Text correction server, terminal device and text correction method

Abstract

The embodiment of the application provides a text error correction server, a terminal device and a text error correction method, and relates to the technical field of speech recognition. The text error correction server comprises: a communicator configured to receive a to-be-corrected text sent by a first terminal device; a controller configured to judge whether a trigger condition for labeling the to-be-corrected text is met, in the case that a first error correction data for correcting the to-be-corrected text is not included in an error correction database of the text error correction server; the communicator is further configured to send a labeling request for requesting to label the to-be-corrected text to at least one terminal device, and receive labeling data sent by the at least one terminal device, in the case that the trigger condition is met; and the controller is further configured to generate error correction data corresponding to the to-be-corrected text according to the labeling data sent by the at least one terminal device, and save the error correction data corresponding to the to-be-corrected text to the error correction database. The embodiment of the application is used for text error correction.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of speech recognition. More particularly, it relates to a text correction server, a terminal device and a text correction method. BACKGROUND

[0002] Text correction refers to correcting errors existing in text. Due to dialects, accents, noise and other reasons, errors often occur in the process of converting speech into text, and the errors in the text will greatly affect the effect of intent recognition, text retrieval and other tasks. Therefore, text correction is one of the research hotspots in the field of speech recognition.

[0003] At present, the text correction scheme commonly used in the industry is a text correction scheme based on natural language processing (NLP). For example: a text correction model is obtained by training sample data, then the text to be corrected is input into the text correction model, and the output of the text correction model is obtained as the correction result of the text to be corrected. However, the development cycle of the text correction scheme based on NLP is generally very long, and often cannot meet the time limit requirements of related products. SUMMARY

[0004] Exemplary embodiments of the present application provide a text correction server, a terminal device and a text correction method for solving the problem of long development cycle of the text correction scheme.

[0005] The technical scheme provided by the embodiments of the present application is as follows:

[0006] In a first aspect, the embodiments of the present application provide a text correction server, comprising:

[0007] A communicator configured to receive text to be corrected sent by a first terminal device;

[0008] A controller configured to determine whether a trigger condition for labeling the text to be corrected is met, in a case that a first correction data for correcting the text to be corrected is not included in a correction database of the text correction server;

[0009] The communicator is further configured to send a labeling request for requesting to label the text to be corrected to at least one terminal device, and receive labeling data sent by the at least one terminal device, in a case that the trigger condition is met.

[0010] The controller is further configured to generate correction data corresponding to the text to be corrected according to the labeling data sent by the at least one terminal device, and save the correction data corresponding to the text to be corrected to the correction database.

[0011] In a second aspect, an embodiment of the present application provides a terminal device, comprising:

[0012] a communicator configured to receive a labeling request sent by a text correction server, the labeling request being used to request labeling of a text to be corrected;

[0013] a display configured to display a labeling interface according to the labeling request;

[0014] a user interface configured to receive a labeling operation input by a user in the labeling interface, and generate labeling data according to the labeling operation;

[0015] the communicator is further configured to send the labeling data to the text correction server.

[0016] In a third aspect, an embodiment of the present application provides a text correction method applied to a text correction server, the method comprising:

[0017] receiving a text to be corrected sent by a first terminal device;

[0018] in a case where a first correction data used to correct the text to be corrected is not included in a correction database of the text correction server, judging whether a trigger condition of labeling the text to be corrected is met;

[0019] in a case where the trigger condition is met, sending a labeling request used to request labeling of the text to be corrected to at least one terminal device, and receiving labeling data sent by the at least one terminal device;

[0020] generating correction data corresponding to the text to be corrected according to the labeling data sent by the at least one terminal device, and saving the correction data corresponding to the text to be corrected to the correction database.

[0021] In a fourth aspect, an embodiment of the present application provides a text correction method applied to a terminal device, the method comprising:

[0022] receiving a labeling request sent by a text correction server, the labeling request being used to request labeling of a text to be corrected;

[0023] displaying a labeling interface according to the labeling request;

[0024] receiving a labeling operation input by a user in the labeling interface, and generating labeling data according to the labeling operation;

[0025] sending the labeling data to the text correction server.

[0026] In a fifth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program. When the computer program is executed by a computing device, the computing device implements the text correction method in the third aspect or the fourth aspect.

[0027] In a sixth aspect, an embodiment of the present application provides a computer program product, which, when running on a computer, causes the computer to implement the text correction method in the third aspect or the fourth aspect.

[0028] According to the above technical solutions, the text correction server and the text correction method provided in the embodiments of the present application, when receiving the to-be-corrected text sent by the first terminal device, determine whether the first correction data for correcting the to-be-corrected text is included in the correction database, and in the case that the first correction data for correcting the to-be-corrected text is not included in the correction database, determine whether a trigger condition for labeling the to-be-corrected text is met; in the case that the trigger condition is met, send a labeling request for requesting to label the to-be-corrected text to at least one terminal device, and receive the labeling data sent by the at least one terminal device; finally, generate the correction data corresponding to the to-be-corrected text according to the labeling data sent by the at least one terminal device, and save the correction data corresponding to the to-be-corrected text to the correction database. Since the correction data is used to correct the to-be-corrected text in the embodiments of the present application, and the correction data is generated according to the labeling data sent by the terminal device, the embodiments of the present application can quickly obtain the correction data of the to-be-corrected text that meets the trigger condition through user labeling, and then correct the to-be-corrected text, so that the embodiments of the present application can solve the problem of long development cycle of the text correction scheme. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present application or the implementation manners in the related art, the drawings needed in the following embodiment or related art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0030] Figure 1 A scene architecture diagram of a text correction method in some embodiments is shown;

[0031] Figure 2 A scene architecture diagram of a terminal device in some embodiments is shown;

[0032] Figure 3 A hardware configuration block diagram of a terminal device in some embodiments is shown;

