A method for analyzing user dining habits based on big data analysis
Through multi-dimensional data fusion and social network analysis, combining user terminals and POS system data, multi-dimensional user portraits are built and model training is carried out, which solves the limitations of single data sources and static analysis in the existing technology, and achieves a more comprehensive and personalized Analysis of user catering habits.
Patent Information
- Application Number
- CN202411224833.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-09-03
AI Technical Summary
The existing user's catering habit analysis methods rely on a single data source and are unable to integrate social media, fitness and health data. They lack hyperparameter tuning and dynamic adjustment mechanisms, ignore the impact of social networks, resulting in a lack of comprehensiveness and personalization of the analysis results.
Using multi-dimensional data fusion, social network analysis and model training methods, data is collected in real time through user terminals and POS systems, data cleaning and formatting are carried out, and multi-dimensional user portraits are constructed and convolutional nerves are used. The network and long-term memory network are trained for model and dynamically adjust the recommendation strategy.
It realizes more accurate multi-dimensional user behavior characteristics capture, improves the accuracy and user satisfaction of personalized recommendations, and can flexibly respond to changes in user needs and external environments.
Smart Images

Figure CN118761796B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of big data analysis, and in particular to a method for analyzing user dining habits based on big data analysis. Background Art
[0002] Existing methods for analyzing user dining habits mainly rely on simple statistical analysis based on historical data. Typically, these methods use a single data source, such as a user's dining consumption data, and ignore other potentially useful information such as the user's social media data, fitness data, and health data. In addition, traditional methods rely on basic statistical methods for analysis and prediction.
[0003] However, the inventors have discovered that this technical solution still has at least the following defects:
[0004] First, existing methods for analyzing user dining habits mainly rely on simple statistics and analysis based on historical data, usually using only a single data source, such as user dining consumption data. This single data source approach cannot integrate users’ social media data, fitness data, and health data, resulting in a lack of comprehensiveness in the analysis results, making it difficult to accurately reflect users’ multi-dimensional behavioral characteristics and preferences.
[0005] Second, traditional methods lack systematic hyperparameter tuning and dynamic adjustment mechanisms during model training and optimization, and usually adopt fixed recommendation strategies that cannot be flexibly adjusted based on users' real-time behavior data and changes in the external environment. This static approach limits the accuracy and practicality of personalized recommendations and makes it difficult to meet users' ever-changing needs.
[0006] Third, in addition, existing methods ignore the influence of social networks and fail to fully consider the potential influence of users' friends and family on their dining habits. This method, which lacks social network analysis, cannot provide truly personalized and targeted dining recommendations, resulting in insufficient relevance of recommended content and user satisfaction. Summary of the invention
[0007] Based on the above objectives, the present invention provides a method for analyzing user dining habits based on big data analysis.
[0008] A method for analyzing user dining habits based on big data analysis comprises the following steps:
[0009] S1, data collection: through the user terminal equipment and the POS system of the catering merchant, the user's catering consumption data is collected in real time, including consumption time, consumption location, consumption amount and order content;
[0010] S2, data preprocessing: cleaning, deduplication, and formatting the collected raw data to generate a standardized user dining data set;
[0011] S3, multi-dimensional data fusion: Integrate the user's dining data with the user's social media data, fitness data, and health data to form a multi-dimensional user portrait, add contextual awareness information, and optimize the user portrait;
[0012] S4, social network analysis: based on multi-dimensional data fusion, further analyze the user's social network data to evaluate the influence of the user's friends and family on their dining habits;
[0013] S5, behavioral feature extraction: extracting user behavioral features from the fused multidimensional data, including consumption frequency, consumption time period preference, and dish preference, and using the results of social network analysis to refine and optimize these features;
[0014] S6, model training: build and train a user dining habit analysis model, model and analyze user behavior characteristics, and comprehensively consider social network analysis results;
[0015] S7, Habit Prediction: Based on the trained model, predict the user's future dining habits and generate personalized dining recommendations.
