Intelligent response method and device, electronic equipment and storage medium
Through the proposed intelligent response method, through intent analysis, feature extraction and compliance evaluation of user questions, the problem of possible learning bias during training of intelligent response models is solved, achieving higher replies compliance and user experience.
Patent Information
- Application Number
- CN202411587258.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-07
- Publication Date
- 2025-02-28
AI Technical Summary
When using data containing biases during training, the intelligent response model may learn and amplify these biases, resulting in unfair decision-making and cannot ensure that the responses received by the user comply with relevant specifications and industry standards, which may cause customer losses.
An intelligent response method is proposed, by obtaining the question text data of the target user, performing intent analysis, generating initial reply text data, extracting reply characteristics, using a preset reply evaluation model for compliance evaluation, performing text modification to improve compliance, and ultimately providing answers to users.
Through this method, the compliance of intelligent replies can be improved, ensuring that replies comply with relevant specifications and industry standards, and reducing the risk of customer losses.
Smart Images
Figure CN119537538A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of text processing technology, applicable to the field of financial technology, and in particular to an intelligent response method and device, an electronic device and a storage medium. Background Art
[0002] Intelligent answering technology can be used to automatically reply to the questioner, thereby responding to the questions or needs asked. For example, in the financial field, intelligent answering technology can be used to quickly respond to users' insurance product inquiries and quickly resolve users' consultation issues.
[0003] At present, smart answering technology usually uses smart answering models to respond to users' questions. However, when the smart answering model is being trained, if the training data contains biases, the smart answering model may learn and amplify these biases, leading to unfair decisions. It cannot guarantee that the answers users receive comply with relevant specifications and industry standards, which can easily cause customer losses. Therefore, how to improve the compliance of smart answers has become a technical problem that needs to be solved urgently. Summary of the invention
[0004] The main purpose of the embodiments of the present application is to propose an intelligent response method and device, an electronic device and a storage medium, aiming to improve the compliance of intelligent replies.
[0005] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes an intelligent response method, the method comprising:
[0006] Obtain the target user's question text data;
[0007] Performing intent analysis on the question text data to obtain text intent;
[0008] Providing a text reply to the question text data based on the text intention to obtain initial reply text data;
[0009] Extracting features from the initial reply text data to obtain reply features;
[0010] Based on the response characteristics and a preset response evaluation model, the response compliance evaluation is performed on the initial response text data to obtain target evaluation data;
[0011] Based on the target evaluation data, modify the initial reply text data to obtain target reply text data;
[0012] Based on the target answer text data, the question is answered to the target user.
[0013] In some embodiments, extracting features from the initial reply text data to obtain reply features includes:
[0014] Performing word segmentation processing on the initial reply text data to obtain reply phrases;
[0015] Based on the initial reply text data, context understanding is performed on the reply phrase to obtain phrase context information;
[0016] Based on the phrase context information, keywords are extracted from the reply phrase to obtain the reply features.
[0017] In some embodiments, the reply features include behavioral features and background features, and the reply compliance evaluation is performed on the initial reply text data based on the reply features and a preset reply evaluation model to obtain target evaluation data, including:
[0018] Performing behavior prediction based on the background features to obtain target behavior features;
[0019] Based on the response evaluation model, the behavior feature is compared with the target behavior feature to obtain first evaluation data;
[0020] Based on the background features, predicting the response of the behavior features to obtain predicted response;
[0021] Based on the reply evaluation model and the predicted response, performing feature evaluation on the initial reply text data to obtain second evaluation data;
[0022] The first evaluation data and the second evaluation data are merged to obtain the target evaluation data.
[0023] In some embodiments, the step of modifying the initial reply text data based on the target evaluation data to obtain the target reply text data includes:
[0024] Performing data analysis on the target evaluation data to obtain feature evaluation data, wherein the feature evaluation data is used to represent evaluation data of the behavior feature;
[0025] Based on the feature evaluation data, the initial reply text data is phrase-adjusted to obtain the target reply text data.
[0026] In some embodiments, the phrase adjustment of the initial reply text data based on the feature evaluation data to obtain the target reply text data includes:
[0027] Based on the feature evaluation data, feature adjustment is performed on the behavior feature to obtain a corrected behavior feature;
[0028] Based on the correction behavior characteristics and the background characteristics, a text answer is generated for the question text data to obtain the target answer text data.
