Question and answer intent recognition method, server, device and storage medium

Abstract

The application relates to the technical field of intelligent question and answer, and discloses a question and answer intention recognition method, which comprises the following steps: obtaining current voice information based on a node question of a current node, performing intention recognition on the current voice information based on the node question of the current node through an intention recognition model to obtain a current node answer; if the current node is configured with a future node question, performing intention recognition on the current voice information based on the future node question through the intention recognition model to obtain a future node answer; selecting a subsequent node according to the current node answer, judging whether the node question of the subsequent node is contained in the future node question, if not, reacquiring user voice information, if yes, obtaining a subsequent node answer corresponding to the node question of the subsequent node from the future node answer, and stopping the question and answer when there is no subsequent node for the current question and answer node or communication with a communication terminal is ended. The question and answer intention recognition method avoids repeated inquiries during multiple question and answer processes, and improves the experience of users during the question and answer process.

Description

Technical Field

[0001] This invention relates to the field of intelligent question-answering technology, and in particular to a question-answering intent recognition method, server, device, and storage medium. Background Technology

[0002] Intelligent outbound calling is an intelligent customer service robot product that integrates Automatic Speech Recognition (ASR), Text-to-Speech (TTS), and Natural Language Understanding (NLU) technologies and is offered to enterprise customers. In the outbound calling robot framework, the dialogue transition logic is generally pre-defined; that is, based on the user's current intent, the system selects what question to ask, and the direction of the dialogue transition is determined by the recognition result of the current user intent.

[0003] Current intent recognition solutions primarily assign similar phrases to each intent, then semantically match the user's current utterance with these similar phrases to determine the most similar intent. In other words, each intent recognition operation identifies a single system intent corresponding to the user's utterance. However, in some cases, a user's statement may contain multiple pieces of information, answering not only the current question but also potential future questions from the outbound call robot. Under the current intent matching logic, the outbound call robot ignores this extraneous information, focusing only on the answer to the current question. Subsequent inquiries, following predetermined conversation transitions, may result in redundant questions, degrading the user experience when engaging in intelligent question-and-answer sessions with the outbound call robot. Summary of the Invention

[0004] In view of the above, it is necessary to provide a question-and-answer intent recognition method to avoid repeated questions in multiple question-and-answer sessions.

[0005] To achieve the above objectives, the present invention provides a question-and-answer intent recognition method, the method comprising:

[0006] The current voice information input by the user at the communication terminal is obtained based on the node question of the current node. The current voice information is then used to perform intent recognition based on the node question of the current node to obtain the answer of the current node.

[0007] Determine whether the current node is configured with a future node question; otherwise, end the intent recognition of the current node; if so, use the intent recognition model to sequentially perform intent recognition on the current voice information based on the future node question to obtain the future node answer.

[0008] Based on the current node answer, a subsequent node is selected. It is determined whether the node question of the subsequent node is included in the future node question. If not, the user's voice information for the node question of the subsequent node is re-acquired. If so, the pending node answer corresponding to the node question of the subsequent node is obtained from the future node answers. The user is then confirmed via the communication terminal whether the pending node answer is the answer of the subsequent node. If not, the user's voice information for the node question of the subsequent node is re-acquired. If so, the pending node answer is used as the answer of the subsequent node, and the selection of a subsequent node continues until the current question and answer node has no subsequent node or the communication with the communication terminal ends, at which point the question and answer process stops.

[0009] Optionally, the step of obtaining the current node answer by performing intent recognition on the current speech information based on the node question of the current node using an intent recognition model includes:

[0010] The current voice information is converted into text information and concatenated with the node question of the current node, and then converted into a number string and input into the intent recognition model;

[0011] Obtain the output of the intent recognition model, and obtain the answer for the current node from the current speech information based on the output.

[0012] Optionally, the step of converting the current speech information into text information and concatenating it with the node question of the current node, then converting it into a number string and inputting it into the intent recognition model, includes:

[0013] Convert the current voice information into text information;

[0014] Obtain the preset dictionary corresponding to the intent recognition model, and concatenate the text information and the node question of the current node according to the preset dictionary to obtain the concatenated text;

[0015] The pre-defined dictionary maps each character of the concatenated text to its corresponding ID to obtain the corresponding numeric string.

[0016] The number string is input into the intent recognition model.

