Intelligent information allocation system based on big data

Abstract

The invention discloses an intelligent information allocation system based on big data, and relates to the technical field of mobile phone information processing. The system comprises a multi-source data acquisition and integration module and a user behavior and scene analysis module which is used for analyzing integrated data through a machine learning algorithm, establishing a user behavior model and a scene identification model, and identifying the user behavior and scene by analyzing information use habits of a user at different time periods and different geographic positions. Identifying a scene where the user is located; the information priority division module is used for performing priority division on various types of information in the mobile phone according to the user behavior model, the scene recognition result and the attributes of the information; and the intelligent allocation strategy module is used for formulating an intelligent allocation strategy according to the priority of the information and the current scene of the user, allocating the information according to the generated intelligent allocation strategy, and feeding back and optimizing the information in real time. According to the system, intelligent allocation of information can be realized, and the efficiency and experience of a user in the aspect of mobile phone information processing are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mobile phone information processing, and particularly to an information intelligent allocation system based on big data. Background Art

[0002] With the popularization of smart phones, the types and quantities of information generated and received by users on mobile phones are increasing, including text messages, call records, various application notifications, and file transfers. Traditional mobile phone information management methods simply arrange and display information in chronological order or by application category, lacking intelligent allocation capabilities. For example, when users are busy, a large number of low-priority advertisement pushes and social software notifications continuously disturb, affecting the user's concentration; while important work emails or urgent messages may be buried in a large number of information, resulting in users missing key content. In addition, when the storage space of the mobile phone is limited, it is impossible to intelligently manage the storage of various types of information, causing space waste or loss of important data. Currently, some existing simple information management functions cannot deeply analyze and intelligently allocate the massive mobile phone information, and it is difficult to meet the growing needs of users for convenient and efficient use of mobile phones. Summary of the Invention

[0003] The technical problem to be solved by the present invention is how to provide an information intelligent allocation system that can intelligently analyze and process various types of information in a mobile phone, and realize the intelligent allocation of information according to the user's usage habits, scenarios, and needs, so as to improve the efficiency and experience of users in mobile phone information processing.

[0004] To solve the above technical problems, the technical solution adopted by the present invention is: an information intelligent allocation system based on big data, including:

[0005] Multi-source data collection and integration module: used to collect various types of information data in the mobile phone by using the API interface of the mobile phone system, integrate the collected data in different formats and sources, and uniformly store them in the local database of the mobile phone;

[0006] User behavior and scenario analysis module: used to analyze the integrated data through machine learning algorithms, establish user behavior models and scenario recognition models, and identify the scenarios where the user is located by analyzing the user's information usage habits at different time periods and different geographical locations;

[0007] Information priority division module: used to divide the priority of various types of information in the mobile phone according to the user behavior model, scenario recognition results, and the attributes of the information itself;

[0008] Intelligent allocation strategy module: used to formulate intelligent allocation strategies according to the information priority and the user's current scenario, and allocate information according to the generated intelligent allocation strategies;

[0009] Real-time feedback and optimization module: used to collect user feedback on information allocation results in real time, optimize user behavior models, scene recognition models and intelligent allocation strategies based on feedback data, and then use the optimized intelligent allocation strategies to allocate information accordingly.

[0010] The beneficial effects of adopting the above technical solution are: 1) The system described in this application comprehensively collects various types of information in the mobile phone and conducts in-depth integration. By mining the relationships between these data, it can understand the user's behavior patterns and information needs more comprehensively and accurately.

[0011] 2) This application performs personalized data preprocessing based on the usage habits and data characteristics of different users. For some users with more missing data, a more suitable filling method is adopted; for users with more noisy data, data cleaning is strengthened to improve data quality and provide a more reliable data foundation for subsequent model construction and information allocation.

[0012] 3) This application uses a variety of machine learning algorithms to build a user behavior model and a scene recognition model. The K-Means clustering algorithm is used to classify user behavior, and the logistic regression algorithm is used to identify scenes. By combining the advantages of different models, it is possible to more accurately analyze the user's information usage habits and scenes in different time periods and geographical locations.

