Task execution method and device, computer device and storage medium thereof

Abstract

The application relates to a task execution method and device, computer equipment and a storage medium thereof, and relates to the technical field of artificial intelligence. The method comprises the following steps: performing segmentation processing on a task demand voice input by a target user to obtain at least two voice segments; determining at least two prediction results corresponding to each voice segment through at least two hidden Markov (HMM) models, wherein each prediction result corresponding to each voice segment comprises a category of key information in the voice segment predicted by each HMM model and a confidence degree of the category; determining key information contained in each voice segment based on the at least two prediction results corresponding to each voice segment; and determining a task execution strategy of a to-be-executed task based on the key information. The application realizes the determination of the task execution strategy of the to-be-executed task under the premise of reducing the time and energy consumption of the user.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a task execution method, apparatus, computer device and its storage medium. Background Technology

[0002] With the continuous development and progress of computer systems, computer systems are able to handle more and more tasks. When a user wants to perform a task, they need to manually input the key information of the task to be performed into the computer system according to the actual needs, and select the target strategy that meets the requirements from multiple recommended strategies fed back by the computer system based on the key information.

[0003] However, in the process of determining the target strategy according to the above method, users not only need to manually input key information into the computer system, but also need to select the target strategy from multiple recommended strategies according to requirements. This process consumes a lot of users' time and energy. Summary of the Invention

[0004] Therefore, it is necessary to provide a task execution method, apparatus, computer device and its storage medium that can quickly determine the task execution strategy of the task to be executed, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a task execution method. The method includes:

[0006] The task requirement speech input by the target user is segmented to obtain at least two speech segments;

[0007] Using at least two Hidden Markov Models (HMMs), at least two prediction results are determined for each speech segment. Each prediction result for each speech segment includes the category of the key information in the speech segment predicted by each HMM model and the confidence level of the category.

[0008] Based on at least two prediction results corresponding to each speech segment, determine the key information contained in each speech segment;

[0009] Based on key information, determine the task execution strategy for the task to be executed.

[0010] In one embodiment, based on at least two prediction results corresponding to each speech segment, the key information contained in each speech segment is determined, including:

[0011] Based on the confidence of the category to which each speech segment belongs in at least two prediction results, the target prediction result for each speech segment is determined from the at least two prediction results corresponding to each speech segment.

[0012] Based on the category of key information in each speech segment and the target prediction results of each speech segment, determine the key information contained in each speech segment.

[0013] In one embodiment, based on the confidence level of the category to which each speech segment belongs in at least two prediction results, the target prediction result for each speech segment is determined from the at least two prediction results corresponding to each speech segment, including:

[0014] The confidence scores of the categories to which each speech segment belongs are scored using the support vector machine (SVM) model, thus obtaining the confidence scores of the categories to which each prediction result belongs.

[0015] Based on the confidence score of the category to which each prediction result belongs, the target prediction result for each speech segment is determined from at least two prediction results corresponding to each speech segment.

[0016] In one embodiment, at least two prediction results corresponding to each speech segment are determined using at least two Hidden Markov Models (HMMs), including:

[0017] Based on at least two Hidden Markov Models (HMMs), at least two candidate results are determined for each speech segment; wherein, each candidate result for each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model to belong to each candidate category;

[0018] Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, the category to which the key information in each speech segment belongs and the confidence level of the category are determined.

[0019] In one embodiment, the task execution strategy for the task to be executed is determined based on key information, including:

[0020] Based on key information, the task to be executed is processed for approval, and the approval result is obtained.

[0021] If the approval result is passed, the user's historical task execution information corresponding to the key information will be determined;

[0022] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed.

[0023] In one embodiment, based on key information, a task approval process is performed on the task to be executed to obtain an approval result, including:

[0024] Based on key information and application templates, generate approval applications for tasks to be executed and send them to the approval terminal for processing.

[0025] Receive the approval results sent by the review terminal.

[0026] In one embodiment, the user's historical task execution information includes historical execution strategies and satisfaction scores corresponding to the historical execution strategies;

[0027] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed, including:

[0028] Based on the satisfaction scores corresponding to historical execution strategies, determine the target users' interest index and disinterest index regarding historical execution strategies;

[0029] Based on the interest index and the disinterest index, identify the reference users corresponding to the target users;

[0030] Based on historical execution strategies and reference execution strategies of reference users, determine the task execution strategy for the task to be executed.

[0031] In one embodiment, the task execution strategy for the task to be executed is determined based on historical execution strategies and reference execution strategies of reference users, including:

[0032] From the reference execution strategies of the reference users, remove the historical execution strategies to obtain the candidate execution strategies;

[0033] Based on the candidate execution strategies, the task execution strategy for the task to be executed is determined.

[0034] Secondly, this application also provides a task execution apparatus. The apparatus includes:

[0035] The processing module is used to segment the task requirement voice input by the target user to obtain at least two voice segments;

[0036] The first determining module is used to determine at least two prediction results corresponding to each speech segment through at least two Hidden Markov Models (HMMs). Each prediction result corresponding to each speech segment includes the category to which the key information in the speech segment predicted by each HMM model belongs and the confidence level of the category.

[0037] The second determining module is used to determine the key information contained in each speech segment based on at least two prediction results corresponding to each speech segment;

[0038] The third determination module is used to determine the task execution strategy for the task to be executed based on key information.

[0039] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the task execution method as described in any of the embodiments of the first aspect above.

[0040] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the task execution method as described in any of the embodiments of the first aspect above.

[0041] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the task execution method as described in any of the embodiments of the first aspect above.

[0042] The aforementioned task execution method, apparatus, computer equipment, and storage medium segment the task-requirement speech to ensure that at least two prediction results corresponding to each speech segment can be determined using two Hidden Markov Models (HMMs). Based on these at least two prediction results, key information is determined, and then a task execution strategy is determined based on this key information. Since the user only needs to input speech to obtain the task execution strategy, without manual input to the computer equipment or selection from multiple recommended strategies, this application achieves the determination of the task execution strategy while reducing user time and effort. Furthermore, by determining the category and confidence level of the key information in the speech segment, the accuracy of the key information determination is ensured. Determining the task execution strategy based on the key information ensures that the strategy meets the actual needs of the user and the task, enabling the strategy to be used to execute the task. Attached Figure Description

[0043] Figure 1 An application environment diagram of a task execution method provided in an embodiment of this application;

[0044] Figure 2 A flowchart of a task execution method provided in an embodiment of this application;

[0045] Figure 3 A flowchart illustrating the steps for determining key information provided in this application embodiment;

[0046] Figure 4 A flowchart illustrating the steps for determining a prediction result is provided in this application embodiment;

[0047] Figure 5 A flowchart illustrating the steps for determining an execution strategy is provided in an embodiment of this application.

