Intelligent interaction method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of interaction processing, and provides an intelligent interaction method and device, electronic equipment and a storage medium. The method comprises the following steps: in response to interaction information sent by a user for a target object, performing feature extraction on the interaction information to obtain key features corresponding to the interaction information; obtaining knowledge data corresponding to the target object according to the key features; determining an information generation model according to the key features, inputting the key features and the knowledge data into the information generation model, and enabling the information generation model to generate response information corresponding to the interaction information according to the key features and the knowledge data; and interacting with the user through the response information. The problem that in the prior art, the interaction process is lack of personalization is solved.

Description

Technical Field

[0001] The present application relates to the field of interactive processing technology, and in particular to an intelligent interactive method, device, electronic device and storage medium. Background Art

[0002] The real estate brokerage industry has long relied on manual work for customer communication, market analysis, document processing, etc. At present, the real estate economic industry has introduced intelligence and applied it to customer communication and market analysis, but these technologies are often not mature enough. For example, in terms of customer communication, market analysis, document processing, service standards, etc., they still rely on manual services. These tasks are tedious and time-consuming. Traditional real estate display services based on manual services are also relatively limited and cannot fully display the characteristics and advantages of real estate. The customer experience is limited, resulting in the lack of personalization in the interactive tasks of existing technologies. Summary of the invention

[0003] In view of this, the present application provides an intelligent interaction method, device, electronic device and storage medium to solve the problem of lack of personalization in the interaction process in the prior art.

[0004] The first aspect of the present application provides an intelligent interaction method, including: in response to interaction information sent by a user to a target object, extracting features of the interaction information to obtain key features corresponding to the interaction information; acquiring knowledge data corresponding to the target object based on the key features; determining an information generation model based on the key features, inputting the key features and the knowledge data into the information generation model, so that the information generation model generates response information corresponding to the interaction information based on the key features and the knowledge data; and interacting with the user through the response information.

[0005] According to a second aspect of the present application, an intelligent interaction device is provided, comprising: a feature extraction module, configured to extract features of the interaction information in response to the interaction information sent by the user to the target object, and obtain key features corresponding to the interaction information; a data acquisition module, configured to obtain knowledge data corresponding to the target object according to the key features; a response module, configured to determine an information generation model according to the key features, and input the key features and the knowledge data into the information generation model, so that the information generation model generates response information corresponding to the interaction information according to the key features and the knowledge data; and an interaction module, configured to interact with the user through the response information.

[0006] According to a third aspect of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0007] According to a fourth aspect of the present application, a storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0008] At least one of the above technical solutions adopted in this application can achieve the following beneficial effects:

[0009] By responding to the interactive information sent by the user to the target object, the interactive information is feature extracted to obtain the key features corresponding to the interactive information. The key features can represent the key information in the interactive information, and then the knowledge data corresponding to the target object is obtained according to the key features. The knowledge data corresponding to the target object in the corresponding professional knowledge field can be extracted from the interactive information to provide support for the subsequent information generation. After obtaining the key features and knowledge data in the interactive information input by the user, the system can select an information generation model suitable for the current interactive task, input the key features and the corresponding knowledge data into the information generation model, and the information generation model generates the corresponding response information. The response information can be based on user needs and is targeted. Finally, the response information generated by the system is returned to the user to complete the interactive task. In this way, personalized response information can be generated by obtaining interactive information and determining the information generation model, so as to provide users with more accurate and more personal services, which greatly enhances the personalization ability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0011] Figure 1 It is a flowchart of an intelligent interaction method provided in an embodiment of the present application;

[0012] Figure 2 is a flowchart of another intelligent interaction method provided in an embodiment of the present application;

[0013] Figure 3 It is a flowchart of another intelligent interaction method provided in an embodiment of the present application;

[0014] Figure 4 is a structural diagram of an intelligent interactive device provided in an embodiment of the present application;

