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

Abstract

The application relates to a business 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: acquiring user voice input by a user and containing a target business function, the target business function being at least one optional business function in a business system; extracting a function voice segment of the target business function from the user voice through a hidden Markov (HMM) model; the HMM model being obtained through training according to function voice data of each optional business function supported by the business system; analyzing at least one to-be-executed business corresponding to the function voice segment; acquiring a target business link corresponding to each to-be-executed business, and executing each to-be-executed business based on the target business link. According to the application, the user does not need to spend a lot of time learning the triggering method of each function in the system, and the time consumption of the user is reduced.

Description

Technical Field

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

[0002] As computer technology continues to develop, the functions of the system are becoming increasingly rich and complete. Furthermore, with the continuous updates and iterations of various functions within the system, the labor costs required to execute business transactions through the system are decreasing.

[0003] However, as the system's functions continue to increase, its operation becomes increasingly cumbersome. When a user wants to perform a certain business function in the system, they need to spend a certain amount of time learning how to trigger each function to prevent triggering the wrong function during operation. Therefore, learning how to trigger each function consumes a lot of the staff's time. Summary of the Invention

[0004] Therefore, it is necessary to provide a business execution method, apparatus, computer equipment, and storage medium that can reduce the time required for users to learn the triggering methods of various functions in the system, in order to address the above-mentioned technical problems.

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

[0006] Acquire user voice input containing the target business function, where the target business function is at least one optional business function in the business system;

[0007] The Hidden Markov Model (HMM) is used to extract functional speech segments of the target business function from the user's speech; the HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0008] At least one pending service corresponding to the parsed voice segment;

[0009] Obtain the target business link corresponding to each business to be executed, and execute each business to be executed based on the target business link.

[0010] In one embodiment, at least one service to be executed corresponding to the parsing function voice segment includes:

[0011] The parsing function retrieves the text data corresponding to the speech segments.

[0012] The text data corresponding to the functional speech segments are quantized to obtain the text word vectors of the functional speech segments;

[0013] By using a Support Vector Machine (SVM) model, feature parsing is performed on text word vectors to obtain at least one business function to be executed corresponding to the target business function; wherein, the SVM model is trained based on the text data of each optional business function supported by the business system.

[0014] In one embodiment, the functional speech segments of the target business function are extracted from the user's speech using an Hidden Markov Model (HMM), including:

[0015] The user's voice is segmented to obtain at least two sub-voice segments;

[0016] The feature extraction network of the HMM model is used to extract the feature values ​​of each sub-speech segment. Then, the function prediction network of the HMM model is used to determine the probability that each sub-speech segment belongs to the optional service function based on the feature values ​​of each sub-speech segment and the value range of the function features corresponding to the optional service functions.

[0017] Based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional business function, the functional voice segment of the target business function is determined.

[0018] In one embodiment, the training process of the HMM model includes:

[0019] The voice data of each optional business function supported by the business system is denoised to obtain the denoised sample voice.

[0020] The feature extraction network of the HMM model is used to extract feature values ​​of the sample speech.

[0021] Based on the relationship between the feature values ​​and functional feature value ranges of the sample speech, adjust the model parameters of the feature extraction network of the HMM model.

[0022] In one embodiment, if the number of target service functions contained in the user's voice is at least two, then the target service link corresponding to each service to be executed is obtained, including:

[0023] Based on each pending business, the sub-business links corresponding to each pending business are selected from the historical link database;

[0024] Based on the user's voice, determine the execution order of each pending business function;

[0025] Based on the execution order of each pending business, the sub-business links corresponding to each pending business are combined to obtain the target business link corresponding to each pending business.

[0026] In one embodiment, based on each service to be executed, the sub-service links corresponding to each service to be executed are selected from the historical link database, including:

[0027] From the historical link database, filter the candidate business links associated with each business to be executed, and determine the maximum matching degree between the business to be executed and the candidate business links;

[0028] Determine if the maximum matching degree is less than the matching degree threshold;

[0029] If so, then display the candidate business link to the user;

[0030] Obtain the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

[0031] In one embodiment, each service to be executed is performed based on the target service link, including:

[0032] Based on the target business chain, output the operation instructions corresponding to the business to be executed;

[0033] Receive execution instructions from users based on operation guidelines, triggering pending business operations, and respond to the execution instructions to execute each pending task.

[0034] In one embodiment, it further includes:

[0035] If the execution instruction for a user-triggered pending service is found to be inconsistent with the operation guide, the actual service link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual service link and the pending service.

[0036] Secondly, this application also provides a business execution apparatus. The apparatus includes:

[0037] The first acquisition module is used to acquire user voice input containing the target business function, wherein the target business function is at least one optional business function in the business system.

[0038] The extraction module is used to extract functional speech segments of the target business function from the user's speech using a Hidden Markov Model (HMM). The HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0039] The parsing module is used to parse at least one service to be executed corresponding to a functional voice segment;

[0040] The second acquisition module is used to acquire the target business link corresponding to each business to be executed, and to execute each business to be executed based on the target business link.

[0041] 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 business execution method as described in any of the embodiments of the first aspect above.

[0042] 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 business execution method as described in any of the embodiments of the first aspect above.

[0043] 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 business execution method as described in any of the embodiments of the first aspect above.

[0044] The aforementioned service execution method, apparatus, computer equipment, and storage medium obtain functional speech segments from the user's voice by inputting the acquired user voice into an HMM model, and parse at least one service to be executed corresponding to each functional speech segment. Based on the target service link corresponding to each service to be executed, each service to be executed is then executed. Since the user only needs to input voice during the process, there is no need for the user to spend a significant amount of time learning the triggering methods of various functions in the system before executing the services to be executed, thus reducing the user's time consumption. Furthermore, since the execution of each service to be executed is based on the target service link corresponding to each service to be executed, it ensures that no incorrect functions are triggered when executing services to be executed based on the user's voice, thereby improving the accuracy of the operation during the execution of each service to be executed. Attached Figure Description

[0045] Figure 1 An application environment diagram for a business execution method provided in an embodiment of this application;

[0046] Figure 2 A flowchart of a business execution method provided in an embodiment of this application;

[0047] Figure 3 A flowchart for parsing a business process to be executed is provided in an embodiment of this application;

[0048] Figure 4 A flowchart illustrating the extraction of speech segments is provided in an embodiment of this application.