[0033] Figure 4 A software configuration block diagram of a terminal device in some embodiments is shown;

[0034] Figure 5 A voice interaction network architecture schematic diagram in some embodiments is shown;

[0035] Figure 6 A configuration block diagram of a text correction server in some embodiments is shown;

[0036] Figure 7 One of the interaction flow diagrams of the text correction method provided by the embodiments of the present application is shown;

[0037] Figure 8 One of the schematic diagrams of the marking interface provided by the embodiments of the present application is shown;

[0038] Figure 9 The second interaction flow diagram of the text correction method provided by the embodiments of the present application is shown;

[0039] Figure 10 The second schematic diagram of the marking interface provided by the embodiments of the present application is shown;

[0040] Figure 11 The third interaction flow diagram of the text correction method provided by the embodiments of the present application is shown;

[0041] Figure 12 The flow diagram of the user emotion judgment method provided by the embodiments of the present application is shown. DETAILED DESCRIPTION

[0042] In order to make the purpose and the embodiments of the present application more clear, the exemplary embodiments of the present application will be described clearly and completely below in combination with the drawings in the exemplary embodiments of the present application. Obviously, the described exemplary embodiments are only some of the embodiments of the present application, but not all the embodiments.

[0043] It should be noted that the brief description of the terms in the present application is only for the convenience of understanding the following described embodiments, and is not intended to limit the embodiments of the present application. Unless otherwise specified, these terms should be understood according to their ordinary and general meanings.

[0044] The terms "include" and "have" and any variations thereof are intended to cover but not exclusively include, for example, a product or device containing a series of components, which is not necessarily limited to all the components listed clearly, but can include other components not listed clearly or inherent to these products or devices.

[0045] Figure 1 The scene architecture schematic diagram of the text correction method provided by the embodiments of the present application is shown. As Figure 1As shown, the scenario architecture provided by the embodiments of the present application includes a terminal device 100 having a voice receiving function and configured to receive voice input of a user, a semantic recognition server 200, a text correction server 300, and at least one terminal device 400 configured to label a text to be corrected. Figure 1 The three terminal devices 400 are taken as an example in the middle.

[0046] The terminal device 100 can receive a user's voice input and send the received voice input to the semantic recognition server 200. The semantic recognition server 200 is configured to convert the voice input sent by the terminal device 100 into voice text and acquire a user's intention according to the converted voice text. In the process of acquiring the user's intention according to the converted voice text, if there is no error in the converted voice text, the semantic recognition server 200 can acquire the user's intention, and then the semantic recognition server 200 directly invokes a corresponding business system according to the acquired user's intention and returns business data returned by the business system to the terminal device 100. However, if there is an error in the converted voice text, the semantic recognition server 200 cannot acquire the user's intention, and then the semantic recognition server 200 returns the converted voice text to the terminal device 100. After receiving the voice text sent by the semantic recognition server 200, the terminal device 100 sends the voice text as a to-be-corrected text to the error correction server 300, so that the error correction server 300 corrects the to-be-corrected text. After receiving the to-be-corrected text sent by the terminal device 100, the error correction server 300 first determines whether the error correction database includes error correction data for correcting the to-be-corrected text; if the error correction database of the text error correction server 300 includes error correction data for correcting the to-be-corrected text, the error correction server 300 corrects the to-be-corrected text based on the error correction data and returns the error correction result to the terminal device 100; and the terminal device 100 re-performs a business request to the semantic recognition server 200 according to the error correction result. If the error correction database of the text error correction server 300 does not include error correction data for correcting the to-be-corrected text, the text error correction server 300 returns indication information that text error correction cannot be performed to the terminal device 100, and determines whether a trigger condition for labeling the to-be-corrected text is met; if the trigger condition for labeling the to-be-corrected text is not met, only the to-be-corrected text is cached, and no other processing is performed; and if the trigger condition for labeling the to-be-corrected text is met, the text error correction server 300 sends a labeling request for requesting to label the to-be-corrected text to each terminal device 400. After receiving the labeling request sent by the text error correction server 300, the terminal device 400 displays a labeling interface according to the labeling request, receives a labeling operation input on the labeling interface, generates labeling data according to the labeling operation, and returns the labeling data to the text error correction server 300. After receiving the labeling data returned by each terminal device 400, the text error correction server generates error correction data corresponding to the to-be-corrected text according to the labeling data returned by each terminal device 400, saves the error correction data corresponding to the to-be-corrected text to the error correction database, so as to correct the to-be-corrected text in the subsequent process.

[0047] Figure 2 The architecture schematic diagram of the terminal device provided in the embodiments of the present application is shown in FIG. 1. Figure 1As shown, the architecture of the terminal device provided in the embodiments of the present application includes a terminal device 100, a control device 500, and a smart device 600.

[0048] The user can control and operate the terminal device 100 through the smart device 1002 or the control device 500, and can receive the voice input of the user through the smart device 1002 or the control device 500. The display device provided in the embodiments of the present application can have various implementation forms, for example, can be a television, a smart speaker, a refrigerator, a curtain with a display function, a personal computer (PC), a laser projection device, a monitor, an electronic bulletin board, a wearable device, a vehicle-mounted device, an electronic table, etc.

[0049] In some embodiments, the control device 500 can be a remote controller, and the communication between the remote controller and the terminal device 100 includes infrared protocol communication or Bluetooth protocol communication, and other short-distance communication modes, to control the terminal device 100 through wireless or wired mode. The user can input the user instruction through the button on the remote controller, voice input, control panel input, etc., to control the terminal device 100.

[0050] In some embodiments, the smart device 600 (such as a mobile phone, a tablet computer, a notebook computer, etc.) can also be used to control the terminal device 100. For example, an application running on the smart device 600 is used to control the terminal device 100.

