Intelligent response method and device and storage medium

Abstract

The invention discloses an intelligent response method and device and a storage medium. The intelligent response method comprises the steps of obtaining similarity between an input question of a user and each standard question in a question and answer knowledge base; the dynamic threshold value of each standard question is obtained, a candidate question set is determined according to the dynamic threshold value and the similarity, and the dynamic threshold value dynamically changes according to the response feedback condition of each standard question; and determining a target question from the candidate question set, and taking a standard answer corresponding to the target question as response information of the input question. The dynamic threshold value is determined according to the response feedback condition of each standard question, the target question is determined by referring to the dynamic threshold value of each standard question, and the dynamic threshold value matched with each standard question is adopted for judgment aiming at each standard question, so that the problem that a scene is limited when a fixed threshold value is adopted is avoided; and each dynamic threshold is determined by performing dynamic change according to the response feedback condition of each standard question, so that the accuracy of the dynamic threshold is ensured.

Description

Technical Field

[0001] This invention relates to the field of financial technology, and in particular to an intelligent response method, device, and storage medium. Background Technology

[0002] Currently, intelligent customer service systems are widely used in the financial sector to provide users with self-service business consultation and business processing services. The intelligent customer service system retrieves the target question from the question-and-answer knowledge base based on the user's input question and provides the user with the standard answer corresponding to the target question.

[0003] Currently, when identifying target questions from the question-and-answer knowledge base, a uniform fixed threshold is usually set globally. This threshold is typically determined manually by operations personnel based on their experience. Using manual threshold determination reduces the accuracy of the threshold, and applying a uniform fixed threshold to various standard questions in different question-and-answer knowledge bases can affect the accuracy of intelligent responses. Summary of the Invention

[0004] This invention provides an intelligent response method, apparatus, and storage medium to achieve accurate and intelligent responses to user-input questions.

[0005] According to a first aspect of this invention, an intelligent response method is provided, comprising: obtaining the similarity between a user's input question and standard questions in a question-and-answer knowledge base, wherein the question-and-answer knowledge base includes multiple standard questions and standard answers corresponding to each of the standard questions;

[0006] The dynamic threshold of each standard question is obtained, and a candidate question set is determined based on the dynamic threshold and the similarity, wherein the dynamic threshold changes dynamically according to the response feedback of each standard question;

[0007] The target question is determined from the set of candidate questions, and the standard answer corresponding to the target question is used as the response information for the input question.

[0008] According to another aspect of the present invention, an intelligent response device is provided, the device comprising: a similarity acquisition module, configured to acquire the similarity between a user's input question and standard questions in a question-and-answer knowledge base, wherein the question-and-answer knowledge base includes multiple standard questions and standard answers corresponding to each of the standard questions;

[0009] The candidate question set determination module is used to obtain the dynamic threshold of each standard question and determine the candidate question set based on the dynamic threshold and the similarity, wherein the dynamic threshold changes dynamically according to the response feedback of each standard question;

[0010] The response information acquisition module is used to determine the target question from the candidate question set and use the standard answer corresponding to the target question as the response information of the input question.

[0011] According to another aspect of the present invention, a terminal device is provided, the terminal device comprising: one or more processors;

[0012] Storage device for storing one or more programs.

[0013] When the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any embodiment of the present invention.

[0014] According to another aspect of the present invention, a storage medium for computer-executable instructions is provided, on which a computer program is stored, which, when executed by a processor, implements the method as described in any of the embodiments of the present invention.

[0015] The technical solution of this invention determines a dynamic threshold based on the response feedback of each standard question, determines the target question by referring to the dynamic threshold of each standard question, and makes a judgment on each standard question using a dynamic threshold that matches itself, thereby avoiding the problem of limited scenarios when using a fixed threshold. Furthermore, each dynamic threshold is dynamically determined based on the response feedback of each standard question, thus ensuring the accuracy of the dynamic threshold.

