Event reminding method and device, electronic device, and storage medium

Abstract

The application provides an event reminding method and device, an electronic device and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: in response to the active reminding mode of a first device being turned on and the network connection between the first device and a second device being interrupted, performing the step of offline extraction of reminding information, obtaining the to-do event of a target user and the corresponding to-do time; the target user is a user using the first device; and event reminding is performed based on the to-do event of the target user and the corresponding to-do time. The event reminding method and device, the electronic device and the storage medium provided by the application can improve the intelligent degree of a wearable device.

Description

Technical Field

[0001] This application belongs to the field of artificial intelligence technology, and more specifically, relates to an event reminder method and device, electronic device, and storage medium. Background Technology

[0002] With the development of artificial intelligence technology, by inputting users' dialogue information into AI models, to-do items can be automatically extracted. The way to automatically extract to-do items changes the traditional manual setting and event reminder mode, further improving the intelligence of electronic devices.

[0003] Wearable devices such as smart audio glasses and smartwatches are small in size and low in power consumption, making it impossible to directly deploy large AI models. Therefore, there is an urgent need to propose an event reminder method that can automatically extract to-do items in wearable devices to improve their intelligence. Summary of the Invention

[0004] The purpose of this application is to provide an event reminder method, device, electronic device, and storage medium to improve the intelligence level of wearable devices.

[0005] A first aspect of this application provides an event notification method, including:

[0006] In response to the activation of the active reminder mode of the first device and the interruption of the network connection between the first device and the second device, the step of offline extraction of reminder information is performed to obtain the to-do events and corresponding to-do times of the target user; the target user is the user using the first device;

[0007] Event reminders are generated based on the target user's to-do events and corresponding to-do times.

[0008] The offline extraction of reminder information includes the following steps:

[0009] Obtain dialogue information;

[0010] Convert the dialogue information into text information;

[0011] The text information is segmented into words to obtain multiple words and the character index range of each word; the text information is then split into sentences to obtain multiple sentence units and the character index range of each sentence unit.

[0012] Filter multiple word segments that match a preset first keyword library from the multiple word segments; sort the multiple matching word segments in ascending order of character index to obtain the target word segment sequence;

[0013] The target word segmentation sequence is segmented based on the character index range of each sentence unit to obtain multiple target word segmentation segments. The sentence unit corresponding to the character index range of each target word segment is taken as the first text segment.

[0014] The second text segment is selected from multiple first text segments based on a preset semantic template;

[0015] Based on the second text fragment, extract the target user's to-do events and corresponding to-do times.

[0016] A second aspect of this application provides an event reminder device, comprising:

[0017] The information extraction module is used to respond to the activation of the active reminder mode of the first device and the interruption of the network connection between the first device and the second device, and to perform the step of offline extraction of reminder information to obtain the to-do events and corresponding to-do times of the target user; the target user is the user using the first device;

[0018] The event reminder module is used to provide event reminders based on the target user's to-do events and corresponding to-do times;

[0019] The offline extraction of reminder information includes the following steps:

[0020] Obtain dialogue information;

[0021] Convert the dialogue information into text information;

[0022] The text information is segmented into words to obtain multiple words and the character index range of each word; the text information is then split into sentences to obtain multiple sentence units and the character index range of each sentence unit.

[0023] Filter multiple word segments that match a preset first keyword library from the multiple word segments; sort the multiple matching word segments in ascending order of character index to obtain the target word segment sequence;

[0024] The target word segmentation sequence is segmented based on the character index range of each sentence unit to obtain multiple target word segmentation segments. The sentence unit corresponding to the character index range of each target word segment is taken as the first text segment.

[0025] The second text segment is selected from multiple first text segments based on a preset semantic template;

[0026] Based on the second text fragment, extract the target user's to-do events and corresponding to-do times.

[0027] If individual clients require that the invention description also include the accessory claims, the following description can be considered:

[0028] A first aspect of this application provides an event notification method, including:

[0029] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the event notification method described above.

[0030] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the event notification method described above.

[0031] The beneficial effects of the event alerting method and apparatus, electronic device, and storage medium provided in this application embodiment are as follows:

[0032] In this embodiment, when the network connection between the first device and the second device is interrupted, the step of extracting reminder information offline can be performed to extract the target user's to-do events and corresponding to-do times from the user's dialogue information, and to actively remind the user of events based on the target user's to-do events and corresponding to-do times.

[0033] The offline extraction of reminder information includes the following steps: First, the dialogue information is converted into text information. The text information is then segmented to obtain multiple word segments and their character index ranges. Simultaneously, the text information is split into multiple sentence units according to rules such as punctuation and logical conjunctions, and the character index range of each sentence unit is saved. Next, based on a preset first keyword library, word segments related to the to-do event are filtered. The filtered word segments are sorted in ascending order of character index to ensure that the target word segment sequence is consistent with the semantic order of the original text. Based on this, the target word segment sequence is segmented according to the index range of the sentence units. Words belonging to the same sentence unit are divided into segments. Then, based on the index range of this segment, the corresponding complete sentence unit is associated as the first text fragment, achieving coarse screening of the text information. Next, based on a preset semantic template, the logical integrity of the filtered first text fragments is verified, and valid second text fragments are selected, achieving fine screening of the text information. Finally, based on the second text fragments, the to-do events and corresponding to-do times of the target user are extracted.