[0051] In some embodiments, the terminal device 100 can also be controlled in a manner other than the control device 500 and the smart device 600, for example, the voice instruction of the user can be directly received through the module configured inside the terminal device 100 to acquire the voice instruction to control.

[0052] Figure 3 A hardware configuration block diagram of the terminal device in the example embodiments is shown. As shown in the figure, Figure 3 The terminal device 100 shown includes a communicator 220, a detector 230, an external device interface 240, a controller 250, a display 260, an audio output interface 270, at least one of a memory, a power supply, and a user interface 280.

[0053] The controller 250 includes a central processing unit, an audio processor, a RAM, a ROM, and a first interface to an n-th interface for input / output.

[0054] The communicator 220 is a component for communicating with external devices or servers according to various communication protocol types. For example, the communicator can include at least one of a Wifi module, a Bluetooth module, a wired Ethernet module, other network communication protocol chips or near field communication protocol chips, and an infrared receiver. The terminal device can establish transmission and reception of control signals and data signals with the semantic recognition server 200 through the communicator 220.

[0055] The user interface 280 can be used to receive external control signals.

[0056] The detector 230 is used to collect signals of the external environment or interaction with the outside. For example, the detector 230 includes a light receiver for collecting ambient light intensity; or the detector 230 includes an image collector such as a camera, which can be used to collect external environmental scenes, user attributes or user interaction gestures; or the detector 230 includes a sound collector such as a microphone, etc., for receiving external sounds.

[0057] The sound collector can be a microphone, also known as a "microphone", "sound collector", which can be used to receive user sound and convert sound signals into electrical signals. The terminal device can be provided with at least one microphone. In some embodiments, the terminal device can be provided with two microphones, which can not only collect sound signals, but also achieve noise reduction functions. In some embodiments, the terminal device can also be provided with three, four or more microphones, which can collect sound signals, reduce noise, and also identify sound sources to achieve directional recording functions, etc.

[0058] In addition, the microphone can be built-in on the terminal device, or the microphone can be connected to the terminal device through a wired or wireless manner. Of course, the position of the microphone on the terminal device is not limited in the embodiments of the present application. Alternatively, the terminal device can not include the microphone, i.e., the microphone is not provided in the terminal device. The terminal device can be externally connected to the microphone (also referred to as the microphone) through the interface (such as the USB interface 130). The externally connected microphone can be fixed on the terminal device through an external fixing member (such as a camera support with a clamp).

[0059] The controller 250 controls the operation of the display device and responds to the user's operation through various software control programs stored in the memory. The controller 250 controls the overall operation of the terminal device.

[0060] In some embodiments, the controller includes at least one of a Central Processing Unit (CPU), a video processor, an audio processor, a Random Access Memory (RAM), a Read-Only Memory (ROM), a first interface to an n-th interface for input / output, a communication bus, and the like.

[0061] In some examples, the operating system of the terminal device 100 is an Android system. As shown in Figure 4 , the terminal device 100 can be logically divided into an Applications layer (referred to as "application layer" for short) 21, a kernel layer 22, and a hardware layer 23.

[0062] As shown in Figure 4 , the hardware layer can include Figure 2 a communicator 220, a controller 250, a detector 230, a display 260, and the like. The application layer 21 includes one or more applications. The applications can be system applications or third-party applications. For example, the application layer 21 includes a speech recognition application, which can provide a voice interaction interface and service for connection between the terminal device 100 and the semantic recognition server 200.

[0063] The kernel layer 22 is a software middleware between the hardware layer and the application layer 21, and is used to manage and control hardware and software resources.

[0064] In some examples, the kernel layer 22 includes a detector driver, which is used to send voice data collected by the detector 230 to the speech recognition application. For example, when the speech recognition application in the terminal device 100 is started and the terminal device 100 has established a communication connection with the semantic recognition server 200, the detector driver is used to send voice data input by a user and collected by the detector 230 to the speech recognition application. Then, the speech recognition application sends query information including the voice data to the communication control module 101 in the semantic recognition server 200. The communication control module 101 is used to input the voice data sent by the terminal device 100 to the intent recognition module 102 and the data storage module 103.

[0065] To clearly illustrate the embodiments of the present application, the following describes a speech recognition network architecture provided by the embodiments of the present application. Figure 5

[0066] As shown in Figure 5 , Figure 5 a speech interaction network architecture provided by the embodiments of the present application. Figure 5 ​In some embodiments, the voice interaction device is configured to receive input information and output processing results of the information. The voice recognition module is configured to deploy a voice recognition service to recognize audio as text. The semantic understanding module is configured to deploy a semantic understanding service to perform semantic analysis on the text. The business management module is configured to deploy a business instruction management service to provide business instructions. The language generation module is configured to deploy a language generation service (NLG) to convert instructions for the voice interaction device to execute into text language. The voice synthesis module is configured to deploy a text-to-speech (TTS) service to process the text language corresponding to the instructions and send the processed text language to a loudspeaker for broadcasting. In one embodiment, Figure 5 There can be multiple entity service devices in the architecture shown in FIG. 1, each of which is configured to deploy different business services. One or more entity service devices can also be configured to deploy one or more functional services.

[0067] In some embodiments, the following describes a process of processing information of an input voice interaction device based on the architecture shown in FIG. 1. The information of the input voice interaction device is taken as an example of a voice instruction input by voice: Figure 5

[0068] [Voice Recognition]

[0069] After the voice interaction device receives the voice instruction input by voice, the voice interaction device can perform noise reduction processing and feature extraction on the audio of the voice instruction. The noise reduction processing can include steps such as removing echo and environmental noise.

