Information consultation service system based on big data

By building an information consulting service system based on big data, the shortcomings of the existing system in data collection and analysis have been solved, enabling real-time data processing and personalized services, thereby improving user experience and enterprise decision support capabilities.

CN118916549APending Publication Date: 2024-11-08BEIJING ZHONGJING FENGRONG TECHNOLOGY CO LTD
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411331568.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-24
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing information consulting service systems are unable to effectively cope with the demand for real-time data streams, especially in terms of data collection, processing timeliness, and data integration. They are unable to deeply analyze user behavior and market trends, making it difficult for services to meet users' personalized needs and provide in-depth market insights.

Method used

A big data-based information consulting service system was designed, which includes modules for data collection, data processing, user behavior analysis, context awareness, market trend analysis, and personalized recommendation. Through real-time data collection from multiple data sources, advanced data cleaning and preliminary analysis, user profile construction, market trend prediction, and personalized service generation, combined with technologies such as graph neural networks and logistic regression models, the system can dynamically adjust service output.

Benefits of technology

It enables real-time and accurate responses to user behavior and market dynamics, provides in-depth user and market insights, generates highly personalized information and consulting, improves service relevance and user satisfaction, and enhances the company's market competitiveness.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916549A_ABST
    Figure CN118916549A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of information consultation, in particular to an information consultation service system based on big data, which comprises a data acquisition module, a data processing module, a user behavior analysis module, a situation awareness module, a market trend analysis module and a personalized recommendation module, the data acquisition module collects real-time data; the data processing module performs data cleaning, preliminary analysis and data integration on the real-time data; the user behavior analysis module analyzes historical behaviors and preferences of the user; the context awareness module utilizes context information to adjust and optimize information service output; the market trend analysis module identifies a market trend and influence factors; and the personalized recommendation module generates personalized information consultation and suggestions. According to the invention, the pertinence and effect of information service are enhanced, and data support is provided for strategy adjustment and decision making of enterprises.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of information consulting technology, and in particular to an information consulting service system based on big data. Background Art

[0002] In today's digital age, the rapid development of information technology has led to a sharp increase in the amount of data. Enterprises and individuals are faced with the challenge of how to extract valuable information from large-scale data. Although existing information consulting service systems have made certain progress in processing and analyzing data, these systems are often unable to effectively respond to the needs of real-time data streams, especially in terms of data collection, processing timeliness and data integration. There are obvious deficiencies, which limits the ability of enterprises to respond promptly in a rapidly changing market environment.

[0003] In addition, traditional information consulting services are also incapable of user behavior analysis and market trend insights. They mainly focus on superficial user activities, such as clicks and browsing history, but ignore deep behavioral motivations and complex user preference analysis, which makes it difficult for services to truly meet users' personalized needs. At the same time, these systems are also relatively limited in their ability to integrate and interpret market data, especially in predicting market trends and movements, and are unable to provide companies with in-depth market insights to support decision-making. Summary of the invention

[0004] Based on the above objectives, the present invention provides an information consulting service system based on big data.

[0005] An information consulting service system based on big data, including a data collection module, a data processing module, a user behavior analysis module, a situational awareness module, a market trend analysis module and a personalized recommendation module, wherein;

[0006] The data collection module is used to collect real-time data from different types of data sources, and the real-time data includes user data and environmental data;

[0007] The data processing module is used to perform data cleaning, preliminary analysis and data integration on real-time data;

[0008] The user behavior analysis module is used to analyze the user's historical behavior and preferences, form a user portrait, and use the real number data obtained from the data acquisition module to understand the driving factors of user behavior;

[0009] The context-aware module uses context information (such as the user's geographic location, the type of device used and the time, etc.) to adjust and optimize the information service output. The context-aware module dynamically adjusts the service output by analyzing the user's current context in real time;

[0010] The market trend analysis module is used to analyze market data and identify market trends and influencing factors;

[0011] The personalized recommendation module generates personalized information consultation and suggestions based on user portraits and market trend data. The personalized recommendation module integrates the analysis results of user behavior and market trends to provide information services.

[0012] Furthermore, the data acquisition module includes:

[0013] Data source and interface configuration: The data acquisition module is equipped with multiple interface units, each of which is connected to different types of data sources, including online social platforms, IoT devices, mobile applications, and environmental monitoring devices;

[0014] Real-time data collection: User data and environmental data are collected in real time from different types of data sources. User data includes users' online behavior, geographic location information, device usage information, and interaction records. Environmental data includes temperature, humidity, air quality, and noise level.

[0015] Furthermore, the data processing module includes:

[0016] Data cleaning: Data cleaning is used to remove invalid data, duplicate data, and erroneous data generated during the collection process, including identifying and correcting outliers in the data, filling in missing values, and standardizing data formats and types;

[0017] Preliminary data analysis: The data processing module performs preliminary analysis on the cleaned real-time data. The preliminary analysis includes calculating statistical indicators and performing correlation analysis and trend analysis. The statistical indicators include mean, median, and standard deviation.