[0048] Figure 6 A flowchart illustrating the steps of another task execution method provided in this application embodiment;

[0049] Figure 7 A structural block diagram of the first task execution device provided in the embodiments of this application;

[0050] Figure 8 This is a structural block diagram of a second type of task execution device provided in an embodiment of this application;

[0051] Figure 9 A structural block diagram of the third task execution device provided in the embodiments of this application;

[0052] Figure 10 A structural block diagram of the fourth task execution device provided in the embodiments of this application;

[0053] Figure 11 A structural block diagram of the fifth task execution device provided in the embodiments of this application;

[0054] Figure 12 A structural block diagram of the sixth task execution device provided in the embodiments of this application;

[0055] Figure 13 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0057] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application. In the description of this application, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0058] Based on the above, the task execution method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, in one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows. Figure 1As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data acquired by the task execution method. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements a task execution method.

[0059] This application discloses a task execution method, apparatus, computer device and its storage medium. The computer device segments the speech of the task requirement to obtain at least two speech segments, and determines at least two prediction results corresponding to each speech segment through at least two HMM models. Based on the at least two prediction results corresponding to each speech segment, key information is determined, and based on the key information, a task execution strategy for the task to be executed is determined.

[0060] In one embodiment, such as Figure 2 As shown, Figure 2 This is a flowchart illustrating a task execution method provided in an embodiment of this application. Figure 1 The task execution method performed by the computer device in the middle may include the following steps:

[0061] Step 201: Segment the voice input from the target user to obtain at least two voice segments.

[0062] It should be noted that there are many methods to obtain user input of task requirements in voice. For example, the task requirements can be received through a voice acquisition device set on a computer device, or the user's input of task requirements can be determined from the digital information sent by the user. In other words, there are many methods to obtain user input of voice, which will not be elaborated here. The following will provide a detailed explanation of the two methods mentioned above for obtaining user input of task requirements.

[0063] As one implementation method, if the computer device that determines the task execution method is equipped with a voice acquisition device for acquiring the voice required for the task, then the user can input the user's voice into the computer device within the effective acquisition range of the voice acquisition device, so that the voice acquisition device can acquire the user's input voice required for the task.

[0064] As another implementation, when a user needs to input task requirement voice into a computer device that performs the task execution method, but the computer device is not equipped with a voice acquisition device, the user can convert the audio signal of the task requirement voice into a digital signal and input the digital signal corresponding to the task requirement voice into the computer device.

[0065] The digital signal corresponding to the task requirement voice can be input to the computer signal in various forms, such as email, SMS, or compressed file. No limitation is made on the input form of the digital signal corresponding to the task requirement voice to the computer signal here.

[0066] To further explain, when it is necessary to obtain the task requirement voice, the text data of the task requirement voice input by the user can be obtained, and then the text data of the task requirement voice can be converted into audio data. The converted audio data is used as the task requirement voice input by the user, thereby ensuring the clarity of the task requirement voice and preventing large noise and interference in the surrounding environment when the user inputs the task requirement voice, which would affect the smooth determination of the task execution strategy for subsequent tasks.

[0067] In one embodiment of this application, an input field can be set in the interface of the computer device that performs the task execution method. When it is necessary to obtain the task requirement voice, the text data of the task requirement voice can be entered in the input field. The computer device will convert the received text data into audio data, which is the task requirement voice entered by the user.

[0068] In one embodiment of this application, when it is necessary to segment the task requirement speech, segmentation tools such as jieba segmentation, ltp segmentation, and ir segmentation can be used to segment the speech, thereby obtaining at least two speech segments. Specifically, the task requirement speech can be segmented using jieba (a segmentation tool) to obtain at least two segmented speech segments; or, the task requirement speech can be segmented using ltp (a Chinese language processing system) to obtain at least two segmented speech segments; or, the task requirement speech can be segmented using ir (a segmentation tool) to obtain at least two segmented speech segments.

[0069] Therefore, there are many ways to segment the voice required for a task, and we will not limit the segmentation methods here.

[0070] Step 202: Using at least two Hidden Markov Models (HMMs), determine at least two prediction results for each speech segment. Each prediction result for each speech segment includes the category of the key information in the speech segment predicted by each HMM model and the confidence level of the category.

[0071] The category of key information varies depending on the task to be executed corresponding to the voice request. For example, if the task to be executed corresponding to the voice request is a business trip, the category of key information for the business trip includes: departure time, return time, and business trip location. If the task to be executed corresponding to the voice request is a procurement task, the category of key information for the procurement task includes: purchased items, purchased quantity, and purchased time.

[0072] This can be understood as the fact that different tasks to be executed correspond to different categories of key information. Therefore, there is no limitation on the content included in the category of key information, and the category of key information can be changed according to the task to be executed and the actual situation.

[0073] In one embodiment of this application, at least two speech segments are input into at least two HMM models respectively, and at least two prediction results corresponding to each speech segment output by at least two HMM models can be obtained.

[0074] For example, if there are two speech segments and two Hidden Markov Models (HMMs), and the two speech segments are speech segment A and speech segment B, then inputting speech segment A into the two HMMs will yield two prediction results for speech segment A output by the two HMMs; inputting speech segment B into the two HMMs will yield two prediction results for speech segment B output by the two HMMs; wherein, the two prediction results for speech segment A and the two prediction results for speech segment B are the at least two prediction results for each speech segment in this application.

[0075] It should be noted that when a speech segment is input into the HMM model, the HMM model can determine the confidence level of the speech segment belonging to each candidate category; then, based on the confidence level of the speech segment belonging to each candidate category, the candidate category with the highest confidence level is selected as the category to which the key information in the speech segment belongs, and the highest confidence level is the confidence level of the category.