[0070] [Semantic Understanding]

[0071] The acoustic model and the language model are used to perform natural language understanding on the recognized candidate text and the associated context information, to analyze the text into structured, machine-readable information, such as business domain, intent, slot, and semantic expression. An executable intent is determined, and an intent confidence score is obtained. The semantic understanding module selects one or more candidate executable intents based on the determined intent confidence score,

[0072] [Business Management]

[0073] The semantic understanding module issues an execution instruction to the corresponding business management module according to the semantic analysis result of the text of the voice instruction, to execute an operation corresponding to the voice instruction, and to complete the operation requested by the user. The semantic understanding module also feeds back an execution result of the operation corresponding to the voice instruction.

[0074] In some examples, as shown in FIG. 2, Figure 6 Figure 6 ​​A configuration block diagram of the text error correction server 300. When the semantic recognition server 200 converts the audio into text, there are errors in the text, and the semantic recognition server 200 returns the converted text to the terminal device 100. Then, the terminal device 100 sends the converted text to the communication control module 301 in the text error correction server 300 as the text to be corrected, and the communication control module 301 outputs the received text to be corrected to the error correction module 302; the error correction module 302 is configured to query whether the error correction database 303 contains error correction data of the text to be corrected, if the error correction database 303 contains error correction data of the text to be corrected, the error correction data is used to correct the text to be corrected, and the error correction result is returned to the terminal device 100 through the communication control module 301. If the error correction database 303 does not contain error correction data of the text to be corrected, the decision module 304 is called to determine whether the trigger condition for labeling the text to be corrected is met. In the case where the trigger condition is met, the voting module 305 sends a labeling request for requesting to label the text to be corrected to the terminal device 400, and receives the labeling data sent by the terminal device 400. The decision module 304 is also configured to generate error correction data corresponding to the text to be corrected according to the labeling data sent by the at least one terminal device, and save the error correction data corresponding to the text to be corrected into the error correction database 303.

[0075] In some embodiments, as shown in Figure 5 The text error correction method provided by the embodiments of the present application comprises the following steps:

[0076] S701, the first terminal device sends the text to be corrected to the text error correction server.

[0077] Correspondingly, the text error correction server receives the text to be corrected sent by the first terminal device.

[0078] In some embodiments, the first terminal device sends the text to be corrected to the text error correction server in the following manner: the first terminal device sends the voice to be recognized to the semantic recognition server, and receives the semantic recognition result returned by the semantic recognition server; if the semantic recognition result only includes the text corresponding to the voice to be recognized, but does not include the user intent corresponding to the voice to be recognized, the text corresponding to the voice to be recognized is sent to the text error correction server as the text to be corrected.

[0079] For example, the user inputs a voice input corresponding to a text content of "how to make preserved eggs" to the first terminal device, the first terminal device sends the voice input to the semantic recognition server, the semantic recognition server converts the voice input into a text "how to make preserved eggs", and because the semantic recognition server cannot obtain the user intent corresponding to the voice input according to the text "how to make preserved eggs", the semantic recognition server returns the text "how to make preserved eggs" to the first terminal device, and the first terminal device sends the text "how to make preserved eggs" as the text to be corrected to the text correction server.

[0080] In some embodiments, the first terminal device can send the text to be corrected to the text correction server through the communicator, and the text correction server can receive the text to be corrected sent by the first terminal device through the communicator.

[0081] S702, the text correction server determines whether the first correction data for correcting the text to be corrected is included in the correction database of the text correction server.

[0082] In step S702, if the first correction data for correcting the text to be corrected is not included in the correction database of the text correction server, the following step S703 is performed.

[0083] S703, the text correction server determines whether the trigger condition for labeling the text to be corrected is met.

[0084] That is, in the embodiments of the present application, all texts without obtaining the user intent will not be labeled, but only texts meeting the trigger condition will be labeled.

[0085] For example, the trigger condition is that the cumulative number of unrecognized times exceeds 5, the text to be corrected is "how to make preserved eggs", and the first correction data for correcting "how to make preserved eggs" is not included in the correction database. When the text correction server determines whether the number of times the terminal device sends the text to be corrected "how to make preserved eggs" to the text correction server exceeds 5, if it exceeds 5, it is determined that the trigger condition for labeling "how to make preserved eggs" is met; if it does not exceed 5, it is determined that the trigger condition for labeling "how to make preserved eggs" is not met.

[0086] In some embodiments, the text correction server can execute the above steps S702 and S703 (determine whether the first correction data for correcting the text to be corrected is included in the correction database of the text correction server, and if the first correction data for correcting the text to be corrected is not included in the correction database of the text correction server, determine whether the trigger condition for labeling the text to be corrected is met) through the controller.

[0087] In step S703, if the text correction server determines that the trigger condition for labeling the text to be corrected is met, the following step S704 is performed.

[0088] S704, the text correction server sends a labeling request for requesting labeling of the text to be corrected to at least one terminal device.

[0089] Correspondingly, the at least one terminal device respectively receives the labeling request for requesting labeling of the text to be corrected sent by the text correction server.

[0090] In some embodiments, the at least one terminal device can include the first terminal device. That is, the text correction server sends a labeling request for requesting labeling of the text to be corrected to the terminal device that sends the text to be corrected. For example, the terminal device A sends the text to be corrected "How to send a flower dan" to the text correction server, and the text correction server sends a labeling request for requesting labeling of the text to be corrected "How to send a flower dan" to a plurality of terminal devices including the terminal device A.

[0091] In some embodiments, the text correction server can perform the above step S704 (send a labeling request for requesting labeling of the text to be corrected to at least one terminal device) through the communicator.

[0092] S705, the at least one terminal device respectively displays a labeling interface according to the labeling request.

[0093] For example, referring to FIG. 8, Figure 8 Figure 8 The labeling interface is described taking the text to be corrected "How to send a flower dan" as an example, as shown in FIG. 8, the labeling interface can display a prompt text 81 and a text to be corrected 82. The prompt text 81 can be "Please enter the following text meaning", and the text to be corrected 82 is "How to send a flower dan". Figure 8

[0094] In some embodiments, the at least one terminal device can respectively perform the above step S705 (display a labeling interface according to the labeling request) through the display.