[0034] Therefore, this embodiment achieves the filtering of the first text fragment by performing operations such as word segmentation, sentence splitting, keyword matching, and character index range segmentation on the first device, and filters the second text fragment from multiple first text fragments through semantic template matching, thereby realizing the function of extracting to-do events and to-do events based on dialogue information and actively reminding events, thus improving the intelligence level of the first device. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 A flowchart illustrating an event notification method provided in an embodiment of this application;

[0037] Figure 2 This is a structural block diagram of an event reminder device provided in an embodiment of this application;

[0038] Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0039] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0040] It is understood that in the embodiments of this application, data such as user information are involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0041] It should be noted that the terms "first," "second," etc., used in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in sequences other than those illustrated or described herein.

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0043] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an event notification method provided in an embodiment of this application. The event notification method provided in this embodiment can be executed by an electronic device, and the method may include:

[0044] S101: In response to the activation of the active reminder mode of the first device and the interruption of the network connection between the first device and the second device, the step of offline retrieval of reminder information is performed to obtain the to-do events and corresponding to-do times of the target user; the target user is the user using the first device.

[0045] The steps for retrieving reminder information offline include:

[0046] Obtain dialogue information;

[0047] Convert dialogue information into text information;

[0048] The text information is segmented into words to obtain multiple words and the character index range of each word; the text information is split into sentences to obtain multiple sentence units and the character index range of each sentence unit.

[0049] Filter multiple word segments that match the preset first keyword library from multiple word segments; sort the matching multiple word segments in ascending order of character index to obtain the target word segment sequence;

[0050] The target word segmentation sequence is segmented based on the character index range of each sentence unit to obtain multiple target word segmentation segments. The sentence unit corresponding to the character index range of each target word segment is taken as the first text segment.

[0051] The second text segment is selected from multiple first text segments based on a preset semantic template;

[0052] Extract the target user's to-do events and corresponding to-do times based on the second text fragment.

[0053] In this embodiment, the first device can be a wearable device, the second device can be a server, and the target user is the user using the first device. When the active reminder mode of the first device is enabled and the network connection between the first device and the second device is interrupted, the offline reminder information extraction step can be performed locally on the first device to extract the target user's to-do events and corresponding to-do times from the user's conversation information.

[0054] Specifically, the user's dialogue information includes the dialogue information between the target user and other users. The user's dialogue voice can be acquired through the microphone set on the first device, the dialogue voice can be divided into multiple voice segments, and then the features of each voice segment can be extracted to obtain the voice features of each voice segment. The voice features of each voice segment are input into a speech-to-text model (such as Wav2Vec 2.0 Tiny / Base quantized version) to obtain the text information corresponding to each voice segment.

[0055] Furthermore, a distilled Hidden Markov Model (HMM) or Conditional Random Field (CRF) mini-model can be used to segment the text information, resulting in multiple segments. Each segment contains multiple characters, and each character has a corresponding character index to represent the position of the character in the text information. The range of the character index of each segment is determined by the starting character index and the ending character index of the segment.

[0056] For example, the text message is: "Go to the company for a project meeting at 3 o'clock tomorrow." This text message contains 11 characters, with the character index starting from 0 and increasing sequentially. The index of each character is shown in Table 1 below:

[0057] Table 1 - Example of Character Index

[0058]

[0059] After performing word segmentation on the above text information, the resulting word segments and their corresponding character index ranges are: tomorrow (character index range 0-1), 3 o'clock (character index range 2-3), company (character index range 5-6), open (character index range 7-7), and project meeting (character index range 8-10).

[0060] Simultaneously, text information can be segmented into sentences according to rules such as punctuation and logical conjunctions, resulting in multiple sentence units. Using the same processing method as word segmentation, the character index range corresponding to each sentence unit can be obtained. Other methods can also be used for sentence segmentation, as detailed in the following examples.

[0061] Next, multiple word segments matching a preset first keyword library are selected from a pool of word segments. This library contains multiple preset first keywords, categorized into time-based, action-based, and object-based keywords. For example, time-based keywords might include "today," "tomorrow," "the day after tomorrow," "Monday to Sunday," "this month," "next month," "first half of the year," and "end of the year." Action-based keywords might include "submit," "complete," "attend a meeting," "hold a meeting," "report," "feedback," "follow up," "review," "approval," "send," "receive," "collect," and "return." Object-based keywords might include "plan," "report," "document," "material," "contract," "work order," "approval form," "meeting minutes," and "quotation." Each word segment is then subjected to synonym expansion matching against the first keywords in the first keyword library, resulting in a pool of matching words. Finally, the matched words are sorted to obtain the target word segmentation sequence.