[0013] 4) The model of this application has the ability to update and optimize in real time. By regularly collecting new data, retraining and adjusting the model, it can promptly reflect changes in user behavior and scenarios. According to the user behavior model, scenario recognition results, and the attributes of the information itself, the various types of information in the mobile phone are comprehensively prioritized. Fully consider the information needs and preferences of users in different scenarios to make the priority division more in line with the actual situation.

[0014] 5) This application develops an exclusive intelligent allocation strategy based on the user's personalized information usage habits and current scenarios. Different users may have different information needs in the same scenario, and this application can accurately allocate information based on the characteristics of each user. Through intelligent allocation strategies, low-priority irrelevant information is shielded or delayed according to user scenarios and information priorities, ensuring that users only receive important information in specific scenarios, reducing information interference and improving user concentration.

[0015] 6) This application pushes important information to users in a timely and accurate manner, adjusts the information display method and location according to user scenarios and needs, and improves the efficiency of users in obtaining information. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0017] Figure 1 is the principle block diagram of the system according to the embodiment of the present invention;

[0018] Figure 2 is the principle block diagram of the multi-source data collection and integration module in the system according to the embodiment of the present invention;

[0019] Figure 3 is the principle block diagram of the user behavior and scenario analysis module in the system according to the embodiment of the present invention;

[0020] Figure 4 is the principle block diagram of the information priority division module in the system according to the embodiment of the present invention;

[0021] Figure 5 is the principle block diagram of the intelligent allocation strategy module in the system according to the embodiment of the present invention. Specific embodiments

[0022] Next, in combination with the accompanying drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar promotions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0024] As Figure 1 shown, the embodiment of the present invention discloses an information intelligent allocation system based on big data. The system includes:

[0025] Multi-source data collection and integration module 101: used to collect various information data in the mobile phone by using the API interface of the mobile phone system, integrate the collected data in different formats and sources, and uniformly store them in the database on the local mobile phone;

[0026] User behavior and scenario analysis module 102: used to analyze the integrated data through machine learning algorithms, establish a user behavior model and a scenario recognition model, and identify the scenario where the user is located by analyzing the user's information usage habits at different times and different geographical locations;

[0027] Information priority division module 103: used to divide the priority of various information in the mobile phone according to the user behavior model, the scenario recognition result, and the attributes of the information itself;

[0028] Intelligent allocation strategy module 104: used to formulate an intelligent allocation strategy according to the information priority and the user's current scenario, and allocate information according to the generated intelligent allocation strategy;

[0029] Real-time feedback and optimization module 105: collect the user's feedback on the information allocation result in real time, optimize the user behavior model, scenario recognition model and intelligent allocation strategy according to the feedback data, and then allocate information accordingly using the optimized intelligent allocation strategy.

[0030] Furthermore, as Figure 2 shown, the multi-source data collection and integration module 101 includes:

[0031] Task initialization unit 1011: First, check the permission settings of the mobile phone system to ensure that the user has legal data collection permissions, such as permissions to obtain call records, text message content, application usage frequency, etc. If the permissions are insufficient, prompt the user to perform an authorization operation, and trigger the initialization of the data collection unit when the verification is passed;

[0032] Collection source and period determination unit 1012: Determine the collection sources of different types of data according to the preset collection strategy. For example, call records are obtained from the call record database of the mobile phone, text message content is read from the text message database, and application usage frequency is obtained through the application management interface of the system; at the same time, to ensure the timeliness and accuracy of the data, set different collection periods. For call records and text message content, monitor data changes in real time and perform incremental collection; for application usage frequency and geographical location information, perform a full-scale collection every 5 minutes (can be adjusted according to actual needs);

[0033] Data collection unit 1013: Collect data according to the set collection source and period by calling the API interface of the mobile phone system. During the collection process, perform preliminary legality verification on the data, such as checking whether the number format in the call record is correct and whether the text message content is complete. For illegal data, record the log and mark it;

[0034] Data format conversion unit 1014: Since the collected data comes from different data sources, the formats may vary. The time in the call record may be stored in the form of a timestamp, while the time in the text message content may be in the form of a string. Therefore, it is necessary to convert the collected data into a standard data format. For example, convert all time data into a unified timestamp format to facilitate subsequent data integration and analysis.