[0017] Optionally, the preset dictionary includes a "start character" for identifying the beginning of a sentence, a "end character" for identifying the end of a sentence, and a numeric ID corresponding to each character;

[0018] The method for concatenating the text information and the node problem of the current node according to the preset dictionary to obtain the concatenated text is as follows:

[0019] Concatenate text = "start character" + text information + "end character" + node issue + "end character".

[0020] Optionally, obtaining the output of the intent recognition model includes:

[0021] The number string is input into the intention recognition model to obtain the start sequence and the end sequence, and the start sequence and the end sequence are converted into start probability and end probability through the softmax layer of the intention recognition model;

[0022] The starting position is determined by taking the position corresponding to the maximum value of the initial probability, and the ending position is determined by taking the position corresponding to the maximum value of the ending probability. The initial probability includes the probability that each character of the concatenated text is the starting position of the answer, and the ending probability includes the probability that each character of the concatenated text is the ending position of the answer.

[0023] Optionally, obtaining the answer for the current node from the current speech information based on the output result includes:

[0024] If the starting position is before the ending position, the answer to the undetermined node is obtained from the current voice information based on the starting position and the ending position, and the product of the starting probability and the ending probability is taken as the probability of the answer to the undetermined node. The probability of the answer to the undetermined node that is greater than a preset threshold is taken as the answer to the current node.

[0025] If the starting position is after the ending position or the probability of the answer to the pending node is less than a preset threshold, it is determined that the current voice information does not contain the answer to the current node, and the voice information is reacquired for intent recognition.

[0026] Optionally, the question-answering intent recognition method further includes:

[0027] If the node question of the subsequent node is in the future node question and the answer of the future node does not contain the node answer corresponding to the node question of the subsequent node, then the subsequent node voice information input by the user on the communication terminal is obtained based on the node question of the subsequent node, and the node answer of the subsequent node is obtained by performing intent recognition on the subsequent node voice information based on the node question of the subsequent node through an intent recognition model.

[0028] Furthermore, to achieve the above objectives, the present invention also provides a question-and-answer intent recognition device, the question-and-answer intent recognition device comprising:

[0029] The current node answer acquisition module is used to acquire the current voice information input by the user on the communication terminal based on the node question of the current node, and to obtain the current node answer by performing intent recognition on the current voice information based on the node question of the current node.

[0030] The future node answer acquisition module is used to determine whether the current node is configured with a future node question. If not, the intent recognition of the current node ends; if so, the intent recognition model is used to sequentially perform intent recognition on the current voice information based on the future node question to obtain the future node answer.

[0031] The subsequent node answer acquisition module is used to select a subsequent node based on the current node answer, determine whether the node question of the subsequent node is included in the future node questions, if not, reacquire the user's voice information for the node question of the subsequent node, if so, obtain the pending node answer corresponding to the node question of the subsequent node from the future node answers; confirm with the user through the communication terminal whether the pending node answer is the subsequent node answer, if not, reacquire the user's voice information for the node question of the subsequent node, if so, use the pending node answer as the subsequent node answer and continue to select a subsequent node until the current question and answer node has no subsequent node or the communication with the communication terminal ends and the question and answer stops.

[0032] Furthermore, to achieve the above objectives, the present invention also provides an electronic device, the electronic device comprising:

[0033] Memory, storing at least one computer program; and

[0034] The processor executes the program stored in the memory to implement the question-and-answer intent recognition method described above.

[0035] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the question-and-answer intent recognition method described above.

[0036] This invention configures future node questions for each node in the session jump logic and obtains the node answer corresponding to the current node and the node answer corresponding to the future node question through an intent recognition model, thereby avoiding repeated questioning of subsequent nodes and improving the user experience when conducting intelligent question and answer with outbound call robots. Attached Figure Description

[0037] Figure 1 This is a flowchart of an embodiment of the question-and-answer intent recognition method of the present invention;

[0038] Figure 2 This is a schematic diagram of an embodiment of the electronic device of the present invention;

[0039] Figure 3 This is a schematic diagram of an embodiment of the question-and-answer intent recognition program of the present invention.

[0040] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0042] It should be noted that the descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0043] like Figure 1 The diagram shown is a flowchart of an embodiment of the question-and-answer intent recognition method of the present invention. The question-and-answer intent recognition method is applied to a question-and-answer intent recognition device and includes steps S1-S3.

[0044] S1. Obtain the current voice information input by the user at the communication terminal based on the node question of the current node, and obtain the current node answer by performing intent recognition on the current voice information based on the node question of the current node through the intent recognition model.