[0016] Optionally, S1 includes:
[0017] S11, collecting data through user terminal equipment, specifically including:
[0018] Users actively record consumption information while dining through a mobile app, and the device automatically detects the user's dining environment;
[0019] The terminal device records the consumption location through the GPS module;
[0020] The terminal device records the consumption time through the timestamp function;
[0021] Users manually enter or scan restaurant receipts through the app to record the amount of consumption and the content of the order;
[0022] S12, collect data through the POS system of catering merchants, including:
[0023] When a user checks out, the restaurant merchant records the consumption information through the POS system, including the consumption time, consumption location, consumption amount and the content of the order;
[0024] The POS system automatically stores the above data and uploads it to the central server in real time;
[0025] S13, data synchronization and integration, specifically including:
[0026] The user's terminal device and the restaurant merchant's POS system synchronize data through the cloud server;
[0027] The central server receives and integrates consumption data from user terminal devices and POS systems, eliminates duplicate records, and forms a unified consumption data set.
[0028] Optionally, S2 includes:
[0029] S21, data cleaning:
[0030] S211, check and process missing values: for missing consumption time, consumption location, consumption amount and order content, fill in the missing values through interpolation;
[0031] S212, check and process outliers: use the Zscore method to detect and process outlier data points;
[0032] S22, data deduplication:
[0033] Use hash functions to uniquely identify consumption records, identify and delete duplicate records;
[0034] Calculate a unique identifier through a hash function to identify duplicate data;
[0035] S23, data formatting:
[0036] Convert data from different sources into a unified format and standardize text data;
[0037] S24, generate standardized data set:
[0038] The cleaned, deduplicated and formatted data are integrated into a standardized data structure, where each record contains the time, location, amount and order content of the consumption.
[0039] Optionally, S3 includes:
[0040] S31, high-dimensional feature extraction: Use convolutional neural networks to extract high-dimensional features from each data source, use high-dimensional features as input data, and construct tensor representation;
[0041] S32, construct tensor representation: construct the processed data into a third-order tensor ;
[0042] S33, Tensor Decomposition and Regularization: Use CANDECOMP / PARAFAC decomposition method to decompose tensors Decompose and add regularization terms;
[0043] S34, Multi-task learning: Combining multi-task learning methods, multiple related tasks are optimized simultaneously during tensor decomposition, including dining habit prediction and health recommendations;
[0044] S35, online learning and dynamic updating: Introducing online learning mechanism, dynamically updating the tensor model as new data is continuously added to reflect the user's latest behaviors and habits in real time;
[0045] S36, Context-aware fusion: In the tensor fusion process, context-aware information is added, including weather, time and location.
[0046] Optionally, the S36 includes:
[0047] S361, Data Collection: Real-time collection of situational awareness information, including current weather, time and location, through user terminal devices and environmental sensors;
[0048] S362, data preprocessing: cleaning and formatting the collected situational data;
[0049] S363, multi-source data integration: Integrate the processed contextual data with the user’s dining data, social media data, fitness data, and health data to form a multi-dimensional context-aware data set;
[0050] S364, Context Feature Extraction: Extract context features from the integrated context-aware dataset and analyze the impact of context data on user dining habits;
[0051] S365, tensor fusion: In the tensor fusion process, the context data is input into the tensor model together with the user's data in other dimensions in combination with the extracted context features;
[0052] S366, Context-aware optimization: Dynamically adjust user profiles and recommendation models based on changes in context-aware data.
[0053] Optionally, S4 includes:
[0054] S41, social network data collection: obtaining users’ social relationship data and interactive behavior data from their social media platforms, including likes, comments, shares, and tag information;
[0055] S42, social relationship construction: Use social network analysis technology to construct a user's social relationship graph, identify the user's friends, family members and their interaction frequency and intensity, and form a social relationship network graph;
[0056] S43, influence assessment: Analyze users’ social network data to assess the influence of friends and family on users’ dining habits, including assessing the frequency of interaction and the emotional tendency of the interaction content;
[0057] S44, influencing factor extraction: Extract influencing factors related to catering from social network data, including restaurants recommended by friends, dining places frequented by family members, and dietary preferences of friends and family members.