[0035] Data integration unit 1015: Store the data after format conversion in the local database of the mobile phone according to the data type and association relationship. For example, create a comprehensive data table containing information such as call records, text message content, and application usage frequency, and associate and integrate different types of data through unique identifiers (such as user ID, device ID, etc.). At the same time, to improve the efficiency of data query and analysis, create indexes for key fields, such as creating indexes for the phone number field in call records and the sender field in text message content.

[0036] Furthermore, as Figure 3 shown, the user behavior and scenario analysis module 102 includes:

[0037] 1) Data preprocessing unit 1021: Check the integrated data, find and remove duplicate records, outliers, and error data. For missing data, select a filling method according to the data characteristics to fill in the missing values; for example, for the partially missing sending time in text message content, the median of the user's text message sending time can be used for filling.

[0038] 2) Feature selection and extraction unit 1022: Select features related to user behavior and scenarios from the integrated data. For user behavior analysis, select features such as call duration, call frequency, application usage duration, and text message sending quantity; for scenario recognition, select geographical location, time, and weather features; transform and extract the original features to obtain more valuable features. For example, convert the time feature into time periods of a day (morning, afternoon, evening), or extract keywords from text message content as text features.

[0039] 3) Model selection and training unit 1023:

[0040] 3-1) User behavior model:

[0041] Use the K-Means algorithm to perform clustering analysis on the user's behavior patterns, which is used to divide data points into K different clusters, so that the data points within the cluster have high similarity and the data points between clusters have low similarity;

[0042] Initialization: Determine the number of clusters K and randomly initialize K centroids μ 1 , μ 2 , … μ K ;

[0043] Assign data points: Calculate the distance of each data point x = (x 1 , x 2 , … x n ) to each centroid μ j = (μ j1 , μ j2 , … μ jn) Euclidean distance d(x, μ j ), and the formula is:

[0044]

[0045] Assign the data point to the cluster C where the nearest centroid is located j ;

[0046] Update the centroid: Update the position of the centroid according to the newly assigned data points. The centroid update formula is

[0047]

[0048] where |C j | is the number of data points in cluster C j . Continuously repeat the process of assigning data points and updating the centroid until the position of the centroid no longer changes significantly;

[0049] 3 - 2) Scene recognition model

[0050] Use the logistic regression algorithm to classify different scenes. Logistic regression linearly combines the input features and then maps the result to a probability value between 0 and 1 using the logistic function to determine the probability that the sample belongs to each class:

[0051] For the input feature x = (x 1 , x 2 , … x n ), obtain z = θ 0 + θ 1 x 1 + θ 2 x 2 + … + θ n x n = + θ T x, where θ = (θ 0 , θ 1 , + θ 2 , … θ n ) is the parameter vector of the model;

[0052] Map the result z of the linear combination to a probability value between 0 and 1 through the logistic function ;

[0053] Use the logarithmic loss function J(θ) to train the model:

[0054]

[0055] where m is the number of training samples, y (i) is the true label of the i-th sample, and x (i)is the feature vector of the i-th sample. The parameters θ of the model are continuously adjusted by the gradient descent method to minimize J(θ);

[0056] 4) Model evaluation and optimization unit 1024:

[0057] For the user behavior model, the silhouette coefficient is used to evaluate the quality of the clustering result. The calculation formula of the silhouette coefficient s(i) is:

[0058]

[0059] where a(i) is the average distance from data point i to other data points in the same cluster, and b(i) is the average distance from data point i to all data points in the nearest neighboring cluster. The closer the silhouette coefficient is to 1, the better the clustering effect;

[0060] For the scene recognition model, accuracy, recall, and F1 value are used to evaluate the classification performance of the model;

[0061] The accuracy is:

[0062]

[0063] F1 value:

[0064] where TP is the true positive, TN is the true negative, FP is the false positive, and FN is the false negative;

[0065] The model is optimized according to the evaluation results. If the model performance is not ideal, the model parameters are adjusted;

[0066] 5) Model deployment unit 1025: Used to deploy the trained user behavior model and scene recognition model to the mobile phone system to analyze and identify the user's behavior and scene in real time.

[0067] The above modules are described in detail below with specific embodiments:

[0068] Through various sensors and application interfaces of the mobile phone system, relevant data of the user in the past month is collected, including call records, text message sending and receiving situations, start and end times and durations of application usage, GPS location information of the mobile phone, and corresponding timestamps, etc. Duplicate call records and abnormal location data are removed. For example, if a certain location shows that the user is in the deep sea area, which is obviously not in line with the actual situation, it is excluded.