[0018] Data integration: Merging data from different types of data sources into a unified data set, including associating and mapping various data.

[0019] Furthermore, the user behavior analysis module includes:

[0020] Historical behavior data collection: Receive the user's historical behavior data from the data collection module. The historical behavior data includes the user's browsing history, purchase history, search queries, application usage, and social media activities;

[0021] User portrait construction: Based on the collected historical behavior data, the user's behavior patterns and preferences are identified through the latent class analysis (LCA) model, and then the user portrait is constructed. The user portrait includes the user's interests, activity frequency, and preferred content type;

[0022] Behavior Driver Analysis: Analyzes real-time data obtained from the data acquisition module and uses the Adaptive Behavior Influence Network (ABIN) algorithm to identify the drivers that influence user behavior.

[0023] Further, the latent class analysis (LCA) model includes:

[0024] Define latent categories and feature parameters: Suppose there are K latent categories, and each user u is represented by a probability vector π u Description, then π uk represents the probability that user u belongs to latent category k, and each latent category corresponds to a set of parameters θ k , the calculation formula is:

[0025]

[0026] Among them, γ uk is the log odds that user u belongs to latent class k, K is the total number of latent classes, γ uj is the log odds that user u belongs to latent class j;

[0027] Probabilistic definition of the model: User u’s behavior X on content category c uc , according to the potential category it belongs to, there are different probability distributions, and the calculation formula is:

[0028] P(X uc =x|Z u = k, θ) = f(x; θ kc );

[0029] Among them, Z u represents the potential category to which user u belongs, Θ is the set of all parameters of the latent category analysis model, f(x; θ kc ) is the probability density function;

[0030] Application of expectation maximization EM algorithm:

[0031] Step E: Calculate the expected probability (posterior probability) that each user belongs to each potential category. The calculation formula is:

[0032]

[0033] Among them, X u is all the behavior data of user u, P(Z u =k|X u ,Θ) is the given user behavior data X u and latent class analysis model parameters Θ, the posterior probability that user u belongs to latent class k;

[0034] M step: Update parameters π and θ to maximize the log-likelihood of the data, calculated as:

[0035]

[0036]

[0037] in, and θ o ld are the newly calculated parameter values ​​and the old parameter values ​​used in the EM algorithm, respectively, and N is the total number of users.

[0038] Further, the Adaptive Behavior Influence Network (ABIN) algorithm includes:

[0039] Data input and preprocessing: collect user behavior data in real time, including clicks, browsing, and purchases, and format the user behavior data;

[0040] Network initialization: Each user behavior type is initialized as a node, and the initial impact relationship between user behaviors is estimated based on historical data, and the edge weight is set. The edge weight calculation formula is:

[0041]

[0042] Among them, w ij is the edge weight, I i,t represents the indicator function of the user's behavior i at time t, I j,t+Δt represents the indicator function of the user performing behavior j at time t+Δt, where Δt is the time interval and T is the total observation time period;

[0043] Real-time update mechanism: As new user behavior data arrives, the nodes and edges of the network are updated in real time. The online gradient descent method is applied to dynamically adjust the weights of the edges according to the new behavior data. The calculation formula is:

[0044]

[0045] in, is the updated edge weight, is the edge weight before updating, η is the learning rate;

[0046] Behavior impact analysis: Apply graph neural networks (GNNs) to analyze the relationships between nodes, identify behavior drivers, and predict future user behavior based on the results of graph neural networks.

[0047] Furthermore, the graph neural network (GNN) includes:

[0048] Build a graph structure: define each user behavior or behavior type as a node in the graph, and define the relationship between behaviors (such as sequence, similarity, or other logical associations) as edges;

[0049] Node feature initialization: Initialize node feature vector

[0050] Information aggregation and node update: For each node in the graph, aggregate information from neighboring nodes and update the state;

[0051] The calculation formula of the information aggregation is:

[0052] in, is the information that node v aggregates from neighboring nodes at layer l, N(v) is the set of neighboring nodes of node v, is the feature representation of the neighbor node at layer l;

[0053] The calculation formula for the node update is:

[0054] Among them, W (l) and b (l) are the weight matrix and bias vector of the lth layer, φ is the activation function, and ‖ represents the connection operation;

[0055] Read out the global information of the graph: Use the updated features of all nodes to predict the final task. The calculation formula is:

[0056] in, Yes, the predicted results, Represents the features of all nodes in the graph at the last layer.