[0076] For example, if a speech segment has three candidate categories, namely category A, category B, and category C, and the speech segment is input into the HMM model, the HMM model outputs that the speech segment has a 30% confidence level for belonging to category A, a 40% confidence level for belonging to category B, and a 90% confidence level for belonging to category C. Among these, the candidate category with the highest confidence level is category C. Therefore, the key information of this speech segment belongs to category C, and the confidence level of the category is 90%.

[0077] Step 203: Based on at least two prediction results corresponding to each speech segment, determine the key information contained in each speech segment.

[0078] Among them, the key information is the information in the task requirement voice input by the target user that can affect the determination of the task execution strategy; for example, if the task to be executed corresponding to the task requirement voice is a business trip task, then the key information could be: departure on March 22 and return on March 23. Furthermore, the key information "departure on March 22" belongs to the category of "departure time"; the key information "return on March 23" belongs to the category of "return time".

[0079] It should be noted that when it is necessary to determine the key information contained in each speech segment, the target prediction result of each speech segment can be determined based on the confidence of the category to which it belongs in at least two prediction results; and then the key information contained in each speech segment can be determined based on the target prediction result.

[0080] To further explain, there are many methods for determining the target prediction results for each speech segment. Two methods for determining the target prediction results will be described below:

[0081] As one implementation method, the confidence of the category to which at least two prediction results belong can be scored according to the SVM model, and then the target prediction result of each speech segment can be determined from the at least two prediction results corresponding to each speech segment based on the scoring results.

[0082] As another implementation method, the prediction result with the highest confidence among the at least two prediction results corresponding to each speech segment can be used as the target prediction result for each speech segment based on the confidence of the category to which the at least two prediction results belong.

[0083] Step 204: Based on key information, determine the task execution strategy for the task to be executed.

[0084] It should be noted that before determining the task execution strategy for the task to be executed, an application template can be set in advance, key information can be filled into the application template to obtain an approval application; and the approval application can be sent to the review terminal for approval processing; if the approval result sent by the review terminal is approved, then the step of determining the task execution strategy based on the key information is executed.

[0085] In one embodiment of this application, when it is necessary to determine a task execution strategy: based on key information, the user's historical task execution information can be determined, and based on the user's historical task execution information, the target user's interest index and disinterest index for the historical execution strategy can be determined; based on the interest index and disinterest index, the reference user corresponding to the target user can be determined; and based on the reference execution strategy of the reference user and the historical execution strategy, the task execution strategy of the task to be executed can be determined.

[0086] Among them, the interest and disinterest indices of reference users regarding historical execution strategies are similar to those of target users regarding historical execution strategies.

[0087] Furthermore, when determining the reference users corresponding to the target users, users with similar interest and disinterest indices for historical execution strategies can be selected from multiple users based on the target users' interest and disinterest indices for historical execution strategies. These users are the reference users.

[0088] The aforementioned task execution method segments the voice request for the task, ensuring that at least two prediction results for each voice segment can be determined using two Hidden Markov Models (HMMs). Based on these prediction results, key information is identified, and then a task execution strategy is determined. Since the user only needs to input voice to obtain the task execution strategy, without manual input to the computer or selection from multiple recommended strategies, this method achieves task execution strategy determination while reducing user time and effort. Furthermore, by determining the category and confidence level of the key information in the voice segment, the accuracy of key information determination is ensured. Using key information to determine the task execution strategy ensures that the strategy meets the actual needs of the user and the task, enabling the strategy to be used to execute the task.

[0089] To ensure that the task execution strategy for the task to be performed can be determined based on the key information contained in each speech segment, it is necessary to determine the target prediction result based on at least two prediction results corresponding to each speech segment. However, since the key information contained in the speech segment can belong to any category, determining the key information is quite difficult. Therefore, the computer device in this embodiment can, as shown in the example, Figure 3 The method shown identifies the key information contained in each speech segment, specifically including the following steps:

[0090] Step 301: Based on the confidence of the category to which each speech segment belongs in at least two prediction results, determine the target prediction result for each speech segment from the at least two prediction results corresponding to each speech segment.

[0091] It should be noted that when determining the target prediction result for each speech segment, the categories to which the key information in each speech segment belongs can be sorted according to the confidence level of the categories in at least two prediction results corresponding to each speech segment. Based on the sorting results of the categories to which the key information in each speech segment belongs, the category to which the key information belongs with the highest confidence level is determined. Text parsing is performed on each speech segment to obtain the text data corresponding to each speech segment. Based on the category to which the key information belongs with the highest confidence level of each speech segment and the text data corresponding to each speech segment, the key information contained in each speech segment is determined.

[0092] As one implementation, if we pre-define two speech segments, Speech Speech A and Speech Speech B, each segment contains two prediction results. The two prediction results for Speech Speech A are: the key information belongs to category m with a confidence level of 80%; and the key information belongs to category n with a confidence level of 20%. Similarly, the two prediction results for Speech Speech B are: the key information belongs to category m with a confidence level of 20%; and the key information belongs to category n with a confidence level of 80%. Furthermore, the key information with the highest confidence level in category m for Speech Speech A is classified as category m; the key information with the highest confidence level in category n for Speech Speech B is classified as category n. Therefore, the target prediction result for Speech Speech A is the key information belonging to category m with a confidence level of 80%; and the target prediction result for Speech Speech B is the key information belonging to category n with a confidence level of 80%.

[0093] It should be noted that when it is necessary to determine the target prediction result for each speech segment, the following may also be included: using a support vector machine (SVM) model, scoring the confidence of the category to which each speech segment belongs in at least two prediction results to obtain the confidence score result of the category to which each prediction result belongs; based on the confidence score result of the category to which each prediction result belongs, determining the target prediction result for each speech segment from at least two prediction results to which each speech segment belongs.

[0094] The scoring result represents the accuracy of the predicted confidence level of the category to which each prediction result belongs. If the confidence level score of a prediction result belongs to a category is high, it means that the SVM model has a high accuracy in predicting the confidence level of the category to which the prediction result belongs. If the confidence level score of a prediction result belongs to a category is low, it means that the SVM model has a low accuracy in predicting the confidence level of the category to which the prediction result belongs.