[0095] S706, the at least one terminal device respectively receives the labeling operation input by the user in the labeling interface, and generates labeling data according to the labeling operation.

[0096] In some embodiments, the labeling operation input by the user in the labeling interface can be a text input operation. For example, the user inputs the text "How to make a pine-flower egg" in the labeling interface through a text input operation, and generates labeling data according to the text "How to make a pine-flower egg". ​​

[0097] In some embodiments, the at least one terminal device can perform the above step S706 (receive the labeling operation input by the user in the labeling interface, and generate labeling data according to the labeling operation) respectively through a user interface.

[0098] S707, the at least one terminal device respectively sends the labeling data to the text correction server.

[0099] Correspondingly, the text correction server receives the labeling data sent by the at least one terminal device.

[0100] In some embodiments, the at least one terminal device can send the labeling data to the text correction server respectively through a communicator, and the text correction server can receive the labeling data sent by the at least one terminal device through a communicator.

[0101] S708, the text correction server generates the correction data corresponding to the text to be corrected according to the labeling data sent by the at least one terminal device.

[0102] For example, the text to be corrected is "how to make a flower egg", and the labeling data sent by the at least one terminal device to the text correction server is "how to make a flower egg", and the generated correction data can be "how to make a flower egg".

[0103] S709, the text correction server saves the correction data corresponding to the text to be corrected to the correction database.

[0104] In some embodiments, the text correction server can perform the above steps S708 and S709 (generate the correction data corresponding to the text to be corrected according to the labeling data sent by the at least one terminal device, and save the correction data corresponding to the text to be corrected to the correction database) through a controller.

[0105] The text correction method provided in the embodiments of the present application comprises the following steps: when a text correction server receives a to-be-corrected text sent by a first terminal device, it is determined whether the first correction data for correcting the to-be-corrected text is included in a correction database; if the first correction data for correcting the to-be-corrected text is not included in the correction database, it is determined whether a trigger condition for labeling the to-be-corrected text is met; if the trigger condition is met, a labeling request for requesting to label the to-be-corrected text is sent to at least one terminal device, and labeling data sent by the at least one terminal device is received; finally, the correction data corresponding to the to-be-corrected text is generated according to the labeling data sent by the at least one terminal device, and the correction data corresponding to the to-be-corrected text is saved to the correction database. Since the to-be-corrected text is corrected based on the correction data in the embodiments of the present application, and the correction data is generated according to the labeling data sent by the terminal device, the embodiments of the present application can quickly obtain the correction data of the to-be-corrected text that meets the trigger condition through user labeling, and then correct the to-be-corrected text, so that the problem of long development cycle of the text correction scheme can be solved.

[0106] As an extension and refinement of the above embodiments, the embodiments of the present application provide another text correction method, as shown in Figure 9 The text correction method comprises the following steps:

[0107] S901, the first terminal device receives a voice input of a user.

[0108] In some embodiments, after the first terminal device receives the voice input of the user, the voice input can also be subjected to noise reduction processing.

[0109] S902, the first terminal device sends the voice input to a semantic recognition server.

[0110] Correspondingly, the semantic recognition server receives the voice input sent by the first terminal device.

[0111] S903, the semantic recognition server converts the voice input into a text, and acquires a user intention according to the text.

[0112] In the above step S903, if the semantic recognition server successfully acquires the user intention according to the text, the semantic recognition server directly calls a related business system to acquire business data, and returns the acquired business data to the first terminal device.

[0113] For example, if the semantic recognition server converts the text obtained from the voice input into "play the poem Quiet Night Thoughts", the business system is called to obtain audio resources or video resources related to the poem Quiet Night Thoughts, and the audio resources or video resources related to the poem Quiet Night Thoughts obtained are sent to the first terminal device, so that the first terminal device plays the audio resources or video resources related to the poem Quiet Night Thoughts.

[0114] In step S903 described above, if the semantic recognition server fails to obtain the user's intention according to the text, the semantic recognition server performs the following step S904.

[0115] S904, the semantic recognition server sends the text converted from the voice input to the first terminal device.

[0116] Correspondingly, the first terminal device receives the text converted from the voice input sent by the semantic recognition server.

[0117] For example, if the semantic recognition server converts the text obtained from the voice input into "where is the capital of the Netherlands?", the text converted from the voice input "where is the capital of the Netherlands?" is returned to the first terminal device because the user's intention cannot be obtained according to the text "where is the capital of the Netherlands?".

[0118] S905, the first terminal device determines the text converted from the voice input as the text to be corrected, and sends the text to be corrected to the text correction server.

[0119] Correspondingly, the text correction server receives the text to be corrected sent by the first terminal device.

[0120] S906, the text correction server determines whether the first correction data for correcting the text to be corrected is included in the correction database of the text correction server.

[0121] In some embodiments, the first correction data is correction data for correcting the text to be corrected sent by all terminal devices.

[0122] For example, if the text to be corrected is "how tall is Hu Jian?", and the correction database includes correction data "how tall is Fujian?" for correcting the text "how tall is Hu Jian?" sent by all terminal devices, it can be determined that the first correction data for correcting the text to be corrected is included in the correction database, and the first correction data is correction data for correcting the text to be corrected sent by all terminal devices.

[0123] In some embodiments, the first correction data is correction data configured to be used only for correcting the text to be corrected sent by the first terminal device.

[0124] For example, the text to be corrected is "zhi bu dao", and the error correction database includes error correction data "bubudao" for correcting the text "zhi bu dao" sent by the first terminal device. It can be determined that the error correction database includes first error correction data for correcting the text to be corrected, and the first error correction data is error correction data for correcting only the text to be corrected sent by the first terminal device.