[0057] Furthermore, the context awareness module includes:

[0058] Contextual data collection: Collect contextual data, which includes the user's geographic location, the type of device used, the current time, and contextual information related to the environment. Contextual data is obtained through real-time communication with the user's device or directly extracted from the device's information;

[0059] Contextual information analysis: Based on the collected contextual data, context-aware algorithms are applied to analyze the user's current context, including identifying the user's physical environment (such as at home, at work, outdoors, etc.), device usage (such as using a mobile phone or a desktop), and the time of the user's activity (such as morning, midnight);

[0060] Dynamic adjustment of service output: Based on the results of contextual information analysis, the output of information services is dynamically adjusted, including pushing relevant weather forecasts or local news based on the user's geographic location, adjusting the display format of content to adapt to different screen sizes and operating devices based on the type of device used, and pushing schedule reminders or activity suggestions based on the time of the user's activity.

[0061] Furthermore, the context-aware algorithm includes:

[0062] Initial processing of situational data: standardizing all collected situational data into a unified scope or format;

[0063] Feature extraction and encoding: Construct features based on the collected contextual data and use one-hot encoding to encode categorical features. The calculation formula is:

[0064] One-Hot Encoding:x encoded =one_hot(x category );

[0065] Among them, x encoded is the data after one-hot encoding, x category is a feature in categorical data;

[0066] Situation model construction: Based on the extracted features, a logistic regression model is used to predict user needs or behaviors. The logistic regression model is expressed as:

[0067] y=ψ(W·x+b);

[0068] Among them, y is the prediction result, x is the input data, W and b are the logistic regression model parameters, and ψ is the sigmoid function.

[0069] Furthermore, the market trend analysis module includes:

[0070] Market data collection: Collect various market-related data, including sales data, market share, consumer behavior data, economic indicators and competitor activities;

[0071] Market data processing: cleaning, formatting and standardizing the collected relevant data;

[0072] Trend identification and analysis: Use the time series analysis ARIMA model to analyze the time trend of sales and market data. The time series analysis ARIMA model is expressed as:

[0073]

[0074] Among them, X t is the data point at time t, β and δ are model parameters, ζ is the error term, p and q are the order of the model, and c is the constant term;

[0075] Analysis of influencing factors: Use regression analysis models to analyze the impact of various influencing factors (such as price changes, promotional activities, seasonal changes, etc.) on market dynamics;

[0076] Market forecasting: Based on the identified trends and influencing factors, the random forest model is used to predict future market trends and changes in user demand.

[0077] Beneficial effects of the present invention:

[0078] The present invention, through its data acquisition module and data processing module, can collect and quickly process large-scale data in real time from a wide range of data sources. The data acquisition module covers a variety of data sources from social platforms to Internet of Things devices, ensuring the diversity and comprehensiveness of information. The data processing module uses advanced data cleaning and preliminary analysis techniques to not only improve the quality and availability of data, but also ensure the timeliness and accuracy of information services. This efficient collection and processing of real-time data provides a solid data foundation for subsequent in-depth analysis, enabling the system to respond instantly and accurately to user behavior and market dynamics.

[0079] The present invention provides in-depth user and market insights by integrating the user behavior analysis module and the market trend analysis module. The user behavior analysis module uses historical behavior data and real-time data to build a detailed user portrait and effectively predict the driving factors and future trends of user behavior. The market trend analysis module conducts in-depth analysis of market data to identify key market trends and influencing factors, and provides companies with valuable insights into market trends. This analysis capability that combines the two dimensions of users and markets not only enhances the pertinence and effectiveness of information services, but also provides data support for corporate strategy adjustments and decision-making.

[0080] The present invention generates highly personalized information consultation and suggestions by integrating user portraits and market trend data. The context awareness module dynamically adjusts service output by real-time analysis of the user's current context, thereby improving the personalization and practicality of the service. This dynamic and highly customized service output not only improves the user experience, but also enhances the relevance and actual utility of the service. In the fierce market competition, this refined service adjustment mechanism helps enterprises interact with users at the right time and in an appropriate manner, greatly improving user satisfaction and the market competitiveness of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] 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.

[0082] Figure 1 A schematic diagram of system function modules according to an embodiment of the present invention;

[0083] Figure 2 Schematic diagram of a user behavior analysis module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0084] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0085] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0086] like Figure 1-Figure 2 As shown, an information consulting service system based on big data includes a data collection module, a data processing module, a user behavior analysis module, a situational awareness module, a market trend analysis module and a personalized recommendation module, wherein;

[0087] The data collection module is used to collect real-time data from different types of data sources, including user data and environmental data;

[0088] The data processing module is used to clean, perform preliminary analysis and integrate real-time data to ensure data quality and availability;

[0089] The user behavior analysis module is used to analyze the user's historical behavior and preferences, form a user profile, and use the real data obtained from the data acquisition module to understand the driving factors of user behavior;

[0090] The context-aware module uses contextual information (such as the user's geographic location, the type of device used, and the time) to adjust and optimize the information service output. The context-aware module dynamically adjusts the service output by analyzing the user's current context in real time to enhance the personalization and practicality of the service.