[0095] In one embodiment of this application, the confidence scores of the categories to which at least two prediction results for each speech segment belong are input into an SVM model to obtain a score of the confidence scores of the categories to which at least two prediction results for each speech segment belong, as output by the SVM model. Based on the score of the confidence scores of the categories to which at least two prediction results for each speech segment belong, the confidence score of the category with the highest score among the at least two prediction results for each speech segment is determined. The prediction result corresponding to the confidence score of the category with the highest score among the at least two prediction results for each speech segment is the target prediction result of the speech segment.

[0096] To further explain, when inputting the confidence scores of the categories to which at least two prediction results for each speech segment belong into the SVM model, it is necessary to first normalize the confidence scores of the categories to which at least two prediction results for each speech segment belong, obtain the normalized results, and then input the normalized results for each speech segment into the SVM model.

[0097] The normalization formula (1) is shown below:

[0098]

[0099] Where y refers to the normalization result, x refers to the confidence of the category of each speech segment in at least two prediction results, min(x) refers to the minimum confidence of the category of each speech segment in at least two prediction results, and max(x) refers to the maximum confidence of the category of each speech segment in at least two prediction results.

[0100] Step 302: Determine the key information contained in each speech segment based on the category to which the key information in the target prediction results of each speech segment belongs.

[0101] It should be noted that when it is necessary to determine the key information contained in each speech segment, the text of each speech segment can be parsed to determine the text data corresponding to each speech segment. The combination of the text data of each speech segment and the category to which the key information in the target prediction result of each speech segment belongs is taken as the key information contained in each speech segment.

[0102] For example, if the key information category in the target prediction result of speech segment A is "business trip time", and the key information category in the target prediction result of speech segment B is "business trip location", then by parsing the text of the two speech segments, the text data of speech segment A is "March 20th morning"; the text data of speech segment B is "city M". Combining the text data of speech segment A with the "business trip time" category in the target prediction result of speech segment A, the key information is determined to be: the business trip time is March 20th morning; combining the text data of speech segment B with the "business trip location" category in the target prediction result of speech segment B, the key information is determined to be: the business trip location is in city M.

[0103] The above task execution method determines the target prediction result of each speech segment, which enables the subsequent determination of the key information contained in each speech segment based on the target prediction result. Since the target prediction result of each speech segment is determined based on the confidence of the category of at least two prediction results corresponding to each speech segment, the accuracy of key information determination is high, and the subsequent task execution strategy based on key information is consistent with the actual situation of the task to be executed.

[0104] To ensure the accuracy of identifying key information, it is necessary to determine the category to which the key information belongs for each speech segment. However, determining the category of key information for each speech segment is quite difficult, making it hard to guarantee the accuracy of this determination. Therefore, the computer device in this embodiment can, as shown in the example... Figure 4 The method shown determines at least two prediction results for each speech segment using at least two Hidden Markov Models (HMMs), specifically including the following steps:

[0105] Step 401: Based on at least two HMM models, determine at least two candidate results corresponding to each speech segment; wherein, each candidate result corresponding to each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model belonging to each candidate category.

[0106] In one embodiment of this application, if a certain speech segment can be any one of multiple candidate categories, the speech segment is input into a certain HMM model to obtain the confidence level of the speech segment belonging to each candidate category output by the HMM model.

[0107] For example, if a speech segment C can be any of three candidate categories, namely candidate category a, candidate category b, and candidate category c, by inputting the speech segment C into a certain HMM model, the confidence scores of the speech segment C belonging to candidate category a, candidate category b, and candidate category c output by the HMM model can be obtained.

[0108] Step 402: Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, determine the category to which the key information in each speech segment belongs and the confidence level of the category.

[0109] It should be noted that when it is necessary to determine the category to which the key information in each speech segment belongs and the confidence level of that category, the maximum confidence level of the key information in the candidate results corresponding to each speech segment belongs to each candidate category can be determined based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category. The candidate category corresponding to the maximum confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category is taken as the category to which the key information in each speech segment belongs, and the maximum confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category is the confidence level of the category to which the key information in each speech segment belongs.

[0110] For example, if the confidence level of speech segment C output by the HMM model is determined to be 80% for belonging to candidate category a, 20% for belonging to candidate category b, and 20% for belonging to candidate category c; then, the 80% confidence level of speech segment C belonging to candidate category a is the maximum confidence level of the key information in the candidate results corresponding to speech segment C belonging to each candidate category. Therefore, candidate category a is the category to which the key information in speech segment C belongs, and 80% is the confidence level of the category to which each speech segment C belongs.

[0111] To further explain, when training an HMM model, the sample speech to be used for training the HMM model and the category of the key information corresponding to each sample speech can be determined in advance. By inputting the sample speech into the HMM model and adjusting the parameter library of the HMM model based on the difference between the prediction result output by the HMM model and the category of the key information corresponding to the sample speech, the trained HMM model can be obtained.

[0112] Furthermore, during the training of the HMM model, the HMM model outputs candidate results based on the sample speech, and the candidate results include the confidence level of the sample speech belonging to each candidate category; by determining the maximum value among the confidence levels of the sample speech belonging to each candidate category, the prediction result output by the HMM model based on the sample speech is obtained.

[0113] The above task execution method uses at least two Hidden Markov Models (HMMs) to determine at least two candidate results for each speech segment. It then determines the category and confidence level of the key information in the speech segment based on the maximum confidence level among the candidate results for each candidate category. This improves the accuracy of determining the category and confidence level of the key information in the speech segment and provides a foundation for determining the task execution strategy for the task to be executed.

[0114] Because target users need to manually filter from multiple recommended strategies before determining the task execution strategy, they consume a significant amount of time and effort during the filtering process. Furthermore, when target users lack the time or find it inconvenient to filter, it can also affect the determination of the task execution strategy. Therefore, the computer device in this embodiment can, as follows: Figure 5 The method shown determines the task execution strategy for the task to be executed based on key information, specifically including the following steps:

[0115] Step 501: Based on the key information, perform task approval processing on the task to be executed and obtain the approval result.

[0116] It should be noted that there are many methods for processing task approvals. For example, tasks can be approved using computer equipment to obtain approval results, or approval applications can be generated based on key information and manually reviewed to obtain approval results. In short, there are many methods for processing task approvals, which will not be elaborated upon here. The following will explain the two methods mentioned above in detail:

[0117] One approach is to send key information directly to a computer device, enabling the device to process the task to be executed based on that information and obtain the approval result from the computer.