[0049] Figure 5 A flowchart for obtaining a target service link is provided in an embodiment of this application;

[0050] Figure 6 A flowchart for executing various pending business operations is provided as an embodiment of this application;

[0051] Figure 7 A flowchart illustrating another business execution method provided in this application embodiment;

[0052] Figure 8A structural block diagram of a first type of service execution device provided in the embodiments of this application;

[0053] Figure 9 A structural block diagram of a second type of service execution device provided in the embodiments of this application;

[0054] Figure 10 A structural block diagram of a third service execution device provided in the embodiments of this application;

[0055] Figure 11 A structural block diagram of the fourth service execution device provided in the embodiments of this application;

[0056] Figure 12 A structural block diagram of the fifth type of service execution device provided in the embodiments of this application;

[0057] Figure 13 A structural block diagram of the sixth service execution device provided in the embodiments of this application;

[0058] Figure 14 A structural block diagram of the seventh service execution device provided in the embodiments of this application;

[0059] Figure 15 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0060] 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.

[0061] 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.

[0062] Based on the above, the business execution method provided in this application embodiment can be applied to, for example, Figure 1In 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 1 As 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 in the non-volatile storage media. The database stores data acquired by the business 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 business execution method.

[0063] This application discloses a service execution method, apparatus, computer device and its storage medium. The computer device acquires user voice input, extracts functional voice segments of target service functions from the user voice through an HMM model, and determines at least one service to be executed corresponding to the voice segments; and executes each service to be executed according to the target service link corresponding to each service to be executed.

[0064] In one embodiment, such as Figure 2 As shown, Figure 2 This application provides a flowchart of a business execution method according to an embodiment of the present application. Figure 1 The business execution method performed by the computer device in the process may include the following steps:

[0065] Step 201: Obtain user voice input containing the target business function, where the target business function is at least one optional business function in the business system.

[0066] It should be noted that there are many ways to obtain user voice input. For example, user voice can be received through a voice acquisition device set up on a computer device, or the user voice input can be determined from the digital information sent by the user. In other words, there are many ways to obtain user voice input, which will not be elaborated here. The following will provide a detailed explanation of the two methods mentioned above.

[0067] As one implementation method, if the computer device executing the business execution method is equipped with a voice acquisition device for collecting user voice, then the user can input user 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.

[0068] As another implementation method, when a user needs to input user voice into a computer device that performs a business execution method, but the computer device that performs the business execution method is not equipped with a voice acquisition device, the user can convert the audio signal of the user voice into a digital signal and input the digital signal corresponding to the user voice into the computer device.

[0069] The input of digital signals corresponding to user voice into computer signals can take many forms, such as email, SMS, or compressed file. No limitation is placed on the input form of digital signals corresponding to user voice into computer signals here.

[0070] To further explain, when it is necessary to obtain user voice, the text data containing the target business function input by the user can be obtained, and then the text data containing the target business function can be converted into audio data. The converted audio data is used as the user's voice input, thereby ensuring the clarity of the user's voice and preventing large noise and interference from the surrounding environment when the user inputs the voice, which would affect the subsequent execution of each business step.

[0071] In one embodiment of this application, an input field can be set in the interface of a computer device that performs a business execution method. When it is necessary to obtain user voice, text data containing the target business function can be entered in the input field. The computer device will then convert the received text data into audio data, which is the user voice input by the user.

[0072] Step 202: Extract functional speech segments of the target business function from the user's speech using a Hidden Markov Model (HMM); the HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0073] Since user voice is usually composed of multiple audio characters, in order to ensure that the business to be executed in the user voice can be accurately located in the subsequent process, it is necessary to perform word segmentation on the user voice containing multiple audio characters to ensure the accuracy of the subsequent determination of the business to be executed.

[0074] In one embodiment of this application, when it is necessary to determine the functional voice segment of the target business function, the user's voice can be pre-segmented to obtain at least two sub-voice segments after segmentation; then each sub-voice segment is input into the HMM model to extract the functional voice segment of the target business function from each sub-voice segment of the user's voice.

[0075] In this context, word segmentation refers to dividing the audio of each character in the user's speech into independent phrase audio; character audio refers to the audio corresponding to each character, while phrase audio refers to the audio corresponding to each word.

[0076] To further explain, when training the HMM model, sample speech can be input into the feature extraction network of the HMM model to extract the feature values ​​of the sample speech. Based on the feature values ​​of the sample speech and the feature value range of the corresponding optional business function, the parameters of the feature extraction network of the HMM model are adjusted to ensure that the feature values ​​of the sample speech are input into the feature value range of the corresponding optional business function.

[0077] Among them, sample speech refers to the speech data of each optional business function supported by the business system after noise reduction processing.

[0078] Step 203: Analyze at least one service to be executed corresponding to the functional voice segment.

[0079] It should be noted that when it is necessary to parse at least one service to be executed in a functional voice segment, the functional voice segment can be parsed and processed based on the SVM model to determine at least one service to be executed in the functional voice segment.

[0080] In the process of parsing and processing functional speech segments based on the SVM model, it is necessary to convert the functional speech segments into text to determine the corresponding text data.

[0081] In one embodiment of this application, parsing at least one service to be executed may include the following steps: converting a functional speech segment into text, determining the text data corresponding to the functional speech segment, determining at least one service to be executed corresponding to the text data corresponding to the functional speech segment based on an SVM model, and then determining at least one service to be executed corresponding to the functional speech segment.

[0082] Step 204: Obtain the target business link corresponding to each business to be executed, and execute each business to be executed based on the target business link.