[0016] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of an intelligent response method provided according to Embodiment 1 of the present invention;

[0019] Figure 2 This is a flowchart of an intelligent response method provided according to Embodiment 2 of the present invention;

[0020] Figure 3 This is a schematic diagram of the structure of an intelligent response device according to Embodiment 3 of the present invention;

[0021] Figure 4 This is a structural block diagram of a terminal device provided in Embodiment 4 of the present invention. Specific Implementation

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be used interchangeably where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or terminal device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or terminal devices. Moreover, the information collected in this embodiment is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, necessary confidentiality measures have been taken, public order and good morals have not been violated, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0024] Example 1

[0025] Figure 1 This is a flowchart illustrating an intelligent response method provided in an embodiment of the present invention. This embodiment is applicable to situations where intelligent responses are provided to user-input questions. The method can be executed by an intelligent response device, which can be implemented in hardware and / or software, and can be integrated into a terminal device. Figure 1 As shown, the method includes:

[0026] Step S101: Obtain the similarity between the user's input question and each standard question in the question-answering knowledge base.

[0027] Optionally, the similarity between the user's input question and each standard question in the question-answering knowledge base is obtained, including: obtaining a first vector of the input question and a second vector of the standard questions; calculating the cosine similarity between the first vector and the second vector, and using the cosine similarity as the similarity between the input question and the standard questions.

[0028] Specifically, the question-and-answer knowledge base includes multiple standard questions and corresponding standard answers. Standard questions are designed for specific business scenarios, and their design is based on the collection of actual user questions within those scenarios. For example, a standard question might be "Why is my account locked when transferring funds?" or "How do I perform a transfer?" This is merely an example and does not limit the specific content of the standard questions. Each standard question in this implementation has a corresponding standard answer, which primarily addresses the questions posed in the standard questions. In this implementation, the standard answer is mainly used in intelligent response scenarios to identify user intent. Intent identification here refers to the process of determining the correspondence between the user's input question and the standard answers in the question-and-answer knowledge base.

[0029] In this embodiment, when identifying users, the main task is to obtain the similarity between the user's input question and each standard question. When calculating the question similarity, the main task is to obtain the first vector of the input question. Specifically, the input question is segmented into words to obtain word vectors, and the average vector of all word vectors is calculated. This average vector is then used as the first vector of the input question. Similarly, the second vector of the standard question is obtained, and the cosine similarity between the first and second vectors is calculated. This cosine similarity is then used as the similarity between the input question and the standard question. The result is calculated with a value between [0,1]. The higher the score, the closer the meaning of the two texts is. Of course, this embodiment only uses cosine similarity as an example for explanation and does not limit the specific calculation method of question similarity.

[0030] Step S102: Obtain the dynamic threshold of each standard question, and determine the candidate question set based on the dynamic threshold and similarity.

[0031] Optionally, obtaining the dynamic threshold for each standard question includes: obtaining the number of positive and negative feedbacks from the user's historical inputs for the standard answer corresponding to each standard question through a periodic scheduler, and determining the total number of feedbacks based on the number of positive and negative feedbacks; determining the smoothing ratio based on the number of positive and negative feedbacks and the smoothing parameter, and calculating the confidence weight based on the smoothing ratio and the total number of feedbacks; calculating the threshold adjustment amount based on the confidence weight and the smoothing ratio, and adjusting the current threshold based on the threshold adjustment amount to obtain the dynamic threshold.

[0032] Optionally, after adjusting the current threshold according to the threshold adjustment amount to obtain the dynamic threshold, the method further includes: obtaining the dynamic threshold standard range; detecting the dynamic threshold, and generating an alarm prompt when the dynamic threshold exceeds the dynamic threshold standard range.

[0033] Specifically, in this implementation, a corresponding dynamic threshold is obtained for each standard answer. The dynamic threshold can be calculated weekly using a periodic scheduler. For example, within a statistical period, for standard question i, the number of positive feedback entries from the user's historical inputs is obtained for the standard answer corresponding to standard question i using the periodic scheduler. and the number of negative feedback The number of positive feedbacks indicates the number of times the intelligent feedback for standard question i meets the user's needs, while the number of negative feedbacks indicates the number of times the intelligent feedback for standard question i does not meet the user's needs. The total number of feedbacks is determined based on the number of positive and negative feedbacks. In addition, this embodiment also determines the smoothing ratio by the number of positive feedbacks, the number of negative feedbacks, and the smoothing parameter, specifically by setting... Where α and β are smoothing parameters, the smoothing ratio is: Of course, this embodiment is merely an example and does not limit the specific method for determining the smoothing ratio. In this embodiment, the smoothing ratio can be based on... Total number of feedback Calculate confidence weights, for example, confidence weights Where C is a constant, the threshold adjustment can be calculated based on the confidence weights and smoothing ratios obtained above. Where k is the step size, which is a constant, and 0.5 is the baseline for satisfaction. but Lower the threshold to improve recall; if The threshold is then increased to suppress false hits. Given a current threshold of t_i, the current threshold can be adjusted based on the threshold adjustment amount obtained above to obtain a dynamic threshold. For example, the dynamic threshold can be updated as follows: That is, t_i is updated to Meanwhile, the threshold needs to be within Within the range, Since the lower and upper limits of the threshold are preset, the dynamic threshold in this embodiment changes dynamically according to the response feedback of each standard question. Of course, this embodiment is only an example and does not limit the specific adjustment method of the dynamic threshold.