[0069] Divide a day into four time periods: morning (6:00 - 12:00), afternoon (12:00 - 18:00), evening (18:00 - 24:00), and early morning (0:00 - 6:00). According to the location information, the geographical locations frequently visited by the user are divided into four categories: workplace, residence, shopping mall, and restaurant. Extract the usage habit features, such as the number of times each application is used, the total usage duration, the number of call times, and the number of text messages sent and received within each time period. Perform one-hot encoding on the time periods and geographical locations. For example, "morning" is encoded as [1, 0, 0, 0], and "workplace" is encoded as [1, 0, 0, 0].

[0070] User behavior model: Determine the number of clusters K: Use the elbow method to calculate the sum of squared errors within clusters (SSE) for different K values. It is found that when K = 4, the rate of decrease in SSE begins to slow down, so K = 4 is determined.

[0071] Clustering process: Randomly initialize 4 centroids. Calculate the Euclidean distance from each data point to each centroid, and assign the data point to the cluster where the nearest centroid is located. Recalculate the centroid of each cluster based on the newly assigned data points. Continuously repeat these two steps until the positions of the centroids no longer change significantly.

[0072] Analysis of clustering results:

[0073] Cluster 1: Most data points are concentrated in the morning and afternoon on weekdays, and the geographical location is the workplace. The application usage is mainly office applications (such as Word, Excel, and enterprise communication software). The call objects are mostly colleagues and customers, and the text messages are also mostly related to work. It can be judged that this cluster represents the user behavior pattern in the work scenario.

[0074] Cluster 2: The data points mainly appear in the evening and on weekends, and the geographical location is the residence. The application usage is mainly video applications (such as iQIYI, Tencent Video) and social applications (such as WeChat, Weibo). The calls and text messages are mainly for communication with family and friends. This cluster represents the rest and entertainment scenario.

[0075] Cluster 3: Concentrated in the noon time period on weekdays, and the geographical location is mostly the restaurant. The application usage is less, and occasionally the user will check the food delivery software or group buying software. This cluster represents the dining scenario.

[0076] Cluster 4: Appears more in the afternoon on weekends, and the geographical location is the shopping mall. The application usage is mainly shopping applications (such as Taobao, JD.com) and payment applications (such as Alipay, WeChat Pay). This cluster represents the shopping scenario.

[0077] Analysis of the scenario recognition model:

[0078] The preprocessed data is divided into a training set and a test set according to a ratio of 7:3. The logistic regression model is trained using the training set, and the parameters of the model are adjusted by minimizing the logarithmic loss function. At 3 pm one day, the current data of the user is collected: located at the workplace, using office software, and having call records with colleagues. These data are input into the trained logistic regression model, and the model outputs the probabilities of each scenario. The probability of the work scenario is the highest, so it is recognized that the user is currently in the work scenario. At 8 pm, the user is at home, watching Douyin videos, and the model recognizes that the user is in the rest and entertainment scenario at this time.

[0079] Comprehensively analyze information usage habits:

[0080] Work scenario (weekday mornings and afternoons, workplace): The user mainly uses office applications for document processing, data statistics, and communication. The average daily usage duration of office software in the morning is about 3 hours, and about 2.5 hours in the afternoon. The number of calls is relatively large, mainly concentrated on work-related matters.

[0081] Rest and entertainment scenario (evenings and weekends, residence): The user spends a lot of time on video and social applications. The average evening usage duration of video applications is 2 hours, and 1.5 hours for social applications. There are frequent text messages and calls with family and friends.

[0082] Dining scenario (weekday noon, restaurant): The user has a relatively high frequency of using food delivery software or group buying software, and occasionally browses news applications while waiting for meals. The usage duration is short, generally not exceeding 15 minutes.

[0083] Shopping scenario (weekend afternoons, mall): The user will frequently open shopping applications to compare products and place orders, and at the same time use payment applications to complete payments. During the shopping process, the total usage duration of shopping applications can reach 1 - 2 hours.

