do item in the interactive data input by the user.

[0134] In the above embodiment, the output result can indicate whether target item data exists in the interaction data.

[0135] Here, the target item data may include items that have been processed, to-do items that the user needs to process in the future, and information associated with these items, such as person information, location information, time information, etc. related to the to-do items.

[0136] Through the above-mentioned first output result, second output result and third output result, it can be gradually determined whether there are matters related to life events in the interactive data, whether there are events that users need to deal with in the future, whether there are to-do items and related related information.

[0137] It can be seen from the above embodiments that in the process of determining whether there are matters that need to be reminded in the interactive data input by the target user, since the first output result, the second output result and the third output result are all clear model inference results output according to the preset model output format, it can help the automated application to more accurately understand and analyze the inference results of the target processing model, determine the model processing target of the next stage based on the inference results of the previous stage, generate corresponding prompt word data according to the specified target field format, and thus determine whether there are reminders according to the executable model processing steps.

[0138] In some embodiments, the target event data is generated as a target reminder event according to the target output format using the target processing model, that is, the above step S11 can be implemented as at least one of the following steps S111 to S113:

[0139] Step S111, when the third output result represents that the interaction data includes target item data that can constitute a reminder event, the interaction data and the third output result are processed into fifth prompt word data according to the fifth field format and input into the target processing model, and a first reminder event is generated according to the target output format.

[0140] Here, the target item data that can constitute the reminder event refers to items that need to be processed by the user in the future. In some embodiments, items that the target user has started but not yet completed or items that the target user has newly created and needs to be processed in the future can be output as to-do items in the third output result.

[0141] In some embodiments, when the target event data included in the third output result that can constitute a reminder event includes time information, the interaction data and the third output result are processed into fifth prompt word data according to the fifth field format.

[0142] In some embodiments, the fifth field format includes a personality setting field, a model processing target field, a model output format field and a reference data field; wherein the personality setting field indicates that the current personality of the target processing model is a schedule assistant; the model processing target field indicates that the target processing model reminds the user of to-do items at the appropriate time; the model output format field indicates that the output format of the target processing model is "Whether to include time: "", "Time to remind the user: "", "To-do content: """; the reference data includes the interaction data input by the target user and the third output result.

[0143] In some embodiments, when the third output result includes multiple to-do items (i.e., target item data that can constitute a reminder event), corresponding fifth prompt word data is generated for each to-do item, and the generated multiple fifth prompt word data are input into the target processing model in sequence to output the corresponding first reminder event in accordance with the target output format, i.e., the model output format in the above-mentioned fifth prompt word data.

[0144] Step S112: when the third output result does not include the time information required to constitute a reminder event, use the target processing model to obtain second time information to generate a second reminder event based on the second time information.

[0145] Here, when the third output result includes target matter data that can constitute a reminder event, and the target matter data does not include the time information required to constitute the reminder event, the target processing model is used to obtain the reminder time information corresponding to the target matter data, that is, the second time information, so as to generate a second reminder event based on the second time information and the corresponding target matter data.

[0146] For example, the third output result includes target event data of ordering flowers, but the target event data does not include corresponding ordering time data. The target processing model is used to inquire the target user about the flower ordering time information, thereby determining the second time information, and generating a second reminder event about ordering flowers based on the second time information.

[0147] In some embodiments, the to-do items that do not include the time information required to constitute the reminder event are stored in the to-do item database. In this way, the to-do items can be obtained one by one from the to-do item database and the corresponding time query information can be sent to the target user. In some embodiments, if the target user does not return the time information corresponding to the to-do item, the to-do item can be saved in the to-do item database, so that when the target user mentions information related to the to-do item, the to-do item can be directly obtained from the to-do item database.

[0148] Step S113, when the first time information of the first reminder event generated by the target processing model is unavailable, obtaining second time information by using the target processing model to generate a second reminder event based on the second time information.

[0149] Here, the first time information of the first reminder event is unavailable, which means that in the process of the target processing model generating the first reminder event based on the interaction data input by the target user, it is determined that the time information in the first reminder event is unavailable. For example, when the first time information is false time information, expired time information, or time information that does not conform to the composition of the event to be reminded, it is determined that the first time information is unavailable.

[0150] In some embodiments, based on the third output result, it is determined that the interaction data input by the target user includes a to-do item, and the to-do event has corresponding first time information, then fifth prompt word data is generated based on the to-do item and the corresponding first time information, and the fifth prompt word data is input into the target processing model; in the process of the target processing model generating a first reminder event corresponding to the to-do item based on the fifth prompt word data, the target processing model determines that the first time information is not available based on context information.