[0034] It should be noted that while dynamic threshold adjustment is constrained by upper and lower limits, problems may arise due to software or hardware malfunctions. Therefore, this embodiment sets a standard range for dynamic thresholds. If the final obtained dynamic threshold exceeds this standard range, it is determined that the obtained dynamic threshold is incorrect, and an alarm will be generated to alert maintenance personnel that the current dynamic threshold adjustment is incorrect. This allows maintenance personnel to make timely adjustments to the software or hardware based on the alert information. Of course, this embodiment is only an example and does not limit the specific method of dynamic threshold detection.

[0035] Optionally, a candidate question set can be determined based on a dynamic threshold and similarity, including: taking standard questions with a similarity greater than their own dynamic threshold as candidate questions; and constructing a candidate question set based on all candidate questions.

[0036] Specifically, in this embodiment, given that the dynamic thresholds of each standard problem and its similarity to the input problem are known, the obtained similarity for each standard problem is compared with its own dynamic threshold. For example, the dynamic threshold 1 for standard problem 1 is 0.6, the dynamic threshold 2 for standard problem 2 is 0.8, and the dynamic threshold 3 for standard problem 3 is 0.6. When the similarity between standard problem 1 and the input problem is 0.7, the similarity between standard problem 2 and the input problem is 0.7, and the similarity between standard problem 3 and the input problem is 0.7, standard problem 1 is selected as a candidate problem because its similarity is greater than its own dynamic threshold. Standard problem 2 is not selected as a candidate problem because its similarity is less than its own dynamic threshold. Standard problem 3 is selected as a candidate problem because its similarity is greater than its own dynamic threshold. Thus, although standard problem 1 and standard problem 2 have the same similarity, because their corresponding dynamic thresholds are different, only standard problem 1 is retained as a candidate problem. Therefore, the constructed candidate problem set can be determined as X{standard problem 1}. Standard question 3}, of course, this embodiment is only an example and does not limit the number of standard questions included in the candidate question set. Therefore, it can be seen that in this embodiment, when determining candidate questions, not all standard questions are judged using a uniform and fixed threshold, but rather the initial judgment of user intent is based on the dynamic threshold of each standard question itself. This allows for the identification of user intent from multiple angles and dimensions, making the identification results more accurate.

[0037] Step S103: Determine the target question from the candidate question set, and use the standard answer corresponding to the target question as the response information for the input question.

[0038] Optionally, the target problem is determined from the candidate problem set, including: determining whether the number of candidate problems contained in the candidate problem set is one; if so, the candidate problem is directly used as the target problem; otherwise, the candidate problem with the highest similarity is used as the target problem.

[0039] Specifically, in this embodiment, after obtaining a candidate question set through preliminary screening based on the dynamic thresholds of each standard question, the target question can be further determined from the candidate question set. For example, when the candidate question set contains only one candidate question, it is directly used as the target question to accurately identify the user's intent; while when the candidate question set contains multiple candidate questions, the candidate question with the highest similarity is directly used as the target question. Therefore, in this embodiment, the determination of the target question mainly involves two screenings: first, preliminary screening of candidate questions from the question-and-answer knowledge base based on dynamic thresholds; and second, determining the target question from the candidate question set based on the similarity. The first screening ensures the comprehensiveness of the screening, while the second screening ensures the accuracy of the screening. Of course, this embodiment is only an example and does not limit the specific method of determining the target question.

[0040] The technical solution of this invention determines a dynamic threshold based on the response feedback of each standard question, determines the target question by referring to the dynamic threshold of each standard question, and makes a judgment on each standard question using a dynamic threshold that matches itself. This avoids the problem of limited scenarios when using a fixed threshold. Furthermore, each dynamic threshold is dynamically determined based on the response feedback of each standard question, thereby ensuring the accuracy of the dynamic threshold.