[0084] Furthermore, as Figure 4 shown, the information priority division module 103 includes:

[0085] 1) User behavior model and scenario recognition result analysis unit 1031:

[0086] For analyzing the information usage preferences of users in each cluster based on the clustering results of the user behavior model. For example, users in a certain cluster frequently use office applications during working hours and have a high degree of attention to work-related information. Mining association rules in user behavior data based on the association rule mining algorithm to find potential relationships between information. For example, it is found that there is a high probability that a user will open a social media application after browsing a news application, indicating an association between these two types of information; determining the current scene where the user is located according to the output of the scene recognition model, and obtaining the information needs and attention degrees of users in different scenes. For example, in a work scene, users are more concerned about work-related emails, notifications, etc.; in a rest scene, users may be more inclined to entertainment and social information.

[0087] 2) Information attribute analysis unit 1032:

[0088] Used to classify information sources into different categories, including important contacts (which can be family members, bosses, and customers), ordinary contacts, system application notifications, and third-party application notifications. Information from different sources has different importance. For example, information from important contacts is usually more important than information from ordinary contacts; judging the urgency of information according to the urgency flag or content in the information. For example, if a text message contains keywords such as "urgent" or "reply quickly", or the subject of an email has the words "important and urgent", this type of information can be marked as high urgency; evaluating the timeliness of information, and giving higher priority to information with strong timeliness. For example, the closer the time is to the deadline for event registration, the higher the priority of this information.

[0089] 3) Feature extraction and quantification unit 1033:

[0090] Extracting relevant features from the user behavior model, scene recognition results, and information attributes and quantifying them into numerical values; using the Analytic Hierarchy Process (AHP) to determine the weights of each feature, and calculating the priority score of information by the method of weighted summation. Let the feature vector of the information be x = (x 1 , x 2 , …, x n ), and the corresponding weight vector be w = (w 1 , w 2 , …, w n ), then the calculation formula for the priority score P of the information is:

[0091]

[0092] For example, assume that the features of the information include the usage frequency x 1 (weight w 1 = 0.2), the urgency of the information x 2 (weight w 2= 0.5), the importance x of the information source 3 (weight w 3 = 0.3), then the priority score P = 0.2x 1 + 0.5x 2 + 0.3x 3 。

[0093] 4) Information priority division unit 1034:

[0094] According to the calculated priority score, the information is divided into different priority levels. Information with a priority score greater than 80 points is of high priority, information with a score between 30 - 80 points is of medium priority, and information with a score less than 30 points is of low priority; Different processing strategies are adopted for information of different priorities. For high-priority information, it is promptly pushed and the user is reminded; For medium-priority information, it is displayed at an appropriate time; For low-priority information, it is displayed in batches or temporarily blocked;

[0095] Furthermore, as Figure 5 shown, the intelligent allocation strategy module 104 includes the following steps:

[0096] 1) Allocation rule definition unit 1041 under different scenarios:

[0097] 1 - 1) Work scenario rules

[0098] High-priority information: For urgent work emails, instant messaging messages, etc. from important customers and superior leaders, the user is notified by simultaneously turning on pop-up reminders, sounds, and vibrations to ensure that the user can be informed in a timely manner;

[0099] Medium-priority information: For general notifications within the department and regular work report reminders, when the user is in an idle state (judged by the mobile phone operation frequency), the user is reminded by sound or vibration;

[0100] Low-priority information: Industry news and advertising promotions are temporarily blocked and only displayed in batches during work breaks (such as lunch breaks);

[0101] 1 - 2) Rest scenario rules

[0102] High-priority information: For calls and text messages from family members and emergency contacts, a loud ringtone, strong vibration, and pop-up reminder are used;

[0103] Medium-priority information: For important messages from friends on social software, a soft sound and pop-up reminder are used to avoid disturbing the user too much;

[0104] Low-priority information: All notifications of promotional activities for various applications, non-urgent system update prompts, etc. are blocked and will be displayed after the user finishes resting;

[0105] 1 - 3) Travel scenario rules

[0106] High - priority information: For information such as route change reminders and traffic congestion warnings in navigation software, use voice announcements and pop - up reminders.

[0107] Medium - priority information: For reservation confirmation information related to travel, recommended information around the destination, etc., display it in the form of a pop - up window during vehicle stops or walking intervals.

