Informatization intelligent interaction system
By acquiring historical and real-time data of target users in households, personalized user profiles and demand models are constructed, solving the problems of inaccurate data collection and response delays, and realizing efficient and personalized management of smart home devices and improved user satisfaction.
Patent Information
- Application Number
- CN202510791207.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-13
- Publication Date
- 2025-11-07
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Inaccurate data collection and processing during the construction of the user preference database can affect the system's understanding of users' real needs. Personalized profiles rely on a large amount of user data, which raises privacy and data security issues. Response delays or incomplete functionality of smart home devices can also negatively impact user experience.
By acquiring multi-dimensional data on the historical behavioral habits of target users in the home, utilizing edge heterogeneous sensors to obtain real-time behavioral action data, and combining random forest and deep learning algorithms, personalized user profiles and demand models are constructed, demand instruction sets are generated, and home device interaction tasks are optimized.
It improves the quality of life, enhances the responsiveness and intelligence of home appliances, and enables smarter home management and higher user satisfaction.
Smart Images
Figure CN120909416A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of information-based intelligent interaction technology, and in particular to an information-based intelligent interaction system. BACKGROUND
[0002] The technical defects of an information-based intelligent interaction system can include: in the process of constructing a user preference database, there can be inaccurate data collection and processing, affecting the system's understanding of the user's real needs; the construction of a personalized portrait relies on a large amount of user data, which can involve privacy protection and data security issues; when matching smart home devices with user needs, there can be device response delays or device functions that do not fully meet the needs, affecting user experience. SUMMARY
[0003] To solve the above technical problems, an information-based intelligent interaction system is provided, which solves the above problems of inaccurate data collection and processing in the process of constructing a user preference database, affecting the system's understanding of the user's real needs; the construction of a personalized portrait relies on a large amount of user data, which can involve privacy protection and data security issues; when matching smart home devices with user needs, there can be device response delays or device functions that do not fully meet the needs, affecting user experience.
[0004] To achieve the above purposes, the technical solution adopted by the present application is: An information-based intelligent interaction method, comprising: Obtaining historical behavior habit multi-dimensional data of a target user, quantifying the personalized portrait of the target user; Based on edge heterogeneous sensors, obtaining real-time behavior action data of the target user, analyzing the target user demand vector in the real-time behavior action data; Based on the association analysis of the target user's personalized user portrait and the target user's demand vector, generating a target user demand instruction set; According to the priority analysis of the target user demand instruction set and the smart home device, generating a home device interaction task for the target user, and generating an information-based intelligent interaction system.
[0005] Preferably, based on the historical behavior habit multi-dimensional data of the target user, the internal data and external data of the target user are integrated, multiple imputations are used for missing value processing, the processed data is normalized, and the behavior characteristics of the target user are extracted; Based on the behavior characteristics of the family target user, a random forest algorithm is used to generate multiple subsets of behavior characteristic data of the family target user through BBootstrap sampling, train the corresponding decision tree of the behavior characteristic data set of the family target user, repeatedly train to generate multiple decision trees of the behavior characteristic data set of different family target users, and assemble the behavior characteristic random forest model of the family target user. Based on the behavior characteristic random forest model of the family target user, the behavior characteristics of the family target user are classified, and the individualized portrait of the family target user is quantified.
[0006] Preferably, based on edge heterogeneous sensor data acquisition, data preprocessing is performed; According to the data preprocessing, the real-time action of the family target user is obtained by using a lightweight spatio-temporal graph convolution network method.
[0007] Preferably, based on the real-time action of the family target user, the behavior demand of the family target user is constructed, the fuzzy reasoning system is used to process the real-time action fuzzy problem of the family target user, and the time sequence demand model of the family target user is constructed; Based on the time sequence demand model of the family target user, the demand vector of the family target user in the real-time behavior action data is analyzed.
[0008] Preferably, based on the individualized user portrait of the family target user and the demand vector of the family target user, a dynamic time warping (DTW) algorithm is used to construct a distance matrix of the individualized user portrait of the family target user and the demand vector of the family target user, to calculate the cumulative distance, find the minimum cumulative path from the starting point to the ending point of the distance matrix of the individualized user portrait of the family target user and the demand vector of the family target user, and align the individualized user portrait of the family target user and the demand of the family target user.