[0058] Optionally, S5 includes:
[0059] S51, consumption frequency extraction: Calculate the user's consumption frequency from the standardized data set, including the average daily consumption frequency, the average weekly consumption frequency and the average monthly consumption frequency;
[0060] S52, consumption time period preference extraction: analyzing the user's consumption behavior in different time periods, and counting the user's consumption frequency and amount in the breakfast, lunch, dinner and supper periods;
[0061] S53, extracting dish preferences: based on the user's historical consumption records, counting the dishes and cuisines that the user frequently orders, and analyzing the user's dish preferences, including the preferred dish types, flavors, and price ranges;
[0062] S54, integration of social network analysis results: Combined with the results of social network analysis, the influence of friends and family on the user's dining habits is taken into consideration to refine the user's consumption frequency, time period preference and dish preference;
[0063] S55, feature optimization: optimize the extracted behavioral features, remove redundant and noisy data, and retain key features.
[0064] Optionally, the S6 includes:
[0065] S61, data preparation: Integrate the extracted user behavior features, fused multidimensional data and social network analysis results into a training data set;
[0066] S62, feature selection: Use recursive feature elimination to select features from training data, retain key features, and remove redundant and irrelevant features;
[0067] S63, model selection: select long short-term memory network to build a user dining habit analysis model;
[0068] S64, model training: Use the prepared training data set to train the user dining habit analysis model, adjust the model parameters to improve the model performance, and use the cross-validation method to verify the stability and generalization ability of the model;
[0069] S65, model optimization: perform hyperparameter tuning based on the cross-validation results to optimize the model’s hyperparameters;
[0070] S66, Model Evaluation: Evaluate the trained model using the test dataset, including calculating the accuracy of the model , Accuracy , recall rate and F1 score ;
[0071] S67, Model Deployment: Deploy the trained and optimized model to the system for real-time prediction and personalized recommendation, and dynamically update the model based on the user's latest behavior and data.
[0072] Optionally, the S65 includes:
[0073] S651, determine the hyperparameter range: define the hyperparameters to be tuned and their possible value ranges, and create a hyperparameter grid. The hyperparameters include learning rate, regularization coefficient, number of hidden layers, and number of neurons in each layer;
[0074] S652, Grid Search: Perform grid search, traverse each combination in the hyperparameter grid by exhaustive method, train and verify the model for each combination, and record the performance indicators;
[0075] S653, hyperparameter combination evaluation: conduct a comprehensive evaluation of the verification results of each hyperparameter combination, calculate the average verification error and standard deviation, and select the best hyperparameter combination;
[0076] S654, Model Retraining: Retrain the model on the entire training dataset using the best hyperparameter combination.
[0077] Optionally, S7 includes:
[0078] Behavior prediction: Use the trained model to predict user behavior and output the user's future dining habits, including consumption frequency, consumption time period preference, and dish preference;
[0079] Personalized recommendation generation: Generate personalized dining recommendations based on predicted dining habits, including recommended restaurants, recommended dishes, and nutritional recommendations;
[0080] Real-time feedback and adjustment: Collect user feedback on recommendation results, analyze feedback data, and dynamically adjust recommendation strategies and model parameters;
[0081] Visual presentation: Personalized dining recommendations are presented to users in an intuitive manner through the user interface, including charts, lists, and interactive interfaces.
[0082] Beneficial effects of the present invention:
[0083] The present invention integrates the user's dining data, social media data, fitness data and health data through multi-dimensional data fusion to form a comprehensive user portrait. This method makes up for the limitations of single data source analysis in the existing technology and can more accurately capture the user's multi-dimensional behavioral characteristics and preferences, thereby improving the accuracy of personalized recommendations.
[0084] This invention introduces a systematic hyperparameter tuning and dynamic adjustment mechanism, optimizes the model's hyperparameters through grid search, and dynamically adjusts the recommendation strategy based on the user's real-time behavior data and external environment changes. This method ensures the model's prediction accuracy and generalization ability, enabling the recommendation system to flexibly respond to changes in user needs and provide more accurate and practical dining recommendations.