[0083] It should be noted that when it is necessary to obtain the target business link corresponding to the business to be executed, a historical link database can be predetermined. Then, by filtering in the historical link database, the target business link corresponding to the business to be executed can be determined. The historical link database contains the sub-business links corresponding to each optional business function in the business system.

[0084] Furthermore, since the historical link database contains sub-business links corresponding to each optional business function in the business system, when the target business link corresponding to the business to be executed is determined based on the historical link database, the sub-business links corresponding to the business to be executed can be determined by filtering the historical link database based on the business to be executed.

[0085] In one embodiment of this application, if the target service function contained in the user's voice is not unique, it means that the sub-service links determined according to the historical link database are also not unique. Therefore, the execution order of each sub-service function can be determined based on the user's voice. Then, according to the execution order of each sub-service function, the sub-service links corresponding to each service to be executed are combined to obtain the target service link corresponding to the service to be executed. If the user's voice contains only one target service function, then the sub-service link corresponding to that service to be executed is taken as the target service link.

[0086] As one implementation method, when the target business function contained in the user's voice is not unique, the execution of each business to be executed can be carried out in the order of the sub-business links in the target business link, so as to ensure that the execution order of the business to be executed matches the user's input voice.

[0087] The above-described business execution method involves inputting the acquired user voice into an HMM model to obtain functional voice segments from the user's voice, parsing at least one business to be executed corresponding to each functional voice segment, and executing each business according to the target business link corresponding to each business. Since the user only needs to input voice during the above process, the user does not need to spend a lot of time learning the triggering methods of each function in the system before executing the business, thus reducing the user's time consumption. Furthermore, the execution of each business is based on the target business link corresponding to each business, thus ensuring that no incorrect functions are triggered when executing each business according to the user's voice, thereby improving the accuracy of the operation during the execution of each business.

[0088] To ensure that each pending service can be executed according to the target service link, it is necessary to parse at least one pending service corresponding to the functional voice segment. Since the service system contains a large number of optional service functions, parsing each pending service from the functional voice segment is quite difficult. Therefore, the computer device in this embodiment can, as shown in the example... Figure 3 The method shown parses at least one pending service corresponding to a voice segment, specifically including the following steps:

[0089] Step 301: Parse the text data corresponding to the function speech segment.

[0090] It should be noted that since the input to the SVM model needs to be in the vector format corresponding to the text data, in order to ensure that at least one business to be executed corresponding to the target business function can be determined according to the SVM model, it is necessary to first parse the function speech segment into text, and then determine the text data corresponding to the function speech segment.

[0091] One approach is to use a trained text parsing model that can parse the input speech into corresponding text data, and then use this model to parse functional speech segments into corresponding text data.

[0092] The text parsing model can be trained based on training speech and the corresponding text data. Specifically, the training speech is input into the text parsing model to obtain the parsing result output by the text parsing model. The parameters of the text parsing model are adjusted based on the difference between the parsing result and the text data corresponding to the training speech until the similarity between the parsing result output by the text parsing model and the text data corresponding to the training speech is greater than a preset similarity threshold, thereby obtaining the text parsing model that has been trained.

[0093] In one embodiment of this application, when it is necessary to determine the text data corresponding to a functional speech segment, the functional speech segment can be input into a trained text parsing model to obtain the parsing result output by the text parsing model, which is the text data corresponding to the functional speech segment.

[0094] Step 302: Quantize the text data corresponding to the functional speech segment to obtain the text word vector of the functional speech segment.

[0095] It should be noted that, in order to ensure the efficiency of determining at least one pending business corresponding to the target business function and to prevent the SVM model from being unable to parse and process the text data corresponding to the functional speech segment, it is necessary to quantize the functional speech segment. This ensures that by inputting the quantized text word vectors into the SVM model, at least one pending business corresponding to the target business function can be obtained.

[0096] The methods for quantizing functional speech segments include vector quantization, scalar quantization, etc. Furthermore, by quantizing functional speech segments, the encoded result of the functional speech segments after data compression is obtained, which is the text word vector of the functional speech segments.

[0097] Step 303: Use the Support Vector Machine (SVM) model to perform feature parsing on the text word vectors to obtain at least one business function to be executed corresponding to the target business function.

[0098] In one embodiment of this application, text word vectors are input into an SVM model to obtain the business to be executed corresponding to the text word vectors output by the SVM model. When there are multiple text word vectors input into the SVM model, the business to be executed corresponding to the multiple text word vectors output by the SVM model is at least one business to be executed corresponding to the target business function.

[0099] The SVM model is trained based on text data of each optional business function supported by the business system.

[0100] Furthermore, when training the SVM model, the text data corresponding to each optional business function can be determined, and the text data corresponding to each optional business function can be quantized to obtain the sample vector of the text data corresponding to each optional business function. The sample vector is input into the SVM model to obtain the prediction result output by the SVM model based on the sample vector. Based on the relationship between the prediction result and the optional business function corresponding to the sample vector, the parameters of the SVM model are adjusted to complete the training of the SVM model.

[0101] The above-mentioned business execution method, by parsing the text data corresponding to the functional voice segments and determining the text word vectors corresponding to the text data, ensures that at least one business to be executed corresponding to the target business function can be determined through the SVM model. This ensures that users do not need to spend a lot of time learning the triggering methods of each function in the system before executing the business to be executed, thus reducing the user's time consumption; and it also reduces the occurrence of system triggering incorrect functions, improving the accuracy of operation during the execution of each business to be executed.

[0102] Because user-inputted voice contains a lot of information, it is difficult to identify the functional voice segments containing the target business function from the user's voice. Therefore, the computer device in this embodiment can, as shown in the example... Figure 4 The method shown extracts functional voice segments of the target business function from the user's voice, specifically including the following steps:

[0103] Step 401: Perform speech segmentation on the user's speech to obtain at least two sub-speech segments.

