A Personalized Prediction Method and System for Residents' Travel Demand Based on User Profiles

Abstract

This invention discloses a method and system for personalized prediction of residents' travel needs based on user profiles, belonging to the field of digital data processing technology. The method includes the following steps: user information collection effect analysis; profile construction accuracy analysis; and personalized decision interference analysis. This invention analyzes the collection effect of specified user information data using acquired collection effect data, determines whether data collection optimization is needed based on the results, then determines whether profile construction optimization is needed based on the accuracy analysis results, and finally determines whether decision interference optimization is needed based on the interference analysis results. This achieves the effect of improving the accuracy of residents' commuting decision prediction under foggy weather events, solving the problem of low accuracy in commuting decision prediction during user profile analysis under foggy weather events.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a method and system for personalized prediction of residents' travel needs based on user profiles. Background Technology

[0002] As digital life becomes increasingly saturated, residents' travel needs are becoming more diversified and personalized. Traditional travel services are struggling to meet users' expectations for intelligent travel, necessitating the use of user profiling technology to uncover potential user needs and optimize the travel experience. On one hand, by collecting and analyzing multi-dimensional user data, travel characteristics can be analyzed to construct accurate user profiles. Combined with user segmentation based on these profiles, personalized travel plan recommendations can be implemented. On the other hand, personalized travel demand prediction can optimize traffic resource allocation, reduce travel delays, and improve the overall efficiency of urban transportation systems. Traditional travel systems rely on preset rules and cannot dynamically adjust to real-time environments. Against this backdrop, personalized prediction of residents' travel needs based on user profiles provides a more accurate predictive solution for users' travel decisions.

[0003] Existing technologies collect demographic data, geographic information data, transportation facility data, and travel content published on social media platforms. After data cleaning, data fusion technology is used to integrate and link multi-source data to users. Algorithm models are used to build user travel preference profiles. Based on these user profiles, parameter optimization suggestions are generated to predict user travel needs and provide users with travel advice.

[0004] For example, a user-demand-driven service matching method published in the invention patent announcement CN113139125B includes: receiving user input requirements, which at least include keywords, type, domain, and tags of the service name; matching similar users of the user, supplementing the user's tags from the tags of similar users, and assigning weights to the supplemented tags to obtain a tag weight set; matching the service with the three types of requirements layer by layer, obtaining the matched services, and sorting them according to their weights.

[0005] For example, the invention patent announcement CN116932853B discloses a method for obtaining user needs based on APP review data, which includes: crawling APP review data using a Python web crawler and preprocessing it to obtain preprocessed text data; using the preprocessed text data to jointly pre-train a BERT model on SRP and LAP tasks; constructing a user needs prediction model, using the pre-trained BERT model to construct training data encoding for the preprocessed text data, and using the encoded training data to train the user needs prediction model; the pre-trained BERT model and the trained user needs prediction model together form a user needs generation model, and the user needs generation model obtains the data to be processed and inputs it to obtain the corresponding user needs generation result.

[0006] However, in the process of implementing the inventive technical solution in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems:

[0007] In existing technologies, profile building focuses primarily on trajectory data calculation, lacking effective extraction and utilization of unstructured data, thus failing to achieve semantic-level understanding. Furthermore, profile building systems rely on centralized computing, lacking the integration of real-time data, resulting in insufficient real-time performance and ignoring the impact of unexpected events on users' travel decisions. This leads to issues such as low accuracy in predicting commuting decisions during user profile analysis in foggy weather events. Summary of the Invention

[0008] This application provides a personalized prediction method and system for residents' travel needs based on user profiles, which solves the problem of low accuracy in predicting commuting decisions during user profile analysis in foggy weather events, and improves the accuracy of commuting decision indicators prediction in foggy weather events.