[0041] Example 2

[0042] Figure 2 This is a flowchart of an intelligent response method provided by an embodiment of the present invention. Based on the above embodiments, this embodiment, after using the standard answer corresponding to the target question as the response information for the input question, further includes: displaying the standard answer as response information on a human-computer interaction interface; receiving user feedback on the standard answer; and writing the feedback result into the response feedback table of the periodic scheduler. Figure 2 As shown, the method includes:

[0043] Step S201: Obtain the similarity between the user's input question and each standard question in the question-answering knowledge base.

[0044] Optionally, the similarity between the user's input question and each standard question in the question-answering knowledge base is obtained, including: obtaining a first vector of the input question and a second vector of the standard questions; calculating the cosine similarity between the first vector and the second vector, and using the cosine similarity as the similarity between the input question and the standard questions.

[0045] Step S202: Obtain the dynamic threshold of each standard question, and determine the candidate question set based on the dynamic threshold and similarity.

[0046] Optionally, obtaining the dynamic threshold for each standard question includes: obtaining the number of positive and negative feedbacks from the user's historical inputs for the standard answer corresponding to each standard question through a periodic scheduler, and determining the total number of feedbacks based on the number of positive and negative feedbacks; determining the smoothing ratio based on the number of positive and negative feedbacks and the smoothing parameter, and calculating the confidence weight based on the smoothing ratio and the total number of feedbacks; calculating the threshold adjustment amount based on the confidence weight and the smoothing ratio, and adjusting the current threshold based on the threshold adjustment amount to obtain the dynamic threshold.

[0047] Optionally, after adjusting the current threshold according to the threshold adjustment amount to obtain the dynamic threshold, the method further includes: obtaining the dynamic threshold standard range; detecting the dynamic threshold, and generating an alarm prompt when the dynamic threshold exceeds the dynamic threshold standard range.

[0048] Optionally, a candidate question set can be determined based on a dynamic threshold and similarity, including: taking standard questions with a similarity greater than their own dynamic threshold as candidate questions; and constructing a candidate question set based on all candidate questions.

[0049] Step S203: Determine the target question from the candidate question set, and use the standard answer corresponding to the target question as the response information for the input question.

[0050] Optionally, the target problem is determined from the candidate problem set, including: determining whether the number of candidate problems contained in the candidate problem set is one; if so, the candidate problem is directly used as the target problem; otherwise, the candidate problem with the highest similarity is used as the target problem.

[0051] Step S204: Display the standard answer as response information on the human-computer interaction interface, receive the user's feedback on the standard answer, and write the feedback result into the response feedback table of the periodic scheduler.

[0052] Specifically, in this implementation, after identifying the target question from the question-and-answer knowledge base, the corresponding standard answer is displayed on the human-computer interaction interface. The interface also includes a comment entry point for the standard answer, such as a pop-up window or a button, allowing users to input their feedback. Users can provide feedback on the standard answer through this comment entry point. When the comment entry point is a pop-up window, users can indicate whether the displayed standard answer answers their question. If the user indicates a successful answer, the corresponding number of positive feedback entries for that standard answer is increased. 1. When the user inputs "unsuccessful answer," the negative feedback count for the corresponding standard answer is incremented by 1. Similarly, when the comment entry is in the form of buttons, the displayed buttons include two buttons: a "successful answer" button and a "unsuccessful answer" button. Users can select different buttons according to their satisfaction with the standard answer. When the user selects the "successful answer" button, the positive feedback count for the corresponding standard answer is incremented by 1. When the user selects the "unsuccessful answer" button, the negative feedback count for the corresponding standard answer is incremented by 1. Of course, this implementation is only an example and does not limit the specific form of user feedback.

[0053] It should be noted that in this implementation, after receiving the user's feedback, the feedback result will be written to the response feedback table of the periodic scheduler. The response feedback table contains the feedback results corresponding to each standard answer in the question-and-answer knowledge base, as shown in Table 1 below as an example of the response feedback table:

[0054]

[0055] Whenever a feedback result is received from the front-end interface, the number of positive and negative feedback results corresponding to the standard answer will be updated accordingly. Of course, Table 1 is only used as an example with two standard answers, and in actual applications, there is no limit to the specific number of standard answers contained in the question-and-answer knowledge base.