[0009] Preferably, based on the alignment of the individualized user portrait of the family target user and the demand of the family target user, a deep learning algorithm is used to construct the individualized user portrait of the family target user and the demand model of the family target user; Based on the individualized user portrait of the family target user and the demand model of the family target user, expert evaluation of the relative importance of the demand of the family target user is represented by triangular fuzzy numbers, a fuzzy judgment matrix of the demand of the family target user is established, the numerical values of the fuzzy numbers are converted by the barycentric method, the weight vector of the demand of the family target user is calculated, and the demand instruction set of the family target user is generated.
[0010] Preferably, based on the individualized portrait of the family target user, a family target user preference database is constructed; Based on the household target user preference database, the proportion of each preference data in the personalized portrait of the household target user to the overall household target user preference is calculated to obtain a proportion index of each preference data in the personalized portrait of the household target user; Based on the information entropy formula, the discrete degree of each preference data proportion index in the personalized portrait of the household target user to the overall household target user preference is calculated to obtain the information entropy of each preference data in the personalized portrait of the household target user. According to the weighted method, the information entropy of each preference data in the personalized portrait of the household target user is normalized to obtain the weight coefficient of each preference data in the personalized portrait of the household target user.
[0011] Preferably, based on the weight coefficient of each preference data in the personalized portrait of the household target user, the priority sequence of the preference data in the personalized portrait of the household target user is obtained. The priority sequence of the preference data in the personalized portrait of the household target user is matched with the executable tasks of the smart home device based on the household target user demand instruction set to generate a home device interaction task of the household target user, and an information-based intelligent interaction system is generated.
[0012] Further, an information-based intelligent interaction system is used to implement the above-mentioned information-based intelligent interaction method, which comprises: a personalized portrait module of a household target user, a household target user demand vector module, a household target user demand instruction set module, and a home device interaction task module The personalized portrait module of the household target user is used to obtain historical behavior habit multidimensional data of the household target user and quantify the personalized portrait of the household target user. The household target user demand vector module is used to obtain real-time behavior action data of the household target user based on edge heterogeneous sensors and analyze the household target user demand vector in the real-time behavior action data. The household target user demand instruction set module is electrically connected with the personalized portrait module of the household target user and the household target user demand vector module, and is used to perform correlation analysis based on the household target user personalized user portrait and the household target user demand vector to generate a household target user demand instruction set. The home device interaction task module is electrically connected with the personalized portrait module of the household target user, and is used to perform priority analysis on the household target user demand instruction set and the smart home device to generate a home device interaction task of the household target user and generate an information-based intelligent interaction system.
[0013] Compared with the prior art, the present application has the beneficial effects that: The present application proposes an information-based intelligent interaction scheme. The system optimizes according to the specific needs and preferences of users, not only improving the quality of life, but also enhancing the responsiveness and intelligent degree of home equipment, thereby realizing more intelligent home management and higher user satisfaction. BRIEF DESCRIPTION OF DRAWINGS
[0014] Figure 1 For an information-based intelligent interaction method flowchart; Figure 2 For an information-based intelligent interaction system framework diagram. DETAILED DESCRIPTION
[0015] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only as examples, and other obvious modifications can be thought of by those skilled in the art.
[0016] Referring to Figure 1 As shown in the figure, an information-based intelligent interaction system includes: Step one, obtaining the historical behavior habit multi-dimensional data of the target user of the family, quantifying the personalized portrait of the target user of the family; The step one includes the following contents: Step 101, based on the historical behavior habit multi-dimensional data of the target user of the family, integrating the internal data and external data of the target user of the family, using multiple imputation for missing value processing, normalizing the processed data, and extracting the behavior characteristics of the target user of the family; As further content, the historical behavior habit multi-dimensional data of the target user of the family includes: family structure, content preference, product purchase record, seasonal change pattern; Step 102, based on the behavior characteristics of the target user of the family, using a random forest algorithm, generating multiple behavior characteristic data subsets of the target user of the family through information-based intelligent interaction Bootstrap sampling, training the corresponding behavior characteristic data set decision tree of the target user of the family, repeatedly training to generate multiple different behavior characteristic data set decision trees of the target user of the family, and establishing a behavior characteristic random forest model of the target user of the family; Step 103, based on the behavior characteristic random forest model of the target user of the family, classifying the behavior characteristics of the target user of the family, and quantifying the personalized portrait of the target user of the family.