[0108] Low - priority information: Information unrelated to travel, including game updates, entertainment news, etc., is temporarily blocked.

[0109] 2) Intelligent allocation strategy unit 1042 formulated based on the rule engine:

[0110] 2 - 1) Rule engine architecture

[0111] Build a rule engine. The rule engine includes a rule base, a fact base, and an inference engine. The rule base stores information allocation rules for different scenarios; the fact base stores the current information priority and user scenario information; the inference engine makes inferences based on the rule base and the fact base to generate specific allocation strategies.

[0112] 2 - 2) Rule matching algorithm

[0113] Adopt a forward reasoning algorithm. The inference engine obtains information priority and user scenario information from the fact base, and then matches the rules one by one in the rule base. For example, if the fact base shows that the user is in a work scenario, a certain piece of information is of high priority and is an urgent work email from an important customer, the inference engine will match the allocation rule for high - priority information in the work scenario.

[0114] 2 - 3) Strategy generation

[0115] According to the matched rules, generate specific allocation strategies. The strategies include the reminder method of information (such as pop - up window, sound, vibration, voice announcement), reminder time (immediate reminder, reminder when idle, batch reminder), and display location (top of the screen, lock screen interface, etc.).

[0116] 3) Information allocation execution unit 1043:

[0117] 3 - 1) Reminder method setting: According to the allocation strategy, set the reminder method of information. If the strategy requires a pop - up reminder, call the pop - up interface of the mobile phone system to display the information in the form of a pop - up window on the screen; if the strategy requires a sound reminder, set an appropriate ringtone and play it; if the strategy requires a vibration reminder, turn on the vibration function of the mobile phone.

[0118] 3-2) Reminder time control: For information that requires immediate reminder, the reminder operation is triggered immediately; for information that requires idle reminder, the user is judged to be idle by monitoring the operation frequency of the mobile phone, and the reminder is issued when the idle state is detected; for batch reminder information, multiple low-priority information is displayed in a centralized manner within the specified time period;

[0119] 3-3) Display position adjustment: According to the allocation strategy, medium and high priority information are displayed at the top of the screen or on the lock screen to attract the user's attention; low priority information is displayed in a specific message list;

[0120] 4) Real-time monitoring and feedback optimization unit 1044:

[0121] Monitor the effect of information allocation in real time, collect user feedback data, and optimize the allocation rules and strategies based on the monitoring and feedback data; if it is found that users frequently close a certain reminder method, it means that the reminder method may not be suitable, then adjust the rules; if it is found that certain information is not well displayed in a specific scenario, then modify the corresponding allocation strategy.

[0122] Through the above detailed steps, the system can formulate intelligent allocation strategies according to information priority and user's current scenario, and effectively allocate information to improve the user's experience and efficiency in mobile phone information processing.

[0123] In summary, the system can intelligently analyze and process various types of information in the mobile phone, realize intelligent allocation of information according to the user's usage habits, scenarios and needs, and improve the user's efficiency and experience in mobile phone information processing.

Claims

1. An intelligent information allocation system based on big data, characterized in that include: Multi-source data collection and integration module: It is used to collect various information data in the mobile phone by using the API interface of the mobile phone system, integrate the collected data in different formats and sources, and store them uniformly in the local database of the mobile phone; User behavior and scenario analysis module: used to analyze the integrated data through machine learning algorithms, establish user behavior models and scenario recognition models, and identify the scenarios in which users are located by analyzing the information usage habits of users in different time periods and geographical locations; Information priority classification module: used to prioritize various types of information in the mobile phone based on the user behavior model, scene recognition results and the attributes of the information itself; Intelligent allocation strategy module: used to formulate intelligent allocation strategies according to information priority and user current scenarios, and allocate information according to the generated intelligent allocation strategies; Real-time feedback and optimization module: used to collect user feedback on information allocation results in real time, optimize user behavior models, scene recognition models and intelligent allocation strategies based on feedback data, and then use the optimized intelligent allocation strategies to allocate information accordingly.