[0151] In this way, the target processing model is used to obtain the reminder time information corresponding to the to-do item, that is, the second time information, so as to generate a second reminder event based on the second time information and the corresponding target item data.

[0152] For example, the third output result includes data on ordering flowers and the order quantity. When the order quantity is identified as the ordering time information (i.e., the first time information), fifth prompt word data is generated for the matter of ordering flowers, and the fifth prompt word data is input into the target processing model; in the process of the target processing model generating a first reminder event corresponding to the matter of ordering flowers based on the fifth prompt word data, the target processing model determines based on the context information that the first time information is the order quantity information of the flowers, rather than the ordering time information, then the target processing model determines the first time information in the reminder event of ordering flowers as unavailable, and further obtains the second time information to generate a second reminder event based on the second time information.

[0153] In some embodiments, the to-do items corresponding to the first reminder time for which the first time information is unavailable are stored in a to-do item database. In this way, the to-do items can be obtained one by one from the to-do item database and the corresponding time inquiry information can be sent to the target user. In some embodiments, if the target user does not return the time information corresponding to the to-do item, the to-do item can be saved in the to-do item database, so that when the target user mentions information related to the to-do item, the to-do item can be directly obtained from the to-do item database.

[0154] In some embodiments, the step S112 or S113 of acquiring the second time information by using the target processing model to generate the second reminder event based on the second time information may be implemented as the following steps S114 to S116:

[0155] Step S114, controlling the target processing model to output question information regarding the time information required to constitute a reminder event.

[0156] Here, for a to-do item that does not include time information required to constitute a reminder event or a to-do item for which the first time information of the first reminder event is unavailable, question information is generated using a target processing model and sent to a target user.

[0157] In some embodiments, sixth prompt word data is generated in accordance with a sixth field format based on a to-do item that does not include the time information required to constitute a reminder event or a to-do item for which the first time information is unavailable, and the interaction data input by the target user; wherein the sixth field format includes a persona setting field, a model processing target field, and a reference data field; wherein the persona setting field is used to instruct the target processing model not to repeat the model persona; the model processing target field is used to instruct the target processing model to communicate with the user in colloquial language and ask the user when he wants to be reminded to process the to-do items in the reference data; and the reference data field is used to indicate the corresponding to-do items and the interaction data input by the target user.

[0158] In this way, the sixth prompt word data is input into the target processing model so that the target processing model generates question information.

[0159] Step S115: using the target processing model to obtain the second time information from the interactive feedback data for the question information.

[0160] Here, the question information generated by the target processing model is sent to the target user, and the corresponding time feedback information is obtained from the user; the user's time feedback information is input into the target processing model so that the model can identify the second time information from the user's time feedback information, that is, the reminder time information of the corresponding to-do event.

[0161] Step S116: input the second time information and the third output result into the target processing model, and generate a second reminder event according to the target output format.

[0162] Here, the to-do items that do not include the time information required to constitute the reminder event or the to-do items for which the first time information is unavailable and the corresponding second time information are input into the target processing model to generate the second reminder event using the target processing model.

[0163] In some embodiments, the second time information and the third output result are processed into seventh prompt word data according to a seventh field format, and input into a target processing model to generate a second reminder event.

[0164] In some embodiments, the seventh field format includes a model processing target field, a model output format field and a reference data field; wherein, the model processing target field is used to indicate that the target processing model generates a reminder event based on the second time information and to-do items in the reference data; the model output format field is used to indicate that the target processing model outputs the data processing results in the format of "whether to include time: "", "time to remind the user: """ and "to-do content: """; the reference data includes the interaction data input by the target user, the third output result and the second time information.

[0165] In this way, after the seventh prompt word data is input into the target processing model, the target processing model is used to generate a second reminder event according to the target output format, that is, the model output format in the seventh field format.

[0166] In some embodiments, the method further includes at least one of the following steps S117 and S118:

[0167] Step S117, when the third output result indicates that the interaction data does not include the target item data, processing the third output result into fourth prompt word data in a fourth field format, so that the target processing model responds to the fourth prompt word data in a conversation mode;

[0168] Step S118, when the time information of the second reminder event generated by the target processing model is unavailable, processing the third output result into fourth prompt word data in a fourth field format, so that the target processing model responds to the fourth prompt word data in a conversation mode.