[0017] In use, the contents of steps 101 to 103 are combined, By integrating the historical behavior data of the target user of the family, and using multiple interpolation and normalization processing to clean and standardize the data, the key behavior characteristics of the user are extracted; the random forest algorithm is used to generate multiple data subsets through informationized intelligent interaction Bootstrap sampling, different decision trees are trained, and a behavior characteristic random forest model of the target user of the family is constructed; the model can effectively classify user behavior characteristics, and then quantify the user's personalized portrait, thereby providing data support for subsequent personalized recommendation, precision marketing and other applications; its beneficial effects include improving the integrity and accuracy of the data, deeply mining the user behavior mode through the model, accurately identifying the user demand, and providing a scientific basis for personalized service and strategy formulation.
[0018] Step two, based on edge heterogeneous sensors, real-time behavior action data of the target user of the family is obtained, and the demand vector of the target user of the family in the real-time behavior action data is analyzed; The step two includes the following contents: Step 201, based on edge heterogeneous sensor data collection, data preprocessing is performed; According to the data preprocessing, a lightweight spatio-temporal graph convolution network method is used to obtain the real-time action of the target user of the family, and the formula is as follows,
[0019] Among them, is the real-time action of the target user of the family, is the i-th target joint of the target user of the family, is the adjacent joint of the i-th target joint of the target user of the family, is the joint directly connected to the adjacent joint of the i-th target joint of the target user of the family, is a preset constant, is the learnable weight matrix corresponding to the i-th target joint of the target user of the family, is the division label of the i-th target joint of the target user of the family, is the feature vector of the i-th target joint of the target user of the family; Step 202, based on the real-time action of the target user of the family, the behavior demand of the target user of the family is constructed, the fuzzy reasoning system is used to process the real-time action fuzziness of the target user of the family, and a time sequence demand model of the target user of the family is constructed; Step 203, based on the time sequence demand model of the target user of the family, the demand vector of the target user of the family in the real-time behavior action data is analyzed.
[0020] In use, the contents of steps 201 to 203 are combined, The behavior data of the target user at home is collected by an edge heterogeneous sensor, and the action features of the user are extracted by using a light spatio-temporal graph convolution network (ST-GCN) for preprocessing; then, the fuzziness in real-time actions is processed by using a fuzzy inference system, and a time-series demand model is constructed, so as to accurately predict the behavior demand of the user; this process helps to improve the response capability and resource scheduling effect of the home intelligent system, and optimizes the personalized service.
[0021] Step three, correlation analysis is performed based on the personalized user portrait of the target user at home and the demand vector of the target user at home, and a demand instruction set of the target user at home is generated; The step three includes the following contents: Step 301, based on the personalized user portrait of the target user at home and the demand vector of the target user at home, a distance matrix of the personalized user portrait of the target user at home and the demand vector of the target user at home is constructed by using a dynamic time warping (DTW) algorithm, the cumulative distance is calculated, the minimum cumulative path from the starting point to the ending point of the distance matrix of the personalized user portrait of the target user at home and the demand vector of the target user at home is found, and the personalized user portrait of the target user at home and the demand of the target user are aligned; Step 302, based on the alignment of the personalized user portrait of the target user at home and the demand of the target user, a personalized user portrait of the target user at home and a demand model of the target user at home are constructed by using a deep learning algorithm; Step 303, based on the personalized user portrait of the target user at home and the demand model of the target user at home, an expert evaluation of the relative importance between the demands of the target user at home is represented by using a triangular fuzzy number, a fuzzy judgment matrix of the demand of the target user at home is established, the numerical value of the fuzzy number is converted by using a barycenter method, the demand weight vector of the target user at home is calculated, and a demand instruction set of the target user at home is generated; In use, the contents of steps 301 to 303 are combined, The long-term portrait of the user and the short-term demand vector are aligned in the time dimension by using a dynamic time warping (DTW) algorithm, the mapping relationship between them is established, the deep learning model is used to mine the deep non-linear relationship between the portrait features and the demand, and finally the fuzzy analytic hierarchy process (FAHP) is used to quantify the demand priority, and the expert experience and data-driven analysis are combined to generate a personalized instruction set; its beneficial effects are: 1) the time scale inconsistency problem of multi-source data is solved by spatio-temporal alignment, and the matching accuracy is improved; 2) the deep learning model can automatically extract the hidden features between the portrait and the demand, and the intelligent level of the instruction generation is improved; 3) the triangular fuzzy number handles the uncertainty in demand evaluation, so that the instruction priority ranking is more in line with the actual scene demand, and finally the end-to-end accurate conversion from the user portrait to the executable instruction is realized.