[0062] Based on the obtained target word segmentation sequence, the target word segmentation sequence is divided into segments according to the character index range of the sentence unit. One target word segment corresponds to one sentence unit, and one sentence unit corresponding to one target word segment is used as a first text fragment.

[0063] For example, suppose the text message is: "We need to visit a client at 2 PM next Monday. This client doesn't like being late, so we need to be on time; we plan to hold a new product launch event next week, and the event plan needs to be submitted by Sunday."

[0064] The above text information is segmented into words, and the segmentation is filtered based on the preset first keyword library. The target segmentation sequence is as follows: ["Next Monday", "2 PM", "Visit", "Customer", "Next week", "Hold", "New product launch", "Event", "Sunday", "Submit", "Event plan"]. The character index range corresponding to each segmentation is: 0-2, 3-5, 8-9, 10-11, 31-32, 33-34, 35-40, 41-42, 44-45, 47-48, 49-52.

[0065] The target word segmentation sequence is segmented based on the character index range of each sentence unit, resulting in:

[0066] Target segmentation 1: ["Next Monday", "2 PM", "Visit", "Client"];

[0067] Target segmentation 2: ["next week", "hold", "new product launch", "event"];

[0068] Target segmentation 3: ["Sunday", "submit", "activity plan"];

[0069] Among them, target segment 1 corresponds to sentence 1 (character index range 0-11): I need to visit a client at 2 PM next Monday; target segment 2 corresponds to sentence 4 (character index range 27-42): We plan to hold a new product launch event next week; target segment 3 corresponds to sentence 5 (character index range 44-52): The event plan needs to be submitted on Sunday. Through the above filtering process, three first text fragments, sentence 1, sentence 4, and sentence 5, are obtained.

[0070] Furthermore, considering that the first keyword library filters matching word segments from a single dimension, that is, it retains any word segment that matches time / action / object keywords without verifying whether they belong to the same valid to-do event, it will result in the filtering of invalid word segments and first text fragments.

[0071] To address the aforementioned issues, this embodiment can pre-construct a semantic template. By using the semantic template, the logical order and combination requirements of time words, action words, and object words can be clearly defined. This allows verification of whether the first text fragment satisfies the "time + action + object" to-do logic. The first text fragment containing single-dimensional word segmentation but not the complete to-do event is eliminated, resulting in a second text fragment containing the complete to-do event.

[0072] In this embodiment, the word segmentation in the text information is first coarsely screened based on the first keyword library, retaining all word segments that match time / action / object keywords to avoid information omission. Based on this, the semantic template focuses on the validity of the logic, verifying whether the first text fragment selected based on the first keyword meets the "time + action + object" to-do logic. First text fragments containing single-dimensional word segments but not complete to-do events are removed, thus ensuring the validity of the extracted information.

[0073] Based on this, extracting the target user's to-do events and corresponding to-do times from the second text fragment can ensure the accuracy and efficiency of the extraction of to-do events and to-do times.

[0074] S102: Provide event reminders based on the target user's to-do events and corresponding to-do times.

[0075] In this embodiment, based on the extracted to-do events and corresponding to-do times of the target user, an event reminder task can be automatically generated in the first device. When the to-do time arrives, the first device can remind users of the to-do events through voice broadcast, vibration, or pop-up window.

[0076] As can be seen from the above, when the network connection between the first device and the second device is interrupted, this embodiment can extract the target user's to-do events and corresponding to-do times from the user's dialogue information by performing the offline extraction of reminder information, and actively provide event reminders based on the target user's to-do events and corresponding to-do times.

[0077] The offline extraction of reminder information includes the following steps: First, the dialogue information is converted into text information. The text information is then segmented to obtain multiple word segments and their character index ranges. Simultaneously, the text information is split into multiple sentence units according to rules such as punctuation and logical conjunctions, and the character index range of each sentence unit is saved. Next, based on a preset first keyword library, word segments related to the to-do event are filtered. The filtered word segments are sorted in ascending order of character index to ensure that the target word segment sequence is consistent with the semantic order of the original text. Based on this, the target word segment sequence is segmented according to the index range of the sentence units. Words belonging to the same sentence unit are divided into segments. Then, based on the index range of this segment, the corresponding complete sentence unit is associated as the first text fragment, achieving coarse screening of the text information. Next, based on a preset semantic template, the logical integrity of the filtered first text fragments is verified, and valid second text fragments are selected, achieving fine screening of the text information. Finally, based on the second text fragments, the to-do events and corresponding to-do times of the target user are extracted.

[0078] Therefore, this embodiment achieves the filtering of the first text fragment by performing operations such as word segmentation, sentence splitting, keyword matching, and character index range segmentation on the first device, and filters the second text fragment from multiple first text fragments through semantic template matching, thereby realizing the function of extracting to-do events and to-do events based on dialogue information and actively reminding events, thus improving the intelligence level of the first device.

[0079] In one embodiment of this application, text information is processed by sentence segmentation to obtain multiple sentence units and the character index range of each sentence unit, including:

[0080] The text information is divided into multiple basic sentence units based on preset boundary symbols, and the character index range of each basic sentence unit is determined.