[0085] The present invention fully considers the influence of social networks, and incorporates these factors into the user portrait by analyzing the influence of the user's friends and family on their dining habits. In this way, it can not only provide more targeted personalized recommendations, but also improve the user's satisfaction with the recommended content. Overall, the innovation of the present invention in data fusion, model optimization and social network analysis makes the user dining habit analysis method more comprehensive, accurate and personalized. BRIEF DESCRIPTION OF THE DRAWINGS
[0086] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0087] Figure 1 A schematic diagram of a method for analyzing user dining habits based on big data analysis according to an embodiment of the present invention;
[0088] Figure 2 Schematic diagram of model training according to an embodiment of the present invention. DETAILED DESCRIPTION
[0089] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.
[0090] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0091] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.
[0092] like Figure 1-Figure 2 As shown, a method for analyzing user dining habits based on big data analysis includes the following steps:
[0093] S1, data collection: through the user terminal equipment and the POS system of the catering merchant, the user's catering consumption data is collected in real time, including consumption time, consumption location, consumption amount and order content;
[0094] S2, data preprocessing: cleaning, deduplication, and formatting the collected raw data to generate a standardized user dining data set;
[0095] S3, multi-dimensional data fusion: Integrate the user's dining data with the user's social media data, fitness data, and health data to form a multi-dimensional user portrait, add contextual awareness information, and optimize the user portrait;
[0096] S4, social network analysis: based on multi-dimensional data fusion, further analyze the user's social network data to evaluate the influence of the user's friends and family on their dining habits;
[0097] S5, behavioral feature extraction: extracting user behavioral features from the fused multidimensional data, including consumption frequency, consumption time period preference, and dish preference, and using the results of social network analysis to refine and optimize these features;
[0098] S6, model training: build and train a user dining habit analysis model, model and analyze user behavior characteristics, comprehensively consider social network analysis results, and improve the prediction accuracy of the model;
[0099] S7, Habit Prediction: Based on the trained model, predict the user's future dining habits and generate personalized dining recommendations.
[0100] S1 includes:
[0101] S11, collecting data through user terminal equipment, specifically including:
[0102] Users actively record consumption information while dining through a mobile app, and the device automatically detects the user's dining environment;
[0103] The terminal device records the consumption location through the GPS module;
[0104] The terminal device records the consumption time through the timestamp function;
[0105] Users manually enter or scan restaurant receipts through the app to record the amount of consumption and the content of the order;
[0106] S12, collect data through the POS system of catering merchants, including:
[0107] When a user checks out, the restaurant merchant records the consumption information through the POS system, including the consumption time, consumption location, consumption amount and the content of the order;
[0108] The POS system automatically stores the above data and uploads it to the central server in real time;
[0109] S13, data synchronization and integration, specifically including:
[0110] The user's terminal device and the restaurant merchant's POS system synchronize data through the cloud server;
[0111] The central server receives and integrates consumption data from user terminal devices and POS systems, eliminates duplicate records, and forms a unified consumption data set.
[0112] S2 includes:
[0113] S21, data cleaning:
[0114] S211, check and process missing values: for missing consumption time, consumption location, consumption amount and order content, fill in the missing values through interpolation;
[0115] S212, check and process outliers: use the Zscore method to detect and process outlier data points, expressed as:
[0116] ;
[0117] in, is the data point, is the mean of the data set, is the standard deviation;
[0118] when When it is greater than the set threshold, it is considered an outlier and processed or deleted;
[0119] S22, data deduplication:
[0120] Use the hash function to uniquely identify consumption records, identify and delete duplicate records, expressed as:
[0121] ;
[0122] in, For consumption records, and is a random constant, is a large prime number;
[0123] Calculate a unique identifier through a hash function to identify duplicate data;
[0124] S23, data formatting:
[0125] Convert data from different sources into a unified format and standardize text data;
[0126] S24, generate standardized data set:
[0127] The cleaned, deduplicated and formatted data are integrated into a standardized data structure, where each record contains the time, location, amount and order content of the consumption.
[0128] Through systematic cleaning, deduplication, formatting and standardization, we ensure the accuracy, consistency and integrity of the data. We eliminate missing values and outliers in the data to improve data quality; the deduplication step avoids data redundancy; and unified formatting makes the data easier to process and analyze. Together, these steps ensure that the generated standardized data set can provide a reliable foundation for subsequent multi-dimensional data fusion and behavioral feature extraction, and improve the accuracy and stability of model analysis and prediction.