[0125] For example, the text to be corrected is "liang zhi huang lie meng cui liu", and the error correction database includes error correction data "liang ge huang lie meng cui liu" for correcting the text "liang zhi huang lie meng cui liu" sent by the second terminal device. It can be determined that the error correction database does not include first error correction data for correcting the text to be corrected.

[0126] In step S906, if the error correction database of the text error correction server includes first error correction data for correcting the text to be corrected, the following steps S907-S910 are executed.

[0127] S907, the text error correction server corrects the text to be corrected according to the first error correction data to obtain a second text.

[0128] For example, the text to be corrected is "hujian you duo da?", and the error correction database includes first error correction data "fujian you duo da?" for correcting "hujian you duo da?". The text to be corrected can be corrected according to the first error correction data to obtain a second text "fujian you duo da?".

[0129] In some embodiments, the text error correction server can execute the above step S907 (the text error correction server corrects the text to be corrected according to the first error correction data to obtain a second text) through the controller.

[0130] S908, the text error correction server sends the second text to the first terminal device.

[0131] Correspondingly, the first terminal device receives the second text sent by the text error correction server.

[0132] S909, the first terminal device sends the second text to the semantic recognition server.

[0133] Correspondingly, the semantic recognition server receives the second text sent by the first terminal device.

[0134] S910, the semantic recognition server obtains the user's intention according to the second text.

[0135] For example, if the second text is "How big is Fujian?", the semantic recognition server can obtain the user intent "query the area of Fujian Province" according to the second text, and then call the relevant business system to obtain the area of Fujian Province, and return the queried area of Fujian Province to the first terminal device.

[0136] In step S906, if the first error correction data for correcting the to-be-corrected text is not included in the error correction database of the text error correction server, the following steps S911 to S918 are executed.

[0137] S911, the text error correction server determines whether the trigger condition for labeling the to-be-corrected text is met.

[0138] In step S911, if the text error correction server determines that the trigger condition for labeling the to-be-corrected text is met, the following step S912 is executed.

[0139] S912, the text error correction server sends a labeling request to at least one terminal device.

[0140] Correspondingly, the at least one terminal device respectively receives the labeling request sent by the text error correction server.

[0141] The labeling request is used to request labeling of the to-be-corrected text, and the labeling request carries at least one labeling option.

[0142] S913, the at least one terminal device respectively displays a labeling interface according to the labeling request.

[0143] For example, referring to FIG. 10, Figure 10 Figure 10 The labeling interface is described taking the to-be-corrected text "this I know not" as an example, as shown in FIG. 10, Figure 10 The labeling interface can display a prompt text 1001, a to-be-corrected text 1002, and a labeling option 1003 and a labeling option 1004. The prompt text 1001 can be "Please select the correct meaning of the following text", the to-be-corrected text 1002 is "this I know not", the content of the labeling option 1003 is "this I know", and the content of the labeling option 1004 is "this I don't know".

[0144] S914, the at least one terminal device respectively receives a user's selection operation on the at least one labeling option, and generates labeling data according to the user's selected labeling option.

[0145] ​For example, the at least one terminal device 1 receives the selection operation of the user on the annotation option A in the at least one annotation option, and the terminal device 1 generates annotation data according to the annotation option A; the at least one terminal device 2 receives the selection operation of the user on the annotation option B in the at least one annotation option, and the terminal device 2 generates annotation data according to the annotation option B.

[0146] S915, the at least one terminal device respectively sends the annotation data to the text correction server.

[0147] Correspondingly, the text correction server receives the annotation data sent by the at least one terminal device.

[0148] S916, the text correction server judges whether there is annotation data with a proportion exceeding a threshold proportion in the annotation data respectively sent by the at least one terminal device to the text correction server.

[0149] For example, the threshold proportion is 60%, the terminal device includes terminal device 1 to terminal device 10, and the annotation data returned by the terminal device 1 to the terminal device 10 is shown in Table 1 as follows:

[0150] Table 1

[0151]

[0152] From the above Table 1, it can be seen that the proportion of annotation data A is 70%, the proportion of annotation data B is 10%, and the proportion of annotation data C is 20%, wherein the proportion (70%) of annotation data A is greater than the threshold proportion (60%), so the text correction server determines that there is annotation data with a proportion exceeding the threshold proportion in the annotation data respectively sent by the at least one terminal device to the text correction server, and the annotation data exceeding the threshold proportion is annotation data A.

[0153] For example, the threshold proportion is 75%, the terminal device includes terminal device 1 to terminal device 10, and the annotation data returned by the terminal device 1 to the terminal device 10 is shown in Table 2 as follows:

[0154] Table 2

[0155]

[0156] As shown in Table 2 above, the proportion of labeled data A is 30%, the proportion of labeled data B is 20%, the proportion of labeled data C is 20%, the proportion of labeled data D is 10%, and the proportion of labeled data E is 20%. The proportions of labeled data A, B, C, D, and E are all less than the threshold proportion (75%). Therefore, the text correction server determines that at least one terminal device can send to the text correction server the labeled data that does not contain labeled data with a proportion exceeding the threshold proportion.

[0157] In step S916 above, if the proportion of target annotation data in the annotation data sent by the at least one terminal device exceeds a threshold proportion, then step S917 is executed as follows:

[0158] S917. The text correction server generates correction data based on the target annotation data for correcting the text to be corrected sent by all terminal devices.

[0159] As shown in Table 1 above, the proportion of annotation data A in the annotation data sent by terminal devices 1 to 10 is greater than the threshold proportion. Therefore, error correction data is generated based on annotation data A to correct the text to be corrected sent by all terminal devices.

[0160] In step S916 above, if the proportion of each labeled data in the labeled data sent by the at least one terminal device is less than the threshold proportion, then step S918 is executed as follows:

[0161] S918. The text correction server generates correction data based on the annotation data of each terminal device in the at least one terminal device, which is used only to correct the text to be corrected sent by the corresponding terminal device.