[0091] The market trend analysis module is used to analyze market data, identify market trends and influencing factors, and help predict market trends and changes in user needs;

[0092] The personalized recommendation module generates personalized information consultation and suggestions based on user portraits and market trend data. The personalized recommendation module integrates the analysis results of user behavior and market trends to provide information services;

[0093] Through the above content, integrated multi-module collaborative processing has achieved efficient collection and in-depth analysis of real-time data, thereby significantly improving the accuracy and personalization of information services. By dynamically adjusting service output to adapt to the user's specific situation, the system not only enhances the user experience, but also improves the practicality and relevance of the service. In addition, the system can accurately predict market trends and changes in user demand, and provide strong data support for decision-making, thus having a significant advantage in the highly competitive market.

[0094] The data acquisition module includes:

[0095] Data source and interface configuration: The data acquisition module is equipped with multiple interface units, each of which connects to different types of data sources, including online social platforms, IoT devices, mobile applications, and environmental monitoring devices, ensuring that the system can cover a wide range of data inputs and collect data from various backgrounds and conditions;

[0096] Real-time data collection: User data and environmental data are collected in real time from different types of data sources. User data includes users' online behavior, geographic location information, device usage information, and interaction records. Environmental data includes temperature, humidity, air quality, and noise level. Real-time collection of this data is achieved through efficient data transmission protocols and network connections to ensure the timeliness and integrity of the data.

[0097] The data processing module includes:

[0098] Data cleaning: Data cleaning is used to remove invalid data, duplicate data, and erroneous data generated during the collection process, including identifying and correcting outliers in the data, filling in missing values, and standardizing data formats and types to ensure data accuracy and consistency;

[0099] Preliminary data analysis: The data processing module performs preliminary analysis on the cleaned real-time data. The preliminary analysis includes calculating statistical indicators and performing correlation analysis and trend analysis. The statistical indicators include mean, median, and standard deviation to reveal the basic properties of the data and potential information patterns.

[0100] The mean calculation formula is:

[0101] Among them, x i represents the i-th observation, and n is the total number of observations;

[0102] The formula for calculating standard deviation is:

[0103] Where σ is the standard deviation, μ is the mean, and x i is each observation, and n is the total number of observations;

[0104] Correlation analysis: The linear correlation between two variables is measured by calculating the Pearson correlation coefficient. The calculation formula is:

[0105]

[0106] Where n is the number of data points, x i and i are the x-value and y-value of the ith data point, and is the mean of the x and y data series;

[0107] The value of the correlation coefficient r ranges from -1 to +1, where +1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no correlation;

[0108] Trend analysis: Trend analysis uses a linear regression model to evaluate the trend of data over time. The calculation formula of the linear regression model is:

[0109] y=β0+β1x+;

[0110] Among them, y is the dependent variable, x is the independent variable (usually time), β0 is the intercept, β1 is the slope (indicating the direction and strength of the trend), and is the error term;

[0111] The slope β1 is calculated by the least squares method, and the calculation formula is:

[0112] The calculation formula for intercept β0 is:

[0113] Data integration: Merging data from different types of data sources into a unified data set, including associating and mapping various data, such as combining a user’s geographic location data with their online behavior data to form a more complete user profile;

[0114] Through the above content, not only the availability and quality of the data are improved, but also a solid data foundation is provided for subsequent in-depth analysis and personalized services. Through systematic data cleaning, preliminary analysis and integration steps, the system can effectively improve data processing efficiency and the accuracy of analysis results.

[0115] User behavior analysis modules include:

[0116] Historical behavior data collection: Receive the user's historical behavior data from the data collection module. The historical behavior data includes the user's browsing history, purchase history, search queries, application usage, and social media activities. The historical behavior data provides a preliminary view of the user's past behavior patterns and preferences;

[0117] User portrait construction: Based on the collected historical behavior data, the user's behavior patterns and preferences are identified through the latent class analysis (LCA) model, and then the user portrait is constructed. The user portrait includes the user's interests, activity frequency, and preferred content type. This information is the basis for customized personalized services and recommendations;

[0118] Behavior Driver Analysis: Analyze the real-time data obtained from the data acquisition module and use the Adaptive Behavior Influence Network (ABIN) algorithm to identify the drivers that influence user behavior;

[0119] Through these steps, the user behavior analysis module can not only build in-depth user portraits, but also understand and predict changing trends in user behavior, thus laying a solid foundation for providing more accurate and personalized information services.