[0029] In some embodiments, providing an intelligent response to the target user based on the target reply text data includes:
[0030] Binding the target answer text data with the question text data to obtain a dialogue text;
[0031] Based on the conversation text, a text is transmitted to the target user to obtain a reply text.
[0032] In some embodiments, after providing an intelligent response to the target user based on the target reply text data, the method further includes:
[0033] Acquiring user response data transmitted by the target user based on the reply text;
[0034] Based on the user response, the parameters of the reply evaluation model are adjusted to obtain a target evaluation model.
[0035] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides an intelligent response device, the device comprising:
[0036] A question text acquisition module is used to obtain question text data of target users;
[0037] A text intent analysis module is used to perform intent analysis on the question text data to obtain text intent;
[0038] A text preliminary reply module, used for making a text reply to the question text data based on the text intention to obtain initial reply text data;
[0039] A reply feature extraction module, used for extracting features from the initial reply text data to obtain reply features;
[0040] A reply feature evaluation module, used to perform reply compliance evaluation on the initial reply text data based on the reply feature and a preset reply evaluation model to obtain target evaluation data;
[0041] A reply text modification module, used to modify the initial reply text data based on the target evaluation data to obtain target reply text data;
[0042] The question text reply module is used to provide an intelligent reply to the target user based on the target reply text data.
[0043] To achieve the above objectives, a third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in the first aspect is implemented.
[0044] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.
[0045] The intelligent answering method and device, electronic device and storage medium proposed in the present application obtain the question text data of the target user, then perform intent analysis on the question text data to obtain the text intent, and perform a text reply to the question text data based on the text intent to obtain initial reply text data, so that the initial reply text can solve the problem raised in the question text data, thereby improving the accuracy of the intelligent answering question. Secondly, feature extraction is performed on the initial reply text data to obtain reply features, and based on the reply features and a pre-trained reply evaluation model, a reply compliance evaluation is performed on the initial reply text data to obtain target evaluation data, and then based on the target evaluation data, the initial reply text data is modified to obtain target reply text data, so that the target reply text data is more standardized. Finally, based on the target reply text data, a text reply is made to the target user, which can improve the compliance of the reply text. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 is a flow chart of the intelligent response method provided in an embodiment of the present application;
[0047] Figure 2 yes Figure 1 Flow chart of step S104 in FIG.
[0048] Figure 3 yes Figure 1 Flow chart of step S105 in FIG.
[0049] Figure 4 yes Figure 1 Flow chart of step S106 in FIG.
[0050] Figure 5 yes Figure 4 Flow chart of step S402 in FIG.
[0051] Figure 6 yes Figure 1 Flow chart of step S107 in FIG.
[0052] Figure 7 is a flow chart of an intelligent response method provided by another embodiment of the present application;
[0053] Figure 8 is a schematic diagram of the structure of the intelligent response device provided in an embodiment of the present application;
[0054] Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION
[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0056] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.
[0058] First, some nouns involved in this application are analyzed:
[0059] Smart Response: Smart response refers to a system that uses artificial intelligence technology, especially natural language processing and machine learning, to automatically understand and respond to user queries or instructions. This system can simulate human conversations, provide real-time information, answer questions or perform tasks, aiming to improve efficiency and enhance user experience, and is used in many fields such as customer service, personal assistants, education and entertainment. The smart response system analyzes user input, identifies intent, and generates appropriate responses to interact with users.
[0060] Intelligent answering technology can be used to automatically reply to the questioner, thereby responding to the questions or needs asked. For example, in the financial field, intelligent answering technology can be used to quickly respond to users' insurance product inquiries and quickly resolve users' consultation issues.
[0061] At present, smart answering technology usually uses smart answering models to respond to users' questions. However, when the smart answering model is being trained, if the training data contains biases, the smart answering model may learn and amplify these biases, leading to unfair decisions. It cannot guarantee that the answers users receive comply with relevant specifications and industry standards, which can easily cause customer losses. Therefore, how to improve the compliance of smart answers has become a technical problem that needs to be solved urgently.
[0062] Based on this, the embodiments of the present application provide an intelligent response method and device, an electronic device and a storage medium, aiming to improve the compliance of intelligent replies.