[0104] It should be noted that there are many methods for segmenting user speech, such as jieba segmentation, ltp segmentation, and ir segmentation. Specifically, user speech can be segmented using jieba (a segmentation tool) to obtain at least two sub-speech segments after segmentation; or, user speech can be segmented using ltp (a Chinese language processing system) to obtain at least two sub-speech segments after segmentation; or, user speech can be segmented using ir (a segmentation tool) to obtain at least two sub-speech segments after segmentation.

[0105] Furthermore, to ensure that a complete phrase is not split into two sub-language segments when segmenting user speech, for example, splitting the user speech "This morning is sunny" into "Today morning" and "On sunny day", where "morning" is split into two sub-language segments, a segmentation model for segmenting user speech can be pre-trained, and the user speech can be segmented according to the segmentation model.

[0106] In one embodiment of this application, the training process of the segmentation model may include the following: pre-setting multiple sample speech as training samples, and manually segmenting each sample speech to obtain the manual segmentation result corresponding to each sample speech; inputting each training sample into the segmentation model to obtain the model segmentation result corresponding to each segmentation model output by the segmentation model; adjusting the parameters of the segmentation model based on the difference between the model segmentation result and the manual segmentation result until the model segmentation result output by the segmentation model meets the preset conditions, thereby completing the training of the segmentation model.

[0107] Step 402: Extract feature values ​​of each sub-speech segment through the feature extraction network of the HMM model, and determine the probability that each sub-speech segment belongs to an optional service function through the function prediction network of the HMM model based on the feature values ​​of each sub-speech segment and the range of function feature values ​​corresponding to the optional service functions.

[0108] It should be noted that the training process of the HMM model may include the following steps: denoising the speech data of each optional business function supported by the business system to obtain denoised sample speech; extracting feature values ​​of the sample speech through the feature extraction network of the HMM model; and adjusting the model parameters of the feature extraction network of the HMM model according to the relationship between the feature values ​​of the sample speech and the value range of the functional features.

[0109] Step 403: Based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional service function, determine the functional voice segment of the target service function.

[0110] It should be noted that when it is necessary to determine the functional voice segment of the target business function, the maximum probability of each sub-voice segment belonging to the optional business function can be determined based on the probability of each sub-voice segment belonging to the optional business function. It is then determined whether the maximum probability of each sub-voice segment belonging to the optional business function is greater than a preset probability threshold. If so, the optional business function corresponding to the maximum probability of the sub-voice segment belonging to the optional business function is the optional business function to which the sub-voice segment belongs, and the sub-voice segment is the functional voice segment of the target business function.

[0111] For example, if a probability threshold of 50% is preset, and the speech contains two sub-speech segments and two optional service functions; the two sub-speech segments are sub-speech segment A and sub-speech segment B, and the two optional service functions are optional service function M and optional service function N, the probability that sub-speech segment A belongs to optional service function M is determined to be 70%, and the probability that sub-speech segment A belongs to optional service function N is determined to be 30%. Furthermore, the probability that sub-speech segment B belongs to optional service function M is determined to be 20%, and the probability that sub-speech segment A belongs to optional service function N is determined to be 10%. The 70% probability of sub-speech segment A belonging to optional service function M is the maximum probability value, and 70% is greater than the probability threshold of 50%. Therefore, sub-speech segment A is determined to be a functional speech segment, and the target service function of sub-speech segment A is optional service function M. Since the 20% probability of sub-speech segment B belonging to optional service function M is the maximum probability value, and 20% is less than the probability threshold of 50%, sub-speech segment B is not a functional speech segment.

[0112] It should be noted that if a sub-language segment is not a functional speech segment, it means that the sub-speech segment does not contain optional business functions.

[0113] The above-mentioned business execution method provides a foundation for successfully determining the functional voice segments of the target business function by processing the user's voice through voice segmentation. By using the HMM model, the probability of each sub-voice segment belonging to the optional business function is determined, thereby determining the functional voice segment of the target business function. This ensures that no system erroneous functions are triggered when executing each business to be executed based on the user's voice, and improves the accuracy of the operation during the execution of each business to be executed.

[0114] When the number of target service functions contained in the user's voice is at least two, it is necessary to consider the execution order of each target service function when determining the target service link corresponding to the service to be executed. Therefore, the computer device in this embodiment can, as shown in the example... Figure 5 The method shown above obtains the target business link corresponding to each business to be executed, specifically including the following steps:

[0115] Step 501: Based on each business to be executed, filter the sub-business links corresponding to each business to be executed from the historical link database.

[0116] It should be noted that the specific steps of filtering the sub-business links corresponding to each business to be executed from the historical link database may include the following: filtering the candidate business links associated with each business to be executed from the historical link database, and determining the maximum matching degree between the business to be executed and the candidate business links; determining whether the maximum matching degree is less than the matching degree threshold; if so, displaying the candidate business links to the user; and obtaining the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

[0117] Furthermore, if not, the candidate business link with the highest matching degree among the candidate business links associated with each business to be executed will be taken as the sub-business link corresponding to each business to be executed.

[0118] For example, if a service to be executed is associated with three candidate service links, namely candidate service link A, candidate service link B, and candidate service link C, and the matching degree threshold is 0.6, one implementation method is as follows: determine the matching degree between the service to be executed and the three candidate service links as follows: the matching degree of candidate service link A is 0.2, the matching degree of candidate service link B is 0.1, and the matching degree of candidate service link C is 0.7. Among them, candidate service link C has the highest matching degree with the service to be executed, and the matching degree of candidate service link C is greater than the matching degree threshold. Therefore, the sub-service link corresponding to the service to be executed is candidate service link C. As another implementation method: the matching degree between the business to be executed and the three candidate business links is determined as follows: the matching degree of candidate business link A is 0.5, the matching degree of candidate business link B is 0.3, and the matching degree of candidate business link C is 0.2. Among them, the matching degree of candidate business link A with the business to be executed is the largest. However, the matching degree of candidate business link A with the business to be executed is less than the matching degree threshold. Therefore, candidate business link A, candidate business link B, and candidate business link C are displayed to the user to obtain the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

[0119] Step 502: Determine the execution order of each service to be executed based on the user's voice.