[0169] Here, it is determined based on the third output result of the target processing model that the interactive data does not include to-do items. For example, if the items that the user needs to handle in the future in the second output result are all items that the user has already started to handle, then the third output result will not include the relevant to-do items. At this time, the fourth prompt word data is generated based on the third output result in the fourth field format to make the target processing model enter the conversation mode; or,

[0170] When the time information of the second reminder event is not available, that is, when the time information in the time feedback information retrieved from the target user for the to-do item is not available, fourth prompt word data is generated in a fourth field format based on the third output result to enable the target processing model to enter a conversation mode.

[0171] In some embodiments, the fourth field format includes a model processing target field and a reference data field; wherein, the model processing target field is used to indicate that the model asks questions to the user with the reference data as the background; the reference data includes the third output result and the interaction data input by the target user.

[0172] In this way, for at least one to-do item that does not contain available time information, a dialog message is generated using the target processing model, and the dialog message is sent to the target user to enter a chat mode or a question-and-answer mode.

[0173] In some embodiments, the data processing method further includes at least one of the following steps S13 to S16:

[0174] Step S13: outputting a prompt of whether to add the target reminder event to the to-do schedule, and adding the target reminder event to the to-do schedule or not responding to the prompt based on the feedback information of the target user.

[0175] Here, after determining the target reminder event, a prompt is output to the target user as to whether to add the target reminder event to the to-do schedule (for example, a pop-up prompt); in response to the user's feedback information that determines to add it to the to-do schedule (for example, the user enters "yes" in the pop-up window or clicks a pop-up control that displays "yes"), the target reminder event is added to the to-do schedule; in response to the user's feedback information that determines not to add it to the to-do schedule (for example, the user enters "no" in the pop-up window or clicks a pop-up control that displays "no"), the target reminder event is not responded to.

[0176] In some embodiments, when there are multiple target reminder events, reminders are output respectively for the multiple target reminder events.

[0177] Step S14, sending the target reminder event to the target terminal device of the target user, so as to synchronize the target reminder event to the schedule management application of the target terminal device or reject the prompt of the target reminder event based on the user feedback information.

[0178] Here, when the application used to execute the data processing method provided in this application is AI Center, or any other application running on the cloud or edge server, the application is used to send the target reminder event to the target terminal device of the target user, so that the target user can obtain the target reminder event information through the target terminal device.

[0179] In some embodiments, the target terminal device can be a user's laptop computer, tablet computer, desktop computer, smart TV, set-top box, mobile device (such as a mobile phone, portable video player, personal digital assistant, dedicated messaging device, wireless headset) and other devices with data processing capabilities.

[0180] After the target reminder event is sent to the target terminal device of the target user, the user can enter feedback information on the target terminal device regarding whether to add the target reminder event to the to-do schedule; then, the target terminal device sends the user's feedback information to the AI ​​Center or the application running in the cloud or edge device.

[0181] In this way, in response to the user's feedback on adding the to-do schedule, AI Center or the application running in the cloud or edge device synchronizes the target reminder event to the schedule management application of the target terminal device; in response to the user's feedback information on not adding the to-do schedule, AI Center or the application running in the cloud or edge device does not respond to the target reminder event.

[0182] Step S15: storing at least one inference result output by the target processing model for the interaction data into a corresponding target storage area.

[0183] Here, in the process of processing interactive data using the target processing model, at least one inference result output by the target processing model is stored in the corresponding target storage area, so that the target processing model can query the inference result information in the target storage area according to the data processing process.

[0184] For example, the first output result, the second output result and / or the third output result generated by the target processing model can be stored in the corresponding storage area, so that the target processing model can read the matter information from the corresponding storage area in sequence and perform model reasoning for each matter information.

[0185] Step S16, using the target processing model to evaluate whether the matter data in the interaction data constitutes the risk and / or value of a reminder event, so as to determine the target matter data from the interaction data based on the risk and / or value.

[0186] Here, whether the matter data in the interaction data constitutes the risk and / or value of the reminder event means determining whether the matter data in the interaction data will bring risks and / or value to the user if not executed by judging the association between the matter data in the interaction data and the historical data of the target user.

[0187] For example, if the event data in the interaction data is associated with the user's historical data, and if the event data in the interaction data is not executed, the user will suffer a loss of available benefits, then the event data is considered to have the risk and / or value of constituting a reminder event;

[0188] For another example, if the event data in the interaction data is not associated with the user's historical data, it is considered that the event data has no value in constituting a reminder event.

[0189] In some embodiments, the data processing method provided by the present application further includes the following steps S17 to S18:

[0190] Step S17, verifying the output format of the inference result to be output obtained by the target processing model for the interactive data processing and / or the configuration parameters of the target processing model;

[0191] Step S18, based on the verification result, the target processing model is controlled to output the corresponding reasoning result, perform re-reasoning, or not output the reasoning result.