2. The information intelligent deployment system based on big data according to claim 1, characterized in that: The multi-source data collection and integration module includes: Task initialization unit: First, check the permission settings of the mobile phone system to ensure that the user has legal data collection permissions. If the permissions are insufficient, prompt the user to perform authorization operations. After passing the verification, the data collection unit initialization is triggered; Collection source and cycle determination unit: determines the collection sources of different types of data according to the pre-set collection strategy, and sets different collection cycles; Data collection unit: by calling the API interface of the mobile phone system, data is collected according to the set collection source and cycle. During the collection process, the data is preliminarily checked for legality. For illegal data, logs are recorded and marked; Data format conversion unit: convert the collected data into a standard data format; Data integration unit: stores the format-converted data in the local database of the mobile phone according to data types and associations, and at the same time, creates indexes for key fields.

3. The information intelligent allocation system based on big data according to claim 1, characterized in that: The user behavior and scenario analysis module includes: 1) Data preprocessing unit: Check the integrated data, find and remove duplicate records, outliers and erroneous data. For missing data, select a filling method based on the data characteristics and fill in the missing values. 2) Feature selection and extraction unit: select features related to user behavior and scenarios from the integrated data. For user behavior analysis, select call duration, call frequency, application usage time, and number of text messages sent; for scenario recognition, select geographic location, time, and weather features; 3) Model selection and training unit: 3-1) User behavior model: The K-Means algorithm is used to perform cluster analysis on the user's behavior pattern, which is used to divide the data points into K different clusters, so that the similarity of the data points within the cluster is high and the similarity of the data points between clusters is low; Initialization: First determine the number of clusters K and randomly initialize K centroids μ1, μ2, …μ K ; Assign data points: Calculate each data point x=(x1,x2,…x n ) to each center of mass μ j =(μ j1 ,μ j2 ,…μ jn ) of the Euclidean distance d(x,μ j ), the formula is: Assign the data point to the cluster C with the closest centroid j middle; Update the centroid: Update the position of the centroid according to the newly assigned data points. The centroid update formula is: Where |C j | is cluster C j The number of data points in the data set is calculated, and the process of allocating data points and updating the centroid is repeated until the position of the centroid no longer changes significantly; 3-2) Scene Recognition Model Use the logistic regression algorithm to classify different scenarios. Logistic regression performs a linear combination of input features and then uses a logical function to map the results to a probability value between 0 and 1 to determine the probability of the sample belonging to each category: For input feature x=(x1,x2,…x n ), through linear combination we get z = θ0 + θ1x1 + θ2x2 + … + θ n x n =+θ T x, where θ=(θ0,θ1,+θ2,…θ n ) is the parameter vector of the model; The result of the linear combination z is passed through the logistic function Mapped to a probability value between 0 and 1; Use the logarithmic loss function J(θ) to train the model: Where m is the number of training samples, y (i) is the true label of the i-th sample, x (i) is the feature vector of the i-th sample. The model parameters θ are continuously adjusted by the gradient descent method to minimize J(θ); 4) Model evaluation and optimization unit: For the user behavior model, the silhouette coefficient is used to evaluate the quality of the clustering results. The calculation formula of the silhouette coefficient s(i) is: Where a(i) is the average distance from data point i to other data points in the same cluster, b(i) is the average distance from data point i to all data points in the nearest neighbor cluster. The closer the silhouette coefficient is to 1, the better the clustering effect. For the scene recognition model, the accuracy, recall and F1 value are used to evaluate the classification performance of the model; The accuracy is: Recall: F1 value: Among them, TP is a true positive example, TN is a true negative example, FP is a false positive example, and FN is a false negative example; Optimize the model based on the evaluation results. If the model performance is not ideal, adjust the model parameters. 5) Model deployment unit: used to deploy the trained user behavior model and scene recognition model into the mobile phone system, and analyze and identify user behavior and scenes in real time.