[0063] The intelligent response method and device, electronic device and storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the intelligent response method in the embodiments of the present application is described.
[0064] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0065] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0066] The intelligent response method provided in the embodiment of the present application relates to the field of text processing technology and is applicable to the field of financial technology. The intelligent response method provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the intelligent response method, etc., but is not limited to the above forms.
[0067] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.
[0068] Figure 1 is an optional flow chart of the intelligent response method provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S107.
[0069] Step S101, obtaining question text data;
[0070] Step S102, performing intent analysis on the question text data to obtain the text intent;
[0071] Step S103, providing a text reply to the question text data based on the text intent to obtain initial reply text data;
[0072] Step S104, extracting features from the initial reply text data to obtain reply features;
[0073] Step S105, based on the reply characteristics and the preset reply evaluation model, the initial reply text data is evaluated for reply compliance to obtain target evaluation data;
[0074] Step S106, based on the target evaluation data, modify the initial reply text data to obtain target reply text data;
[0075] Step S107, answering the question text data based on the target answer text data.
[0076] Steps S101 to S107 shown in the embodiment of the present application are to perform intent analysis on the acquired target user's question text data to obtain the text intent, and then perform a text reply to the question text data according to the text intent to obtain initial reply text data. Furthermore, feature extraction is performed on the initial reply text data to obtain reply features, and then, based on a preset reply evaluation model and the reply features, a reply compliance assessment is performed on the initial reply text data to obtain target evaluation data. Secondly, based on the target evaluation data, the initial reply text data is modified to obtain target reply text data. Finally, based on the target reply text data, an intelligent reply is given to the target user to make the intelligent reply more standardized and improve the user's consulting experience. Therefore, the present application obtains the target user's question text data, then performs intent analysis on the question text data to obtain the text intent, and makes a text reply to the question text data based on the text intent to obtain initial reply text data, so that the initial reply text can solve the problem raised in the question text data, thereby improving the accuracy of the intelligent answering question. Secondly, feature extraction is performed on the initial reply text data to obtain reply features, and based on the reply features and a pre-trained reply evaluation model, a reply compliance evaluation is performed on the initial reply text data to obtain target evaluation data. Based on the target evaluation data, the initial reply text data is modified to obtain target reply text data, so that the target reply text data is more standardized. Finally, based on the target reply text data, a text reply is made to the target user, which can improve the compliance of the reply text.
[0077] In step S101 of some embodiments, the target user may be a user who has questions to consult, for example, in an insurance sales scenario, the target user may be a user who wants to know the product features of an insurance product. The question text data may be textual data of the question consulted by the target user.
[0078] The present application can obtain question voice data by recording when the target user asks a question, and then convert the question voice data into text to obtain question text data, which can improve the speed at which the target user can conduct question consultation.
[0079] In step S102 of some embodiments, the grammatical structure of the question text data can be parsed to obtain the text grammatical structure, and then the question text data can be segmented according to the text grammatical structure to obtain the question phrases. Secondly, the question phrases can be understood to obtain the phrase meanings. Finally, according to the phrase meanings and the text grammatical structure, the question text data can be contextually understood to obtain the user intention.
[0080] This application parses the question text data to obtain the text intent, which facilitates the understanding of the questions in the question text data, thereby improving the accuracy of the reply.
[0081] In step S103 of some embodiments, entity extraction is performed on the question text data according to the text intent to obtain entity information, and then a reply text is generated according to the entity information. For example, in the field of financial consulting, the question text data of the target user is "How is the stock market of Company A today?" By performing intent analysis on the question text data, the text intent can be obtained: asking about the stock market. According to the text intent, entity extraction is performed on the question text data to obtain the enterprise entity (Company A) and the time entity (today), so that it can be clearly determined that the information to be found is the stock market of Company A today. Finally, the queried stock market of Company A today is output.
[0082] In step S104 of some embodiments, the reply features include behavioral features and background features, wherein the behavioral features refer to the behavioral features implied by the initial reply text data. For example, in an insurance product consultation scenario, the behavioral features in the initial reply text data may be a recommendation of insurance product B or an introduction to insurance product B, and the background features may be the basic background of the target user himself. For example, in a loan consultation scenario, features such as stable income and large expenditure may be background features.