[0081] Based on the character index range of each basic sentence unit, multiple word segments are obtained in each basic sentence unit;

[0082] The scene classification corresponding to each basic sentence unit is determined based on multiple word segments in each basic sentence unit.

[0083] For each pair of adjacent basic sentence units, based on the scene classification of each pair of adjacent basic sentence units, determine the scene consistency evaluation value between each pair of adjacent basic sentence units; calculate the word segmentation overlap between each pair of adjacent basic sentence units; and perform a weighted sum of the word segmentation overlap and the scene consistency evaluation value to obtain the correlation between the two adjacent basic sentence units.

[0084] Multiple consecutive basic sentence units with a correlation degree greater than the correlation degree threshold are merged to obtain multiple sentence units;

[0085] The character index ranges of all basic sentence units contained in each sentence unit are merged to obtain the character index range of each sentence unit.

[0086] In this embodiment, when performing sentence segmentation on text information, the text information can first be preliminarily segmented based on preset boundary symbols (such as commas, periods, semicolons, exclamation marks, line breaks, and other commonly used Chinese sentence segmentation symbols) to obtain multiple basic sentence units; at the same time, the starting character index and ending character index of each basic sentence unit in the original text are saved to determine its position range in the original text.

[0087] Then, for each basic sentence unit, based on the character index range of that basic sentence unit, all word segments falling within that range are selected from the multiple word segments obtained in the word segmentation processing steps of the aforementioned embodiment, resulting in multiple word segments in that basic sentence unit. The word segments in each basic sentence unit are matched with a preset scene feature lexicon to determine the scene category corresponding to that basic sentence unit. For example, scene categories may include work scenes, life scenes, or shopping scenes, etc.

[0088] Specifically, scenario keywords for various scenarios such as work, life, and shopping can be collected in advance to construct a scenario feature lexicon. For example, scenario keywords for work scenarios could be meetings, reports, clients, conferences, submissions, or plans; keywords for life scenarios could be picking up packages, grocery shopping, cooking, shopping, parties, or seeing a doctor; and keywords for shopping scenarios could be buying, placing an order, paying, express delivery, goods, or returns. Based on this, for each basic sentence unit, the word segments in the basic sentence unit are compared with the pre-set scenario feature lexicon for synonym expansion matching. The number of successfully matched word segments is counted, and the scenario with the most successfully matched word segments is selected as the scenario category for that basic sentence unit.

[0089] Next, for each pair of adjacent basic sentence units, a corresponding scene consistency evaluation value is given based on their scene classification. Specifically, if the scene classifications of the two adjacent basic sentence units are the same, the scene consistency evaluation value is set to the larger first evaluation value (e.g., 1); if the scene classifications of the two adjacent basic sentence units are different, the scene consistency evaluation value is set to the smaller second evaluation value (e.g., 0). Simultaneously, the number of words that appear together in the two adjacent basic sentence units is counted, and divided by the total number of words in the two units (the total number of words after deduplication) to obtain the word segmentation overlap (value range 0-1). The word segmentation overlap and the scene consistency evaluation value are weighted and summed to obtain the correlation between the two adjacent basic sentence units.

[0090] Finally, two basic sentence units with a correlation score greater than a preset correlation score threshold (e.g., 0.6) are merged. If multiple consecutive basic sentence units exist, and the correlation score between any two adjacent basic sentence units is greater than the preset correlation score threshold, then these multiple consecutive basic sentence units are merged into one sentence unit. Simultaneously, the starting index of all basic sentence units contained in the merged sentence unit is used as the starting index of the merged unit, and the ending index is used as the ending index of the merged unit, resulting in the final sentence unit and its corresponding character index range.

[0091] As can be seen from the above, this embodiment obtains basic sentence units based on preset boundary symbols, and judges the semantic correlation between adjacent basic sentence units by scene consistency evaluation value and word segmentation overlap. The semantically coherent basic sentence units are merged, and the merged sentence units contain more complete to-do related semantics (such as time, action, and object association information), which can provide more comprehensive contextual support for subsequent first keyword screening and semantic template matching, and reduce the omission or misjudgment of to-do element extraction caused by context fragmentation.

[0092] In one embodiment of this application, the preset semantic template includes multiple semantic elements, including the responsible person, the task to be done, the task content, and the execution time;

[0093] The second text segment is selected from multiple first text segments based on a preset semantic template, including:

[0094] For each first text segment, the element matching degree between the first text segment and the semantic template is determined based on the number of semantic elements contained in the first text segment; the semantic fluency of the first text segment is calculated, and the semantic matching degree between the first text segment and the semantic template is determined based on the semantic fluency and the element matching degree; wherein, the number of semantic elements contained in the first text segment is positively correlated with the element matching degree;

[0095] The first text fragment with a semantic matching degree greater than the matching degree threshold is used as the third text fragment. From multiple third text fragments, text fragments containing the responsible person element, including the target user, are selected as the second text fragments.

[0096] In this embodiment, considering that the responsible person, the task to be done, the task content, and the execution time are necessary components of a task event, the preset semantic template can include multiple semantic elements such as the responsible person, the task to be done, the task content, and the execution time.