[0129] S3 includes:
[0130] S31, high-dimensional feature extraction: Use convolutional neural network (CNN) to extract high-dimensional features from each data source, take the high-dimensional features as input data, and construct tensor representation;
[0131] S32, construct tensor representation: construct the processed data into a third-order tensor , expressed as:
[0132] ;
[0133] in, It is rank, , , Represents the number of features of restaurant data, social media data, fitness and health data respectively;
[0134] S33, Tensor Decomposition and Regularization: Use CANDECOMP / PARAFAC (CP) decomposition method to decompose tensors Decompose and add regularization terms to prevent overfitting, expressed as:
[0135] ;
[0136] in, , , is a factor matrix, is the regularization parameter, represents the outer product;
[0137] S34, Multi-task learning: Combined with multi-task learning methods, multiple related tasks are optimized simultaneously during tensor decomposition, including dining habit prediction and health advice, to improve the effect of data fusion;
[0138] S35, online learning and dynamic updating: Introducing online learning mechanism, dynamically updating the tensor model as new data is continuously added to reflect the user's latest behaviors and habits in real time;
[0139] S36, Context-aware fusion: In the tensor fusion process, context-aware information is added, including weather, time, and location, to improve the model's ability to adapt to changes in the external environment.
[0140] S36 includes:
[0141] S361, Data Collection: Real-time collection of situational awareness information, including current weather, time and location, through user terminal devices and environmental sensors;
[0142] S362, data preprocessing: cleaning and formatting the collected situational data to ensure data consistency and accuracy;
[0143] S363, multi-source data integration: Integrate the processed contextual data with the user’s dining data, social media data, fitness data, and health data to form a multi-dimensional context-aware data set;
[0144] S364, Context Feature Extraction: Extract context features from the integrated context-aware dataset and analyze the impact of context data on user dining habits;
[0145] S365, tensor fusion: In the tensor fusion process, the context data is combined with the user's data in other dimensions into the tensor model in combination with the extracted context features to improve the model's ability to adapt to changes in the external environment;
[0146] S366, Context-aware optimization: Dynamically adjust user portraits and recommendation models based on changes in context-aware data to ensure that personalized dining recommendations can adapt to changes in the external environment in real time.
[0147] By introducing external environmental data (such as weather, time, location, etc.), the user profile is made more comprehensive and dynamic, providing personalized and accurate dining recommendations. It improves the model's adaptability to user behavior and prediction accuracy, ensures that the recommendation system responds to environmental changes in real time, and improves user satisfaction and system practicality.
[0148] S4 includes:
[0149] S41, social network data collection: obtaining users’ social relationship data and interactive behavior data from their social media platforms (such as Facebook, Twitter, Instagram, etc.), including likes, comments, shares, and tag information;
[0150] S42, social relationship construction: Use social network analysis technology to construct a user's social relationship graph, identify the user's friends, family members and their interaction frequency and intensity, and form a social relationship network graph;
[0151] S43, influence assessment: Analyze users’ social network data to assess the influence of friends and family on users’ dining habits, including assessing the frequency of interaction and the emotional tendency of the interaction content;
[0152] S44, influencing factor extraction: Extract influencing factors related to catering from social network data, including restaurants recommended by friends, dining places frequented by family members, and dietary preferences of friends and family members.
[0153] By incorporating users' social relationships and interactive behaviors into user portraits through social network analysis, users' dining habits and preferences can be captured more comprehensively. Social influence factors are combined to optimize personalized recommendations, improve the accuracy of recommendations and user satisfaction. This process ensures that the recommendation system not only considers the user's own data, but also the influence of social networks, making the recommendation results more in line with the user's actual needs.
[0154] S5 includes:
[0155] S51, consumption frequency extraction: Calculate the user's consumption frequency from the standardized data set, including the average daily consumption frequency, the average weekly consumption frequency and the average monthly consumption frequency;
[0156] S52, extracting consumption time period preference: analyzing the consumption behavior of users in different time periods, counting the consumption frequency and amount of users in breakfast, lunch, dinner and supper periods, and determining the consumption time period preference of users;
[0157] S53, extracting dish preferences: based on the user's historical consumption records, counting the dishes and cuisines that the user frequently orders, and analyzing the user's dish preferences, including the preferred dish types, flavors, and price ranges;
[0158] S54, integration of social network analysis results: Combined with the results of social network analysis, the influence of friends and family on the user's dining habits is taken into consideration to refine the user's consumption frequency, time period preference and dish preference;
[0159] S55, feature optimization: optimize the extracted behavioral features, remove redundant and noisy data, retain key features, and improve the accuracy of feature extraction.