[0118] As another implementation method, based on key information and application template, an approval application corresponding to the task to be executed is generated, and the approval application is sent to the review terminal for approval processing; the approval result sent by the review terminal is received.

[0119] The application template contains different fields for filling in key information. By filling in the key information in the corresponding fields, the approval application for the task to be executed can be generated. Furthermore, the application templates for different tasks to be executed are different.

[0120] Step 502: If the approval result is passed, determine the user's historical task execution information corresponding to the key information.

[0121] The user's historical task execution information includes historical execution strategies and corresponding satisfaction scores for those strategies.

[0122] It should be noted that the user's historical task execution information corresponding to the key information can be pre-stored in the local storage location of the computer device or in the cloud storage location. If the approval result is passed, the user's historical task execution information corresponding to the key information will be obtained by filtering from the local storage location of the computer device or in the cloud storage location based on the key information.

[0123] Furthermore, since the user's historical task execution information includes historical execution strategies and corresponding satisfaction scores, the user's historical task execution information is updated based on the target user's task execution strategy and corresponding satisfaction score after each target user completes the task to be executed.

[0124] Step 503: Determine the task execution strategy for the task to be executed based on the user's historical task execution information.

[0125] It should be noted that when it is necessary to determine the task execution strategy for the task to be executed, the following may be included: based on the satisfaction scores corresponding to the historical execution strategies, determine the target user's interest index and disinterest index for the historical execution strategies; based on the interest index and disinterest index, determine the reference user corresponding to the target user; based on the historical execution strategies and the reference execution strategies of the reference users, determine the task execution strategy for the task to be executed.

[0126] To further explain, when it is necessary to determine the target user's interest index for historical execution strategies, the satisfaction score corresponding to the historical execution strategies can be substituted into the calculation formula (2) to obtain the target user's interest index for historical execution strategies. The calculation formula 2 is as follows:

[0127]

[0128] Where μ1 refers to the target user's interest index in historical execution strategies, and r ui This refers to the target user's satisfaction rating for the historical implementation strategies. This refers to the average satisfaction rating of target users for historically implemented strategies.

[0129] max

[0130] r u This refers to the maximum satisfaction score of the target user for the historical implementation strategy.

[0131] To further explain, when it is necessary to determine the target user's disinterest index regarding historical execution strategies, the satisfaction score corresponding to the historical execution strategies can be substituted into the calculation formula (3) to obtain the target user's disinterest index regarding historical execution strategies. The calculation formula (3) is as follows:

[0132]

[0133] Where μ2 refers to the target user's disinterest index in historical execution strategies, r ui This refers to the target user's satisfaction rating for the historical implementation strategies. This refers to the average satisfaction rating of target users for historically implemented strategies, r. u min This refers to the minimum satisfaction score given by the target user for the historical implementation strategy.

[0134] It should be noted that when determining the reference user corresponding to the target user based on the interest index and the disinterest index, the similarity of the interest index between the target user and multiple candidate users can be determined based on the calculation formula (4) and the calculation formula (5), respectively; and then the candidate user with the largest similarity of the interest index and the disinterest index is selected as the reference user corresponding to the target user.

[0135] To further explain, when it is necessary to determine the similarity of the interest index between the target user and the candidate user, the interest index can be substituted into the calculation formula (4) to obtain the similarity of the interest index between the target user and the candidate user; wherein, the calculation formula (4) is as follows:

[0136]

[0137] in, This refers to the similarity of interest indices between target users and candidate users, where u represents the target user and v represents the candidate user, μ1(r ui () refers to the interest index of the target user for the historical execution strategy i. Iuv refers to the average interest index of target users for historical execution strategies i; Iuv is the set of execution strategies for which target users and candidate users have jointly conducted satisfaction ratings.

[0138] To further explain, when it is necessary to determine the similarity of the disinterest index between the target user and the candidate user, the disinterest index can be substituted into the calculation formula (5) to obtain the similarity of the disinterest index between the target user and the candidate user; wherein, the calculation formula (5) is as follows:

[0139]

[0140] in, This refers to the similarity of the disinterest index between the target user and the candidate user, where u represents the target user and v represents the candidate user, and μ2(r) represents the disinterest index. ui This refers to the target user's level of disinterest in historically implemented strategy i. Iuv refers to the average interest index of target users for historical execution strategies i; Iuv is the set of execution strategies for which target users and candidate users have jointly conducted satisfaction ratings.

[0141] In one embodiment of this application, when determining the task execution strategy of a task to be executed based on historical execution strategies and reference execution strategies of reference users, the following may be included: removing historical execution strategies from the reference execution strategies of reference users to obtain candidate execution strategies; and determining the task execution strategy of the task to be executed based on the candidate execution strategies.

[0142] To further explain, when determining the task execution strategy for the task to be executed based on the candidate execution strategies, the target user's predicted score for the candidate execution strategies can be calculated based on the calculation formula (6). Based on the predicted score of the candidate execution strategies, the candidate execution strategy with the highest predicted score is selected as the task execution strategy for the task to be executed.

[0143] The calculation formula (6) is shown below:

[0144]

[0145] Where, r 预测 This refers to the target user's predicted score for the candidate execution strategy. This refers to the average satisfaction rating of target users for historical implementation strategies, where k is the number of reference users, u is the target user, v is the candidate user, and r is the number of reference users. vi This refers to referencing user satisfaction ratings for candidate execution strategies, s uv This refers to the similarity between the target user and the reference user.

[0146] Furthermore, s uv The calculation formula (7) can be shown below:

[0147]

[0148] Among them, s uv This refers to the similarity between the target user and the reference user. This refers to the similarity of interest index between the target user and the reference user. This refers to the similarity of the disinterest index between the target user and the reference user.

[0149] The above task execution method, when the approval result is passed, determines the task execution strategy of the task to be executed based on the user's historical task execution information, ensuring that the task execution strategy of the task to be executed can meet the actual needs and selection habits of the target user.