[0162] As shown in Table 1 above, the proportion of each annotation data in the annotation data sent by terminal device 1 to terminal device 10 is less than the threshold proportion. Therefore, based on the annotation data A sent by terminal device 1, error correction data is generated only for correcting the text to be corrected sent by terminal device 1. Based on the annotation data E sent by terminal device 2, error correction data is generated only for correcting the text to be corrected sent by terminal device 2. ... Based on the annotation data D sent by terminal device 10, error correction data is generated only for correcting the text to be corrected sent by terminal device 10.

[0163] S919. The text correction server saves the correction data corresponding to the text to be corrected to the correction database.

[0164] In some embodiments, refer to Figure 11 As shown, in Figure 9On the basis of the embodiment shown, in a case where the first error correction data for correcting the to-be-corrected text is not included in the error correction database of the text error correction server, the text error correction method provided by the embodiment of the application further includes the following steps:

[0165] S111, the text error correction server sends indication information indicating that the to-be-corrected text fails to be corrected to the first terminal device.

[0166] Correspondingly, the first terminal device receives the indication information indicating that the to-be-corrected text fails to be corrected sent by the text error correction server.

[0167] S112, the first terminal device outputs indication information indicating that the speech input recognition fails.

[0168] In some embodiments, the implementation manner of the first terminal device outputting the indication information indicating that the speech input recognition fails can be playing an audio file indicating that the speech input recognition fails. For example, an audio file corresponding to the text "I don't understand what you mean" is played.

[0169] In some embodiments, the implementation manner of the first terminal device outputting the indication information indicating that the speech input recognition fails can be displaying text information indicating that the speech input recognition fails. For example, text information "what you said I haven't learned yet" is displayed.

[0170] In some embodiments, the implementation manner of determining whether the trigger condition of labeling the to-be-corrected text is met can include:

[0171] If the number of times that the first terminal device continuously sends the to-be-corrected text to the text error correction server exceeds a first threshold number of times; or,

[0172] If the number of times that the terminal devices in a target geographic area send the to-be-corrected text to the text error correction server within a first preset time length exceeds a second threshold number of times, the target geographic area is a geographic area to which the first terminal device belongs; or,

[0173] If the number of times that the first terminal device sends text to the text error correction server within a second preset time length exceeds a third threshold number of times; or,

[0174] If the user emotion corresponding to the to-be-corrected text is a negative emotion;

[0175] Then, it is determined that the trigger condition of labeling the to-be-corrected text is met.

[0176] That is, it is determined whether the following conditions 1 to 4 are met, and when at least one of the following conditions 1 to 4 is met, it is determined that the trigger condition of labeling the to-be-corrected text is met:

[0177] Condition 1, the number of times that the first terminal device continuously sends the text to be corrected to the text correction server exceeds a first threshold number of times.

[0178] The first threshold number of times can be configured and issued by the background of the text correction server.

[0179] For example, if the first threshold number of times is configured as 3 times, whether the trigger condition for labeling the text to be corrected is met is determined by judging whether the number of times that the first terminal device continuously sends the text to be corrected to the text correction server exceeds 3 times.

[0180] Condition 2, the number of times that terminal devices in a target geographic area send the text to be corrected to the text correction server within a first preset time period exceeds a second threshold number of times. The target geographic area is the geographic area to which the first terminal device belongs.

[0181] That is, the text correction server divides the geographic area in advance, and then, when determining whether the trigger condition for labeling the text to be corrected is met, first acquires the geographic area to which the first terminal device belongs, then acquires the text correction history data of all terminal devices in the geographic area to which the first terminal device belongs within the first preset time period, and determines whether the trigger condition for labeling the text to be corrected is met by judging whether the number of times that the text to be corrected is sent in the text correction history data of all terminal devices in the geographic area to which the first terminal device belongs within the first preset time period exceeds the second threshold number of times. The geographic area to which the first terminal device belongs can be acquired according to the Internet Protocol (IP) address of the first terminal device.

[0182] For example, if the first preset time period is 24 hours, the geographic area to which the first terminal device belongs is XX town, and the second threshold number of times is 10, whether the trigger condition for labeling the text to be corrected is met is determined by acquiring whether the number of times that all terminal devices in XX town send the text to be corrected to the text correction server within the past 24 hours exceeds 10 times.

[0183] Condition 3, the number of times that the first terminal device sends text to the text correction server within a second preset time period exceeds a third threshold number of times.

[0184] For example, if the second preset time period is 7 days and the third threshold number of times is 5, whether the trigger condition for labeling the text to be corrected is met is determined by judging whether the number of times that the first terminal device sends text to the text correction server within the past 7 days exceeds 5 times.

[0185] Condition 4, the user emotion corresponding to the text to be corrected is a negative emotion.

[0186] That is, the user emotion corresponding to the text to be corrected is obtained, and it is determined whether the trigger condition for labeling the text to be corrected is met by judging whether the user emotion corresponding to the text to be corrected is a negative emotion. In some embodiments, referring to Figure 12 As shown in the implementation manner for obtaining the user emotion corresponding to the text to be corrected can include the following steps a to b:

[0187] S121, performing word segmentation on the text to be corrected to obtain a word segmentation result of the text to be corrected.

[0188] For example, if the text to be corrected is "I like playing games", the word segmentation result of the text to be corrected can include "I", "like", "play", and "games".

[0189] S122, loading a dirty word database.

[0190] S123, determining whether the word segmentation result of the text to be corrected contains dirty words in the dirty word database.

[0191] In step c, if the word segmentation result of the text to be corrected contains dirty words in the dirty word database, the following step d is performed:

[0192] S124, determining that the user emotion corresponding to the text to be corrected is a negative emotion.

[0193] In step c, if the word segmentation result of the text to be corrected does not contain dirty words in the dirty word database, the following step e is performed:

[0194] S125: determining whether the user emotion corresponding to the text to be corrected is a negative emotion based on an emotion analysis model.