[0009] This application provides a method for personalized prediction of residents' travel needs based on user profiles, including the following steps: Step 1, analyzing the collection effect of the specified user information data collection process using the acquired collection effect data to obtain collection effect analysis results. The collection effect analysis is used to measure the quantitative degree of the collection effect data on the collection effect of the specified user information. Step 2, determining whether to optimize data collection based on the collection effect analysis results, and simultaneously performing accuracy analysis on the construction process of the specified user profile based on the acquired profile construction data to obtain accuracy analysis results. Data collection optimization means improving the comprehensiveness of the extracted specified user information data by adjusting correlation analysis and adjusting the data collection frequency. Accuracy analysis measures the accuracy of the analysis of corresponding text and voice data modalities during the construction of a specified user profile. Step three involves determining whether to optimize the profile construction based on the accuracy analysis results. Simultaneously, interference analysis is performed on the prediction results of the specified user's travel demand based on the acquired interference factor data. The interference analysis results are then used to determine whether to optimize the decision interference. Profile construction optimization means improving the accuracy of user profile generation by adjusting the profile update frequency. Interference analysis measures the degree of interference of interference factor data on the prediction results of user travel demand. Decision interference optimization means improving the reliability of user travel decision prediction by adjusting the visibility collection frequency.

[0010] This application provides a personalized prediction system for residents' travel needs based on user profiles, including: a data collection effect analysis module, an accuracy analysis module, and an interference analysis module. The data collection effect analysis module analyzes the data collection process of a specified user information data using acquired data collection effect data, obtaining a data collection effect analysis result. This analysis measures the quantitative degree to which the data collection effect data contributes to the collection of the specified user information. The accuracy analysis module determines whether data collection optimization is needed based on the data collection effect analysis result, and simultaneously performs an accuracy analysis on the construction process of a specified user profile based on acquired profile construction data, obtaining an accuracy analysis result. Data collection optimization refers to improving accuracy by adjusting correlation analysis and adjusting the data collection frequency. The comprehensiveness and accuracy analysis of the extracted user information data is used to measure the accuracy of the analysis of corresponding text and voice data modalities during the construction of a user profile. The interference analysis module is used to determine whether to optimize the profile construction based on the accuracy analysis results. At the same time, it performs interference analysis on the prediction results of the user's travel demand based on the acquired interference factor data, and obtains interference analysis results. Based on the interference analysis results, it determines whether to optimize the decision interference. Profile construction optimization means improving the accuracy of user profile generation by adjusting the profile update frequency. Interference analysis is used to measure the degree of interference of interference factor data on the prediction results of user travel demand. Decision interference optimization means improving the reliability of user travel decision prediction by adjusting the visibility collection frequency.

[0011] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0012] 1. By analyzing the data collection results of the acquired data collection, the system determines whether to optimize the data collection process. Then, based on the accuracy analysis results, it determines whether to optimize the user profile construction. Finally, based on the acquired interference factor data, the system performs interference analysis on the prediction results of the user's travel demand. Based on the interference analysis results, it determines whether to optimize the decision interference. This improves the accuracy of commuting decision prediction under foggy weather events, and effectively solves the problem of low accuracy in commuting decision prediction during user profile analysis under foggy weather events.

[0013] 2. Analyze the collection effect of the specified user information data collection process using the acquired collection effect data, obtain the collection effect analysis results, determine whether to optimize the data collection based on the collection effect analysis results, optimize the extraction of voice information, thereby improving the ability to parse voice data, enhancing data diversity, realizing intermodal information interaction, and further enriching the dimensions of user profiles and improving the comprehensiveness of user preference information extraction.

[0014] 3. By analyzing the accuracy of the user profile construction process based on the acquired profile construction data, the accuracy analysis results are obtained. Based on the accuracy analysis results, it is determined whether to optimize the profile construction. By adjusting the profile update frequency, an optimal solution is provided to balance the construction time and update speed, so as to achieve adaptive acquisition of the profile update frequency adjustment range, thereby improving the accuracy of profile construction.