[0045] In one embodiment, the intent recognition model is an intent recognition model based on a transformer structure trained on a preset training dataset. The intent recognition model includes a corresponding preset dictionary, which includes a "start character" for identifying the beginning of a sentence, a "end character" for identifying the end of a sentence, and a numeric ID corresponding to each character.

[0046] In one embodiment, the step of obtaining the current node answer by performing intent recognition on the current speech information based on the node question of the current node through the intent recognition model includes: converting the current speech information into text information and concatenating it with the node question of the current node, then converting it into a number string and inputting it into the intent recognition model; obtaining the output result of the intent recognition model, and obtaining the current node answer from the current speech information according to the output result.

[0047] Specifically, the step of converting the current speech information into text information and concatenating it with the node question of the current node to convert it into a number string for input into the intent recognition model includes: converting the current speech information into text information; obtaining a preset dictionary corresponding to the intent recognition model; concatenating the text information and the node question of the current node according to the preset dictionary to obtain concatenated text; mapping each character of the concatenated text to its corresponding ID through the preset dictionary to obtain a number string corresponding to the concatenated text; and inputting the number string into the intent recognition model.

[0048] The intent recognition model concatenates text information and node questions according to a preset dictionary to obtain concatenated text in the following way: concatenated text = “start character” + text information + “end character” + node question + “end character”.

[0049] Specifically, obtaining the output of the intent recognition model includes: inputting the number string into the intent recognition model to obtain a start sequence and an end sequence; converting the start sequence and the end sequence into start probabilities and end probabilities through the softmax layer of the intent recognition model; taking the position corresponding to the maximum value of the start probability as the start position and the position corresponding to the maximum value of the end probability as the end position, wherein the start probability includes the probability that each character of the concatenated text is the start position of the answer, and the end probability includes the probability that each character of the concatenated text is the end position of the answer.

[0050] Specifically, obtaining the current node answer from the current voice information based on the output result includes: if the start position is before the end position, obtaining the undetermined node answer from the current voice information based on the start position and the end position, and using the product of the start probability and the end probability as the undetermined node answer probability, and using the undetermined node answer probability greater than a preset threshold as the current node answer; if the start position is after the end position or the undetermined node answer probability is less than the preset threshold, then determining that the current voice information does not contain the current node answer, and re-acquiring voice information for intent recognition.

[0051] S2. Determine whether the current node is configured with a future node question. If not, end the intent recognition of the current node. If so, use the intent recognition model to sequentially perform intent recognition on the current voice information based on the future node question to obtain the future node answer.

[0052] Specifically, the current node is configured with the corresponding issues for other nodes in the future based on the business characteristics.

[0053] Specifically, the step of obtaining the corresponding future node answer by performing intent recognition on the current speech information based on each future node question using the intent recognition model includes the following processing on the current speech information and each future node question:

[0054] The current voice information and future node questions are concatenated according to the preset dictionary to obtain the concatenated text, and each character of the concatenated text is mapped to the corresponding ID to obtain the corresponding number string of the concatenated text.

[0055] The number string is input into the intention recognition model to obtain the start sequence and the end sequence, and the start sequence and the end sequence are converted into start probability and end probability through a softmax layer;

[0056] Use the position of the maximum initial probability as the starting position and the position of the maximum ending probability as the ending position, and determine whether the starting position is before the ending position;

[0057] If so, the answer to the undetermined node is obtained from the voice information based on the start and end positions, and the product of the start probability and the end probability is taken as the probability of the answer to the undetermined node. The probability of the answer to the undetermined node that is greater than the preset threshold is taken as the answer to the future node.

[0058] If the starting position is after the ending position or the probability of the answer to the pending node is less than a preset threshold, then it is determined that the voice information does not contain the answer to the future node question.

[0059] S3. Select a subsequent node based on the current node answer, and determine whether the node question of the subsequent node is included in the future node question. If not, reacquire the user's voice information for the node question of the subsequent node. If yes, obtain the pending node answer corresponding to the node question of the subsequent node from the future node answers. Confirm with the user through the communication terminal whether the pending node answer is the answer of the subsequent node. If not, reacquire the user's voice information for the node question of the subsequent node. If yes, use the pending node answer as the answer of the subsequent node and continue to select a subsequent node until the current question and answer node has no subsequent node or the communication with the communication terminal ends and the question and answering stops.