[0120] As one approach, when it is necessary to determine the execution order of tasks to be performed, the user's speech can be converted into text to determine the text data corresponding to the user's speech, and the order of the tasks to be performed in the text data can be used as the execution order of the tasks to be performed.

[0121] As another implementation method, when it is necessary to determine the execution order of the services to be executed, at least one service to be executed corresponding to the functional voice segments can be parsed based on the SVM model. Based on the position of each functional voice segment in the user's voice, the order of each functional voice segment can be determined. The order of each functional voice segment is the execution order of the services to be executed.

[0122] Step 503: Based on the execution order of each pending business, combine the sub-business links corresponding to each pending business to obtain the target business link corresponding to each pending business.

[0123] It should be noted that when combining the sub-business links corresponding to each business to be executed, the order of each sub-business link in the combined target business link must be the same as the execution order of the business to be executed.

[0124] In one embodiment of this application, if there are three services to be executed, and the sub-service links corresponding to the three services are sub-service link A, sub-service link B, and sub-service link C, the execution order of the three sub-service links is: first execute sub-service link A, then execute sub-service link B, and finally execute sub-service link C. Therefore, when combining the sub-service links corresponding to the three services to be executed, sub-service link B should be combined after sub-service link A, and then sub-service link C should be combined after sub-service link B to finally obtain the target service links corresponding to the three services to be executed.

[0125] The above-mentioned business execution method determines the execution order of each pending business, thereby identifying the target business link based on the execution order of each pending business. This ensures that subsequent pending businesses can be executed in the order described by the user's voice, further improving user satisfaction and reducing the user's time consumption by eliminating the need for the user to input their voice multiple times.

[0126] When executing pending tasks, there may be situations where the execution instruction triggered by the user for the pending task does not match the operational flow of the target task chain. In this case, the computer device in this embodiment can, as follows: Figure 6 The method shown is used to execute each pending business, specifically including the following steps:

[0127] Step 601: Based on the target business link, output the operation instructions corresponding to the business to be executed.

[0128] It should be noted that the operation guide can prompt users according to the operation methods of each business to be executed in the target business chain, so as to ensure that users can execute the business to be executed according to the operation methods recorded in the target business chain.

[0129] In one embodiment of this application, a continuous and uninterrupted operation flow can be generated based on the operation methods for executing each service to be executed and the operation sequence of each service to be executed recorded in the target service link. This operation flow is the operation guide corresponding to the service to be executed.

[0130] Step 602: Receive the execution instructions for the pending business triggered by the user based on the operation guide, and respond to the execution instructions to execute each pending task.

[0131] In one embodiment of this application, if the user is operating based on a computer device, the operation instructions can be displayed on the display interface of the computer device to prompt the user to perform the operation method and operation sequence of the business to be performed.

[0132] Furthermore, if it is detected that the execution instruction of the pending business triggered by the user is inconsistent with the operation guide, the actual business link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual business link and the pending business.

[0133] In one embodiment of this application, if it is detected that the execution instruction and operation guide of the user-triggered pending service are inconsistent, it is determined whether the actual service link corresponding to the execution instruction can realize the pending service. If not, the historical link database is not updated; if so, the actual service link is compared with the target service link to determine whether the actual service link is more convenient to operate than the target service link. If not, the historical link database is not updated; if so, the target service link in the historical link database is replaced with the actual service link to update the historical link database.

[0134] The above-mentioned business execution method provides operation prompts to users by defining operation guidelines, thereby reducing the difficulty of operation for users and ensuring that no system error functions are triggered when executing each pending business according to the user's voice. This improves the accuracy of operation during the execution of each pending business. Furthermore, by detecting whether the execution instructions of the pending business are consistent with the operation guidelines, the historical link database is updated to improve the effectiveness of the candidate business links contained in the historical link database.

[0135] In this embodiment, as Figure 7 As shown, Figure 7 A flowchart of another business execution method provided in this application embodiment, which may include the following when it is necessary to execute the business to be executed:

[0136] Step 701: Obtain user voice input containing the target business function, wherein the target business function is at least one optional business function in the business system.

[0137] Step 702: Perform speech segmentation on the user's speech to obtain at least two sub-speech segments.

[0138] Step 703: Extract feature values ​​of each sub-speech segment through the feature extraction network of the HMM model, and determine the probability that each sub-speech segment belongs to an optional service function through the function prediction network of the HMM model based on the feature values ​​of each sub-speech segment and the range of function feature values ​​corresponding to the optional service functions.

[0139] Step 704: Determine the functional voice segment of the target business function based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional business function.

[0140] Step 705: Parse the text data corresponding to the function speech segment.

[0141] Step 706: Quantize the text data corresponding to the functional speech segment to obtain the text word vector of the functional speech segment.

[0142] Step 707: Use the Support Vector Machine (SVM) model to perform feature parsing on the text word vectors to obtain at least one business function to be executed corresponding to the target business function.

[0143] Step 708: Based on each pending business, filter the sub-business links corresponding to each pending business from the historical link database.

[0144] Step 709: Determine the execution order of each service to be executed based on the user's voice.

[0145] Step 710: Based on the execution order of each pending business, combine the sub-business links corresponding to each pending business to obtain the target business link corresponding to each pending business.

[0146] Step 711: Based on the target business link, output the operation instructions corresponding to the business to be executed.

[0147] Step 712: Receive the execution instructions for the pending business triggered by the user based on the operation guide, and respond to the execution instructions to execute each pending task.

[0148] Step 713: If it is detected that the execution instruction of the pending business triggered by the user is inconsistent with the operation guide, then the actual business link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual business link and the pending business.

[0149] The above-described business execution method involves inputting the acquired user voice into an HMM model to obtain functional voice segments from the user's voice, parsing at least one business to be executed corresponding to each functional voice segment, and executing each business according to the target business link corresponding to each business. Since the user only needs to input voice during the above process, the user does not need to spend a lot of time learning the triggering methods of each function in the system before executing the business, thus reducing the user's time consumption. Furthermore, the execution of each business is based on the target business link corresponding to each business, thus ensuring that no incorrect functions are triggered when executing each business according to the user's voice, thereby improving the accuracy of the operation during the execution of each business.