[0015] Figure 5 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0017] The terms "first", "second", etc. in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0018] In addition, it should be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "includes..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0019] In the real estate brokerage industry, brokers need to handle a large amount of real estate information and communicate with customers. Traditional methods include manually writing texts such as real estate introductions, transaction negotiations and transaction processes. These tasks are cumbersome and time-consuming. The existing real estate brokerage industry has introduced some automated processes, but there are some technical defects, such as low efficiency in solving real estate economic industry problems, insufficient personalization of customer service, cumbersome real estate document processing, low standards of real estate economic services, and a single way of communicating with customers. Therefore, the purpose of the present invention is to provide intelligent real estate economic services by integrating professional knowledge and data in the field of real estate brokerage into a large language model, and to provide comprehensive, accurate, efficient, convenient, personalized and interactive real estate economic services through vertical field knowledge integration, animated customer communication, and interactive display of real estate and other functions. It should be understood that the embodiments of the present application are mainly based on intelligent brokers, but this solution can be applied to any other intelligent interaction scenarios, such as intelligent retrieval services, educational counseling and other fields.

[0020] An intelligent interaction method and device according to an embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0021] Figure 1 is a flow chart of an intelligent interaction method provided by an embodiment of the present application. Figure 1 As shown, the intelligent interaction method includes:

[0022] S101, in response to interaction information sent by a user to a target object, extracting features of the interaction information to obtain key features corresponding to the interaction information;

[0023] S102, acquiring knowledge data corresponding to the target object according to the key features;

[0024] S103, determining an information generation model according to the key features, inputting the key features and knowledge data into the information generation model, so that the information generation model generates response information corresponding to the interactive information according to the key features and knowledge data;

[0025] S104, interacting with the user through response information.

[0026] Specifically, the intelligent interactive system can first extract the most critical features for the problem or need from the interactive information provided by the user for the target object. These key features can be explicit, such as the needs, interests, preferences clearly expressed by the user, etc., or they can be invisible, such as inferring potential needs by analyzing the user's expectations, wording, and areas of frequent concern. The target object can be the main object in the current application field of the intelligent interactive system, such as real estate information in the real estate field, that is, information about the interactive information sent by the user for the real estate information. As an example: the user sends an interactive message "I want to find a three-bedroom apartment close to the school with a budget of about 3 million", the intelligent interactive system can extract key features from it: "close to school", "three bedrooms", "budget 3 million", etc.

[0027] Furthermore, obtaining knowledge data corresponding to the target object based on key features can represent obtaining knowledge data corresponding to user needs from the intelligent interactive system. The knowledge data can represent professional knowledge in the real estate field, such as housing price range and housing supply status corresponding to user needs.

[0028] Furthermore, determining the information generation model based on key features can be understood as different information generation models may correspond to different interaction scenarios and user needs, so as to solve the current task in a targeted manner. For example, the user interaction information mentions: view the VR image of a certain room type. In this case, the intelligent interaction system may not be able to complete the operation of generating a VR image, and it is necessary to generate a VR image by selecting an auxiliary enhancement tool (i.e., the information generation model). This can expand the capabilities of the intelligent interaction system and increase the personalization of the system. For example, when the user needs to view the housing price trend, the chart generation model can be selected as the information generation model to generate the housing price trend corresponding to the user's required time period and room type and other information.

[0029] Afterwards, based on the determined information generation model, targeted response information for user needs is generated, which can answer user questions or provide valuable suggestions, and the response information is returned to the user. The process of interacting with the user can be a single round, for example, the system gives a one-time answer to the user's interactive information, or it can be a multi-round dialogue, for example, the user can further screen the properties based on the aforementioned questions, consult for details, etc., there is no restriction here.

[0030] According to the solution provided in the above embodiment, by acquiring the key features of the interactive information, acquiring knowledge data and determining the information generation model based on the key features, and further generating response information for user needs based on the key features and knowledge data, a personalized response service is provided to enhance the user experience.

[0031] In some embodiments, knowledge data corresponding to the target object is obtained based on key features, including: constructing an initial knowledge model, acquiring professional knowledge to train the initial knowledge model to obtain a target knowledge model, wherein the professional knowledge represents the knowledge in the field to which the target object belongs; inputting key features into the target knowledge model to extract data to obtain knowledge data corresponding to the target object.

[0032] Specifically, knowledge data can be extracted through the target knowledge model, that is, the target knowledge model is used to analyze key features, identify user needs, and extract knowledge data based on user needs.