[0083] The present application can obtain reply phrases by performing word segmentation processing on the initial reply text data, and then understand the meaning of the reply phrases in combination with the context information of the reply phrases in the initial reply text data to obtain the meaning of the phrases, and then perform feature screening based on the meaning of the phrases to obtain reply features.
[0084] For details, see Figure 2 In some embodiments, step S104 may include but is not limited to steps S201 to S203:
[0085] Step S201, performing word segmentation processing on the initial reply text data to obtain reply phrases;
[0086] Step S202, based on the initial reply text data, context understanding is performed on the reply phrase to obtain phrase context information;
[0087] Step S203, based on the phrase context information, keyword extraction is performed on the reply phrase to obtain reply features.
[0088] In step S201 of some embodiments, word segmentation processing can divide continuous initial reply text data into meaningful reply phrases. Specifically, the present application can use a regular word segmenter, a machine learning model, or a deep learning model to identify the word boundaries of each phrase in the initial reply text data, thereby implementing word segmentation processing of the initial reply text data to obtain reply phrases.
[0089] In step S202 of some embodiments, context understanding refers to understanding the meaning of each response phrase in the initial response text data, specifically including semantic role, emotional color, and relationship with other response phrases.
[0090] The present application can understand the complex relationship between each reply phrase by performing dependency syntactic parsing or semantic role labeling on the reply phrase, thereby obtaining the phrase context information of the reply phrase.
[0091] In step S203 of some embodiments, based on the phrase context information, the meaning expressed by each reply phrase can be clarified, and based on the meaning, it can be determined whether the selection conditions of the reply feature are met, and phrase extraction is performed on the reply phrases that meet the selection conditions of the reply feature to obtain the reply feature.
[0092] In steps S201 to S203 of some embodiments, the initial reply text data is segmented to obtain reply phrases. Secondly, the reply phrases are contextually understood based on the initial reply text data to obtain phrase context information. Finally, based on the phrase context information, keywords are extracted from the reply phrases to obtain reply features. This can improve the accuracy of reply feature extraction, thereby facilitating subsequent reply compliance assessment of the initial reply text data.
[0093] In step S105 of some embodiments, the response evaluation model may be a pre-fitted response evaluation function, such as the evaluation formula of consequentialist ethics, the evaluation formula of duty ethics, the evaluation formula of virtue ethics, etc.
[0094] The present application can predict target behavior features that meet compliance based on background features in the reply features, and then perform feature evaluation on the behavior features in the reply features based on the target behavior features and the above-mentioned reply evaluation model, thereby obtaining first evaluation data, and then predict the reverse reaction of the target user in combination with the behavior features and background features in the reply features to obtain predicted response, and based on the predicted response, perform feature evaluation on the initial reply text data to obtain second evaluation data, and finally, combine the above-mentioned first evaluation data and second evaluation data to obtain target evaluation data for the initial reply text data.
[0095] For details, see Figure 3In some embodiments, step S105 may include but is not limited to steps S301 to S305:
[0096] Step S301, predicting behavior based on background features to obtain target behavior features;
[0097] Step S302, based on the response evaluation model, comparing the behavior characteristics with the target behavior characteristics to obtain first evaluation data;
[0098] Step S303, based on the background features, predicting the response of the behavior features to obtain predicted response;
[0099] Step S304, based on the reply evaluation model and the predicted response, feature evaluation is performed on the initial reply text data to obtain second evaluation data;
[0100] Step S305: merge the first evaluation data and the second evaluation data to obtain target evaluation data.
[0101] In step S301 of some embodiments, the decision consequence prediction can be performed on the combination of background features and possible behavioral features according to the evaluation formula of consequentialist ethics to obtain the decision consequence generated by the combination of each behavioral feature and the background feature, and then the target behavioral feature can be screened out from the behavioral features based on the decision consequence.
[0102] In step S302 of some embodiments, the above-mentioned response evaluation model can be used to calculate the decision consequences of the combination of behavioral characteristics and background characteristics, and then the decision consequences can be compared with the target decision consequences generated by the combination of the above-mentioned target behavioral characteristics and background characteristics, so as to achieve feature evaluation of the behavioral characteristics and obtain first evaluation data. Alternatively, the similarity between the behavioral characteristics and the target behavioral characteristics can be calculated, and the behavioral characteristics can be feature evaluated based on the similarity to obtain the first evaluation data.