[0160] S6 includes:
[0161] S61, data preparation: Integrate the extracted user behavior features, fused multidimensional data and social network analysis results into a training data set to ensure data integrity and consistency;
[0162] S62, feature selection: Recursive feature elimination (RFE) is used to select features of training data, retain key features, remove redundant and irrelevant features, and perform feature engineering to enhance model input, expressed as:
[0163] ;
[0164] in, Characterized by is the training data set, For labels, is the importance function used to evaluate features;
[0165] S63, model selection: select the long short-term memory network to build a user dining habit analysis model, expressed as:
[0166] ;
[0167] ;
[0168] ;
[0169] ;
[0170] ;
[0171] ;
[0172] in, are the outputs of different gating units, and is the weight matrix and bias vector;
[0173] S64, model training: Use the prepared training data set to train the user dining habit analysis model, adjust the model parameters to improve the model performance, and use the cross-validation method to verify the stability and generalization ability of the model, expressed as:
[0174] ;
[0175] in, and Respectively Folded training dataset and labels, is the loss function;
[0176] S65, model optimization: Based on the cross-validation results, perform hyperparameter tuning and use GridSearch to optimize the model’s hyperparameters to improve the model’s prediction accuracy.
[0177] S66, Model Evaluation: Evaluate the trained model using the test dataset, including calculating the accuracy of the model , Accuracy , recall rate and F1 score , expressed as:
[0178] ;
[0179] ;
[0180] ;
[0181] ;
[0182] in, , , and They are true positive, true negative, false positive, and false negative;
[0183] S67, Model Deployment: Deploy the trained and optimized model to the system for real-time prediction and personalized recommendation, and dynamically update the model based on the user's latest behavior and data.
[0184] S65 includes:
[0185] S651, determine the hyperparameter range: define the hyperparameters to be tuned and their possible value ranges, and create a hyperparameter grid. The hyperparameters include learning rate, regularization coefficient, number of hidden layers, and number of neurons in each layer;
[0186] S652, Grid Search: Perform grid search, traverse each combination in the hyperparameter grid by exhaustive method, train and verify the model for each combination, and record the performance indicators, expressed as:
[0187] ;
[0188] in, is a set of hyperparameters, is the specific hyperparameter value;
[0189] S653, Hyperparameter combination evaluation: Comprehensively evaluate the verification results of each hyperparameter combination, calculate the average verification error and standard deviation, and select the best hyperparameter combination, expressed as:
[0190] ;
[0191] S654, Model Retraining: Use the best hyperparameter combination to retrain the model on the entire training dataset to ensure the model's optimal performance on new data;
[0192] By determining the scope, grid search, combined evaluation and model retraining, the optimal performance of the model is ensured, the prediction accuracy and generalization ability of the model are improved, making the model more effective and reliable in practical applications.
[0193] S7 includes:
[0194] Behavior prediction: Use the trained model to predict user behavior and output the user's future dining habits, including consumption frequency, consumption time period preference, and dish preference;
[0195] Personalized recommendation generation: Generate personalized dining recommendations based on predicted dining habits, including recommended restaurants, recommended dishes, and nutritional recommendations, ensuring that the recommendations are consistent with the user's preferences and health goals;
[0196] Real-time feedback and adjustment: Collect user feedback on recommendation results, analyze feedback data, dynamically adjust recommendation strategies and model parameters, and continuously optimize personalized recommendation effects;
[0197] Visual presentation: Personalized dining recommendations are presented to users in an intuitive way through the user interface, including charts, lists, and interactive interfaces, to help users understand and adopt the recommendations.
[0198] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.