[0033] The target knowledge model is obtained by constructing an initial knowledge model and training it. The initial knowledge model may include a knowledge base or data set, covering basic information and data structures in the corresponding field, such as property types, market prices, geographical locations, home purchase processes and other professional knowledge in the real estate field.

[0034] The initial knowledge model constructed can be similar to a framework that has not been personalized or optimized. Training it can make it more targeted and have the ability to gather and extract knowledge. Through optimization based on professional knowledge, the knowledge structure and rules of a specific field can be accurately reflected, and more relevant and accurate outputs can be generated for the key features of user input to handle complex market analysis, user preferences and other professional needs, providing support for subsequent knowledge extraction.

[0035] By inputting the key features into the trained target knowledge model for data extraction and analysis, the most relevant knowledge data of the key features can be obtained. These knowledge data can represent the core content of providing personalized services to users, such as generating VR charts or trend charts for each house data, related information for house recommendations, etc.

[0036] According to the solution provided by the above embodiment, by constructing and training the knowledge model, the system can effectively extract knowledge data closely related to user needs, so that the system can filter out the most valuable information from massive data and provide users with accurate and personalized services.

[0037] In some embodiments, an initial knowledge model is constructed, and professional knowledge is acquired to train the initial knowledge model to obtain a target knowledge model, including: constructing an initial knowledge model, tuning the hyperparameters of the initial knowledge model to obtain a tuned initial knowledge model; acquiring professional knowledge, and obtaining a training set, a validation set, and a test set based on the professional knowledge; using the training set to train the tuned initial knowledge model, and using the validation set to adjust the hyperparameters to obtain a trained initial knowledge model; using the test set to test the indicators of the trained initial knowledge model to obtain a target knowledge model, the indicators including the accuracy and recall of the trained initial knowledge model.

[0038] Specifically, the entire training process of the initial knowledge model can be understood as: based on the basic professional knowledge and data structure of the corresponding field, the initial knowledge model is constructed, and the hyperparameters of the initial knowledge model are tuned. The performance of the model is improved by adjusting the hyperparameters (learning rate, regularization parameters, etc.). Parameter tuning can be achieved through grid search, random search or other optimization algorithms. In the construction of the initial knowledge model in the real estate field, the tuning can involve adjusting the parameters in the house price prediction to improve the accuracy and stability of the prediction, which is not limited here. The tuned initial knowledge model can process data more effectively and make data analysis, that is, the initial knowledge model has language processing capabilities. Dividing the acquired professional knowledge into training sets, validation sets and test sets to train the model can enable the target knowledge model to generate accurate, comprehensive and professional response information, so that the model can learn more representative features and have better generalization ability, and integrate the professional rules and concepts of the corresponding field into the model, so that users can more easily understand the decision-making process of the model and enhance trust.

[0039] Furthermore, using the training set to store the tuned initial knowledge model allows the model to learn the patterns and relationships in the training set for accurate application in future data analysis and language processing.

[0040] The validation set is used to further adjust the hyperparameters to prevent the model from overfitting or underfitting. The performance of the validation set is used to evaluate the current hyperparameter settings and help the model find the best parameter combination, so that the initial knowledge model after training can be obtained.

[0041] The test set can also be used to conduct a final evaluation of the performance of the trained initial knowledge model to check the performance of the model on unseen data and evaluate the actual generalization ability of the model. The evaluation process can focus on indicators such as the model's accuracy, recall, and F1 score. For example, in the evaluation process of the real estate model, the accuracy of the model's prediction of housing prices (i.e., accuracy), whether it can correctly identify key properties (i.e., recall), and the comprehensive consideration of the F1 score can be evaluated. The F1 score can be understood as an important indicator for the performance evaluation of machine learning models, which can represent the harmonic mean of accuracy and recall, which will not be expanded here.

[0042] After testing, the target knowledge model is finally obtained, which has high accuracy and reliability and can be applied to practical tasks such as market analysis, real estate recommendations, and price predictions.