[0192] Here, the inference result to be output may include the inference result to be output corresponding to any output result from processing the interaction data by the target processing model to determining the target reminder event.

[0193] In some embodiments, verifying the output format of the inference result to be output may include verifying whether the output format of the result to be output satisfies the model output format specified in the corresponding prompt word data.

[0194] In some embodiments, verifying the configuration parameters of the target processing model may include verifying whether the inference result to be output generated by the target processing model satisfies the output format information configured for the model, or the output format information of the model itself, etc. For example, when the data output format of the model is configured as JSON format, if the data format of the inference result to be output is in JSON format, it is determined that the verification is passed, otherwise it is determined that the verification is not passed.

[0195] In this way, when the inference result to be output passes the verification, the control target processing model outputs the corresponding inference result; when the inference result to be output does not pass the verification, the control target processing model re-reasons or does not output the inference result.

[0196] In some embodiments, for the inference results to be output that have not passed the verification, the prompt word information corresponding to the inference results to be output is returned to the target processing model to re-execute the model inference.

[0197] In some embodiments, the data processing method provided by the present application further includes at least one of the following steps S19 to S110:

[0198] Step S19, determining the correlation between the inference result to be output and the interaction data; based on the correlation, controlling the target processing model to output the corresponding inference result, perform re-inference, or not output the inference result.

[0199] Here, the relevance between the inference result to be output and the interaction data input by the target user is determined, and whether to output the corresponding inference result is determined according to the relevance. For example, when the relevance is higher than a specified relevance threshold, the corresponding inference result is output; when the relevance is not higher than the specified relevance threshold, the prompt word data corresponding to the corresponding inference result to be output is sent to the target processing model for re-inference, or the corresponding inference result is not output.

[0200] In some embodiments, the relevance of the inference result to be output and the interaction data may be determined based on the relevance of the inference result to be output and the corresponding prompt word data.

[0201] In some embodiments, the relevance between the inference result to be output and the prompt word data is determined based on the context perception machine. For example, based on the context perception machine, it is determined whether the output format of the inference result to be output meets the requirements of the model output format field in the prompt word data; or, based on the context perception machine, it is determined the relevance between the inference result to be output and the model person setting, model processing target and / or reference data in the prompt word data.

[0202] In this way, through the context perception machine, the output results can be traced back according to the context, thereby improving the accuracy of the model output results.

[0203] Step S110, based on multiple sample data, using the model to be trained, generate multiple prediction outputs; based on the sample outputs and prediction outputs corresponding to the multiple sample data, calculate loss information; based on the loss information, update the model to be trained to obtain the target processing model; wherein the sample output representation is obtained after processing the sample data in a specified format.

[0204] Here, the plurality of sample data are data used to train the model to be trained. In some embodiments, each sample data is a randomly generated text segment, and each sample data has a corresponding sample output in a specified format.

[0205] In some embodiments, the specified format may be a JSON format. In this way, the output of each inference result of the trained target processing model is in a JSON format, which can help the model to more accurately understand and analyze each event, thereby better executing subsequent model processing steps.

[0206] In some embodiments, multiple sample data may be input into a JSON format validation generator to generate corresponding sample outputs.

[0207] In this way, multiple sample data are input into the model to be trained, and the model to be trained outputs multiple prediction outputs; wherein the multiple prediction outputs are output by the model to be trained based on the prompt words in a specified format.

[0208] In some embodiments, the loss information between the predicted output and the sample output may be calculated using a cross-entropy loss function, a mean square error loss (MSE Loss) function, a mean absolute error loss (MAE Loss) function, or any type of loss function commonly used in the art.

[0209] In this way, the loss information is used to update the model to be trained until the preset convergence conditions are met to obtain the target processing model.

[0210] In some embodiments, the model to be trained can be trained using a reinforcement learning mechanism. In this way, when the loss between the predicted output and the sample output is small, the model to be trained is given extra points, and when the loss is large, the model to be trained is deducted points, thereby strengthening the weight of the model to be trained for the output of the specified format.

[0211] In the above embodiment, by using multiple sample data to train the model to be trained, the trained model can output the inference result data in a specified output format, thereby ensuring the consistency and interpretability of the model output format, solving the problem that the model output in the related technology is too broad, unstable and complex, and allowing the automation program to determine the reasoning logic for fixed tasks based on the output results, improve the engineering effect of the model, and thereby improve the reliability and efficiency of the model application.