[0060] Specifically, if the node question of the subsequent node is in the future node question and the future node answer does not contain the node answer corresponding to the node question of the subsequent node, then the user's voice information of the subsequent node is obtained based on the node question of the subsequent node, and the node answer of the subsequent node is obtained by performing intent recognition on the voice information of the subsequent node based on the node question of the subsequent node.

[0061] Specifically, after the question-and-answer session ends, a question-and-answer document is generated based on the node questions and node answers corresponding to each node in this question-and-answer session and stored in a preset question-and-answer database for subsequent querying and use.

[0062] As can be seen from the above embodiments, the question-and-answer intent recognition method proposed in this invention configures future node questions for each node in the session jump logic, and obtains the node answer corresponding to the current node and the node answer corresponding to the future node question based on the intent recognition model, thereby avoiding repeated questioning of subsequent node questions and improving the user experience when conducting intelligent question-and-answer with outbound call robots.

[0063] like Figure 2 The diagram shown is a schematic representation of an embodiment of the electronic device of the present invention. The electronic device 1 is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions. The electronic device 1 can be a computer, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing, which is a type of distributed computing consisting of a super virtual computer composed of a group of loosely coupled computers.

[0064] In this embodiment, the electronic device 1 includes, but is not limited to, a memory 11, a processor 12, and a network interface 13 that can be interconnected via a system bus. The memory 11 stores a question-and-answer intent recognition program 10, which can be executed by the processor 12. Figure 3 Only the electronic device 1 with components 11-13 and the question-and-answer intent recognition program 10 is shown. Those skilled in the art will understand that... Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0065] The memory 11 includes RAM and at least one type of readable storage medium. The RAM provides a cache for the operation of the electronic device 1; the readable storage medium can be volatile or non-volatile. Specifically, the readable storage medium can be a storage medium such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the readable storage medium can be an internal storage unit of the electronic device 1, such as the hard disk of the electronic device 1; in other embodiments, the storage medium can also be an external storage device of the electronic device 1, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 1. In this embodiment, the readable storage medium of the memory 11 mainly includes a program storage area and a data storage area. The program storage area is typically used to store the operating system and various application software installed on the electronic device 1, such as the code of the question-and-answer intent recognition program 10 in one embodiment of the present invention. The data storage area can store data created based on the use of blockchain nodes, such as various types of data that have been output or will be output.

[0066] In some embodiments, processor 12 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. This processor 12 is typically used to control the overall operation of the electronic device 1, such as performing control and processing related to data interaction or communication with other devices. In this embodiment, the processor 12 is used to run program code stored in the memory 11 or process data, such as running a question-and-answer intent recognition program 10.

[0067] The network interface 13 may include a wireless network interface or a wired network interface, which is used to establish a communication connection between the electronic device 1 and the client (not shown in the figure).

[0068] Optionally, the electronic device 1 may further include a user interface, which may include a display, an input unit such as a keyboard, and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an organic light-emitting diode (OLED) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device 1 and to display a visual user interface.

[0069] In one embodiment of the present invention, when the question-and-answer intent recognition program 10 is executed by the processor 12, it performs the following steps S1-S3.

[0070] S1. Obtain the current voice information input by the user at the communication terminal based on the node question of the current node, and obtain the current node answer by performing intent recognition on the current voice information based on the node question of the current node through the intent recognition model.

[0071] S2. Determine whether the current node is configured with a future node question. If not, end the intent recognition of the current node. If so, use the intent recognition model to sequentially perform intent recognition on the current voice information based on the future node question to obtain the future node answer.

[0072] S3. Select a subsequent node based on the current node answer, and determine whether the node question of the subsequent node is included in the future node question. If not, reacquire the user's voice information for the node question of the subsequent node. If yes, obtain the pending node answer corresponding to the node question of the subsequent node from the future node answers. Confirm with the user through the communication terminal whether the pending node answer is the answer of the subsequent node. If not, reacquire the user's voice information for the node question of the subsequent node. If yes, use the pending node answer as the answer of the subsequent node and continue to select a subsequent node until the current question and answer node has no subsequent node or the communication with the communication terminal ends and the question and answering stops.

[0073] The specific operational steps implemented by the above steps S1-S3 are largely the same as those in an embodiment of the question-and-answer intent recognition method of the present invention, and will not be repeated here.

[0074] In other embodiments, the question-and-answer intent recognition program 10 may be divided into one or more modules, which are stored in the memory 11 and executed by one or more processors (processor 12 in this embodiment) to complete the present invention. The module referred to in the present invention refers to a series of computer program instruction segments that can perform specific functions and are used to describe the execution process of the question-and-answer intent recognition program 10 in the electronic device 1.