[0056] It's worth noting that when a negative feedback result is received, it indicates that the current standard answer does not meet the user's needs and fails to resolve their doubts. In this case, human assistance will be automatically triggered. Simultaneously, the user's input question will be displayed to the human customer service representative to avoid repetitive statements. This allows for a quick and efficient response from a human representative when the user connects. Therefore, this implementation dynamically updates the intent recognition threshold at the question granularity, improving customer experience; it eliminates the need for lengthy retraining cycles, resulting in faster operational response; and it reduces frequent retraining and large-scale annotation, lowering maintenance costs.

[0057] The technical solution of this invention determines a dynamic threshold based on the response feedback of each standard question, determines the target question by referring to the dynamic threshold of each standard question, and makes a judgment on each standard question using a dynamic threshold that matches itself. This avoids the problem of limited scenarios when using a fixed threshold. Furthermore, each dynamic threshold is dynamically determined based on the response feedback of each standard question, thereby ensuring the accuracy of the dynamic threshold.

[0058] Example 3

[0059] Figure 3 This is a schematic diagram of the structure of an intelligent response device provided in an embodiment of the present invention. Figure 3 As shown, the device includes: a similarity acquisition module 310, a question set determination module 320, and a response information acquisition module 330.

[0060] The similarity acquisition module 310 is used to acquire the similarity between the user's input question and each standard question in the question-answering knowledge base. The question-answering knowledge base includes multiple standard questions and standard answers corresponding to each standard question.

[0061] The candidate question set determination module 320 is used to obtain the dynamic threshold of each standard question and determine the candidate question set based on the dynamic threshold and similarity. The dynamic threshold changes dynamically according to the response feedback of each standard question.

[0062] The response information acquisition module 330 is used to determine the target question from the candidate question set and use the standard answer corresponding to the target question as the response information of the input question.

[0063] Optionally, a similarity acquisition module is used to obtain the first vector of the input question and the second vector of the standard question;

[0064] Calculate the cosine similarity between the first and second vectors, and use the cosine similarity as the similarity between the input problem and the standard problem.

[0065] Optionally, the candidate question set determination module includes a dynamic threshold determination unit, which is used to obtain the number of positive and negative feedbacks of the user's historical input through a periodic scheduler for the standard answer corresponding to each standard question, and determine the total number of feedbacks based on the number of positive and negative feedbacks.

[0066] The smoothing ratio is determined based on the number of positive feedbacks, the number of negative feedbacks, and the smoothing parameter, and the confidence weight is calculated based on the smoothing ratio and the total number of feedbacks.

[0067] The threshold adjustment amount is calculated based on the confidence weight and smoothing ratio, and the current threshold is adjusted according to the threshold adjustment amount to obtain a dynamic threshold.

[0068] Optionally, the device also includes a dynamic threshold detection module for obtaining a dynamic threshold standard range;

[0069] The system detects dynamic thresholds and generates an alarm when a dynamic threshold exceeds the standard range.

[0070] Optionally, the candidate question set determination module includes a candidate question set determination unit, which is used to select standard questions that have a similarity greater than their own dynamic threshold as candidate questions;

[0071] A candidate question set is constructed based on all candidate questions.

[0072] Optionally, the response information acquisition module is used to determine whether the number of candidate questions contained in the candidate question set is one. If so, the candidate question is directly used as the target question.

[0073] Otherwise, the candidate problem with the highest similarity is selected as the target problem.

[0074] Optionally, the device also includes a display module for displaying the standard answer as response information on the human-computer interaction interface;

[0075] Receive user feedback on the standard answer and write the feedback to the response feedback table of the periodic scheduler.

[0076] The intelligent response device provided in this embodiment of the invention can execute an intelligent response method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0077] Example 4

[0078] Figure 4 A schematic diagram of a terminal device 10 that can be used to implement embodiments of the present invention is shown. The terminal device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The terminal device can also represent various forms of mobile devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0079] The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or claimed herein.

[0080] like Figure 4As shown, the terminal device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer programs stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the terminal device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0081] Multiple components in terminal device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows terminal device 10 to exchange information / data with other terminal devices through computer networks such as the Internet and / or various telecommunications networks.

[0082] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as intelligent response methods.