[0195] In some embodiments, the emotion analysis model can be a model obtained by training a machine learning model based on sample data, the sample data including sample texts and emotion labels of the sample texts, the emotion labels being negative emotions or non-negative emotions, and the machine learning model can be a deep neural network model, a convolutional neural network model, etc.

[0196] In some embodiments, the present application provides a computer readable storage medium, the computer readable storage medium storing a computer program, when the computer program is executed by a computing device, the computer program causes the computing device to implement the text correction method of any of the above embodiments.

[0197] In some embodiments, the embodiments of the present application provide a computer program product, which, when running on a computer, enables the computer to implement the text correction method of any of the above embodiments.

[0198] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

[0199] For the convenience of explanation, the above description has been made in combination with specific embodiments. However, the above exemplary discussion is not intended to exhaust or limit the embodiments to the specific forms disclosed above. Various modifications and variations can be derived according to the above teachings. The selection and description of the above embodiments are to better explain the principles and practical applications, so that those skilled in the art can better use the embodiments and various different modified embodiments suitable for specific use considerations.

Claims

1. A text correction server, characterized in that, include: The communicator is configured to receive the text to be corrected sent by the first terminal device; The controller is configured to determine whether the triggering condition for annotating the text to be corrected is met if the text correction database of the text correction server does not contain first correction data for correcting the text to be corrected. The communicator is further configured to, upon meeting the triggering condition, send an annotation request to at least one terminal device to request annotation of the text to be corrected, and to receive annotation data sent by the at least one terminal device. The controller is further configured to generate error correction data corresponding to the text to be corrected based on the number of annotations sent by the at least one terminal device, and save the error correction data corresponding to the text to be corrected to the error correction database; The controller is specifically configured as follows: If the proportion of target annotation data in the annotation data sent by at least one terminal device exceeds a threshold proportion, error correction data is generated based on the target annotation data to correct the text to be corrected sent by all terminal devices. If the proportion of each annotation data in the annotation data sent by the at least one terminal device is less than the threshold proportion, then error correction data is generated based on the annotation data of each terminal device in the at least one terminal device, which is only used to correct the text to be corrected sent by the corresponding terminal device.

2. The text correction server according to claim 1, characterized in that, The controller is further configured to, when the error correction database includes the first error correction data, correct the text to be corrected based on the first error correction data to obtain the second text; The communicator is also configured to send the second text to the first terminal device.

3. The text correction server according to claim 1, characterized in that, The first error correction data is error correction data used to correct the text to be corrected sent by all terminal devices; or; The first error correction data is error correction data configured to be used only for correcting the text to be corrected sent by the first terminal device.

4. The text correction server according to claim 1, characterized in that, The controller is specifically configured as follows: If the number of times the first terminal device continuously sends the text to be corrected to the text correction server exceeds a first threshold number; or... If the number of times a terminal device within the target geographical area sends the text to be corrected to the text correction server within a first preset time period exceeds a second threshold number, the target geographical area is the geographical area to which the first terminal device belongs. or, If the first terminal device sends more than a third threshold number of texts to the text correction server within a second preset time period; or, If the user's emotion corresponding to the text to be corrected is negative; Then, it is determined that the triggering condition for annotating the text to be corrected is met.

5. The text correction server according to claim 1, characterized in that, The communicator is further configured to send an indication message to the first terminal device indicating that the text to be corrected has failed to be corrected, if the text correction server's correction database does not contain first correction data for correcting the text to be corrected.

6. A terminal device, characterized in that, include: The communicator is configured to receive annotation requests sent by the text correction server, which request annotation of the text to be corrected. The display is configured to show a labeling interface in response to the labeling request; The user interface is configured to receive annotation operations input by the user in the annotation interface, and to generate annotation data based on the annotation operations; The communicator is also configured to send the annotation data to the text correction server; The annotation request carries at least one annotation option; The display is specifically configured to display the text to be corrected and the at least one annotation option; The user interface is specifically configured to receive a user's selection of at least one annotation option and generate the annotation data based on the annotation option selected by the user.

7. A text error correction method, characterized in that, Applied to a text correction server, the method includes: Receive the text to be corrected sent by the first terminal device; If the error correction database of the text correction server does not contain first error correction data for correcting the text to be corrected, determine whether the triggering condition for annotating the text to be corrected is met. If the triggering condition is met, a labeling request is sent to at least one terminal device to request the labeling of the text to be corrected, and labeling data is received from the at least one terminal device. Based on the number of annotations sent by the at least one terminal device, error correction data corresponding to the text to be corrected is generated, and the error correction data corresponding to the text to be corrected is saved to the error correction database; The step of generating the correction data corresponding to the text to be corrected based on the number of annotations sent by the at least one terminal device includes: If the proportion of target annotation data in the annotation data sent by at least one terminal device exceeds a threshold proportion, error correction data is generated based on the target annotation data to correct the text to be corrected sent by all terminal devices. If the proportion of each annotation data in the annotation data sent by the at least one terminal device is less than the threshold proportion, then error correction data is generated based on the annotation data of each terminal device in the at least one terminal device, which is only used to correct the text to be corrected sent by the corresponding terminal device.

8. A text correction method, characterized in that, Applied to a terminal device, the method includes: Receive a annotation request sent by a text correction server, which requests annotation of the text to be corrected; the annotation request carries at least one annotation option; The annotation interface is displayed according to the annotation request; the annotation interface includes the text to be corrected and at least one annotation option; Receive annotation operations input by the user in the annotation interface, and generate annotation data based on the annotation operations; The annotation data is sent to the text correction server; The step of receiving annotation operations input by the user in the annotation interface and generating annotation data based on the annotation operations includes: receiving the user's selection operation for at least one annotation option, and generating the annotation data based on the annotation option selected by the user.

Images