4. The information intelligent allocation system based on big data according to claim 1, characterized in that: The information prioritization module includes: 1) User behavior model and scene recognition result analysis unit: It is used to analyze the information usage preferences of users in each cluster based on the clustering results of the user behavior model, mine the association rules in the user behavior data based on the association rule mining algorithm, and find out the potential relationship between information; according to the output of the scene recognition model, determine the current scene of the user, and obtain the user's demand and attention to information in different scenes; 2) Information attribute analysis unit: Used to classify information sources into different categories, including important contacts, ordinary contacts, system application notifications, and third-party application notifications; determine the urgency of information based on the urgency mark or content in the information; evaluate the timeliness of information and give higher priority to information with stronger timeliness; 3) Feature extraction and quantization unit: Extract relevant features from user behavior models, scene recognition results and information attributes, and quantify them into numerical values; use the analytic hierarchy process (AHP) to determine the weight of each feature, and use the weighted summation method to calculate the priority score of the information; let the feature vector of the information be x = (x1, x2, ..., x n ), the corresponding weight vector is w=(w1,w2,…,w n ), then the calculation formula of the information priority score P is: 4) Information Prioritization Unit: According to the calculated priority score, the information is divided into different priority levels. Information with a priority score greater than 80 points is high priority, information with a score between 30-80 points is medium priority, and information with a score less than 30 points is low priority. Different processing strategies are adopted for information of different priorities. For high-priority information, it is pushed and reminded to the user in a timely manner; for medium-priority information, it is displayed at an appropriate time; for low-priority information, it is displayed in batches or temporarily blocked.

5. The information intelligent allocation system based on big data according to claim 1, characterized in that: The intelligent deployment strategy module includes the following steps: 1) Deployment rule definition unit in different scenarios: 1-1) Work scene rules High-priority information: For urgent work emails and instant messaging messages from important customers and superiors, users are notified by means of pop-up reminders, sounds, and vibrations. Medium priority information: For general notifications within the department and regular work report reminders, the user is reminded by sound or vibration when the user is idle; Low-priority information: industry news and advertising promotions are temporarily blocked and only displayed in batches during work breaks; 1-2) Rest scene rules High-priority messages: For calls and text messages from family members and emergency contacts, use loud rings, strong vibrations, and pop-up reminders; Medium priority information: For important messages from friends on social software, use soft voice and pop-up reminders to avoid excessively disturbing users; Low-priority information: All promotional notifications for various applications and non-urgent system update reminders are blocked and displayed after the user has finished resting; 1-3) Travel scenario rules High-priority information: For navigation software route change reminders and traffic congestion warning information, voice broadcasts and pop-up reminders are used; Medium priority information: travel-related booking confirmation information and recommended information around the destination are displayed in pop-up windows when the vehicle is parked or walking; Low-priority information: Information not related to travel, including game updates and entertainment news, is temporarily blocked; 2) Formulate intelligent deployment strategy unit based on rule engine: 2-1) Rule Engine Architecture Build a rule engine, which includes a rule base, a fact base, and an inference engine. The rule base stores information allocation rules for different scenarios, and the fact base stores the current information priority and user scenario information. The inference engine performs inference based on the rule base and the fact base to generate a specific allocation strategy. 2-2) Rule matching algorithm Using the forward reasoning algorithm, the inference engine obtains information priority and user scenario information from the fact base, and then matches the rules one by one in the rule base; 2-3) Strategy Generation Generate a specific deployment strategy based on the matched rules, including the reminder method, reminder time and display location of the information; 3) Information deployment execution unit: 3-1) Reminder method setting: according to the deployment strategy, set the information reminder method; if the strategy requires the use of pop-up window reminders, call the pop-up window interface of the mobile phone system to display the information on the screen in the form of pop-up windows; if the strategy requires sound reminders, set a suitable ringtone and play it; If the policy requires a vibration reminder, turn on the phone's vibration function; 3-2) Reminder time control: For information that requires immediate reminder, the reminder operation is triggered immediately; for information that requires idle reminder, the user is judged to be idle by monitoring the operation frequency of the mobile phone, and the reminder is issued when the idle state is detected; for batch reminder information, multiple low-priority information is displayed in a centralized manner within the specified time period; 3-3) Display position adjustment: According to the allocation strategy, medium and high priority information are displayed at the top of the screen or on the lock screen to attract the user's attention; low priority information is displayed in a specific message list; 4) Real-time monitoring and feedback optimization unit: Monitor the effect of information allocation in real time, collect user feedback data, and optimize the allocation rules and strategies based on the monitoring and feedback data; if it is found that users frequently close a certain reminder method, it means that the reminder method may not be suitable, then adjust the rules; if it is found that certain information is not well displayed in a specific scenario, then modify the corresponding allocation strategy.

Images