[0083] In some embodiments, the smart response method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on terminal device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the smart response method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to execute the smart response method by any other suitable means (e.g., by means of firmware).

[0084] Various embodiments of the apparatuses and techniques described above herein can be implemented in digital electronic circuit devices, integrated circuit devices, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), device-on-a-chip (SoCs), complex programmable logic terminal devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable device including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage device, at least one input device, and at least one output device, and transmitting data and instructions to the storage device, the at least one input device, and the at least one output device.

[0085] Computer programs used to implement the intelligent response method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other non-stop data migration device, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on the machine, partially on the machine, or as a standalone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0086] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution apparatus, device, or terminal device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage terminal devices, magnetic storage terminal devices, or any suitable combination thereof.

[0087] To provide interaction with a user, the apparatus and techniques described herein can be implemented on a terminal device having: a display device (e.g., a touchscreen) for displaying information to the user; and buttons through which the user can provide input to the terminal device. Other types of apparatus can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or haptic feedback); and input from the user can be received in any form (including voice input, speech input, or haptic input).

[0088] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0089] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A smart response method, characterized in that, The method includes: The similarity between a user's input question and standard questions in a question-and-answer knowledge base is obtained, wherein the question-and-answer knowledge base includes multiple standard questions and standard answers corresponding to each standard question; The dynamic threshold of each standard question is obtained, and a candidate question set is determined based on the dynamic threshold and the similarity, wherein the dynamic threshold changes dynamically according to the response feedback of each standard question; The target question is determined from the set of candidate questions, and the standard answer corresponding to the target question is used as the response information for the input question.

2. The method according to claim 1, characterized in that, The process of obtaining the similarity between the user's input question and each standard question in the question-answering knowledge base includes: Obtain the first vector of the input problem and the second vector of the standard problem; Calculate the cosine similarity between the first vector and the second vector, and use the cosine similarity as the similarity between the input problem and the standard problem.

3. The method according to claim 1, characterized in that, The process of obtaining the dynamic thresholds for each standard problem includes: For each standard question, the system obtains the number of positive and negative feedbacks from the user's historical inputs through a periodic scheduler, and determines the total number of feedbacks based on the number of positive and negative feedbacks. The smoothing ratio is determined based on the number of positive feedbacks, the number of negative feedbacks, and the smoothing parameter, and the confidence weight is calculated based on the smoothing ratio and the total number of feedbacks. The threshold adjustment amount is calculated based on the confidence weight and the smoothing ratio, and the current threshold is adjusted according to the threshold adjustment amount to obtain the dynamic threshold.

4. The method according to claim 3, characterized in that, After adjusting the current threshold according to the threshold adjustment amount to obtain the dynamic threshold, the method further includes: Obtain the standard range of dynamic thresholds; The dynamic threshold is detected, and an alarm is generated when the dynamic threshold exceeds the standard range of the dynamic threshold.

5. The method according to claim 1, characterized in that, The step of determining the candidate question set based on the dynamic threshold and the similarity includes: Standard questions whose similarity to the input question is greater than their own dynamic threshold are selected as candidate questions. The candidate question set is constructed based on all the candidate questions.

6. The method according to claim 1, characterized in that, The step of determining the target problem from the candidate problem set includes: Determine whether the number of candidate problems contained in the candidate problem set is one. If so, directly use the candidate problem as the target problem. Otherwise, the candidate problem with the highest similarity is selected as the target problem.

7. The method according to any one of claims 1 to 6, characterized in that, After using the standard answer corresponding to the target question as the response information for the input question, the method further includes: The standard answer will be displayed as response information on the human-computer interaction interface; Receive user feedback on the standard answer and write the feedback to the response feedback table of the periodic scheduler.

8. A smart response device, characterized in that, The device includes: The similarity acquisition module is used to acquire the similarity between the user's input question and each standard question in the question-answering knowledge base, wherein the question-answering knowledge base includes multiple standard questions and standard answers corresponding to each standard question; The candidate question set determination module is used to obtain the dynamic threshold of each standard question and determine the candidate question set based on the dynamic threshold and the similarity, wherein the dynamic threshold changes dynamically according to the response feedback of each standard question; The response information acquisition module is used to determine the target question from the candidate question set and use the standard answer corresponding to the target question as the response information of the input question.

9. A terminal device, characterized in that, The terminal device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

10. A storage medium for computer-executable instructions, wherein a computer program is stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images