[0199] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for analyzing user dining habits based on big data analysis, characterized in that: The following steps are involved: S1, data collection: through the user terminal equipment and the POS system of the catering merchant, the user's catering consumption data is collected in real time, including consumption time, consumption location, consumption amount and order content; S2, data preprocessing: cleaning, deduplication, and formatting the collected raw data to generate a standardized user dining data set; S3, multi-dimensional data fusion: Integrate the user's dining data with the user's social media data, fitness data, and health data to form a multi-dimensional user portrait, add contextual awareness information, and optimize the user portrait, including: S31, high-dimensional feature extraction: Use convolutional neural networks to extract high-dimensional features from each data source, use high-dimensional features as input data, and construct tensor representation; S32, construct tensor representation: construct the processed data into a third-order tensor ; S33, Tensor Decomposition and Regularization: Use CANDECOMP / PARAFAC (CP) decomposition method to decompose tensors Decompose and add regularization terms; S34, Multi-task learning: Combining multi-task learning methods, multiple related tasks are optimized simultaneously during tensor decomposition, including dining habit prediction and health recommendations; S35, online learning and dynamic updating: Introducing online learning mechanism, dynamically updating the tensor model as new data is continuously added to reflect the user's latest behaviors and habits in real time; S36, context-aware fusion: In the tensor fusion process, context-aware information is added, including weather, time and location; The S36 includes: S361, Data Collection: Real-time collection of situational awareness information, including current weather, time and location, through user terminal devices and environmental sensors; S362, data preprocessing: cleaning and formatting the collected situational data; S363, multi-source data integration: Integrate the processed contextual data with the user’s dining data, social media data, fitness data, and health data to form a multi-dimensional context-aware data set; S364, Context Feature Extraction: Extract context features from the integrated context-aware dataset and analyze the impact of context data on user dining habits; S365, tensor fusion: In the tensor fusion process, the context data is input into the tensor model together with the user's data in other dimensions in combination with the extracted context features; S366, Context-aware optimization: Dynamically adjust user profiles and recommendation models based on changes in context-aware data; S4, social network analysis: Based on the fusion of multi-dimensional data, further analyze the user's social network data to evaluate the influence of the user's friends and family on their dining habits, including: S41, social network data collection: obtaining users’ social relationship data and interactive behavior data from their social media platforms, including likes, comments, shares, and tag information; S42, social relationship construction: Use social network analysis technology to construct a user's social relationship graph, identify the user's friends, family members and their interaction frequency and intensity, and form a social relationship network graph; S43, influence assessment: Analyze users’ social network data to assess the influence of friends and family on users’ dining habits, including assessing the frequency of interaction and the emotional tendency of the interaction content; S44, influencing factor extraction: extracting influencing factors related to catering from social network data, including restaurants recommended by friends, dining places frequented by family members, and dietary preferences of friends and family members; S5, behavioral feature extraction: extracting user behavioral features from the fused multidimensional data, including consumption frequency, consumption time period preference, and dish preference, and using the results of social network analysis to refine and optimize these features; S6, model training: build and train a user dining habit analysis model, model and analyze user behavior characteristics, and comprehensively consider social network analysis results; S7, Habit Prediction: Based on the trained model, predict the user's future dining habits and generate personalized dining recommendations.
2. According to the method for analyzing user dining habits based on big data analysis according to claim 1, it is characterized in that: The S1 includes: S11, collecting data through user terminal equipment, specifically including: Users actively record consumption information while dining through a mobile app, and the device automatically detects the user's dining environment; The terminal device records the consumption location through the GPS module; The terminal device records the consumption time through the timestamp function; Users manually enter or scan restaurant receipts through the app to record the amount of consumption and the content of the order; S12, collect data through the POS system of catering merchants, including: When a user checks out, the restaurant merchant records the consumption information through the POS system, including the consumption time, consumption location, consumption amount and the content of the order; The POS system automatically stores the above data and uploads it to the central server in real time; S13, data synchronization and integration, specifically including: The user's terminal device and the restaurant merchant's POS system synchronize data through the cloud server; The central server receives and integrates consumption data from user terminal devices and POS systems, eliminates duplicate records, and forms a unified consumption data set.