[0015] 4. By performing interference analysis on the predicted travel demand of specified users based on the acquired interference factor data, the interference analysis results are obtained. Based on the interference analysis results, it is determined whether to perform decision interference optimization. By adjusting the visibility collection frequency, changes in PM2.5 concentration can be captured more promptly. By dynamically adjusting the sampling frequency, data redundancy can be avoided, data accuracy can be ensured, frequent sampling can be avoided, system load and energy consumption can be reduced, data processing efficiency can be improved, and more accurate data support can be provided for commuting decision prediction. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a personalized prediction method for residents' travel needs based on user profiles, provided as an embodiment of this application;

[0017] Figure 2 A logical framework diagram of a personalized prediction method for residents' travel needs based on user profiles provided in this application embodiment;

[0018] Figure 3 A flowchart for analyzing the user information collection effect provided in this application embodiment;

[0019] Figure 4 A flowchart for the accuracy analysis of portrait construction provided in this application embodiment;

[0020] Figure 5 A flowchart of personalized decision interference analysis provided for embodiments of this application;

[0021] Figure 6 This is a schematic diagram of the structure of a personalized prediction system for residents' travel needs based on user profiles, provided in an embodiment of this application. Detailed Implementation

[0022] This application provides a personalized prediction method and system for residents' travel needs based on user profiles. This addresses the problem of low accuracy in predicting commuting decisions during user profile analysis in existing technologies, particularly during foggy weather events. First, the method analyzes the collection effect of the specified user information data using acquired collection effect data, obtaining the collection effect analysis results. Then, based on the collection effect analysis results, it determines whether data collection optimization is needed. Simultaneously, based on the acquired profile construction data, it analyzes the accuracy of the specified user profile construction process, obtaining the accuracy analysis results. Next, based on the accuracy analysis results, it determines whether profile construction optimization is needed. Finally, based on the acquired interference factor data, it analyzes the interference of the predicted travel needs of the specified user, obtaining the interference analysis results. Finally, based on the interference analysis results, it determines whether decision interference optimization is needed. This approach improves the accuracy of predicting residents' commuting decisions during foggy weather events, effectively solving the problem of low accuracy in predicting commuting decisions during user profile analysis in foggy weather events.

[0023] The technical solution in this application aims to address the problem of low accuracy in predicting commuting decisions during user profile analysis under foggy weather events. The overall approach is as follows:

[0024] By analyzing the collection effect data of the specified user information data, determining whether to optimize the data collection based on the collection effect analysis results, determining whether to optimize the profile construction based on the accuracy analysis results, and finally conducting interference analysis on the prediction results of the specified user's travel demand based on the acquired interference factor data, the accuracy of residents' commuting decision prediction under foggy weather events was improved.

[0025] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0026] like Figure 1The diagram shows a flowchart of a personalized prediction method for residents' travel needs based on user profiles, provided in an embodiment of this application. This method includes the following steps: Step 1: Analyzing the collection effect of the specified user information data collection process using acquired collection effect data to obtain collection effect analysis results. The collection effect analysis measures the quantitative degree to which the collection effect data contributes to the collection effect of the specified user information. Step 2: Determining whether to optimize data collection based on the collection effect analysis results. Simultaneously, performing an accuracy analysis on the construction process of the specified user profile based on the acquired profile construction data to obtain accuracy analysis results. Data collection optimization refers to adjusting correlation analysis and adjusting data collection... Frequency is used to improve the comprehensiveness of the extraction of specified user information data. Accuracy analysis is used to measure the analysis accuracy of corresponding text data modalities and voice data modalities during the construction of specified user profiles. Step 3: Based on the accuracy analysis results, it is determined whether to optimize the profile construction. At the same time, based on the acquired interference factor data, interference analysis is performed on the prediction results of the specified user's travel demand to obtain interference analysis results. Based on the interference analysis results, it is determined whether to optimize the decision interference. Profile construction optimization means improving the accuracy of user profile generation by adjusting the profile update frequency. Interference analysis is used to measure the degree of interference of interference factor data on the prediction results of user travel demand. Decision interference optimization means improving the reliability of user travel decision prediction by adjusting the visibility collection frequency.

[0027] The data collection results include social platform information acquisition values, destination extraction values, and image feature extraction values. The social platform information acquisition value is used to quantify the number of valid information successfully collected from the specified user information data at the end of the preset data collection period. The destination extraction value is used to quantify the number of destination keywords extracted from the valid destination information successfully collected from the specified user information data at the end of the preset data collection period. The image feature extraction value is used to quantify the extraction efficiency of valid features from the specified user information data at the end of the preset data collection period.