[0212] Next, combine Figure 2 , the execution sequence of an embodiment of the data processing method provided by the present application is described. Among them, the user inputs the interactive data through the user terminal 210, and the application terminal 220 is used to receive the interactive data sent by the user terminal and call the model 230 to perform data processing.

[0213] Step S21, the user end 210 sends a voice message to the application end 220;

[0214] Here, the user sends voice to the application end 220 through the microphone in the user end 210; the application end 220 converts the voice into corresponding text data.

[0215] Step S22, the application end 220 sends a first prompt message to the model 230;

[0216] Here, the application end 220 generates first prompt information based on the text data; wherein the first prompt information is used to instruct the model 230 to extract all event information related to life events from the text data.

[0217] Step S23, the model 230 returns multiple event information to the application end 220;

[0218] Here, the model 230 extracts multiple life-related event information from the text data according to the first prompt information, and returns the multiple event information to the application end 220 .

[0219] Step S24, the application end 220 stores the event information;

[0220] Here, the application 220 stores the multiple event information identified by the model 230 in the database 240 .

[0221] Step S25, the application end 220 reads event information from the data block 240;

[0222] Step S26, the application end 220 sends a second prompt message to the model 230;

[0223] Here, the application end 220 generates second prompt information based on the read event information; wherein the second prompt information is used to instruct the model 230 to identify events that the user needs to handle in the future from the text data.

[0224] Step S27, the model 230 returns the event that the user needs to process in the future to the application end 220;

[0225] Here, the model 230 determines from the text data an event that the user needs to handle in the future based on the second prompt information.

[0226] Step S28, the application end 220 stores the events that need to be processed by the user in the future in the database 250;

[0227] Step S29, the application end 220 reads the events that the user needs to process in the future from the database 250;

[0228] Step S211, the application end 220 sends a third prompt message to the model 230;

[0229] Here, after the application end 220 reads an event that the future user needs to handle from the database 250, it generates a third prompt information based on the event; wherein the third prompt information is used to instruct the model 230 to determine the to-do items based on the event that the future user needs to handle and the text data.

[0230] Step S212, the model 230 returns the to-do items to the application end 220;

[0231] Here, the model 230 determines whether the text data contains a to-do item based on the third prompt information, and returns the to-do item to the application end 220 .

[0232] Step S213, the application 220 stores the to-do items in the database 260;

[0233] Step S214, the application end 220 reads the first to-do item from the database 260;

[0234] Here, the database 260 stores a plurality of to-do items, and the application end 220 can read the to-do items from the database 260 in sequence and perform subsequent processing.

[0235] Step S215, the application end 220 sends a fourth prompt message to the model 230;

[0236] Here, the first to-do item includes corresponding time information, so the application end 220 generates fourth prompt information based on the first to-do item to prompt the model 230 to generate a corresponding reminder event.

[0237] Step 216, the model 230 returns the third reminder event to the application end 220;

[0238] Step S217, the application end 220 returns the third reminder event to the user end 210;

[0239] Here, if there is no to-do item to be processed in the database 260, step S229 is executed;

[0240] Step S218, the application end 220 reads the second to-do item from the database 260;

[0241] Here, the second to-do item represents an item that has no corresponding reminder time.

[0242] Step S219, the application end 220 sends the fifth prompt information to the model 230;

[0243] Here, the application end 220 generates fifth prompt information based on the second to-do item and the voice information input by the user; wherein the fifth prompt information is used to prompt the model 230 to generate time query information for the second to-do item.

[0244] Step S221, the model 230 sends time query information to the application end 220;

[0245] Here, the model 230 generates time query information based on the fifth prompt information, and sends it to the application end 220 .

[0246] Step S222, the application end 220 sends time query information to the user end 210;

[0247] Step S223, the user returns a time reply to the application end 220 via the user end 210;

[0248] Step S224, the application end 220 sends sixth prompt information to the model 230;

[0249] Here, the application end 220 generates sixth prompt information based on the time reply returned by the user end 210; wherein the sixth prompt information is used to prompt the model 230 to extract time information from the time reply returned by the user.

[0250] Step S225, the model 230 returns the time information to the application end 220;

[0251] Here, the model 230 returns the time information extracted from the time reply returned by the user to the application end 230 .

[0252] When the model 230 determines that the time reply returned by the client 210 is unavailable, step S229 is executed.

[0253] Step S226, the application end 220 sends the seventh prompt information to the model 230;

[0254] Here, the application 220 generates seventh prompt information based on the time information and the second to-do item; wherein the seventh prompt information is used to prompt the model 230 to generate a fourth reminder event based on the second to-do item, time information and text data.