[0043] According to the solution provided in the above embodiment, by constructing an initial knowledge model and selecting professional knowledge to train it, the target knowledge model can be obtained, which can significantly improve the performance, stability and generalization ability of the target knowledge model, while reducing the training time and the risk of overfitting, and making the model more efficient and reliable in actual application.

[0044] In other embodiments, a training set, a validation set, and a test set are obtained based on professional knowledge, including: preprocessing the professional knowledge to obtain knowledge features, the preprocessing including noise removal, normalization, and feature extraction; segmenting the knowledge features to obtain a training set, a validation set, and a test set.

[0045] Specifically, after obtaining professional knowledge and before model training, the professional knowledge can be preprocessed, that is, cleaned and standardized to ensure that the model can learn useful information from it, and the knowledge features obtained by preprocessing can be segmented to obtain the training set, verification set and test set in the training process.

[0046] The preprocessing steps may include 1. Noise removal, that is, removing erroneous, irrelevant or duplicate information in professional knowledge, such as deleting redundant data entries or correcting outliers. 2. Normalization, that is, the data can be converted to the same scale to eliminate model bias caused by different data magnitudes, such as normalizing the value of real estate prices, areas, etc. to a standard range (such as 0 to 1). 3. Feature extraction, that is, the most critical features for model training can be extracted from the noise-removed and normalized data. This can be achieved through statistical analysis, domain knowledge, etc. The extracted features can best represent the inherent patterns of the data. In this way, the preprocessing of professional knowledge can be completed. As an example, in real estate professional knowledge, preprocessing can include removing duplicate real estate records, standardizing price data to the same unit (such as 10,000 yuan), and extracting key features such as real estate location, area, and house age to obtain knowledge features in the real estate field.

[0047] Furthermore, the knowledge features can be segmented to obtain training sets, validation sets, and test sets for model training and evaluation at different stages. The segmentation method can be based on the amount of knowledge feature data, for example, 70% of the total data is used for training, 15% for validation, and 15% for testing, which is not limited here.

[0048] According to the solution provided in the above embodiment, high-quality knowledge features can be extracted by preprocessing professional knowledge, ensuring the quality and consistency of the data. The knowledge features are divided into training sets, verification sets and test sets, and the initial knowledge model is trained. This allows the model to obtain accurate results in the training, tuning and evaluation stages.

[0049] In some embodiments, an information generation model is determined based on key features, including: using a central model to analyze the key features to obtain analysis results, and the central model is used to control the interactive information to perform interactive tasks; based on the analysis results, an information generation model is determined, and the information generation model is used to perform interactive tasks in conjunction with a target knowledge model.

[0050] Specifically, after the key features are obtained, an information generation model can be determined based on the key features, that is, it is determined which information generation model is used to process user needs.

[0051] Furthermore, the central model can be used as the core of the entire system to analyze the key features in the user's interactive information in order to understand the user's needs. The central model can use a large model with the capabilities of dialogue, text understanding, content creation, information extraction, etc., without limitation here. By analyzing the key features using the central model, an analysis result can be obtained. The analysis result can represent an abstract expression of the key features, which may include the user's needs, preferences, and contextual information. The analysis result can be used to determine which information generation model should be used to handle the task in the subsequent steps.

[0052] For example, if the analysis result indicates that the user needs to view a certain chart, then a chart generation tool that meets the needs can be selected to generate the chart. If the analysis result indicates that the user needs to view the virtual reality view (VR) of the room type, then the VR system can be called to generate the VR view based on the knowledge data in the target knowledge model to achieve the ability to dynamically adjust the virtual reality content according to user instructions. In this way, customers can experience the interior and surrounding environment of the property through virtual reality equipment without the need for on-site inspections. In this way, using the information generation model as a third-party enhancement tool for the central model can enhance the business capabilities of the central model and enhance the business experience.

[0053] The information generation model is used to perform interactive tasks in conjunction with the target knowledge model, which can be expressed as a means of expanding the capabilities of the central model using the information generation model. When the target knowledge model cannot answer the user's interactive needs, it assists the target knowledge model in completing the interactive tasks. In other words, the target knowledge model can be regarded as a model that only processes professional knowledge and cannot process and generate information other than professional knowledge, such as implementing news and view requirements.