[0028] The profile construction data includes text information coverage value and voice information coverage value. The text information coverage value is usually used to quantify the availability of effective text data successfully obtained from the text data of a specified user profile at the end of the preset construction period. The voice information coverage value is usually used to quantify the contribution of effective user intent successfully obtained from the voice data of a specified user profile at the end of the preset construction period. Both the collection effect data and the profile construction data are obtained through monitoring by a high-definition camera. The high-definition camera includes image segmentation algorithms, bone conduction microphones, and voiceprint analysis technology.

[0029] In this embodiment, as Figure 2The diagram shown is a logical framework diagram of a personalized prediction method for residents' travel needs based on user profiles provided in this application embodiment. The method analyzes the collection effect of the specified user information data collection process by acquiring the collection effect data, determines whether to optimize the data collection based on the collection effect analysis results, determines whether to optimize the profile construction based on the accuracy analysis results, and finally determines whether to optimize the decision interference based on the interference analysis results.

[0030] Based on the accuracy analysis results of text and voice data modalities, the profile update frequency is dynamically adjusted to improve the profile's responsiveness to user behavior. Real-time analysis of collected data breaks through the limitations of traditional methods that rely solely on historical data for prediction. The data collection strategy is dynamically adjusted to avoid data loss or redundancy caused by traditional static collection methods. Especially in extreme scenarios such as heavy fog, the data source can be quickly focused. At the same time, when PM2.5 concentration changes rapidly, the sampling frequency is automatically increased to capture short-term fluctuations, avoiding decision-making errors due to data lag. Real-time linkage between interference factors and sampling strategies is achieved, thereby improving the accuracy of commuting decision prediction during heavy fog events.

[0031] like Figure 3 The diagram shown is a flowchart of the user information collection effect analysis provided in this application embodiment. The process starts by acquiring collection effect data and analyzing the collection effect. The collection effect analysis result is obtained, and it is determined whether to optimize the data collection based on the collection effect analysis result. If so, the data collection frequency is adjusted; otherwise, the profile construction data is acquired.

[0032] like Figure 4 The diagram shown is a flowchart of the accuracy analysis process for profile construction provided in this application embodiment. The process starts with performing accuracy analysis, obtaining the accuracy analysis results, and determining whether to optimize profile construction based on the accuracy analysis results. If so, the profile update frequency is adjusted; otherwise, interference factor data is obtained and interference analysis is performed.

[0033] like Figure 5 The diagram shown is a flowchart of the personalized decision interference analysis provided in this application embodiment. The process starts with interference analysis, obtains the interference analysis results, and determines whether to optimize decision interference based on the interference analysis results. If so, the visibility collection frequency is adjusted; otherwise, the process of optimizing the accuracy of resident commuting decision prediction under foggy weather events ends.

[0034] The aforementioned database serves as a foundational data repository built prior to the design of a personalized prediction method for residents' travel needs based on user profiles. It stores various preset values, including but not limited to pre-defined data collection effectiveness indicators, pre-defined data dimensional integrity indicators, and decision interference indicators. These values ​​are directly configured by technical personnel according to business needs, while the setting logic of the pre-defined data collection effectiveness indicators can be dynamically adjusted based on specific application scenarios. For example, the pre-defined data collection effectiveness indicators are represented by extracting historical data collection effectiveness indicators recorded at the end of each historical data collection and analysis period from the database, and then summing and averaging them. Furthermore, all values ​​in the database support debugging settings and iterative fine-tuning by technical personnel based on actual operational results.

[0035] In this embodiment, the data collection effect is comprehensively evaluated by multi-dimensional quantitative indicators, the comprehensiveness of user preference extraction is improved by multimodal fusion, the completeness of profile data dimensions is quantified by processing text and voice information coverage values ​​through geometric mean, the profile update frequency is adaptively adjusted to balance construction time and update speed, and the optimization process is ensured not to introduce significant delays by monitoring latency and resource utilization. Finally, a closed-loop, interpretable, and resource-efficient profile construction system is formed, which significantly improves the quality of user profiles.