[0255] Step S227, the model 230 returns the fourth reminder event to the application end 220;

[0256] Here, the model 230 generates a fourth reminder event for the second to-do item based on the seventh reminder information, and returns the fourth reminder event to the application end 220 .

[0257] Step S228, the application end 220 returns the fourth reminder event to the user end 210;

[0258] Step S229, the application end 220 sends the eighth prompt information to the model 230;

[0259] Here, the application determines that the to-do items in the data block 260 have been processed, and sends an eighth prompt message to the model 230; wherein the eighth prompt message is used to generate chat information based on the text data.

[0260] Step S231, the model 230 returns the chat information to the application end 220;

[0261] Step S232 , the application end 220 returns the chat information to the user end 210 .

[0262] It can be seen from the above embodiments that in the data processing method provided by the present application, by specifying the processing target and output format of the model in each reasoning stage of the model 230, the output result of the model 230 has higher interpretability and higher accuracy, so that the reasoning target of the model in the following reasoning stage can be determined based on the output result of the model 230, thereby improving the executableness of the model reasoning.

[0263] Next, combine Figure 3 , an example of training a model according to the data processing method provided in this application is described. Figure 3 As shown, this embodiment includes the following steps S301 to S304:

[0264] Step S301, using a JSON format verification generator to generate multiple sample data; then, executing step S302;

[0265] Step S302, using the model to be trained, generating multiple prediction outputs corresponding to multiple sample data; then, executing step S303;

[0266] Step S303, based on the sample outputs of the multiple sample data and the multiple predicted outputs, using the reinforcement learning mechanism, scoring the output of the model to be trained; then, executing step S304;

[0267] Step S304: based on the scoring result, update the weight parameters of the model to be trained to obtain a trained model.

[0268] Next, combine Figure 4 , an embodiment of verifying the inference result to be output according to the data processing method provided by the present application is described. Figure 4 As shown, this embodiment includes the following steps S401 to S403:

[0269] Step S401, based on the prompt information, generate the inference result to be output; then, execute step S402;

[0270] Step S402, using the context perception machine, determining whether the correlation between the inference result to be output and the prompt information is higher than the correlation threshold; if so, executing step S403; if not, executing step S401;

[0271] Here, if the correlation between the inference result to be output and the prompt information is lower than the correlation threshold, return to step S401 to re-execute model inference; and after returning to step S401 more than 3 times, no longer return to step S401, and output a prompt information of model inference error.

[0272] Step S403: output the inference result to be output.

[0273] Based on the aforementioned embodiments, the present application provides an electronic device. Figure 5 A schematic diagram of the hardware structure of an electronic device provided in this application, such as Figure 5As shown, the electronic device 500 includes at least one processor 510 and at least one processing model that can run on the processor 510, and the processor can perform at least one of the following:

[0274] In response to obtaining interaction data input to a target application, inputting the interaction data to a target processing model, the target application being an application capable of invoking the target processing model;

[0275] In the case where it is identified that the interaction data includes target event data, the target event data is generated into a target reminder event according to a target output format using the target processing model.

[0276] In some embodiments, inputting the interaction data into a target processing model comprises:

[0277] Processing the interaction data into corresponding prompt word data according to the target field format and inputting the data into the target processing model, so as to obtain target item data corresponding to the target field format from the interaction data;

[0278] Among them, the target item data that can be obtained based on prompt word data in different field formats are different.

[0279] In some embodiments, processing the interaction data into corresponding prompt word data according to the target field format and inputting it into the target processing model includes at least one of the following:

[0280] Processing the interaction data into first prompt word data according to a first field format and inputting the data into the target processing model to output a first output result;

[0281] Processing the interaction data and a first output result output by the target processing model for the interaction data into second prompt word data according to a second field format, and inputting the second prompt word data into the target processing model to output a second output result;

[0282] Processing the interaction data and a second output result output by the target processing model for the interaction data into third prompt word data according to a third field format, and inputting the third prompt word data into the target processing model to output a third output result;

[0283] The output result can indicate whether target item data exists in the interaction data.

[0284] In some embodiments, using the target processing model to generate the target event data into a target reminder event according to the target output format includes at least one of the following:

[0285] In a case where the third output result indicates that the interaction data includes target item data that can constitute a reminder event, the interaction data and the third output result are processed according to a fifth field format into fifth prompt word data, which are input into the target processing model, and a first reminder event is generated according to the target output format;

[0286] In a case where the third output result does not include time information required to constitute a reminder event, obtaining second time information using the target processing model to generate a second reminder event based on the second time information;

[0287] In a case where the first time information of the first reminder event generated by the target processing model is unavailable, the target processing model is used to obtain second time information to generate a second reminder event based on the second time information.