[0054] According to the solution provided by the above embodiment, the central model can be used to effectively analyze the key features of user input, and select a suitable information generation model based on these analysis results. The information generation model is combined with the knowledge model to generate accurate and personalized response information, thereby completing user interaction tasks and ensuring that the system can flexibly and efficiently provide valuable services when processing different user requests.

[0055] In some embodiments, the information generation model generates response information corresponding to the interactive information according to the key features and knowledge data, and also includes: obtaining the user's historical behavior, and building a user portrait based on the historical behavior; generating response information based on the user portrait, key features and knowledge data.

[0056] Specifically, a user profile can also be constructed based on the user's historical behavior, and response information can be generated based on the user profile, key features, and knowledge data. First, the user's historical behavior can be obtained. The user's historical behavior can include the preferences, habits, and decisions shown by the user in past interactions. By recording and analyzing these behavioral data, the system can better understand the user's needs and tendencies, such as the user's historical search history, browsing history, interaction history, purchase or rental history, and other information. Constructing a user profile based on historical behavior means that the user's historical behavior data can be integrated to form a comprehensive user profile. For example, the user's preference features (such as favorite property types, geographical preferences, etc.) are extracted from historical behavior, and a user profile is generated based on the extracted preference features. The user's Huaxi profile can include quantitative information (such as price preferences) and qualitative information (such as liking quiet communities).

[0057] Finally, response information is generated based on user portraits, key features and knowledge data. The response information generated in this way not only takes into account the user's current needs (that is, the information displayed by the key features) but also takes into account the user's historical preferences and domain knowledge, thereby providing more personalized suggestions or answers, enabling the system to more intelligently meet the user's complex needs.

[0058] In some embodiments, after interacting with the user through response information, it also includes: collecting user feedback, training the central model and the target knowledge model based on the feedback, and obtaining a new central model and a new target knowledge model; using the new central model to call the new target knowledge model to process the new interaction information, and obtain response information corresponding to the new interaction information.

[0059] Specifically, after the user completes the interaction with the system, the user's feedback can be collected and the interaction process can be improved based on the user's feedback.

[0060] The collected user feedback can be the user's satisfaction with the response information provided by the system, error correction suggestions or further needs. After collecting the user's feedback, the system will use this feedback information to train the central model and target knowledge model. Through such training, the system can continuously improve its own performance so that the generated response information better meets the user's needs.

[0061] The training process may include: integrating the collected user feedback into a data set, which may include ratings, user behaviors, and textual opinions, etc. The central model is adjusted based on the data set, such as adjusting the natural language processing algorithm to better understand user needs. The target knowledge model can be trained based on the data set, updating the internal knowledge base or adjusting features and data processing methods to ensure that the target knowledge model can provide more accurate knowledge data in future interactions.

[0062] According to the solution provided in the above embodiment, by collecting user feedback, the system can continuously optimize itself and improve the central model and the target knowledge model. This not only improves the performance and response quality of the model, but also continuously adapts to changes in user needs, ultimately improving the overall user experience and satisfaction.

[0063] Figure 2 This is a flow chart of another intelligent interaction method provided by an embodiment of the present application. Figure 2 This application is further described.

[0064] In the real estate field, we can first obtain professional knowledge (general real estate information, professional real estate information, and real transaction processes) to train knowledge models and obtain real estate field models (i.e., target knowledge models). We can also connect the programming interfaces (Application Programming Interface, API) of other tools so that the smart brain model (i.e., central model) can call other tools as information generation models to generate response information. Through technical construction, the smart brain model can call other tools (such as VR systems) based on user needs and generate response information based on user preferences and historical behaviors and return it to users. Among them, user needs can include viewing room information through VR, asking for real estate information, etc. In this way, the integration of VR system and central model is improved through technology to enhance the display experience.

[0065] Figure 3 This is a flow chart of another intelligent interaction method provided in the embodiment of the present application. Figure 3 The embodiments of the present application are further described.

[0066] In the process of the intelligent interactive system performing interactive tasks, several main roles may be involved: user, system, large language model (i.e., target knowledge model), and information generation model (this embodiment takes the virtual reality VR system as an example).