[0150] 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.

[0151] Based on the same inventive concept, this application also provides a business execution apparatus for implementing the business 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 business execution apparatus embodiments provided below can be found in the limitations of the business execution method described above, and will not be repeated here.

[0152] In one embodiment, such as Figure 8 As shown, a business execution device is provided, including: a first acquisition module 10, an extraction module 20, a parsing module 30, and a second acquisition module 40, wherein:

[0153] The first acquisition module 10 is used to acquire user voice input containing a target business function, wherein the target business function is at least one optional business function in the business system.

[0154] Extraction module 20 is used to extract functional speech segments of target business functions from user speech using a Hidden Markov Model (HMM); the HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0155] The parsing module 30 is used to parse at least one service to be executed corresponding to a functional voice segment.

[0156] The second acquisition module 40 is used to acquire the target business link corresponding to each business to be executed, and to execute each business to be executed based on the target business link.

[0157] The aforementioned service execution device obtains functional speech segments from the user's voice input into the HMM model, parses at least one service to be executed corresponding to each functional speech segment, and executes each service according to the target service link corresponding to each service. Since the user only needs to input voice during the above process, the user does not need to spend a lot of time learning the triggering methods of each function in the system before executing the service, thus reducing the user's time consumption. Furthermore, the execution of each service is based on the target service link corresponding to each service, thus ensuring that no incorrect functions are triggered when executing services according to the user's voice, improving the accuracy of the operation during the execution of services.

[0158] In one embodiment, such as Figure 9 As shown, a business execution device is provided. The parsing module 30 in this device includes: a first parsing unit 31, a quantization unit 32, and a second parsing unit 33, wherein:

[0159] The first parsing unit 31 is used to parse the text data corresponding to the functional speech segments.

[0160] The quantization unit 32 is used to quantize the text data corresponding to the functional speech segment to obtain the text word vector of the functional speech segment.

[0161] The second parsing unit 33 is used to perform feature parsing on text word vectors through a support vector machine (SVM) model to obtain at least one business function to be executed corresponding to the target business function; wherein, the SVM model is trained based on the text data of each optional business function supported by the business system.

[0162] In one embodiment, such as Figure 10 As shown, a business execution apparatus is provided. The extraction module 20 of the business execution apparatus includes: a segmentation unit 21, a first lifting unit 22, and a first determining unit 23, wherein:

[0163] Segmentation unit 21 is used to perform speech segmentation processing on user speech to obtain at least two sub-speech segments.

[0164] The first lifting unit 22 is used to extract feature values ​​of each sub-speech segment through the feature extraction network of the HMM model, and to determine the probability that each sub-speech segment belongs to the optional service function through the function prediction network of the HMM model based on the feature values ​​of each sub-speech segment and the range of function feature values ​​corresponding to the optional service function.

[0165] The first determining unit 23 is used to determine the functional voice segment of the target service function based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional service function.

[0166] In one embodiment, such as Figure 11 As shown, a business execution device is provided. The extraction module 20 of the business execution device further includes: a noise reduction unit 24, a second extraction unit 25, and an adjustment unit 26, wherein:

[0167] The noise reduction unit 24 is used to perform noise reduction processing on the voice data of each optional business function supported by the business system to obtain the noise-reduced sample voice.

[0168] The second extraction unit 25 is used to extract feature values ​​of sample speech through the feature extraction network of the HMM model;

[0169] The adjustment unit 26 is used to adjust the model parameters of the feature extraction network of the HMM model according to the relationship between the feature values ​​of the sample speech and the range of functional feature values.

[0170] In one embodiment, such as Figure 12 As shown, a business execution apparatus is provided. The second acquisition module 40 in this apparatus includes: a filtering unit 41, a second determining unit 42, and a combination unit 43, wherein:

[0171] The filtering unit 41 is used to filter the sub-business links corresponding to each business to be executed from the historical link database based on each business to be executed.

[0172] The second determining unit 42 is used to determine the execution order of each service to be executed based on the user's voice.

[0173] Combination unit 43 is used to combine the sub-business links corresponding to each business to be executed according to the execution order of each business to be executed, so as to obtain the target business link corresponding to each business to be executed.

[0174] Specifically, the filtering unit is used to filter candidate business links associated with each business to be executed from the historical link database, and determine the maximum matching degree between the business to be executed and the candidate business links; determine whether the maximum matching degree is less than the matching degree threshold; if so, display the candidate business links to the user; and obtain the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

[0175] In one embodiment, such as Figure 13 As shown, a business execution device is provided. The second acquisition module 40 in the business execution device further includes: an output unit 44 and a triggering unit 45, wherein:

[0176] Output unit 44 is used to output operation instructions corresponding to the business to be executed based on the target business link.

[0177] Triggering unit 45 is used to receive the execution instruction of the pending business triggered by the user based on the operation guide, and to execute each pending task in response to the execution instruction.

[0178] In one embodiment, such as Figure 14 As shown, a business execution apparatus is provided. The second acquisition module 40 of this business execution apparatus further includes an update unit 46, wherein:

[0179] Update unit 46: If it detects that the execution instruction and operation guide of the pending business triggered by the user are inconsistent, it obtains the actual business link corresponding to the execution instruction and updates the historical link database based on the actual business link and the pending business.

[0180] Each module in the aforementioned business execution device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, 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.

[0181] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 15 As 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 business 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.

[0182] Those skilled in the art will understand that Figure 15 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.

[0183] 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:

[0184] Acquire user voice input containing the target business function, where the target business function is at least one optional business function in the business system;

[0185] The Hidden Markov Model (HMM) is used to extract functional speech segments of the target business function from the user's speech; the HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0186] At least one pending service corresponding to the parsed voice segment;

[0187] Obtain the target business link corresponding to each business to be executed, and execute each business to be executed based on the target business link.