[0075] like Figure 3 The diagram shown is a schematic representation of an embodiment of the question-and-answer intent recognition program of the present invention.

[0076] In a first embodiment of the present invention, the question-and-answer intent recognition program 10 includes a current node answer acquisition module 110, a future node answer acquisition module 120, and a subsequent node answer acquisition module 130, exemplarily:

[0077] The current node answer acquisition module 110 is used to acquire the current voice information input by the user on the communication terminal based on the node question of the current node, and to obtain the current node answer by performing intent recognition on the current voice information based on the node question of the current node through an intent recognition model.

[0078] The future node answer acquisition module 120 is used to determine whether the current node is configured with a future node question. If not, the intent recognition of the current node ends; if so, the intent recognition model sequentially performs intent recognition on the current voice information based on the future node question to obtain the future node answer.

[0079] The subsequent node answer acquisition module 130 is used to select a subsequent node based on the current node answer, determine whether the node question of the subsequent node is included in the future node question, if not, reacquire the user's voice information for the node question of the subsequent node, if so, obtain the pending node answer corresponding to the node question of the subsequent node from the future node answers; confirm with the user through the communication terminal whether the pending node answer is the subsequent node answer, if not, reacquire the user's voice information for the node question of the subsequent node, if so, use the pending node answer as the subsequent node answer and continue to select a subsequent node until the current question and answer node has no subsequent node or the communication with the communication terminal ends and the question and answer stops.

[0080] The specific operation steps implemented by the current node answer acquisition module 110, the future node answer acquisition module 120, and the subsequent node answer acquisition module 130 when they are executed are largely the same as those in the above embodiments, and will not be repeated here.

[0081] Furthermore, this embodiment of the invention also proposes a computer-readable storage medium, which can be volatile or non-volatile. Specifically, the computer-readable storage medium can be any one or any combination of several of the following: hard disk, multimedia card, SD card, flash memory card, SMC, read-only memory (ROM), erasable programmable read-only memory (EPROM), portable compact disc read-only memory (CD-ROM), USB memory, etc. The computer-readable storage medium stores a question-and-answer intent recognition program 10.

[0082] When the question-answering intent recognition program 10 is executed by the processor, it performs the following operations:

[0083] A1. Current node answer acquisition module, used to acquire the current voice information input by the user on the communication terminal based on the node question of the current node, and to obtain the current node answer by performing intent recognition on the current voice information based on the node question of the current node through an intent recognition model;

[0084] A2. The future node answer acquisition module is used to determine whether the current node is configured with a future node question. If not, the intent recognition of the current node ends; if so, the intent recognition model is used to sequentially perform intent recognition on the current voice information based on the future node question to obtain the future node answer.

[0085] A3. The subsequent node answer acquisition module is used to select a subsequent node based on the current node answer, determine whether the node question of the subsequent node is included in the future node question, if not, reacquire the user's voice information for the node question of the subsequent node, if so, obtain the pending node answer corresponding to the node question of the subsequent node from the future node answers; confirm with the user through the communication terminal whether the pending node answer is the subsequent node answer, if not, reacquire the user's voice information for the node question of the subsequent node, if so, use the pending node answer as the subsequent node answer and continue to select a subsequent node until the current question and answer node has no subsequent node or the communication with the communication terminal ends and the question and answer stops.

[0086] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0087] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, question-and-answer intent recognition device, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, question-and-answer intent recognition device, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, question-and-answer intent recognition device, article, or method that includes that element.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0089] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A question-and-answer intent recognition method, applied to a question-and-answer intent recognition device, characterized in that, The method includes: Based on the node question of the current node, the current voice information input by the user on the communication terminal is obtained, and the current node answer is obtained by performing intent recognition on the current voice information based on the node question of the current node. Determine whether the current node is configured with a future node question; otherwise, end the intent recognition of the current node; if so, use the intent recognition model to sequentially perform intent recognition on the current voice information based on the future node question to obtain the future node answer. Based on the current node answer, a subsequent node is selected. It is determined whether the node question of the subsequent node is included in the future node question. If not, the user's voice information for the node question of the subsequent node is re-acquired. If so, the pending node answer corresponding to the node question of the subsequent node is obtained from the future node answers. The user is then confirmed via the communication terminal whether the pending node answer is the answer of the subsequent node. If not, the user's voice information for the node question of the subsequent node is re-acquired. If so, the pending node answer is used as the answer of the subsequent node, and the selection of a subsequent node continues until the current question and answer node has no subsequent node or the communication with the communication terminal ends, at which point the question and answer process stops.