3. According to the method for analyzing user dining habits based on big data analysis in claim 1, it is characterized in that: The S2 includes: S21, data cleaning: S211, check and process missing values: for missing consumption time, consumption location, consumption amount and order content, fill in the missing values through interpolation; S212, check and process outliers: use the Zscore method to detect and process outlier data points; S22, data deduplication: Use hash functions to uniquely identify consumption records, identify and delete duplicate records; Calculate a unique identifier through a hash function to identify duplicate data; S23, data formatting: Convert data from different sources into a unified format and standardize text data; S24, generate standardized data set: The cleaned, deduplicated and formatted data are integrated into a standardized data structure, where each record contains the time, location, amount and order content of the consumption.
4. The method for analyzing user dining habits based on big data analysis according to claim 1, characterized in that: The S5 includes: S51, consumption frequency extraction: Calculate the user's consumption frequency from the standardized data set, including the average daily consumption frequency, the average weekly consumption frequency and the average monthly consumption frequency; S52, consumption time period preference extraction: analyzing the user's consumption behavior in different time periods, and counting the user's consumption frequency and amount in the breakfast, lunch, dinner and supper periods; S53, extracting dish preferences: based on the user's historical consumption records, counting the dishes and cuisines that the user frequently orders, and analyzing the user's dish preferences, including the preferred dish types, flavors, and price ranges; S54, integration of social network analysis results: Combined with the results of social network analysis, the influence of friends and family on the user's dining habits is taken into consideration to refine the user's consumption frequency, time period preference and dish preference; S55, feature optimization: optimize the extracted behavioral features, remove redundant and noisy data, and retain key features.
5. The method for analyzing user dining habits based on big data analysis according to claim 1 is characterized in that: The S6 includes: S61, data preparation: Integrate the extracted user behavior features, fused multidimensional data and social network analysis results into a training data set; S62, feature selection: Use recursive feature elimination to select features from training data, retain key features, and remove redundant and irrelevant features; S63, model selection: select long short-term memory network to build a user dining habit analysis model; S64, model training: use the prepared training data set to train the user dining habit analysis model, adjust the model parameters to improve the model performance, and use the cross-validation method to verify the stability and generalization ability of the model; S65, model optimization: perform hyperparameter tuning based on the cross-validation results to optimize the model’s hyperparameters; S66, Model Evaluation: Evaluate the trained model using the test dataset, including calculating the accuracy of the model , Accuracy , recall rate and F1 score ; S67, Model Deployment: Deploy the trained and optimized model to the system for real-time prediction and personalized recommendation, and dynamically update the model based on the user's latest behavior and data.
6. The method for analyzing user dining habits based on big data analysis according to claim 5 is characterized in that: The S65 includes: S651, determine the hyperparameter range: define the hyperparameters to be tuned and their possible value ranges, and create a hyperparameter grid. The hyperparameters include learning rate, regularization coefficient, number of hidden layers, and number of neurons in each layer; S652, Grid Search: Perform grid search, traverse each combination in the hyperparameter grid by exhaustive method, train and verify the model for each combination, and record the performance indicators; S653, Hyperparameter combination evaluation: Comprehensively evaluate the verification results of each hyperparameter combination, calculate the average verification error and standard deviation, and select the best hyperparameter combination; S654, Model Retraining: Retrain the model on the entire training dataset using the best hyperparameter combination.
7. The method for analyzing user dining habits based on big data analysis according to claim 1, characterized in that: The S7 includes: Behavior prediction: Use the trained model to predict user behavior and output the user's future dining habits, including consumption frequency, consumption time period preference, and dish preference; Personalized recommendation generation: Generate personalized dining recommendations based on predicted dining habits, including recommended restaurants, recommended dishes, and nutritional recommendations; Real-time feedback and adjustment: Collect user feedback on recommendation results, analyze feedback data, and dynamically adjust recommendation strategies and model parameters; Visual presentation: Personalized dining recommendations are presented to users in an intuitive manner through the user interface, including charts, lists, and interactive interfaces.
Citation Information
Patent Citations
Cloud catering platform based on big data feature analysis and analysis method thereof
CN107481119A
Catering recommendation method and system based on user portrait
CN117670439A
Credit card preferential information recommendation method and system based on user consumption portrait
CN117745348A