[0022] Step four, according to the family target user demand instruction set and the smart home equipment to carry out priority analysis, generate the home equipment interaction task of the family target user, generate an information-based intelligent interaction system; The step four includes the following contents: Step 401, based on the personalized portrait of the family target user, a family target user preference database is constructed; Based on the family target user preference database, the proportion of each preference data in the personalized portrait of the family target user to the overall family target user preference is calculated, and the proportion index of each preference data in the personalized portrait of the family target user is obtained. Based on the information entropy formula, the discrete degree of each preference data proportion index in the personalized portrait of the family target user to the overall family target user preference is calculated, and the information entropy of each preference data in the personalized portrait of the family target user is obtained. According to the weighting method, the information entropy of each preference data in the personalized portrait of the family target user is normalized to obtain the weight coefficient of each preference data in the personalized portrait of the family target user. Step 402, based on the weight coefficient of each preference data in the personalized portrait of the family target user, the priority sequence of the preference data in the personalized portrait of the family target user is obtained. Step 403, combining the family target user demand instruction set, the priority sequence of the preference data in the personalized portrait of the family target user is matched with the executable task of the smart home equipment to generate the home equipment interaction task of the family target user, and an information-based intelligent interaction system is generated.
[0023] In use, the contents of steps 401 to 403 are combined, Based on the personalized portrait of the family target user, the weight and discrete degree of the user's preference data are calculated, and the priority is further sorted; the relative importance of each preference data is evaluated by information entropy method, and the weight coefficient is obtained by normalization processing by weighting method; then, the user demand instruction set and the executable task of the smart home equipment are matched to generate the home equipment interaction task that meets the user's preference, and a smart interaction system is constructed; its beneficial effects include: improving the individualization and intelligence of the home system, making the user's life more convenient and efficient, and improving the adaptability and flexibility of the system.
[0024] Referring to Figure 2 As shown in the figure, an information-based intelligent interaction system comprises: A personalized portrait module of a family target user, a family target user demand vector module, a family target user demand instruction set module and a home equipment interaction task module The personalized portrait module of the family target user is used for acquiring historical behavior habit multidimensional data of the family target user, and quantifying the personalized portrait of the family target user; The family target user demand vector module is used for acquiring real-time behavior action data of the family target user based on edge heterogeneous sensors, and analyzing the family target user demand vector in the real-time behavior action data; The family target user demand instruction set module is electrically connected with the personalized portrait module of the family target user and the family target user demand vector module, the family target user demand instruction set module is used for performing correlation analysis based on the personalized user portrait of the family target user and the family target user demand vector, and generating a family target user demand instruction set; The home equipment interaction task module is electrically connected with the personalized portrait module of the family target user, the home equipment interaction task module is used for performing priority analysis on the smart home equipment according to the family target user demand instruction set, generating a home equipment interaction task of the family target user, and generating an information-based intelligent interaction system.
[0025] The basic principle, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above examples, the above examples and the description in the specification are only the principles of the present application, various changes and improvements can be made without departing from the spirit and scope of the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection claimed by the present application is defined by the appended claims and their equivalents.
Claims
1. An information-based intelligent interaction method, characterized in that, The application relates to an information-based intelligent interaction system. S1, acquiring historical behavior habit multidimensional data of a target user of a family, and quantifying a personalized portrait of the target user of the family; S2, acquiring real-time behavior action data of the target user of the family based on an edge heterogeneous sensor, and analyzing a demand vector of the target user of the family in the real-time behavior action data; S3, performing correlation analysis based on the personalized user portrait of the target user of the family and the demand vector of the target user of the family, and generating a demand instruction set of the target user of the family; S4, performing priority analysis on the demand instruction set of the target user of the family and smart home equipment, generating a home equipment interaction task of the target user of the family, and generating the information-based intelligent interaction system.
2. The information-based intelligent interaction method of claim 1, wherein, The S1 comprises the following steps. Based on the historical behavior habit multidimensional data of the target user of the family, internal data and external data of the target user of the family are integrated, missing value processing is performed by using multiple interpolation, normalized processing is performed on the processed data, and behavior characteristics of the target user of the family are extracted; Based on the behavior characteristics of the target user of the family, a random forest algorithm is used to generate a plurality of behavior characteristic data subsets of the target user of the family through information-based intelligent interaction Bootstrap sampling, a behavior characteristic data set decision tree corresponding to the target user of the family is trained, a plurality of behavior characteristic data set decision trees of different target users of the family are repeatedly trained, and a behavior characteristic random forest model of the target user of the family is established; Based on the behavior characteristic random forest model of the target user of the family, behavior characteristics of the target user of the family are classified, and a personalized portrait of the target user of the family is quantified.