[0097] Based on this, for each first text segment, the MobileBERT model (Universal a Compact Task-Agnostic BERT for Resource-Limited Devices) can be used to extract features from the first text segment, and the feature matching degree between the first text segment and the semantic template is determined based on the number of the aforementioned semantic features contained in the first text segment. The more semantic features contained in the first text segment, the higher the feature matching degree between the first text segment and the semantic template.

[0098] Specifically, the element matching degree between each first text fragment and the semantic template is calculated using the following formula:

[0099] ;

[0100] in, This represents the element matching degree between the i-th first text fragment and the semantic template. This represents the number of the aforementioned semantic elements contained in the i-th first text segment.

[0101] Simultaneously, for each first text segment, the semantic fluency of the first text segment can be calculated based on the MobileBERT model. The semantic fluency is used to characterize the grammatical rationality and semantic coherence of the first text segment. By weighted summing the semantic fluency and feature matching degree corresponding to the first text segment, the semantic matching degree between the first text segment and the semantic template can be obtained.

[0102] Finally, from multiple first text fragments, the first text fragments with a semantic matching degree greater than a preset matching degree threshold (e.g., 0.6) are selected as third text fragments. From multiple third text fragments, text fragments with the responsible person element including the target user are selected as second text fragments.

[0103] As can be seen from the above, this embodiment constructs a semantic template based on multiple semantic elements such as the responsible person, the task to be done, the task content, and the execution time. It also comprehensively considers the element matching degree and semantic fluency to determine the semantic matching degree between the first text fragment and the semantic template. Based on the semantic matching degree and the responsible person element, the second text fragment is selected from multiple first text fragments, which can ensure that the second text fragment containing the task event of the target user is further selected.

[0104] In one embodiment of this application, filtering text fragments containing the responsible party element from multiple third text fragments includes:

[0105] Obtain the voiceprint features corresponding to multiple third-party text fragments, as well as the voiceprint features of the target user;

[0106] Select text fragments containing the preset second keyword from multiple third text fragments to form the first candidate fragment set;

[0107] For each text segment in the first candidate segment set, if the responsible person element corresponding to the text segment is a first-person singular and the voiceprint feature corresponding to the text segment does not match the voiceprint feature of the target user, then the text segment is deleted from the first candidate segment set to obtain the second candidate segment set.

[0108] The second set of candidate fragments is used as the responsible party element, including text fragments from the target user.

[0109] In this embodiment, the second keyword may include words such as "I," "we," "us," and "everyone." If the responsible party element of a third text fragment contains the aforementioned second keyword, it indicates that the third text fragment is likely to contain a to-do event that the target user needs to perform. Accordingly, the third text fragment can be further filtered to remove text fragments that do not contain the target user.

[0110] Specifically, during the process of converting dialogue information into text information, the mapping relationship between multiple speech segments in the dialogue and their corresponding voiceprint features, as well as the text information, can be saved. Simultaneously, the voiceprint features of the target user are pre-saved on the first device. Based on this, text segments containing a preset second keyword as the responsible party element are selected from multiple third text segments as a first candidate segment set. For each text segment in the first candidate segment set, if the responsible party element corresponding to the text segment is a first-person singular (i.e., "I"), the corresponding voiceprint feature is searched from the aforementioned mapping relationship. If the voiceprint feature corresponding to the text segment is inconsistent with the target user's voiceprint feature, it indicates that the speech segment corresponding to the text segment was not issued by the target user, and the text segment does not contain the target user's to-do event.

[0111] As can be seen from the above, this embodiment can accurately extract text fragments containing the target user's to-do events by filtering text fragments containing the responsible person element from multiple third text fragments based on the second keyword and voiceprint features, thus avoiding interference from other users' to-do events.

[0112] In one embodiment of this application, event reminders are provided based on pending events and corresponding pending times, including:

[0113] Output a to-do list based on the to-do events and their corresponding to-do times;

[0114] In response to receiving confirmation information for the to-do list, an event reminder task is generated based on the to-do event and the corresponding to-do time to provide event reminders.

[0115] In this embodiment, a to-do list can be generated based on the extracted to-do events and their corresponding to-do times. The list must include the to-do event name, specific execution time, and unique identifier. Users can edit the to-do list. Once the user confirms that the to-dos are correct, event reminder tasks can be automatically generated based on the to-do list to provide event reminders.

[0116] As can be seen from the above, this embodiment generates a to-do list that users can view and edit before generating event reminder tasks, which can ensure the accuracy of event reminder tasks.

[0117] In one embodiment of this application, the event notification method further includes:

[0118] In response to the activation of the proactive reminder mode on the first device and the normal network connection between the first device and the second device, the step of retrieving reminder information online is executed to obtain the target user's to-do events and corresponding to-do times;

[0119] The steps for retrieving reminder information online include:

[0120] Obtain dialogue information;

[0121] The dialogue information is sent to the second device so that the second device can extract the target user's to-do events and corresponding to-do times from the dialogue information based on the target model;

[0122] Receive the target user's pending events and corresponding pending times from the second device.