[0188] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0189] The parsing function retrieves the text data corresponding to the speech segments.

[0190] The text data corresponding to the functional speech segments are quantized to obtain the text word vectors of the functional speech segments;

[0191] By using a Support Vector Machine (SVM) model, feature parsing is performed on text word vectors to obtain at least one business function to be executed corresponding to the target business function; wherein, the SVM model is trained based on the text data of each optional business function supported by the business system.

[0192] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0193] The user's voice is segmented to obtain at least two sub-voice segments;

[0194] The feature extraction network of the HMM model is used to extract the feature values ​​of each sub-speech segment. Then, the function prediction network of the HMM model is used to determine the probability that each sub-speech segment belongs to the optional service function based on the feature values ​​of each sub-speech segment and the value range of the function features corresponding to the optional service functions.

[0195] Based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional business function, the functional voice segment of the target business function is determined.

[0196] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0197] The voice data of each optional business function supported by the business system is denoised to obtain the denoised sample voice.

[0198] The feature extraction network of the HMM model is used to extract feature values ​​of the sample speech.

[0199] Based on the relationship between the feature values ​​and functional feature value ranges of the sample speech, adjust the model parameters of the feature extraction network of the HMM model.

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

[0201] Based on each pending business, the sub-business links corresponding to each pending business are selected from the historical link database;

[0202] Based on the user's voice, determine the execution order of each pending business function;

[0203] Based on the execution order of each pending business, the sub-business links corresponding to each pending business are combined to obtain the target business link corresponding to each pending business.

[0204] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0205] From the historical link database, filter the candidate business links associated with each business to be executed, and determine the maximum matching degree between the business to be executed and the candidate business links;

[0206] Determine if the maximum matching degree is less than the matching degree threshold;

[0207] If so, then display the candidate business link to the user;

[0208] Obtain the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

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

[0210] Based on the target business chain, output the operation instructions corresponding to the business to be executed;

[0211] Receive execution instructions from users based on operation guidelines, triggering pending business operations, and respond to the execution instructions to execute each pending task.

[0212] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0213] If the execution instruction for a user-triggered pending service is found to be inconsistent with the operation guide, the actual service link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual service link and the pending service.

[0214] 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:

[0215] Acquire user voice input containing the target business function, where the target business function is at least one optional business function in the business system;

[0216] The Hidden Markov Model (HMM) is used to extract functional speech segments of the target business function from the user's speech; the HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0217] At least one pending service corresponding to the parsed voice segment;

[0218] Obtain the target business link corresponding to each business to be executed, and execute each business to be executed based on the target business link.

[0219] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0220] The parsing function retrieves the text data corresponding to the speech segments.

[0221] The text data corresponding to the functional speech segments are quantized to obtain the text word vectors of the functional speech segments;

[0222] By using a Support Vector Machine (SVM) model, feature parsing is performed on text word vectors to obtain at least one business function to be executed corresponding to the target business function; wherein, the SVM model is trained based on the text data of each optional business function supported by the business system.

[0223] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0224] The user's voice is segmented to obtain at least two sub-voice segments;

[0225] The feature extraction network of the HMM model is used to extract the feature values ​​of each sub-speech segment. Then, the function prediction network of the HMM model is used to determine the probability that each sub-speech segment belongs to the optional service function based on the feature values ​​of each sub-speech segment and the value range of the function features corresponding to the optional service functions.

[0226] Based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional business function, the functional voice segment of the target business function is determined.

[0227] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0228] The voice data of each optional business function supported by the business system is denoised to obtain the denoised sample voice.

[0229] The feature extraction network of the HMM model is used to extract feature values ​​of the sample speech.

[0230] Based on the relationship between the feature values ​​and functional feature value ranges of the sample speech, adjust the model parameters of the feature extraction network of the HMM model.

[0231] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0232] Based on each pending business, the sub-business links corresponding to each pending business are selected from the historical link database;

[0233] Based on the user's voice, determine the execution order of each pending business function;

[0234] Based on the execution order of each pending business, the sub-business links corresponding to each pending business are combined to obtain the target business link corresponding to each pending business.

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

[0236] From the historical link database, filter the candidate business links associated with each business to be executed, and determine the maximum matching degree between the business to be executed and the candidate business links;

[0237] Determine if the maximum matching degree is less than the matching degree threshold;

[0238] If so, then display the candidate business link to the user;

[0239] Obtain the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

[0240] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0241] Based on the target business chain, output the operation instructions corresponding to the business to be executed;

[0242] Receive execution instructions from users based on operation guidelines, triggering pending business operations, and respond to the execution instructions to execute each pending task.

[0243] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0244] If the execution instruction for a user-triggered pending service is found to be inconsistent with the operation guide, the actual service link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual service link and the pending service.

[0245] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0246] Acquire user voice input containing the target business function, where the target business function is at least one optional business function in the business system;

[0247] The Hidden Markov Model (HMM) is used to extract functional speech segments of the target business function from the user's speech; the HMM model is trained based on the functional speech data of each optional business function supported by the business system.

[0248] At least one pending service corresponding to the parsed voice segment;

[0249] Obtain the target business link corresponding to each business to be executed, and execute each business to be executed based on the target business link.

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

[0251] The parsing function retrieves the text data corresponding to the speech segments.

[0252] The text data corresponding to the functional speech segments are quantized to obtain the text word vectors of the functional speech segments;

[0253] By using a Support Vector Machine (SVM) model, feature parsing is performed on text word vectors to obtain at least one business function to be executed corresponding to the target business function; wherein, the SVM model is trained based on the text data of each optional business function supported by the business system.

[0254] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0255] The user's voice is segmented to obtain at least two sub-voice segments;

[0256] The feature extraction network of the HMM model is used to extract the feature values ​​of each sub-speech segment. Then, the function prediction network of the HMM model is used to determine the probability that each sub-speech segment belongs to the optional service function based on the feature values ​​of each sub-speech segment and the value range of the function features corresponding to the optional service functions.