3. The information-based intelligent interaction method of claim 1, wherein, The S2 comprises the following steps. Based on edge heterogeneous sensor data acquisition, data preprocessing is performed; According to the data preprocessing, a real-time action of the target user of the family is acquired by using a lightweight spatiotemporal graph convolution network method.
4. The information-based intelligent interaction method of claim 3, wherein, The S2 further comprises the following steps. Based on the real-time action of the target user of the family, a behavior demand of the target user of the family is constructed, a fuzzy inference system is used to process the real-time action fuzziness of the target user of the family, and a time sequence demand model of the target user of the family is constructed; Based on the time sequence demand model of the target user of the family, a demand vector of the target user of the family in the real-time behavior action data is analyzed.
5. The information-based intelligent interaction method of claim 1, wherein, The S3 comprises the following steps. Based on the personalized user portrait of the target user of the family and the demand vector of the target user of the family, a dynamic time warping (DTW) algorithm is used to construct a distance matrix of the personalized user portrait of the target user of the family and the demand vector of the target user of the family, cumulative distance calculation is performed on the distance matrix, a minimum cumulative path from a starting point to an ending point of the distance matrix of the personalized user portrait of the target user of the family and the demand vector of the target user of the family is found, and the personalized user portrait of the target user of the family is aligned with the demand of the target user of the family.
6. The information-based intelligent interactive system according to claim 1, wherein, The S3 further comprises the following steps. Based on the alignment of the personalized user portrait of the target user of the family and the demand of the target user of the family, a deep learning algorithm is used to construct a personalized user portrait demand model of the target user of the family. Based on the personalized user portrait of the family target user and the demand model of the family target user, expert evaluation of the relative importance of the demand of the family target user is represented by triangular fuzzy numbers, a fuzzy judgment matrix of the demand of the family target user is established, the fuzzy numbers are converted into numerical values by the gravity method, the weight vector of the demand of the family target user is calculated, and a demand instruction set of the family target user is generated.
7. The information-based intelligent interaction method of claim 1, wherein, The S4 comprises: Based on the personalized portrait of the family target user, a family target user preference database is constructed; Based on the family target user preference database, the proportion of each preference data in the personalized portrait of the family target user to the overall preference of the family target user is calculated, and a proportion index of each preference data in the personalized portrait of the family target user is obtained; Based on the information entropy formula, the dispersion degree of each preference data in the personalized portrait of the family target user to the overall preference of the family target user is calculated, and the information entropy of each preference data in the personalized portrait of the family target user is obtained. According to the weighted method, the information entropy of each preference data in the personalized portrait of the family target user is normalized to obtain the weight coefficient of each preference data in the personalized portrait of the family target user.
8. The informationized intelligent interaction method according to claim 7, characterized in that, The S4 further comprises: Based on the weight coefficient of each preference data in the personalized portrait of the family target user, the priority of each preference data in the personalized portrait of the family target user is sorted to obtain a priority sequence of the preference data in the personalized portrait of the family target user; The priority sequence of the preference data in the personalized portrait of the family target user is matched with the executable tasks of the smart home device in combination with the demand instruction set of the family target user to generate a home device interaction task of the family target user and generate an information-based intelligent interaction system.
9. An information-based intelligent interactive system, characterized by, The information-based intelligent interaction method according to any one of claims 1-8, comprising: a personalized portrait module of a family target user, a demand vector module of a family target user, a demand instruction set module of a family target user, and a home device interaction task module The personalized portrait module of the family target user is used to obtain historical behavior habit multidimensional data of the family target user and quantify the personalized portrait of the family target user; The demand vector module of the family target user is used to obtain real-time behavior action data of the family target user based on edge heterogeneous sensors and analyze the demand vector of the family target user in the real-time behavior action data; The demand instruction set module of the family target user is electrically connected with the personalized portrait module of the family target user and the demand vector module of the family target user, and is used to perform correlation analysis based on the personalized user portrait of the family target user and the demand vector of the family target user to generate a demand instruction set of the family target user; The home device interaction task module is electrically connected with the personalized portrait module of the family target user, and is used to perform priority analysis on the smart home device according to the demand instruction set of the family target user to generate a home device interaction task of the family target user and generate an information-based intelligent interaction system.