[0150] In one embodiment of this application, such as Figure 6 As shown, Figure 6 This application provides a flowchart of another task execution method, which includes the following steps when determining the task execution strategy for the task to be executed:

[0151] Step 601: Segment the voice input by the target user to obtain at least two voice segments.

[0152] Step 602: Based on at least two HMM models, determine at least two candidate results corresponding to each speech segment; wherein, each candidate result corresponding to each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model belonging to each candidate category.

[0153] Step 603: Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, determine the category to which the key information in each speech segment belongs and the confidence level of the category.

[0154] Step 604: Using the Support Vector Machine (SVM) model, score the confidence of the category to which each speech segment belongs in at least two prediction results, and obtain the confidence score of the category to which each prediction result belongs.

[0155] Step 605: Based on the confidence score of the category to which each prediction result belongs, determine the target prediction result for each speech segment from at least two prediction results corresponding to each speech segment.

[0156] Step 606: Determine the key information contained in each speech segment based on the category to which the key information in the target prediction results of each speech segment belongs.

[0157] Step 607: Based on the key information and application template, generate the approval application corresponding to the task to be executed, and send the approval application to the review terminal for approval processing.

[0158] Step 608: Receive the approval result sent by the review terminal.

[0159] Step 609: If the approval result is passed, determine the user's historical task execution information corresponding to the key information.

[0160] Step 610: Based on the satisfaction scores corresponding to the historical execution strategies, determine the target user's interest index and disinterest index for the historical execution strategies.

[0161] Step 611: Based on the interest index and the disinterest index, determine the reference users corresponding to the target users.

[0162] Step 612: Remove historical execution strategies from the reference execution strategies of the reference users to obtain candidate execution strategies.

[0163] Step 613: Based on the candidate execution strategies, determine the task execution strategy for the task to be executed.

[0164] The aforementioned task execution method segments the voice request for the task, ensuring that at least two prediction results for each voice segment can be determined using two Hidden Markov Models (HMMs). Based on these prediction results, key information is identified, and then a task execution strategy is determined. Since the user only needs to input voice to obtain the task execution strategy, without manual input to the computer or selection from multiple recommended strategies, this method achieves task execution strategy determination while reducing user time and effort. Furthermore, by determining the category and confidence level of the key information in the voice segment, the accuracy of key information determination is ensured. Using key information to determine the task execution strategy ensures that the strategy meets the actual needs of the user and the task, enabling the strategy to be used to execute the task.

[0165] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0166] Based on the same inventive concept, this application also provides a task execution apparatus for implementing the task execution method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more task execution apparatus embodiments provided below can be found in the limitations of the task execution method described above, and will not be repeated here.

[0167] In one embodiment, such as Figure 7 As shown, a task execution device is provided, comprising: a processing module 10, a first determining module 20, a second determining module 30, and a third determining module 40, wherein:

[0168] The processing module 10 is used to segment the task requirement voice input by the target user to obtain at least two voice segments.

[0169] The first determining module 20 is used to determine at least two prediction results corresponding to each speech segment through at least two Hidden Markov Models (HMMs). Each prediction result corresponding to each speech segment includes the category to which the key information in the speech segment is predicted by each HMM model and the confidence level of the category.

[0170] The second determining module 30 is used to determine the key information contained in each speech segment based on at least two prediction results corresponding to each speech segment.

[0171] The third determination module 40 is used to determine the task execution strategy for the task to be executed based on key information.

[0172] The aforementioned task execution device segments the voice request for the task, ensuring that at least two prediction results corresponding to each voice segment can be determined using two Hidden Markov Models (HMMs). Based on these prediction results, key information is determined, and then a task execution strategy is determined. Since the user only needs to input voice to obtain the task execution strategy, without manual input to the computer or selection from multiple recommended strategies, this application achieves task execution strategy determination while reducing user time and effort. Furthermore, by determining the category and confidence level of the key information in the voice segment, the accuracy of key information determination is ensured. Determining the task execution strategy based on key information ensures that the strategy meets the actual needs of the user and the task, enabling the strategy to be used to execute the task.

[0173] In one embodiment, such as Figure 8 As shown, a task execution device is provided. The second determining module 30 in the task execution device includes: a first determining unit 31 and a second determining unit 32, wherein:

[0174] The first determining unit 31 is used to determine the target prediction result of each speech segment from the at least two prediction results corresponding to each speech segment based on the confidence level of the category to which each speech segment belongs in the at least two prediction results corresponding to each speech segment.

[0175] The second determining unit 32 is used to determine the key information contained in each speech segment based on the category to which the key information in the target prediction results of each speech segment belongs.

[0176] Specifically, the second determining unit is used to: score the confidence of the category to which each speech segment belongs in at least two prediction results using a support vector machine (SVM) model, and obtain the confidence score of the category to which each prediction result belongs; and determine the target prediction result of each speech segment from at least two prediction results corresponding to each speech segment based on the confidence score of the category to which each prediction result belongs.

[0177] In one embodiment, such as Figure 9 As shown, a task execution device is provided. The first determining module 20 of the task execution device includes: a third determining unit 21 and a fourth determining unit 22, wherein:

[0178] The third determining unit 21 is used to determine at least two candidate results corresponding to each speech segment based on at least two HMM models; wherein each candidate result corresponding to each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model belonging to each candidate category.

[0179] The fourth determining unit 22 is used to determine the category to which the key information in each speech segment belongs and the confidence level of the category to which the key information belongs in the candidate results corresponding to each speech segment.

[0180] In one embodiment, such as Figure 10 As shown, a task execution device is provided. The third determining module 40 in this device includes: an approval unit 41, a fifth determining unit 42, and a sixth determining unit 43, wherein:

[0181] Approval unit 41 is used to process the task approval based on key information and obtain the approval result.

[0182] The fifth determining unit 42 is used to determine the user's historical task execution information corresponding to the key information if the approval result is passed.

[0183] The sixth determining unit 43 is used to determine the task execution strategy of the task to be executed based on the user's historical task execution information.

[0184] In one embodiment, such as Figure 11As shown, a task execution device is provided, wherein the approval unit 41 of the task execution device includes: a production subunit 411 and a receiving subunit 412, wherein:

[0185] Production subunit 411 is used to generate approval applications corresponding to tasks to be executed based on key information and application templates, and send the approval applications to the review terminal for approval processing.