[0103] In step S303 of some embodiments, a reply response prediction model may be constructed in advance based on historical reply data, and then the background features and behavioral features in the initial reply text data may be input into the reply response prediction model to obtain the predicted response of the target user based on the behavioral features.
[0104] In step S304 of some embodiments, according to the evaluation formula in the above-mentioned prediction response and reply evaluation model, the second evaluation data of the initial reply text under the predicted response can be calculated.
[0105] In step S305 of some embodiments, the target evaluation data of the initial reply text data may be obtained by performing a weighted summation on the first evaluation data and the second evaluation data.
[0106] In steps S301 to S305 of some embodiments, behavior prediction is performed based on background characteristics to obtain target behavior characteristics, and then the behavior characteristics are compared with the target behavior characteristics based on a reply evaluation model to obtain first evaluation data. Furthermore, response prediction is performed on the behavior characteristics based on the background characteristics to obtain predicted response. Next, based on the reply evaluation model and the predicted response, feature evaluation is performed on the initial reply text data to obtain second evaluation data. Finally, the first evaluation data and the second evaluation data are merged to obtain target evaluation data, so that the target evaluation data is more comprehensive, thereby facilitating improving the quality of the reply.
[0107] In step S106 of some embodiments, the target evaluation data is reversely parsed to extract characteristic evaluation data representing the behavioral characteristics, and then the phrases or texts in the initial reply text data are adjusted based on the characteristic evaluation data to obtain the target reply text data.
[0108] For details, see Figure 4 In some embodiments, step S106 may include but is not limited to steps S401 to S402:
[0109] Step S401, performing data analysis on the target evaluation data to obtain feature evaluation data, wherein the feature evaluation data is used to represent evaluation data of behavior characteristics;
[0110] Step S402, based on the feature evaluation data, the initial reply text data is phrase-adjusted to obtain the target reply text data.
[0111] In step S401 of some embodiments, feature evaluation data representing evaluation data of behavioral characteristics may be obtained by reverse parsing the target evaluation data, or feature evaluation data may be obtained by querying the first reply feature evaluation data generated in the above step.
[0112] In step S402 of some embodiments, based on the above-mentioned feature evaluation data, feature search is performed on the initial reply text data to obtain low-evaluation features that need to be modified, and feature adjustment is performed on the low-evaluation features to achieve phrase adjustment of the initial reply text data to obtain target reply text data.
[0113] For details, see Figure 5 In some embodiments, step S402 may include but is not limited to steps S501 to S502:
[0114] Step S501, based on the feature evaluation data, feature adjustment is performed on the behavior feature to obtain a corrected behavior feature;
[0115] Step S502, based on the correction behavior characteristics and background characteristics, generate a text reply to the question text data to obtain target reply text data.
[0116] In step S501 of some embodiments, the feature evaluation data is further analyzed to identify behavioral features with lower evaluation data, i.e., low-evaluation behavioral features. The low-evaluation behavioral features are then feature adjusted to obtain corrected behavioral features.
[0117] In step S502 of some embodiments, a new reply text is generated according to the correction behavior features and the background features in the reply features, namely, the target reply text data.
[0118] In steps S501 to S502 of some embodiments, feature adjustments are made to behavioral features based on feature evaluation data to obtain corrected behavioral features, and then text replies are generated for question text data based on the corrected behavioral features and background features to obtain target reply text data, which can improve the target evaluation data of the reply text and enhance the consulting experience of the target user.
[0119] In steps S401 to S402 of some embodiments, characteristic evaluation data is obtained by performing data analysis on the target evaluation data, wherein the characteristic evaluation data is used to represent evaluation data of behavioral characteristics. Then, based on the characteristic evaluation data, the initial reply text data is phrase-adjusted to obtain target reply text data, which can make the target reply text data more compliant with relevant specifications and industry standards.
[0120] In step S107 of some embodiments, after obtaining the target reply text data, the target reply text data can be bound to the target user's question text data so that the target user can quickly identify the question answered by the target reply text data, thereby improving the target user's consulting experience.
[0121] For details, see Figure 6 In some embodiments, step S107 may include but is not limited to steps S601 to S602:
[0122] Step S601, text binding the target answer text data with the question text data to obtain a dialogue text;
[0123] Step S602: Based on the conversation text, a text is transmitted to the target user to obtain a reply text.