[0123] In this embodiment, when the active reminder mode of the first device is enabled and the network connection between the first device and the second device is normal, the first device can send dialogue information to the second device. The high-performance target model (e.g., a large language model) deployed on the second device analyzes the dialogue information and extracts the target user's to-do events and corresponding to-do times. The second device then sends the extracted target user's to-do events and corresponding to-do times back to the first device, which can then provide event reminders accordingly.

[0124] As can be seen from the above, when the network connection between the first device and the second device is normal, this embodiment can utilize the large model deployed on the second device to extract pending events and pending times, thereby further improving the accuracy of event extraction.

[0125] In one embodiment of this application, the event notification method further includes:

[0126] In response to the activation of the passive reminder mode of the first device and the receipt of event reminder information sent by the third device, an event reminder is issued based on the event reminder information.

[0127] In this embodiment, the third device can be a mobile phone, on which the app corresponding to the first device and other apps are installed. The app corresponding to the first device can communicate with the other apps. When the passive reminder mode of the first device is enabled, push messages from other apps can be sent to the app corresponding to the first device. The third device then sends the push messages to the first device through the app corresponding to the first device, and the first device broadcasts the messages, thereby achieving the reminder of relevant events.

[0128] In one embodiment of this application, event reminders based on event reminder information include:

[0129] If the event notification message contains an emergency notification identifier, a notification title is generated based on the first title template;

[0130] If the event notification message does not contain an emergency notification identifier, a notification title is generated based on the second title template; wherein, the first title template contains more fields than the second title template.

[0131] Event reminders are based on the reminder title.

[0132] In this embodiment, for emergency events, the event reminder information sent by the third device typically includes an emergency reminder identifier. After receiving the event reminder information sent by the third device, the first device can parse the event reminder information. If it contains an emergency reminder identifier, it can generate a reminder title based on a first title template. The first title template can include content such as "Emergency Identifier + Event Name + Execution Time + Responsible Person + Key Location." For example, the corresponding reminder title could be "[Emergency] 3 PM Customer Meeting (Third Floor Conference Room, Contact Person: Zhang San)," which helps users quickly understand the core information of the event.

[0133] Conversely, if the event notification does not include an emergency alert indicator, a notification title can be generated based on the second title template. The second title template can include content such as "Event Name + Execution Time". For example, the corresponding notification title could be "Submit the plan at 10 AM tomorrow", avoiding excessive information that might distract the user.

[0134] Based on the same inventive concept, this application also provides an event reminder device for implementing the event reminder method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more event reminder device embodiments provided below can be found in the limitations of the event reminder method described above, and will not be repeated here.

[0135] This application provides an event reminder device, such as... Figure 2 As shown, the event reminder device 20 includes an information extraction module 21 and an event reminder module 22.

[0136] The information extraction module 21 is used to respond to the activation of the active reminder mode of the first device and the interruption of the network connection between the first device and the second device, and to perform the offline extraction of reminder information to obtain the to-do events and corresponding to-do times of the target user; the target user is the user using the first device;

[0137] Event reminder module 22 is used to provide event reminders based on the target user's to-do events and corresponding to-do times;

[0138] The steps for retrieving reminder information offline include:

[0139] Obtain dialogue information;

[0140] Convert dialogue information into text information;

[0141] The text information is segmented into words to obtain multiple words and the character index range of each word; the text information is split into sentences to obtain multiple sentence units and the character index range of each sentence unit.

[0142] Filter multiple word segments that match the preset first keyword library from multiple word segments; sort the matching multiple word segments in ascending order of character index to obtain the target word segment sequence;

[0143] The target word segmentation sequence is segmented based on the character index range of each sentence unit to obtain multiple target word segmentation segments. The sentence unit corresponding to the character index range of each target word segment is taken as the first text segment.

[0144] The second text segment is selected from multiple first text segments based on a preset semantic template;

[0145] Extract the target user's to-do events and corresponding to-do times based on the second text fragment.

[0146] In one embodiment of this application, the information extraction module 21 is specifically used for:

[0147] The text information is divided into multiple basic sentence units based on preset boundary symbols, and the character index range of each basic sentence unit is determined.

[0148] Based on the character index range of each basic sentence unit, multiple word segments are obtained in each basic sentence unit;

[0149] The scene classification corresponding to each basic sentence unit is determined based on multiple word segments in each basic sentence unit.

[0150] For each pair of adjacent basic sentence units, based on the scene classification of each pair of adjacent basic sentence units, determine the scene consistency evaluation value between each pair of adjacent basic sentence units; calculate the word segmentation overlap between each pair of adjacent basic sentence units; and perform a weighted sum of the word segmentation overlap and the scene consistency evaluation value to obtain the correlation between the two adjacent basic sentence units.

[0151] Multiple consecutive basic sentence units with a correlation degree greater than the correlation degree threshold are merged to obtain multiple sentence units;

[0152] The character index ranges of all basic sentence units contained in each sentence unit are merged to obtain the character index range of each sentence unit.