[0120] The Latent Class Analysis (LCA) model includes:

[0121] Define latent categories and feature parameters: Suppose there are K latent categories, and each user u is represented by a probability vector π u Description, then π uk represents the probability that user u belongs to latent category k, and each latent category corresponds to a set of parameters θ k , the calculation formula is:

[0122]

[0123] Among them, γ uk is the logarithmic probability that user u belongs to latent class k, which can be predicted by the user's basic characteristics, K is the total number of latent classes, γ uj is the log odds that user u belongs to latent class j;

[0124] Probabilistic definition of the model: User u’s behavior X on content category c uc , such as clicks, purchases, etc., have different probability distributions according to the potential categories they belong to. The calculation formula is:

[0125] P(X uc =x|Z u = k, θ) = f(x; θ kc );

[0126] Among them, Z urepresents the latent category to which user u belongs, Θ is the set of all parameters of the latent category analysis model, including θ for all categories k and probability π k ,f(x;θ kc ) is the probability density function, given the latent class k and the associated parameters θ kc , describes the probability distribution of user behavior x;

[0127] Application of expectation maximization EM algorithm:

[0128] Step E: Calculate the expected probability (posterior probability) that each user belongs to each potential category. The calculation formula is:

[0129]

[0130] Among them, X u is all the behavior data of user u, P(Z u =k|X u ,Θ) is the given user behavior data X u and latent class analysis model parameters Θ, the posterior probability that user u belongs to latent class k;

[0131] M step: Update parameters π and θ to maximize the log-likelihood of the data, calculated as:

[0132]

[0133] in, and θ o ld are the newly calculated parameter values ​​and the old parameter values ​​used in the EM algorithm, respectively, and N is the total number of users;

[0134] Through the above steps, the LCA model can not only process and analyze large-scale user behavior data, but also dynamically adjust model parameters according to the user's real-time data, so as to more accurately build user portraits and predict user preferences and behavior patterns. The flexibility and dynamism of this model makes it particularly suitable for data-intensive and rapidly changing information consulting service systems.

[0135] Probability density function f(x;θ kc ) uses normal distribution, Bernoulli distribution, Poisson distribution and multinomial distribution, where;

[0136] The formula for normal distribution (continuous data) is:

[0137] Where μ is the mean and σ is the standard deviation. In the LCA model, if the distribution of category k for variable c is normal, then θ kc It includes μ and σ;

[0138] The calculation formula for Bernoulli distribution (binary data) is: f(x; p) = p x (1-p) 1-x ;

[0139] Where p is the probability of the outcome being 1. In the LCA model, θ kc is p, which describes the probability that variable x takes the value 1 in category k;

[0140] The calculation formula for Poisson distribution (count data) is:

[0141] Among them, λ is the average number of events. In LCA, for the variable c of category k, θ kc It is λ;

[0142] The calculation formula for multinomial distribution (multi-category data) is:

[0143] Here, x is the outcome vector, p is the probability vector for each category outcome, and n is the total number of trials.

[0144] The Adaptive Behavior Influence Network (ABIN) algorithm includes:

[0145] Data input and preprocessing: collect user behavior data in real time, including clicks, browsing, and purchases, and format the user behavior data;

[0146] Network initialization: Each user behavior type is initialized as a node, and the initial impact relationship between user behaviors is estimated based on historical data, and the edge weight is set. The edge weight calculation formula is:

[0147]

[0148] Among them, w ij is the edge weight, which indicates the influence of behavior i on behavior j. i,t The indicator function represents the user's behavior i at time t. If the user performs behavior i at time t, it is 1, otherwise it is 0. j,t+Δt The indicator function represents the user performing behavior j at time t+Δt. If the user performs behavior j at time t+Δt, it is 1, otherwise it is 0. Δt is the time interval and T is the total observation time period;

[0149] Real-time update mechanism: As new user behavior data arrives, the nodes and edges of the network are updated in real time. The online gradient descent method is applied to dynamically adjust the weights of the edges according to the new behavior data. The calculation formula is:

[0150]

[0151] in, is the updated edge weight, is the edge weight before updating, η is the learning rate, which is used to adjust the speed and amplitude of weight updating;

[0152] Behavior impact analysis: Apply graph neural networks (GNNs) to analyze the relationships between nodes, identify behavior drivers, and predict future user behavior based on the results of graph neural networks;

[0153] By applying the Adaptive Behavior Influence Network (ABIN) algorithm, the information consulting service system can analyze and predict user behavior in real time and dynamically. By continuously learning and adapting to changes in user behavior, the algorithm can accurately reveal and respond to the potential drivers of user behavior, thereby providing more accurate personalized services and effective business decision support. This method is particularly suitable for processing and analyzing large-scale real-time data, ensuring that the information consulting service system maintains its advantage in a highly competitive market environment.