2. The question-and-answer intent recognition method as described in claim 1, characterized in that, The step of obtaining the current node answer by performing intent recognition on the current speech information based on the node question of the current node using an intent recognition model includes: The current voice information is converted into text information and concatenated with the node question of the current node, and then converted into a number string and input into the intent recognition model; Obtain the output of the intent recognition model, and obtain the answer for the current node from the current speech information based on the output.

3. The question-and-answer intent recognition method as described in claim 2, characterized in that, The step of converting the current speech information into text information, concatenating it with the node question of the current node, and then converting it into a number string for input into the intent recognition model includes: Convert the current voice information into text information; Obtain the preset dictionary corresponding to the intent recognition model, and concatenate the text information and the node question of the current node according to the preset dictionary to obtain the concatenated text; The pre-defined dictionary maps each character of the concatenated text to its corresponding ID to obtain the corresponding numeric string. The number string is input into the intent recognition model.

4. The question-and-answer intent recognition method as described in claim 3, characterized in that, The preset dictionary includes a "start character" for identifying the beginning of a sentence, a "end character" for identifying the end of a sentence, and a numeric ID corresponding to each character; The method for concatenating the text information and the node problem of the current node according to the preset dictionary to obtain the concatenated text is as follows: Concatenate text="start character" + text information + "end character" + node problem + "end character".

5. The question-and-answer intent recognition method as described in claim 2, characterized in that, The step of obtaining the output result of the intent recognition model includes: The number string is input into the intention recognition model to obtain the start sequence and the end sequence, and the start sequence and the end sequence are converted into start probability and end probability through the softmax layer of the intention recognition model; The starting position is determined by taking the position corresponding to the maximum value of the initial probability, and the ending position is determined by taking the position corresponding to the maximum value of the ending probability. The initial probability includes the probability that each character of the concatenated text is the starting position of the answer, and the ending probability includes the probability that each character of the concatenated text is the ending position of the answer.

6. The question-and-answer intent recognition method as described in claim 5, characterized in that, Based on the output, the answer for the current node is obtained from the current speech information, including: If the starting position is before the ending position, the answer to the undetermined node is obtained from the current voice information based on the starting position and the ending position, and the product of the starting probability and the ending probability is taken as the probability of the answer to the undetermined node. The probability of the answer to the undetermined node that is greater than a preset threshold is taken as the answer to the current node. If the starting position is after the ending position or the probability of the answer to the pending node is less than a preset threshold, it is determined that the current voice information does not contain the answer to the current node, and the voice information is reacquired for intent recognition.

7. The question-and-answer intent recognition method as described in claim 1, characterized in that, The method further includes: If the node question of the subsequent node is included in the future node question and the answer of the future node does not include the node answer corresponding to the node question of the subsequent node, then the user's voice information of the subsequent node is obtained based on the node question of the subsequent node, and the node answer of the subsequent node is obtained by performing intent recognition on the voice information of the subsequent node based on the node question of the subsequent node.

8. A question-and-answer intent recognition device, characterized in that, The question-and-answer intent recognition device includes: The current node answer acquisition module is used to acquire the current voice information input by the user on the communication terminal based on the node question of the current node, and to obtain the current node answer by performing intent recognition on the current voice information based on the node question of the current node. The future node answer acquisition module is used to determine whether the current node is configured with a future node question. If not, the intent recognition of the current node ends; if so, the intent recognition model sequentially performs intent recognition on the current voice information based on the future node question to obtain the future node answer. The subsequent node answer acquisition module is used to select a subsequent node based on the current node answer, determine whether the node question of the subsequent node is included in the future node questions, if not, reacquire the user's voice information for the node question of the subsequent node, if so, obtain the pending node answer corresponding to the node question of the subsequent node from the future node answers; confirm with the user through the communication terminal whether the pending node answer is the subsequent node answer, if not, reacquire the user's voice information for the node question of the subsequent node, if so, use the pending node answer as the subsequent node answer and continue to select a subsequent node until the current question and answer node has no subsequent node or the communication with the communication terminal ends and the question and answer stops.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the question-answering intent recognition method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the question-answering intent recognition method as described in any one of claims 1 to 7.

Images