[0186] The receiving subunit 412 is used to receive the approval results sent by the review terminal.

[0187] In one embodiment, such as Figure 12 As shown, a task execution apparatus is provided, wherein the sixth determining unit 43 of the task execution apparatus includes: a production subunit 411 and a receiving subunit 412, wherein:

[0188] The first determining subunit 431 is used to determine the target user's interest index and disinterest index for the historical execution strategy based on the satisfaction score corresponding to the historical execution strategy.

[0189] The second determining subunit 432 is used to determine the reference user corresponding to the target user based on the interest index and the disinterest index.

[0190] The third determining subunit 433 is used to determine the task execution strategy of the task to be executed based on the historical execution strategy and the reference execution strategy of the reference user.

[0191] Specifically, the third determining subunit is used to: remove historical execution strategies from the reference execution strategies of the reference user to obtain candidate execution strategies; and determine the task execution strategy of the task to be executed based on the candidate execution strategies.

[0192] Each module in the aforementioned task execution device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0193] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 13As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a task execution method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0194] Those skilled in the art will understand that Figure 13 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0195] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0196] The task requirement speech input by the target user is segmented to obtain at least two speech segments;

[0197] Using at least two Hidden Markov Models (HMMs), at least two prediction results are determined for each speech segment. Each prediction result for each speech segment includes the category of the key information in the speech segment predicted by each HMM model and the confidence level of the category.

[0198] Based on at least two prediction results corresponding to each speech segment, determine the key information contained in each speech segment;

[0199] Based on key information, determine the task execution strategy for the task to be executed.

[0200] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0201] Based on the confidence of the category to which each speech segment belongs in at least two prediction results, the target prediction result for each speech segment is determined from the at least two prediction results corresponding to each speech segment.

[0202] Based on the category of key information in each speech segment and the target prediction results of each speech segment, determine the key information contained in each speech segment.

[0203] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0204] By using the Support Vector Machine (SVM) model, the confidence scores of the categories to which each speech segment belongs in at least two prediction results are scored, and the confidence scores of the categories to which each prediction result belongs are obtained.

[0205] Based on the confidence score of the category to which each prediction result belongs, the target prediction result for each speech segment is determined from at least two prediction results corresponding to each speech segment.

[0206] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0207] Based on at least two Hidden Markov Models (HMMs), at least two candidate results are determined for each speech segment; wherein, each candidate result for each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model to belong to each candidate category;

[0208] Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, the category to which the key information in each speech segment belongs and the confidence level of the category are determined.

[0209] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0210] Based on key information, the task to be executed is processed for approval, and the approval result is obtained.

[0211] If the approval result is passed, the user's historical task execution information corresponding to the key information will be determined;

[0212] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed.

[0213] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0214] Based on key information and application templates, generate approval applications for tasks to be executed and send them to the approval terminal for processing.

[0215] Receive the approval results sent by the review terminal.

[0216] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0217] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed, including:

[0218] Based on the satisfaction scores corresponding to historical execution strategies, determine the target users' interest index and disinterest index regarding historical execution strategies;

[0219] Based on the interest index and the disinterest index, identify the reference users corresponding to the target users;

[0220] Based on historical execution strategies and reference execution strategies of reference users, determine the task execution strategy for the task to be executed.

[0221] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0222] From the reference execution strategies of the reference users, remove the historical execution strategies to obtain the candidate execution strategies;

[0223] Based on the candidate execution strategies, the task execution strategy for the task to be executed is determined.

[0224] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0225] The task requirement speech input by the target user is segmented to obtain at least two speech segments;

[0226] Using at least two Hidden Markov Models (HMMs), at least two prediction results are determined for each speech segment. Each prediction result for each speech segment includes the category of the key information in the speech segment predicted by each HMM model and the confidence level of the category.

[0227] Based on at least two prediction results corresponding to each speech segment, determine the key information contained in each speech segment;

[0228] Based on key information, determine the task execution strategy for the task to be executed.

[0229] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0230] Based on the confidence of the category to which each speech segment belongs in at least two prediction results, the target prediction result for each speech segment is determined from the at least two prediction results corresponding to each speech segment.

[0231] Based on the category of key information in each speech segment and the target prediction results of each speech segment, determine the key information contained in each speech segment.

[0232] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0233] The confidence scores of the categories to which each speech segment belongs are scored using the support vector machine (SVM) model, thus obtaining the confidence scores of the categories to which each prediction result belongs.

[0234] Based on the confidence score of the category to which each prediction result belongs, the target prediction result for each speech segment is determined from at least two prediction results corresponding to each speech segment.

[0235] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0236] Based on at least two Hidden Markov Models (HMMs), at least two candidate results are determined for each speech segment; wherein, each candidate result for each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model to belong to each candidate category;

[0237] Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, the category to which the key information in each speech segment belongs and the confidence level of the category are determined.

[0238] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0239] Based on key information, the task to be executed is processed for approval, and the approval result is obtained.

[0240] If the approval result is passed, the user's historical task execution information corresponding to the key information will be determined;

[0241] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed.

[0242] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0243] Based on key information and application templates, generate approval applications for tasks to be executed and send them to the approval terminal for processing.

[0244] Receive the approval results sent by the review terminal.

[0245] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0246] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed, including:

[0247] Based on the satisfaction scores corresponding to historical execution strategies, determine the interest index and disinterest index of target users regarding historical execution strategies;

[0248] Based on the interest index and the disinterest index, identify the reference users corresponding to the target users;

[0249] Based on historical execution strategies and reference execution strategies of reference users, determine the task execution strategy for the task to be executed.

[0250] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0251] From the reference execution strategies of the reference users, remove the historical execution strategies to obtain the candidate execution strategies;

[0252] Based on the candidate execution strategies, the task execution strategy for the task to be executed is determined.

[0253] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0254] The task requirement speech input by the target user is segmented to obtain at least two speech segments;

[0255] Using at least two Hidden Markov Models (HMMs), at least two prediction results are determined for each speech segment. Each prediction result for each speech segment includes the category of the key information in the speech segment predicted by each HMM model and the confidence level of the category.

[0256] Based on at least two prediction results corresponding to each speech segment, determine the key information contained in each speech segment;

[0257] Based on key information, determine the task execution strategy for the task to be executed.