[0067] When the user logs into the system, the system can automatically load the user portrait, and provide personalized services based on the user portrait during the response information generation process. If the user selects a property and asks a question, the system can process the user's question (i.e., interactive information) and generate interactive answers based on the large language model. However, when the large language model cannot directly answer the user's question, it can select an appropriate information generation model based on the content of the user's question to assist in answering. For example, when the user needs to display a VR property, the response information is generated through the virtual reality VR system.

[0068] When the large language model answers a user's question, the answer form can be determined based on the user's question form, user needs and other factors, such as answering in voice, answering in text, or in the form of icons and text, etc., without limitation here. After the interaction is completed, the song model in the system can be updated based on the feedback provided by the user, and the service can be optimized through model update to provide more accurate and personalized response information.

[0069] Of course, in the actual application process, in addition to using the large language model as the target knowledge model, specific tools based on domain expert systems, such as real estate agent tools, can also be developed. These systems can be provided with knowledge and rules in the field of real estate economics by domain experts, and then process real estate information and communicate with customers through reasoning engines. Other machine learning models can also be used to realize the intelligence of real estate economic services. For example, machine learning algorithms can be used to train models to understand real estate data and customer needs and generate corresponding texts. In terms of the expansion of virtual reality technology, the application of other augmented display (AR) or mixed reality (MR) technologies can be considered to combine virtual real estate information with the environment in the real world to provide customers with a more immersive and realistic real estate experience.

[0070] All the above optional technical solutions can be arbitrarily combined to form optional embodiments of the present application, which will not be described one by one here.

[0071] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0072] Figure 4 Schematic diagram of an intelligent interactive device provided in an embodiment of the present application. Figure 4 As shown, the intelligent interactive device includes:

[0073] The feature extraction module 401 is configured to extract features of the interaction information in response to the interaction information sent by the user to the target object, and obtain key features corresponding to the interaction information;

[0074] The data acquisition module 402 is configured to acquire knowledge data corresponding to the target object according to the key features;

[0075] The response module 403 is configured to determine an information generation model according to the key features, input the key features and the knowledge data into the information generation model, so that the information generation model generates response information corresponding to the interactive information according to the key features and the knowledge data;

[0076] The interaction module 404 is configured to interact with the user through response information.

[0077] In some embodiments, the data acquisition module 402 is specifically used to construct an initial knowledge model, acquire professional knowledge to train the initial knowledge model, and obtain a target knowledge model, where the professional knowledge represents the knowledge in the field to which the target object belongs; key features are input into the target knowledge model for data extraction to obtain knowledge data corresponding to the target object.

[0078] In some embodiments, the data acquisition module 402 is specifically used to construct an initial knowledge model, perform hyperparameter tuning on the initial knowledge model, and obtain a tuned initial knowledge model; acquire professional knowledge, and obtain a training set, a validation set, and a test set based on the professional knowledge; use the training set to train the tuned initial knowledge model, and use the validation set to adjust hyperparameters to obtain a trained initial knowledge model; use the test set to test the indicators of the trained initial knowledge model to obtain a target knowledge model, and the indicators include the accuracy and recall rate of the trained initial knowledge model.

[0079] In some embodiments, the data acquisition module 402 is specifically used to preprocess professional knowledge to obtain knowledge features, the preprocessing includes noise removal, normalization and feature extraction; and segment the knowledge features to obtain a training set, a validation set and a test set.

[0080] In some embodiments, the response module 403 is specifically used to analyze key features using the central model to obtain analysis results, and the central model is used to control interactive information to perform interactive tasks; based on the analysis results, the information generation model is determined, and the information generation model is used to perform interactive tasks in conjunction with the target knowledge model.

[0081] In some embodiments, the response module 403 is specifically used to obtain the user's historical behavior, build a user profile based on the historical behavior, and generate response information based on the user profile, key features, and knowledge data.

[0082] In some embodiments, the interaction module 404 is specifically used to collect user feedback, train the central model and the target knowledge model based on the feedback, and obtain a new central model and a new target knowledge model; use the new central model to call the new target knowledge model to process the new interaction information, and obtain response information corresponding to the new interaction information.