[0036] Furthermore, the specific process for analyzing the collection effect of specified user information data using the acquired collection effect data is as follows: First, obtain the degree of difference between the social media platform information acquisition value and the preset social media platform information acquisition value in the database, that is, obtain the difference between the social media platform information acquisition value and the preset social media platform information acquisition value in the database. At the same time, combine the social media platform information acquisition compensation amount in the database to perform compensation calculations to obtain the social media platform information acquisition coefficient. The specific constraint expression is In the formula, This represents the social media information acquisition coefficient corresponding to the specified user information data at the end of the preset data collection period. This indicates the amount of compensation received for information obtained from social media platforms. This indicates the social media information value obtained at the end of the preset data collection period for the specified user information. This indicates the preset value for obtaining information from social media platforms.

[0037] Secondly, the degree of difference between the extracted destination value and the preset extracted destination value in the database is obtained, that is, the difference between the extracted destination value and the preset extracted destination value in the database is taken, and a compensation calculation is performed in combination with the compensation amount of the extracted destination value in the database to obtain the destination extraction coefficient. The specific constraint expression is In the formula, This represents the destination extraction coefficient corresponding to the specified user information data at the end of the preset data collection period. Indicates the amount of compensation to be extracted at the destination. This indicates the destination value extracted from the specified user information data at the end of the preset data collection period. This indicates the preset destination extraction value.

[0038] Next, the degree of difference between the extracted image feature values ​​and the preset extracted image feature values ​​in the database is obtained, that is, the difference between the extracted image feature values ​​and the preset extracted image feature values ​​in the database is obtained. At the same time, a compensation calculation is performed based on the compensation amount of the extracted image feature values ​​in the database to obtain the image feature extraction coefficient. The specific constraint expression is: In the formula, This represents the image feature extraction coefficient corresponding to the specified user information data at the end of the preset data collection period. This represents the compensation amount for image feature extraction. This represents the image feature extraction value corresponding to the specified user information data at the end of the preset data collection period. This represents the preset image feature extraction values.

[0039] Finally, the acquired social platform information acquisition coefficient, destination extraction coefficient, and image feature extraction coefficient are coupled to obtain the acquisition effect index. This index measures the difference between the acquired and preset acquisition effect indices, i.e., the difference between the acquired and preset indexes. The specific constraint expression is: In the formula, This indicates the collection effect metric for the specified user information data at the end of the preset data collection period.

[0040] In this embodiment, the database pre-stores preset compensation amounts that are deeply correlated with the data collection performance indicators. These compensation amounts establish clear mapping rules between themselves and the social media platform information acquisition values, destination extraction values, and image feature extraction values. It is important to emphasize that these mappings are not randomly generated but carefully designed based on business logic and system characteristics. They can be presented as a one-to-one correspondence or a many-to-one association. For example, when conducting data collection evaluation and analysis, the system can directly substitute the real-time collected social media platform information acquisition values, destination extraction values, and image feature extraction values ​​into this preset mapping relationship. This allows for the rapid and accurate location and acquisition of the corresponding compensation amounts for social media platform information acquisition, destination extraction, and image feature extraction values.

[0041] Crucially, to ensure the consistency of the evaluation process and the comparability of the results, this example explicitly stipulates that the value ranges of the compensation amount for social platform information acquisition, destination extraction value, and image feature extraction value are all strictly limited to the range of 0 to 1, and the sum of these three compensation amounts is always equal to 1.

[0042] It is important to understand that the social media platform information acquisition value determines the total amount of effective information that can be used for destination extraction and image feature extraction. The higher the social media platform information acquisition value, the higher the destination extraction value and image feature extraction value. The level of the destination extraction value may affect the subsequent processing and analysis of image data. If the destination extraction value increases, it will help to perform more in-depth feature extraction on the destination image. The image feature extraction value increases, and by improving the image feature extraction value, irrelevant image data can be identified and filtered out, thereby improving the overall quality of data collection and increasing the social media platform information acquisition value.

[0043] In the data collection and analysis process, three indicators quantify the data collection effect from three dimensions: total amount of information, key information content, and information form. This avoids the one-sidedness of evaluation by a single indicator, ensures that content that users may care about is collected, accurately extracts destination information that users care about, identifies key features in images, enhances contextual understanding, and supports personalized recommendations. While ensuring the collection effect, it avoids resource waste and thus effectively solves the problem of low accuracy in predicting commuting decisions during user profile analysis in foggy weather events.