[0258] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0259] Based on the confidence of the category to which each speech segment belongs in at least two prediction results, the target prediction result for each speech segment is determined from the at least two prediction results corresponding to each speech segment.

[0260] Based on the category of key information in each speech segment and the target prediction results of each speech segment, determine the key information contained in each speech segment.

[0261] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0262] The confidence scores of the categories to which each speech segment belongs are scored using the support vector machine (SVM) model, thus obtaining the confidence scores of the categories to which each prediction result belongs.

[0263] Based on the confidence score of the category to which each prediction result belongs, the target prediction result for each speech segment is determined from at least two prediction results corresponding to each speech segment.

[0264] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0265] Based on at least two Hidden Markov Models (HMMs), at least two candidate results are determined for each speech segment; wherein, each candidate result for each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model to belong to each candidate category;

[0266] Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, the category to which the key information in each speech segment belongs and the confidence level of the category are determined.

[0267] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0268] Based on key information, the task to be executed is processed for approval, and the approval result is obtained.

[0269] If the approval result is passed, the user's historical task execution information corresponding to the key information will be determined;

[0270] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed.

[0271] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0272] Based on key information and application templates, generate approval applications for tasks to be executed and send them to the approval terminal for processing.

[0273] Receive the approval results sent by the review terminal.

[0274] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0275] Based on the user's historical task execution information, determine the task execution strategy for the task to be executed, including:

[0276] Based on the satisfaction scores corresponding to historical execution strategies, determine the target users' interest index and disinterest index regarding historical execution strategies;

[0277] Based on the interest index and the disinterest index, identify the reference users corresponding to the target users;

[0278] Based on historical execution strategies and reference execution strategies of reference users, determine the task execution strategy for the task to be executed.

[0279] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0280] From the reference execution strategies of the reference users, remove the historical execution strategies to obtain the candidate execution strategies;

[0281] Based on the candidate execution strategies, the task execution strategy for the task to be executed is determined.

[0282] It should be noted that the target user information (including but not limited to target user device information, target user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the target user or fully authorized by all parties.

[0283] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0284] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0285] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A task execution method, characterized in that, The method includes: The task requirement speech input by the target user is segmented to obtain at least two speech segments; Using at least two Hidden Markov Models (HMMs), at least two prediction results are determined for each speech segment. Each prediction result for each speech segment includes the category of the key information in the speech segment predicted by each HMM model and the confidence level of the category. Based on at least two prediction results corresponding to each speech segment, determine the key information contained in each speech segment; Based on the aforementioned key information, the task to be executed is processed for approval, and an approval result is obtained. If the approval result is passed, the user's historical task execution information corresponding to the key information is determined; the user's historical task execution information includes historical execution strategies and the satisfaction scores corresponding to the historical execution strategies; Based on the satisfaction scores corresponding to the historical execution strategies, the interest index and disinterest index of the target users regarding the historical execution strategies are determined. Based on the interest index and the disinterest index, the reference users corresponding to the target user are determined; Based on the historical execution strategy and the reference execution strategy of the reference user, the task execution strategy of the task to be executed is determined.

2. The method according to claim 1, characterized in that, The determination of key information contained in each speech segment based on at least two prediction results corresponding to each speech segment includes: Based on the confidence of the category to which each speech segment belongs in at least two prediction results, the target prediction result for each speech segment is determined from the at least two prediction results corresponding to each speech segment. Based on the category of key information in each speech segment and the target prediction results of each speech segment, determine the key information contained in each speech segment.

3. The method according to claim 2, characterized in that, The step of determining the target prediction result for each speech segment from the at least two prediction results corresponding to each speech segment based on the confidence level of the category to which each speech segment belongs includes: The confidence scores of the categories to which each speech segment belongs are scored using the support vector machine (SVM) model, thus obtaining the confidence scores of the categories to which each prediction result belongs. Based on the confidence score of the category to which each prediction result belongs, the target prediction result for each speech segment is determined from at least two prediction results corresponding to each speech segment.

4. The method according to claim 1, characterized in that, The process of determining at least two prediction results for each speech segment using at least two Hidden Markov Models (HMMs) includes: Based on at least two Hidden Markov Models (HMMs), at least two candidate results are determined for each speech segment; wherein, each candidate result for each speech segment includes: the confidence level of the key information in the speech segment predicted by each HMM model belonging to each candidate category; Based on the confidence level of the key information in the candidate results corresponding to each speech segment belonging to each candidate category, the category to which the key information in each speech segment belongs and the confidence level of the category are determined.

5. The method according to claim 1, characterized in that, Based on the key information, the task approval process for the task to be executed is performed to obtain the approval result, including: Based on the key information and application template, an approval application corresponding to the task to be executed is generated, and the approval application is sent to the review terminal for approval processing; Receive the approval result sent by the review terminal.

6. The method according to claim 1, characterized in that, The step of determining the task execution strategy for the task to be executed based on the historical execution strategy and the reference execution strategy of the reference user includes: Remove the historical execution strategies from the reference execution strategies of the reference users to obtain candidate execution strategies; Based on the candidate execution strategies, the task execution strategy for the task to be executed is determined.

7. A task execution device, characterized in that, The device includes: The processing module is used to segment the task requirement voice input by the target user to obtain at least two voice segments; The first determining module is used to determine at least two prediction results corresponding to each speech segment through at least two Hidden Markov Models (HMMs). Each prediction result corresponding to each speech segment includes the category to which the key information in the speech segment predicted by each HMM model belongs and the confidence level of the category. The second determining module is used to determine the key information contained in each speech segment based on at least two prediction results corresponding to each speech segment; The third determining module is used to perform task approval processing on the task to be executed based on the key information, and obtain an approval result; if the approval result is passed, the user's historical task execution information corresponding to the key information is determined; the user's historical task execution information includes historical execution strategies and satisfaction scores corresponding to the historical execution strategies; based on the satisfaction scores corresponding to the historical execution strategies, the target user's interest index and disinterest index for the historical execution strategies are determined; based on the interest index and the disinterest index, the reference user corresponding to the target user is determined; based on the historical execution strategies and the reference execution strategies of the reference users, the task execution strategy for the task to be executed is determined.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

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

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images