[0153] In one embodiment of this application, the preset semantic template includes multiple semantic elements, including the responsible person, the task to be done, the task content, and the execution time; the information extraction module 21 is specifically used for:

[0154] For each first text segment, the element matching degree between the first text segment and the semantic template is determined based on the number of semantic elements contained in the first text segment; the semantic fluency of the first text segment is calculated, and the semantic matching degree between the first text segment and the semantic template is determined based on the semantic fluency and the element matching degree; wherein, the number of semantic elements contained in the first text segment is positively correlated with the element matching degree;

[0155] The first text fragment with a semantic matching degree greater than the matching degree threshold is used as the third text fragment. From multiple third text fragments, text fragments containing the responsible person element, including the target user, are selected as the second text fragments.

[0156] In one embodiment of this application, the information extraction module 21 is further configured to:

[0157] Obtain the voiceprint features corresponding to multiple third-party text fragments, as well as the voiceprint features of the target user;

[0158] Select text fragments containing the preset second keyword from multiple third text fragments to form the first candidate fragment set;

[0159] For each text segment in the first candidate segment set, if the responsible person element corresponding to the text segment is a first-person singular and the voiceprint feature corresponding to the text segment does not match the voiceprint feature of the target user, then the text segment is deleted from the first candidate segment set to obtain the second candidate segment set.

[0160] The second set of candidate fragments is used as the responsible party element, including text fragments from the target user.

[0161] In one embodiment of this application, the event notification module 22 is specifically used for:

[0162] Output a to-do list based on the to-do events and their corresponding to-do times;

[0163] In response to receiving confirmation information for the to-do list, an event reminder task is generated based on the to-do event and the corresponding to-do time to provide event reminders.

[0164] In one embodiment of this application, the information extraction module 21 is specifically used for:

[0165] In response to the activation of the proactive reminder mode on the first device and the normal network connection between the first device and the second device, the step of retrieving reminder information online is executed to obtain the target user's to-do events and corresponding to-do times;

[0166] The steps for retrieving reminder information online include:

[0167] Obtain dialogue information;

[0168] The dialogue information is sent to the second device so that the second device can extract the target user's to-do events and corresponding to-do times from the dialogue information based on the target model;

[0169] Receive the target user's pending events and corresponding pending times from the second device.

[0170] In one embodiment of this application, the information extraction module 21 is specifically used for:

[0171] In response to the activation of the passive reminder mode of the first device and the receipt of event reminder information sent by the third device, an event reminder is issued based on the event reminder information.

[0172] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of the information extraction module 21 and the event reminder module 22 are shown.

[0173] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0174] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0175] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store preset information such as a first keyword library and semantic templates.

[0176] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the event reminder method provided in the embodiments of this application, or they can execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0177] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0178] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., provided on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0179] Those skilled in the art will recognize that the modules / units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0180] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0181] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules, units, or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or modules / units, or it may be an electrical, mechanical, or other form of connection.

[0182] The modules / units described as separate components may or may not be physically separate. Similarly, the components shown as modules / units may or may not be physical modules / units; they may be located in one place or distributed across multiple network modules / units. Some or all of the modules / units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0183] Furthermore, the functional modules / units in the various embodiments of this application can be integrated into one processing module / unit, or each module / unit can exist physically separately, or two or more modules / units can be integrated into one module / unit. The integrated modules / units described above can be implemented in hardware or in the form of software functional modules / units.

[0184] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An event reminder method, characterized in that, Applied to the first device, including: In response to the activation of the active reminder mode of the first device and the interruption of the network connection between the first device and the second device, the step of offline extraction of reminder information is performed to obtain the target user's to-do events and corresponding to-do times; the target user is the user using the first device; Event reminders are generated based on the target user's to-do events and corresponding to-do times. The offline extraction of reminder information includes the following steps: Obtain dialogue information; Convert the dialogue information into text information; The text information is segmented into words to obtain multiple words and the character index range of each word; the text information is then split into sentences to obtain multiple sentence units and the character index range of each sentence unit. Filter multiple word segments that match a preset first keyword library from the multiple word segments; sort the multiple matching word segments in ascending order of character index to obtain the target word segment sequence; The target word segmentation sequence is segmented based on the character index range of each sentence unit to obtain multiple target word segmentation segments. The sentence unit corresponding to the character index range of each target word segment is taken as the first text segment. The second text segment is selected from multiple first text segments based on a preset semantic template; Based on the second text fragment, extract the target user's to-do events and corresponding to-do times; The text information is processed by sentence segmentation to obtain multiple sentence units and the character index range of each sentence unit, including: The text information is divided into multiple basic sentence units based on preset boundary symbols, and the character index range of each basic sentence unit is determined. Based on the character index range of each basic sentence unit, multiple word segments are obtained in each basic sentence unit; The scene classification corresponding to each basic sentence unit is determined based on multiple word segments in each basic sentence unit. For each pair of adjacent basic sentence units, based on the scene classification of each pair of adjacent basic sentence units, a scene consistency evaluation value between each pair of adjacent basic sentence units is determined; the word segmentation overlap degree between each pair of adjacent basic sentence units is calculated; the word segmentation overlap degree and the scene consistency evaluation value are weighted and summed to obtain the correlation degree between the two adjacent basic sentence units. Multiple consecutive basic sentence units with a correlation degree greater than the correlation degree threshold are merged to obtain multiple sentence units; The character index ranges of all basic sentence units contained in each sentence unit are merged to obtain the character index range of each sentence unit.