[0288] In some embodiments, obtaining second time information by using the target processing model to generate a second reminder event based on the second time information includes:

[0289] Controlling the target processing model to output question information for time information required to constitute a reminder event;

[0290] Acquire the second time information from the interactive feedback data for the question information using the target processing model;

[0291] The second time information and the third output result are input into the target processing model, and a second reminder event is generated according to the target output format.

[0292] In some embodiments, the processor is further configured to perform at least one of the following:

[0293] In the case where the third output result indicates that the target item data is not included in the interaction data, processing the third output result into fourth prompt word data in a fourth field format, so that the target processing model responds to the fourth prompt word data in a conversation mode;

[0294] When the time information of the second reminder event generated by the target processing model is unavailable, the third output result is processed into fourth prompt word data in a fourth field format so that the target processing model responds to the fourth prompt word data in a conversation mode.

[0295] In some embodiments, the processor is further configured to perform at least one of the following:

[0296] Outputting a prompt of whether to add the target reminder event to the to-do schedule, and adding the target reminder event to the to-do schedule or not responding to the prompt based on the feedback information of the target user;

[0297] Sending the target reminder event to the target terminal device of the target user, so as to synchronize the target reminder event to the schedule management application of the target terminal device or reject the prompt of the target reminder event based on the user feedback information;

[0298] storing at least one inference result output by the target processing model for the interaction data in a corresponding target storage area;

[0299] The target processing model is used to evaluate whether the matter data in the interaction data constitutes the risk and / or value of a reminder event, so as to determine the target matter data from the interaction data based on the risk and / or value.

[0300] In some embodiments, the processor is further configured to execute:

[0301] Verifying the output format of the inference result to be output obtained by the target processing model for the interactive data processing and / or the configuration parameters of the target processing model;

[0302] Based on the verification result, the target processing model is controlled to output the corresponding reasoning result, perform re-reasoning, or not output the reasoning result.

[0303] In some embodiments, the processor is further configured to perform at least one of the following:

[0304] Determine the relevance between the inference result to be output and the interaction data; based on the relevance, control the target processing model to output the corresponding inference result, perform re-inference, or not output the inference result;

[0305] Based on multiple sample data, multiple prediction outputs are generated using the model to be trained; based on the sample outputs and prediction outputs corresponding to the multiple sample data, loss information is calculated; based on the loss information, the model to be trained is updated to obtain the target processing model; wherein the sample output representation is obtained after processing the sample data in a specified format.

[0306] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. In some embodiments, the functions or units included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiment. For technical details not disclosed in the device embodiment of the present application, please refer to the description of the method embodiment of the present application for understanding.

[0307] It should be noted that in the embodiment of the present application, if the above-mentioned data processing method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific hardware, software or firmware, or any combination of hardware, software, and firmware.

[0308] An embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0309] The embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium can be transient or non-transient.

[0310] An embodiment of the present application provides a computer program, including a computer-readable code. When the computer-readable code is run in a computer device, a processor in the computer device executes some or all of the steps for implementing the above method.

[0311] The embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be implemented specifically by hardware, software or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium, and in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK) and the like.

[0312] It should be noted here that the description of the various embodiments above tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. The description of the above device, storage medium, computer program and computer program product embodiments is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the embodiments of the device, storage medium, computer program and computer program product of this application, please refer to the description of the method embodiment of this application for understanding.

[0313] It should be noted that Figure 6 A schematic diagram of a hardware entity of an electronic device in this application, such as Figure 6 As shown, the hardware entity of the electronic device 600 includes: a processor 601, a communication interface 602 and a memory 603, wherein:

[0314] The processor 601 generally controls the overall operation of the electronic device 600 .

[0315] The communication interface 602 enables the electronic device to communicate with other terminals or servers through a network.

[0316] The memory 603 is configured to store instructions and applications executable by the processor 601, and can also cache data to be processed or processed by the processor 601 and each module in the electronic device 600 (for example, image data, audio data, voice communication data, and video communication data), which can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM). Data transmission can be performed between the processor 601, the communication interface 602, and the memory 603 through the bus 604.

[0317] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the serial number of each step / process mentioned above does not mean the order of execution, and the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application. The serial numbers of the embodiments of the present application mentioned above are for description only and do not represent the advantages and disadvantages of the embodiments.