[0257] Based on each sub-voice segment and the probability that each sub-voice segment belongs to an optional business function, the functional voice segment of the target business function is determined.

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

[0259] The voice data of each optional business function supported by the business system is denoised to obtain the denoised sample voice.

[0260] The feature extraction network of the HMM model is used to extract feature values ​​of the sample speech.

[0261] Based on the relationship between the feature values ​​and functional feature value ranges of the sample speech, adjust the model parameters of the feature extraction network of the HMM model.

[0262] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0263] Based on each pending business, the sub-business links corresponding to each pending business are selected from the historical link database;

[0264] Based on the user's voice, determine the execution order of each pending business function;

[0265] Based on the execution order of each pending business, the sub-business links corresponding to each pending business are combined to obtain the target business link corresponding to each pending business.

[0266] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0267] From the historical link database, filter the candidate business links associated with each business to be executed, and determine the maximum matching degree between the business to be executed and the candidate business links;

[0268] Determine if the maximum matching degree is less than the matching degree threshold;

[0269] If so, then display the candidate business link to the user;

[0270] Obtain the sub-business links corresponding to each business to be executed selected by the user from the candidate business links.

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

[0272] Based on the target business chain, output the operation instructions corresponding to the business to be executed;

[0273] Receive execution instructions from users based on operation guidelines, triggering pending business operations, and respond to the execution instructions to execute each pending task.

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

[0275] If the execution instruction for a user-triggered pending service is found to be inconsistent with the operation guide, the actual service link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual service link and the pending service.

[0276] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0277] 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.

[0278] 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.

[0279] 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 business execution method, characterized in that, The method includes: Acquire user voice input containing a target business function, wherein the target business function is at least one optional business function in the business system; The functional speech segments of the target business function are extracted from the user's speech using a Hidden Markov Model (HMM); the HMM is trained based on the functional speech data of each optional business function supported by the business system. Analyze at least one service to be executed corresponding to the aforementioned functional voice segment; Obtain the target service link corresponding to each of the services to be executed, and execute each of the services to be executed based on the target service link; The step of extracting the functional speech segments of the target business function from the user's speech using the HMM model includes: The user's voice is segmented to obtain at least two sub-voice segments; The feature extraction network of the HMM model is used to extract the feature values ​​of each sub-speech segment. Then, the function prediction network of the HMM model is used to determine the probability that each sub-speech segment belongs to the optional service function based on the feature values ​​of each sub-speech segment and the value range of the function features corresponding to the optional service function. Based on each sub-voice segment and the probability that each sub-voice segment belongs to the optional service function, the functional voice segment of the target service function is determined.

2. The method according to claim 1, characterized in that, The parsing of at least one pending service corresponding to the functional voice segment includes: Analyze the text data corresponding to the aforementioned functional speech segments; The text data corresponding to the functional speech segment is quantized to obtain the text word vector of the functional speech segment; The text word vectors are analyzed using a Support Vector Machine (SVM) model to obtain at least one business function to be executed corresponding to the target business function; wherein the SVM model is trained based on the text data of each optional business function supported by the business system.

3. The method according to claim 1, characterized in that, The training process of the HMM model includes: The voice data of each optional business function supported by the business system is subjected to noise reduction processing to obtain noise-reduced sample voice. The feature extraction network of the HMM model is used to extract feature values ​​of the sample speech. The model parameters of the feature extraction network of the HMM model are adjusted based on the relationship between the feature values ​​of the sample speech and the range of the functional features.

4. The method according to claim 1, characterized in that, If the number of target service functions contained in the user's voice is at least two, then obtaining the target service link corresponding to each of the services to be executed includes: Based on each of the pending services, the sub-service links corresponding to each of the pending services are selected from the historical link database; Based on the user's voice, determine the execution order of each of the pending services; Based on the execution order of each of the pending services, the sub-service links corresponding to each pending service are combined to obtain the target service link corresponding to each pending service.

5. The method according to claim 4, characterized in that, The step of filtering sub-service links corresponding to each of the pending services from the historical link database includes: From the historical link database, candidate service links associated with each of the services to be executed are filtered, and the maximum matching degree between the service to be executed and the candidate service links is determined; Determine whether the maximum matching degree is less than the matching degree threshold; If so, then the candidate service link will be displayed to the user; Obtain the sub-service links corresponding to each service to be executed selected by the user from the candidate service links.

6. The method according to claim 1, characterized in that, The execution of each of the pending services based on the target service link includes: Based on the target business link, output the operation instructions corresponding to the business to be executed; Receive the execution instruction of the pending business triggered by the user based on the operation guide, and execute each pending task in response to the execution instruction.

7. The method according to claim 5, characterized in that, Also includes: If it is detected that the execution instruction and operation guide of the pending business triggered by the user are inconsistent, the actual business link corresponding to the execution instruction is obtained, and the historical link database is updated based on the actual business link and the pending business.

8. A business execution device, characterized in that, The device includes: The first acquisition module is used to acquire user voice input containing a target business function, wherein the target business function is at least one optional business function in the business system; The extraction module is used to extract functional speech segments of the target business function from the user's speech using a Hidden Markov Model (HMM); the HMM is trained based on the functional speech data of each optional business function supported by the business system. The parsing module is used to parse at least one service to be executed corresponding to the functional voice segment; The second acquisition module is used to acquire the target service link corresponding to each of the services to be executed, and to execute each of the services to be executed based on the target service link; The extraction module is specifically used for: The user's voice is segmented to obtain at least two sub-voice segments; The feature extraction network of the HMM model is used to extract the feature values ​​of each sub-speech segment. Then, the function prediction network of the HMM model is used to determine the probability that each sub-speech segment belongs to the optional service function based on the feature values ​​of each sub-speech segment and the value range of the function features corresponding to the optional service function. Based on each sub-voice segment and the probability that each sub-voice segment belongs to the optional service function, the functional voice segment of the target service function is determined.

9. 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 7.

10. 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 7.

11. 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 7.

Images