2. The event reminder method as described in claim 1, characterized in that, The preset semantic template includes multiple semantic elements, including the responsible person, the task to be done, the task content, and the execution time. The filtering of second text segments from multiple first text segments based on a preset semantic template includes: For each first text segment, the element matching degree between the first text segment and the semantic template is determined based on the number of semantic elements contained in the first text segment; the semantic fluency of the first text segment is calculated, and the semantic matching degree between the first text segment and the semantic template is determined based on the semantic fluency and the element matching degree; wherein, the number of semantic elements contained in the first text segment is positively correlated with the element matching degree; The first text fragment with a semantic matching degree greater than the matching degree threshold is used as the third text fragment. From the multiple third text fragments, text fragments containing the responsible person element, including the target user, are selected as the second text fragment.

3. The event reminder method as described in claim 2, characterized in that, The text fragments from which the responsible party element is selected from multiple third text fragments include text fragments from which the target user is selected, including: Obtain the voiceprint features corresponding to multiple third text fragments, as well as the voiceprint features of the target user; From the multiple third text fragments, text fragments containing the preset second keyword for the responsible person element are selected as the first candidate fragment set; For each text segment in the first candidate segment set, if the responsible person element corresponding to the text segment is a first-person singular and the voiceprint feature corresponding to the text segment does not match the voiceprint feature of the target user, then the text segment is deleted from the first candidate segment set to obtain the second candidate segment set. The second set of candidate fragments is used as the responsible party element, including text fragments from the target user.

4. The event reminder method as described in claim 1, characterized in that, The event reminder based on the pending event and the corresponding pending time includes: Output a to-do list based on the to-do events and their corresponding to-do times; In response to receiving confirmation information for the to-do list, an event reminder task is generated based on the to-do event and the corresponding to-do time to provide event reminders.

5. The event reminder method as described in claim 1, characterized in that: Also includes: In response to the activation of the active reminder mode of the first device and the normal network connection between the first device and the second device, the step of retrieving reminder information online is executed to obtain the target user's to-do events and corresponding to-do times; The step of retrieving reminder information online includes: Obtain dialogue information; The dialogue information is sent to the second device so that the second device can extract the target user's to-do events and corresponding to-do times from the dialogue information based on the target model; Receive the target user's to-do events and corresponding to-do times returned by the second device.

6. The event reminder method as described in claim 1, characterized in that, Also includes: In response to the activation of the passive reminder mode of the first device and the receipt of event reminder information sent by the third device, an event reminder is issued based on the event reminder information.

7. An event reminder device, characterized in that, Applied to the first device, including: The information extraction module is used to respond to the activation of the active reminder mode of the first device and the interruption of the network connection between the first device and the second device, and to perform the step of offline extraction of reminder information to obtain the to-do events and corresponding to-do times of the target user; the target user is the user using the first device; The event reminder module is used to provide event reminders based on the target user's to-do events and corresponding to-do times; The offline extraction of reminder information includes the following steps: Obtain dialogue information; Convert the dialogue information into text information; The text information is segmented into words to obtain multiple words and the character index range of each word; the text information is then split into sentences to obtain multiple sentence units and the character index range of each sentence unit. Filter multiple word segments that match a preset first keyword library from the multiple word segments; sort the multiple matching word segments in ascending order of character index to obtain the target word segment sequence; The target word segmentation sequence is segmented based on the character index range of each sentence unit to obtain multiple target word segmentation segments. The sentence unit corresponding to the character index range of each target word segment is taken as the first text segment. The second text segment is selected from multiple first text segments based on a preset semantic template; Based on the second text fragment, extract the target user's to-do events and corresponding to-do times; The information extraction module is specifically used for: The text information is divided into multiple basic sentence units based on preset boundary symbols, and the character index range of each basic sentence unit is determined. Based on the character index range of each basic sentence unit, multiple word segments are obtained in each basic sentence unit; The scene classification corresponding to each basic sentence unit is determined based on multiple word segments in each basic sentence unit. For each pair of adjacent basic sentence units, based on the scene classification of each pair of adjacent basic sentence units, determine the scene consistency evaluation value between each pair of adjacent basic sentence units; calculate the word segmentation overlap between each pair of adjacent basic sentence units; and perform a weighted sum of the word segmentation overlap and the scene consistency evaluation value to obtain the correlation between the two adjacent basic sentence units. Multiple consecutive basic sentence units with a correlation degree greater than the correlation degree threshold are merged to obtain multiple sentence units; The character index ranges of all basic sentence units contained in each sentence unit are merged to obtain the character index range of each sentence unit.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6.

Images