[0154] Graph Neural Networks (GNNs) include:

[0155] Build a graph structure: define each user behavior or behavior type as a node in the graph, and define the relationship between behaviors (such as sequence, similarity, or other logical associations) as edges;

[0156] Node feature initialization: Initialize node feature vector Typically, it is a code of behavior or other descriptive data;

[0157] Information aggregation and node update: For each node in the graph, aggregate information from neighboring nodes and update the state;

[0158] The calculation formula for information aggregation is:

[0159] in, is the information that node v aggregates from neighboring nodes at layer l, N(v) is the set of neighboring nodes of node v, is the feature representation of the neighbor node at layer l;

[0160] The calculation formula for node update is:

[0161] Among them, W (l) and b (l) are the weight matrix and bias vector of the lth layer, φ is the activation function, and ‖ represents the connection operation, which combines the node's own features with the aggregated features;

[0162] The calculation formula of the activation function φ is: φ(x) = max(0,x);

[0163] Read out the global information of the graph: Use the updated features of all nodes to predict the final task, such as classification or prediction of user behavior. The calculation formula is:

[0164] in, Yes, the prediction result can be the output of node classification, graph classification or other tasks. Represents the features of all nodes in the graph at the last layer;

[0165] In this way, graph neural networks can effectively analyze and predict user behavior because they integrate the complex interrelationships between behaviors and consider the temporal dependencies of behavioral data. In information consulting service systems, this capability enables the system to provide personalized recommendations and services based on user behavior patterns, thereby improving user satisfaction and the overall effectiveness of the system. By continuously updating node features, GNN can reflect the latest changes in user behavior in real time, thereby providing support for dynamically changing user needs.

[0166] The situational awareness module includes:

[0167] Contextual data collection: Collect contextual data, which includes the user's geographic location, the type of device used, the current time, and contextual information related to the environment. Contextual data is obtained through real-time communication with the user's device or directly extracted from the device's information;

[0168] Contextual information analysis: Based on the collected contextual data, context-aware algorithms are applied to analyze the user's current context, including identifying the user's physical environment (such as at home, at work, outdoors, etc.), device usage (such as using a mobile phone or a desktop), and the time of the user's activity (such as morning, midnight);

[0169] Dynamic adjustment of service output: Dynamically adjust the output of information services based on the results of contextual information analysis, including pushing relevant weather forecasts or local news based on the user's geographic location, adjusting the display format of content to adapt to different screen sizes and operating devices based on the type of device used, and pushing schedule reminders or activity suggestions based on the time of the user's activity;

[0170] Through the above content, the context awareness module not only reflects current user needs, but also predicts possible changes in demand in the future, making the information consulting service system have a significant advantage in providing highly relevant and timely information.

[0171] Context-aware algorithms include:

[0172] Preliminary processing of contextual data: standardize all collected contextual data to a unified scope or format; feature extraction and coding: construct features based on the collected contextual data, such as extracting whether it is a weekday or a weekend from the time data, extracting the user's city or region from the geographic location information, and using one-hot encoding to classify features. The calculation formula is:

[0173] One-Hot Encoding:x encoded =one_hot(x category );

[0174] Among them, x encoded is the data after one-hot encoding, x category is a feature in categorical data;

[0175] Situation model construction: Based on the extracted features, a logistic regression model is used to predict user needs or behaviors. The logistic regression model is expressed as:

[0176] y=ψ(W·x+b);

[0177] Among them, y is the prediction result, x is the input data, W and b are the logistic regression model parameters, ψ is the sigmoid function, which is used to output the prediction result. The calculation formula of the sigmoid function is:

[0178] In the present invention, the context-aware algorithm enables the information consulting service system to adjust the service output in real time to match the user's current context and needs. The implementation of this algorithm not only improves the relevance and personalization of the service, but also enhances the user's satisfaction and engagement. By accurately analyzing the user's contextual data and providing customized services based on this data, the system can interact with the user at the right time and place in an appropriate manner, greatly improving service efficiency and quality. This dynamic service adjustment mechanism is an indispensable part of modern information services, especially in the digital age of information overload, and can ensure the practicality and timeliness of information.

[0179] Market trend analysis modules include:

[0180] Market data collection: Collect various market-related data, including sales data, market share, consumer behavior data, economic indicators and competitor activities;

[0181] Market data processing: Clean, format and standardize the collected relevant data to ensure data quality and consistency and provide accurate input for subsequent analysis;

[0182] Trend identification and analysis: Use the time series analysis ARIMA model to analyze the time trend of sales and market data. The time series analysis ARIMA model is expressed as:

[0183]

[0184] Among them, X t is the data point at time t, β and δ are model parameters, ζ is the error term, p and q are the orders of the model, and c is a constant term representing the intercept in the model;

[0185] Analysis of influencing factors: Use regression analysis models to analyze the impact of various influencing factors (such as price changes, promotional activities, seasonal changes, etc.) on market dynamics;