[0124] In step S601 of some embodiments, text binding may be performed by referencing question text data in the target reply text data to obtain dialogue text, or the target reply text data and question text data may be placed in the same frame to perform text binding to form dialogue text.
[0125] In step S602 of some embodiments, the above-mentioned conversation text can be transmitted to the terminal used by the target user through data transmission, and then the conversation text is visualized to obtain the reply text.
[0126] The conversation text can be displayed to the target user, and the target user's satisfaction with the conversation text can be collected to obtain user feedback.
[0127] In steps S601 to S602 of some embodiments, the target reply text data is bound to the question text data to obtain a dialogue text, and then the dialogue text is transmitted to the target user to obtain a reply text, so that the reply text is more in line with relevant specifications and industry standards, thereby improving the user's consulting experience.
[0128] It should be noted that this application can also obtain the target user's satisfaction with the reply text, and then adjust the target evaluation model, so that the subsequently generated reply text is more in line with relevant specifications and industry standards.
[0129] For details, see Figure 7 In some embodiments, after step S107, the intelligent answering method may further include but is not limited to steps S701 to S702:
[0130] Step S701, obtaining user response data transmitted by the target user based on the reply text;
[0131] Step S702: Based on user feedback, adjust the parameters of the response evaluation model to obtain a target evaluation model.
[0132] In step S701 of some embodiments, after receiving the reply text, the target user can evaluate the satisfaction of the reply text and transmit the satisfaction evaluation data to the replying terminal, thereby obtaining the user response data of the target user.
[0133] In step S702 of some embodiments, the user response data is analyzed to determine whether it is a positive response, and when the user response data is a negative response, the parameters of the reply evaluation model are adjusted to make the evaluation performance of the adjusted target evaluation model better, thereby facilitating the generation of a reply text that satisfies the target user.
[0134] In steps S701 to S702 of some embodiments, by obtaining user response data transmitted by the target user based on the reply text, and then adjusting the parameters of the reply evaluation model according to the user response, a target evaluation model is obtained, which can improve the model performance and further improve the compliance of the intelligent reply.
[0135] The present application obtains the target user's question text data, then performs intent analysis on the question text data to obtain the text intent, and performs a text reply to the question text data based on the text intent to obtain initial reply text data, so that the initial reply text can solve the problem raised in the question text data, thereby improving the accuracy of the intelligent answering question. Secondly, feature extraction is performed on the initial reply text data to obtain reply features, and based on the reply features and a pre-trained reply evaluation model, a reply compliance evaluation is performed on the initial reply text data to obtain target evaluation data. Based on the target evaluation data, the initial reply text data is modified to obtain target reply text data, so that the target reply text data is more standardized. Finally, based on the target reply text data, a text reply is made to the target user, which can improve the compliance of the reply text.
[0136] See also Figure 8 The embodiment of the present application also provides an intelligent response device, which can implement the above-mentioned intelligent response method, and the device includes:
[0137] The question text acquisition module 801 is used to acquire the question text data of the target user;
[0138] The text intention analysis module 802 is used to perform intention analysis on the question text data to obtain the text intention;
[0139] A text preliminary reply module 803 is used to reply to the question text data based on the text intention to obtain initial reply text data;
[0140] The reply feature extraction module 804 is used to extract features from the initial reply text data to obtain reply features;
[0141] A reply feature evaluation module 805 is used to perform reply compliance evaluation on the initial reply text data based on the reply feature and a preset reply evaluation model to obtain target evaluation data;
[0142] A reply text modification module 806 is used to modify the initial reply text data based on the target evaluation data to obtain target reply text data;
[0143] The question text reply module 807 is used to provide intelligent responses to target users based on target reply text data.
[0144] The specific implementation of the intelligent response device is basically the same as the specific implementation of the above-mentioned intelligent response method, and will not be repeated here.
[0145] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned intelligent response method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.
[0146] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:
[0147] The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;
[0148] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the intelligent response method of the embodiment of this application;
[0149] Input / output interface 903, used to implement information input and output;
[0150] Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized by wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);
[0151] A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);
[0152] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .
[0153] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the above-mentioned intelligent response method is implemented.