[0318] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0319] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0320] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0321] In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be a separate unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0322] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, etc., various media that can store program codes.

[0323] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0324] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A data processing method, comprising: In response to obtaining interaction data input to a target application, inputting the interaction data to a target processing model, the target application being an application capable of invoking the target processing model; In the case where it is identified that the interaction data includes target event data, the target event data is generated into a target reminder event according to a target output format using the target processing model.

2. The method according to claim 1, wherein: Inputting the interaction data into a target processing model comprises: Processing the interaction data into corresponding prompt word data according to the target field format and inputting the data into the target processing model, so as to obtain target item data corresponding to the target field format from the interaction data; Among them, the target item data that can be obtained based on prompt word data in different field formats are different.

3. The method according to claim 2, wherein: Processing the interaction data into corresponding prompt word data according to the target field format and inputting the data into the target processing model includes at least one of the following: Processing the interaction data into first prompt word data according to a first field format and inputting the data into the target processing model to output a first output result; Processing the interaction data and a first output result output by the target processing model for the interaction data into second prompt word data according to a second field format, and inputting the second prompt word data into the target processing model to output a second output result; Processing the interaction data and a second output result output by the target processing model for the interaction data into third prompt word data according to a third field format, and inputting the third prompt word data into the target processing model to output a third output result; The output result can indicate whether target item data exists in the interaction data.

4. The method according to claim 3, wherein: Generating the target event data into a target reminder event according to a target output format using the target processing model includes at least one of the following: In a case where the third output result indicates that the interaction data includes target item data that can constitute a reminder event, the interaction data and the third output result are processed according to a fifth field format into fifth prompt word data, which are input into the target processing model, and a first reminder event is generated according to the target output format; In a case where the third output result does not include time information required to constitute a reminder event, obtaining second time information using the target processing model to generate a second reminder event based on the second time information; In a case where the first time information of the first reminder event generated by the target processing model is unavailable, the target processing model is used to obtain second time information to generate a second reminder event based on the second time information.

5. The method according to claim 4, wherein: Acquiring second time information by using the target processing model to generate a second reminder event based on the second time information includes: Controlling the target processing model to output question information for time information required to constitute a reminder event; Acquire the second time information from the interactive feedback data for the question information using the target processing model; The second time information and the third output result are input into the target processing model, and a second reminder event is generated according to the target output format.

6. The method according to claim 5, further comprising at least one of the following: In the case where the third output result indicates that the target item data is not included in the interaction data, processing the third output result into fourth prompt word data in a fourth field format, so that the target processing model responds to the fourth prompt word data in a conversation mode; When the time information of the second reminder event generated by the target processing model is unavailable, the third output result is processed into fourth prompt word data in a fourth field format so that the target processing model responds to the fourth prompt word data in a conversation mode.

7. The method according to claim 1, further comprising at least one of the following: Outputting a prompt of whether to add the target reminder event to the to-do schedule, and adding the target reminder event to the to-do schedule or not responding to the prompt based on the feedback information of the target user; Sending the target reminder event to the target terminal device of the target user, so as to synchronize the target reminder event to the schedule management application of the target terminal device or reject the prompt of the target reminder event based on the user feedback information; storing at least one inference result output by the target processing model for the interaction data in a corresponding target storage area; The target processing model is used to evaluate whether the matter data in the interaction data constitutes the risk and / or value of a reminder event, so as to determine the target matter data from the interaction data based on the risk and / or value.

8. The method according to claim 1, further comprising: Verifying the output format of the inference result to be output obtained by the target processing model for the interactive data processing and / or the configuration parameters of the target processing model; Based on the verification result, the target processing model is controlled to output the corresponding reasoning result, perform re-reasoning, or not output the reasoning result.

9. The method according to claim 8, further comprising at least one of the following: Determine the relevance between the inference result to be output and the interaction data; based on the relevance, control the target processing model to output the corresponding inference result, perform re-inference, or not output the inference result; Based on multiple sample data, using the model to be trained, generate multiple prediction outputs; Calculating loss information based on the sample outputs and predicted outputs corresponding to the plurality of sample data; Based on the loss information, the model to be trained is updated to obtain the target processing model; wherein the sample output representation is obtained by processing the sample data in a specified format.

10. An electronic device comprising at least one processor and at least one processing model capable of running on the processor, wherein the processor is capable of performing at least one of the following: In response to obtaining interaction data input to a target application, inputting the interaction data to a target processing model, the target application being an application capable of invoking the target processing model; In the case where it is identified that the interaction data includes target event data, the target event data is generated into a target reminder event according to a target output format using the target processing model.

Images