[0186] The regression analysis model is expressed as:

[0187] Y=β0+β1X1+β2X2+…+β n X n +;

[0188] Among them, Y is the market response variable (such as sales volume), X1, X2, …, X n is the influencing factor, β is the coefficient, which indicates the influence of each factor, and is the error term;

[0189] Market forecasting: Based on the identified trends and influencing factors, the random forest model is used to predict future market trends and changes in user demand;

[0190] The random forest model is represented as:

[0191] in, is the predicted market response, k is the number of trees, and each tree Tree i Provide a prediction result;

[0192] In the present invention, the market trend analysis module can effectively reveal the laws and trends of market development and the actual impact of various market activities through in-depth analysis and learning of large-scale market data. This analysis and prediction capability enables the information consulting service system to provide customers with data-based insights, helping them make smarter business decisions and optimize market strategies, thereby maintaining a leading edge in a highly competitive business environment. The market trend analysis module can not only identify short-term market changes, but also predict long-term market trends. Such insight enables companies to prepare in advance and adjust their market strategies and resource allocation to adapt to future market changes.

[0193] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0194] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An information consulting service system based on big data, characterized in that: It includes data collection module, data processing module, user behavior analysis module, situational awareness module, market trend analysis module and personalized recommendation module, among which; The data collection module is used to collect real-time data from different types of data sources, and the real-time data includes user data and environmental data; The data processing module is used to perform data cleaning, preliminary analysis and data integration on real-time data; The user behavior analysis module is used to analyze the user's historical behavior and preferences, form a user profile, and use the real data obtained from the data acquisition module to understand the driving factors of user behavior; The context-aware module uses context information to adjust and optimize information service output. The context-aware module dynamically adjusts service output by analyzing the user's current context in real time. The market trend analysis module is used to analyze market data and identify market trends and influencing factors; The personalized recommendation module generates personalized information consultation and suggestions based on user portraits and market trend data. The personalized recommendation module integrates the analysis results of user behavior and market trends to provide information services.

2. According to the big data-based information consulting service system of claim 1, it is characterized in that: The data acquisition module comprises: Data source and interface configuration: The data acquisition module is equipped with multiple interface units, each of which is connected to different types of data sources, including online social platforms, IoT devices, mobile applications, and environmental monitoring devices; Real-time data collection: User data and environmental data are collected in real time from different types of data sources. User data includes users' online behavior, geographic location information, device usage information, and interaction records. Environmental data includes temperature, humidity, air quality, and noise level.

3. According to claim 2, the information consulting service system based on big data is characterized in that: The data processing module comprises: Data cleaning: Data cleaning is used to remove invalid data, duplicate data, and erroneous data generated during the collection process, including identifying and correcting outliers in the data, filling in missing values, and standardizing data formats and types; Preliminary data analysis: The data processing module performs preliminary analysis on the cleaned real-time data. The preliminary analysis includes calculating statistical indicators and performing correlation analysis and trend analysis. The statistical indicators include mean, median, and standard deviation. Data integration: Merging data from different types of data sources into a unified data set, including associating and mapping various data.

4. The information consulting service system based on big data according to claim 3 is characterized in that: The user behavior analysis module includes: Historical behavior data collection: Receive the user's historical behavior data from the data collection module. The historical behavior data includes the user's browsing history, purchase history, search queries, application usage, and social media activities; User portrait construction: Based on the collected historical behavior data, the user's behavior patterns and preferences are identified through the latent class analysis model, and then the user portrait is constructed. The user portrait includes the user's interests, activity frequency, and preferred content type; Behavior Driver Analysis: Analyzes real-time data obtained from the data acquisition module and uses adaptive behavior impact network algorithms to identify drivers that affect user behavior.

5. The information consulting service system based on big data according to claim 4 is characterized in that: The latent class analysis model includes: Define latent categories and feature parameters: Suppose there are K latent categories, and each user u is represented by a probability vector π u Description, then π uk represents the probability that user u belongs to latent category k, and each latent category corresponds to a set of parameters θ k , the calculation formula is: Among them, γ uk is the log odds that user u belongs to latent class k, K is the total number of latent classes, γ uj is the log odds that user u belongs to latent class j; Probabilistic definition of the model: User u’s behavior X on content category c uc , according to the potential category it belongs to, there are different probability distributions, and the calculation formula is: P(X uc =x|Z u =k,Θ)=f(x:θ kc ): Among them, Z u represents the potential category to which user u belongs, Θ is the set of all parameters of the latent category analysis model, f(x; θ kc ) is the probability density function; Application of expectation maximization EM algorithm: Step E: Calculate the expected probability of each user belonging to each potential category. The calculation formula is: Among them, X u is all the behavior data of user u, P(Z u =k|X u ,Θ) is the given user behavior data X u and latent class analysis model parameters Θ, the posterior probability that user u belongs to latent class k; M step: Update parameters π and θ to maximize the log-likelihood of the data, calculated as: in, and θ o ld are the newly calculated parameter values ​​and the old parameter values ​​used in the EM algorithm, respectively, and N is the total number of users.