[0154] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0155] The intelligent response method, intelligent response device, electronic device and storage medium provided in the embodiments of the present application perform intent analysis on the acquired target user's question text data to obtain the text intent, and then perform text response to the question text data according to the text intent to obtain initial response text data. Furthermore, feature extraction is performed on the initial response text data to obtain response features, and then response compliance is evaluated on the initial response text data according to a preset response evaluation model and the response features to obtain target evaluation data. Secondly, according to the target evaluation data, the initial response text data is modified to obtain target response text data. Finally, according to the target response text data, an intelligent response is performed to the target user, so as to make the intelligent response more standardized and improve the user's consulting experience.
[0156] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.
[0157] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.
[0158] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0159] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.
[0160] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0161] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.
[0162] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0163] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0164] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0165] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.
[0166] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.
Claims
1. An intelligent answering method, characterized in that: The method comprises: Obtain the target user's question text data; Performing intent analysis on the question text data to obtain text intent; Providing a text reply to the question text data based on the text intention to obtain initial reply text data; Extracting features from the initial reply text data to obtain reply features; Based on the response characteristics and a preset response evaluation model, the response compliance evaluation is performed on the initial response text data to obtain target evaluation data; Based on the target evaluation data, modify the initial reply text data to obtain target reply text data; Based on the target answer text data, the question is answered to the target user.
2. The method according to claim 1, characterized in that The step of extracting features from the initial reply text data to obtain reply features includes: Performing word segmentation processing on the initial reply text data to obtain reply phrases; Based on the initial reply text data, context understanding is performed on the reply phrase to obtain phrase context information; Based on the phrase context information, keywords are extracted from the reply phrase to obtain the reply features.
3. The method according to claim 1, characterized in that The reply features include behavioral features and background features. Based on the reply features and a preset reply evaluation model, the reply compliance evaluation is performed on the initial reply text data to obtain target evaluation data, including: Performing behavior prediction based on the background features to obtain target behavior features; Based on the response evaluation model, the behavior feature is compared with the target behavior feature to obtain first evaluation data; Based on the background features, predicting the response of the behavior features to obtain predicted response; Based on the reply evaluation model and the predicted response, performing feature evaluation on the initial reply text data to obtain second evaluation data; The first evaluation data and the second evaluation data are merged to obtain the target evaluation data.
4. The method according to claim 3, characterized in that The step of modifying the initial reply text data based on the target evaluation data to obtain the target reply text data comprises: Performing data analysis on the target evaluation data to obtain feature evaluation data, wherein the feature evaluation data is used to represent evaluation data of the behavior feature; Based on the feature evaluation data, the initial reply text data is phrase-adjusted to obtain the target reply text data.
5. The method according to claim 4, characterized in that The step of adjusting the initial reply text data by phrases based on the feature evaluation data to obtain the target reply text data comprises: Based on the feature evaluation data, feature adjustment is performed on the behavior feature to obtain a corrected behavior feature; Based on the correction behavior characteristics and the background characteristics, a text answer is generated for the question text data to obtain the target answer text data.
6. The method according to any one of claims 1 to 5, characterized in that: The step of providing an intelligent response to the target user based on the target reply text data comprises: Binding the target answer text data with the question text data to obtain a dialogue text; Based on the conversation text, a text is transmitted to the target user to obtain a reply text.
7. The method according to claim 6, characterized in that After providing an intelligent response to the target user based on the target reply text data, the method further includes: Acquiring user response data transmitted by the target user based on the reply text; Based on the user response, the parameters of the reply evaluation model are adjusted to obtain a target evaluation model.
8. An intelligent answering device, characterized in that: The device comprises: A question text acquisition module is used to obtain question text data of target users; A text intent analysis module is used to perform intent analysis on the question text data to obtain text intent; A text preliminary reply module, used for making a text reply to the question text data based on the text intention to obtain initial reply text data; A reply feature extraction module, used for extracting features from the initial reply text data to obtain reply features; A reply feature evaluation module, used to perform reply compliance evaluation on the initial reply text data based on the reply feature and a preset reply evaluation model to obtain target evaluation data; A reply text modification module, used to modify the initial reply text data based on the target evaluation data to obtain target reply text data; The question text reply module is used to provide an intelligent reply to the target user based on the target reply text data.
9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the intelligent answering method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the intelligent response method according to any one of claims 1 to 7 is implemented.