[0083] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0084] Figure 5 Schematic diagram of an electronic device 5 provided in an embodiment of the present application. Figure 5As shown, the electronic device 5 of this embodiment includes: a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 501 executes the computer program 503, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0085] The electronic device 5 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 5 may include, but is not limited to, a processor 501 and a memory 502. Those skilled in the art will appreciate that Figure 5 The electronic device 5 is merely an example and does not limit the electronic device 5 , and may include more or less components than those shown in the figure, or different components.

[0086] The processor 501 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0087] The memory 502 may be an internal storage unit of the electronic device 5, for example, a hard disk or memory of the electronic device 5. The memory 502 may also be an external storage device of the electronic device 5, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 5. The memory 502 may also include both an internal storage unit of the electronic device 5 and an external storage device. The memory 502 is used to store computer programs and other programs and data required by the electronic device.

[0088] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units.

[0089] If the integrated module / 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 storage medium. Based on this understanding, the present application implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0090] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. An intelligent interaction method, characterized in that: include: In response to the interaction information sent by the user to the target object, feature extraction is performed on the interaction information to obtain key features corresponding to the interaction information; Acquire the knowledge data corresponding to the target object according to the key features; Determine an information generation model according to the key feature, input the key feature and the knowledge data into the information generation model, so that the information generation model generates response information corresponding to the interactive information according to the key feature and the knowledge data; Interact with the user through the response information.

2. The method according to claim 1, characterized in that: The acquiring the knowledge data corresponding to the target object according to the key feature includes: Constructing an initial knowledge model, acquiring professional knowledge to train the initial knowledge model, and obtaining a target knowledge model, wherein the professional knowledge represents the knowledge in the field to which the target object belongs; The key features are input into the target knowledge model for data extraction to obtain knowledge data corresponding to the target object.

3. The method according to claim 2, characterized in that The step of constructing an initial knowledge model, acquiring professional knowledge to train the initial knowledge model, and obtaining a target knowledge model includes: Constructing the initial knowledge model, and performing hyperparameter tuning on the initial knowledge model to obtain a tuned initial knowledge model; Acquire the professional knowledge, and obtain a training set, a validation set, and a test set based on the professional knowledge; Using the training set to train the tuned initial knowledge model, and using the validation set to adjust hyperparameters to obtain a trained initial knowledge model; The test set is used to test the indicators of the trained initial knowledge model to obtain the target knowledge model, wherein the indicators include the accuracy and recall rate of the trained initial knowledge model.

4. The method according to claim 3, characterized in that The obtaining of a training set, a validation set and a test set based on the professional knowledge includes: Preprocessing the professional knowledge to obtain knowledge features, wherein the preprocessing includes noise removal, normalization and feature extraction; The knowledge features are segmented to obtain the training set, the verification set and the test set.

5. The method according to claim 3, characterized in that: The step of determining the information generation model according to the key features comprises: Analyzing the key features using a central model to obtain analysis results, wherein the central model is used to control the interactive information to perform interactive tasks; The information generation model is determined based on the analysis result, and the information generation model is used to perform an interactive task in conjunction with the target knowledge model.

6. The method according to claim 1, characterized in that The step of causing the information generation model to generate response information corresponding to the interactive information according to the key features and the knowledge data further includes: Acquire the historical behavior of the user, and build a user profile based on the historical behavior; The response information is generated based on the user portrait, the key features and the knowledge data.

7. The method according to claim 5, characterized in that After interacting with the user through the response information, the method further includes: Collecting feedback from the user, and training the central model and the target knowledge model based on the feedback to obtain a new central model and a new target knowledge model; The new central model is used to call the new target knowledge model to process the new interaction information, and response information corresponding to the new interaction information is obtained.

8. An intelligent interactive device, characterized in that: include: A feature extraction module is configured to extract features from the interaction information sent by the user to the target object, and obtain key features corresponding to the interaction information; A data acquisition module, configured to acquire knowledge data corresponding to the target object according to the key features; a response module, configured to determine an information generation model according to the key feature, and input the key feature and the knowledge data into the information generation model, so that the information generation model generates response information corresponding to the interactive information according to the key feature and the knowledge data; The interaction module is configured to interact with the user through the response information.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images