6. The information consulting service system based on big data according to claim 4 is characterized in that: The adaptive behavior affects the network algorithm including: Data input and preprocessing: collect user behavior data in real time, including clicks, browsing, and purchases, and format the user behavior data; Network initialization: Each user behavior type is initialized as a node, and the initial impact relationship between user behaviors is estimated based on historical data, and the edge weight is set. The edge weight calculation formula is: Among them, w ij is the edge weight, I i,t represents the indicator function of the user's behavior i at time t, I j,t+Δt represents the indicator function of the user performing behavior j at time t+Δt, where Δt is the time interval and T is the total observation time period; Real-time update mechanism: As new user behavior data arrives, the nodes and edges of the network are updated in real time. The online gradient descent method is applied to dynamically adjust the weights of the edges according to the new behavior data. The calculation formula is: in, is the updated edge weight, is the edge weight before updating, η is the learning rate; Behavior impact analysis: Apply graph neural networks to analyze the relationships between nodes, identify behavior drivers, and predict future user behavior based on the results of graph neural networks.

7. The information consulting service system based on big data according to claim 6 is characterized in that: The graph neural network comprises: Build a graph structure: define each user behavior or behavior type as a node in the graph, and define the relationship between behaviors as edges; Node feature initialization: Initialize node feature vector Information aggregation and node update: For each node in the graph, aggregate information from neighboring nodes and update the state; The calculation formula of the information aggregation is: in, is the information that node v aggregates from neighboring nodes at layer l, N(v) is the set of neighboring nodes of node v, is the feature representation of the neighbor node at layer l; The calculation formula for the node update is: Among them, W (l) and b (l) are the weight matrix and bias vector of the lth layer, φ is the activation function, and ‖ represents the connection operation; Read out the global information of the graph: Use the updated features of all nodes to predict the final task. The calculation formula is: in, is the prediction result, Represents the features of all nodes in the graph at the last layer.

8. The information consulting service system based on big data according to claim 1 is characterized in that: The context awareness module comprises: Contextual data collection: Collect contextual data, which includes the user's geographic location, the type of device used, the current time, and contextual information related to the environment. Contextual data is obtained through real-time communication with the user's device or directly extracted from the device's information; Contextual information analysis: Based on the collected contextual data, context-aware algorithms are applied to analyze the user's current context, including identifying the user's physical environment, device usage, and the time of user activity; Dynamic adjustment of service output: Dynamically adjust the output of information services based on the results of contextual information analysis, including pushing relevant weather forecasts or local news based on the user's geographic location, adjusting the display format of content to adapt to different screen sizes and operating devices based on the type of device used, and pushing schedule reminders or activity suggestions based on the time of the user's activity.

9. The information consulting service system based on big data according to claim 8 is characterized in that: The context-aware algorithm includes: Initial processing of situational data: standardizing all collected situational data into a unified scope or format; Feature extraction and encoding: Construct features based on the collected contextual data and use one-hot encoding to encode categorical features. The calculation formula is: One-Hot Encoding:x encoded =one_hot(x category ); Among them, x encoded is the data after one-hot encoding, x category is a feature in categorical data; Situation model construction: Based on the extracted features, a logistic regression model is used to predict user needs or behaviors. The logistic regression model is expressed as: y=ψ(W·x+b); Among them, y is the prediction result, x is the input data, W and b are the logistic regression model parameters, and ψ is the sigmoid function.

10. The information consulting service system based on big data according to claim 1, characterized in that: The market trend analysis module includes: Market data collection: Collect various market-related data, including sales data, market share, consumer behavior data, economic indicators and competitor activities; Market data processing: cleaning, formatting and standardizing the collected relevant data; Trend identification and analysis: Use the time series analysis ARIMA model to analyze the time trend of sales and market data. The time series analysis ARIMA model is expressed as: Among them, X t is the data point at time t, β and δ are model parameters, ζ is the error term, p and q are the order of the model, and c is the constant term; Influencing factor analysis: Use regression analysis model to analyze the impact of various influencing factors on market dynamics; Market forecasting: Based on the identified trends and influencing factors, the random forest model is used to predict future market trends and changes in user demand.

Citation Information

Cited By

  • Big data analysis and processing method and system based on deep learning

    CN119226778A

  • Method and system for performing network information security analysis by using big data

    CN119835086A

  • Intelligent promotion comprehensive processing system based on e-commerce information

    CN119884491A