[0044] Furthermore, based on the results of the data collection effect analysis, it is determined whether to optimize the data collection. The specific process is as follows: if the obtained data collection effect indicators do not exceed the preset data collection effect indicators in the database, the data collection effect analysis result is determined to be qualified and an accuracy analysis command is sent; otherwise, the data collection effect analysis result is determined to be unqualified and voice information extraction optimization is performed.

[0045] The specific process for optimizing speech information extraction is as follows: The acquired acquisition effect index deviation and speech information extraction efficiency deviation are jointly input into the SHAP (SHapley Additive exPlanations) value attribution calculation for multimodal feature fusion, realizing intermodal information interaction. The acquisition effect index deviation measures the difference between the acquired acquisition effect index and the preset acquisition effect index, i.e., the difference between the acquired acquisition effect index and the preset acquisition effect index. The speech information extraction efficiency deviation measures the difference between the acquired speech information extraction efficiency and the preset speech information extraction efficiency, i.e., the difference between the speech information extraction efficiency and the preset speech information extraction efficiency. After one speech information extraction optimization, the multimodal fusion delay time is obtained. If the obtained multimodal fusion delay time is not greater than the multimodal fusion delay time set in the database, it indicates that the current speech information extraction optimization is effective; otherwise, based on the obtained CPU (Central Processing Time) value... The deviation between the memory usage rate and the reacquired acquisition effect indicators is mapped in the database to obtain the reduction in data acquisition frequency. If the multimodal fusion delay time of the reacquired data within the number of times the data acquisition frequency is reduced is not greater than the multimodal fusion delay time set in the database, and the number of times the data acquisition frequency is reduced is obtained by monitoring through a counter, then the voice information extraction optimization is completed and an accuracy analysis command is sent; otherwise, an acquisition effect warning command is sent.

[0046] In this embodiment, precise optimization decisions are made by determining whether the data collection effect meets the standards. When the data collection is unqualified, the correlation between the collection deviation and the speech extraction efficiency deviation is quantified. Combined with multimodal feature fusion to enhance modal interaction, the comprehensiveness of speech information extraction is improved and the real-time performance of multimodal fusion is guaranteed. At the same time, through a dual monitoring mechanism of latency and resource utilization, the data collection frequency is dynamically adjusted to balance the optimization effect and system load. This not only improves the data collection quality and the accuracy of user preference understanding, but also avoids the waste of resources caused by ineffective optimization through the early warning mechanism.

[0047] Furthermore, the specific process for analyzing the accuracy of the user profile construction process based on the acquired profile construction data is as follows: The acquired text information coverage value and voice information coverage value are geometrically averaged to obtain a data dimension integrity index. This index reflects the quantitative degree of accuracy of the profile construction data in constructing the specified user profile. The ratio of the number of covered text features to the total number of text features required for the profile is recorded as the text information coverage value, and the ratio of the number of effective features parsed from voice to the total number of voice features required for the profile is recorded as the voice information coverage value. The square root of the product of the text information coverage value and the voice information coverage value is taken to obtain the data dimension integrity index. The geometric average processing is used to reflect the product effect and suppress the influence of extreme values, and to more accurately quantify the integrity of the user profile construction data.

[0048] In this embodiment, when the text information coverage value decreases, the voice information provides additional context or details, helping to improve the user profile, thus increasing the voice information coverage value. Conversely, if the voice information coverage value decreases, text information can also serve as a primary basis for constructing the user profile, thus increasing the text information coverage value. This example effectively balances the weights of different data dimensions by using a geometric average of the text and voice information coverage values. This reflects the importance of both in user profile construction and accurately quantifies the completeness of the profile construction data. Therefore, it provides a scientific and comprehensive quantitative indicator for evaluating the accuracy of constructing a specified user profile, helping to accurately identify data shortcomings, optimize collection strategies, and ultimately improve the quality and application value of user profiles.

[0049] Furthermore, the specific process for determining whether to optimize the profile construction based on the accuracy analysis results is as follows: if the data dimension integrity index obtained exceeds the preset data dimension integrity index in the database, the accuracy analysis result is determined to be qualified data collection and an interference analysis command is sent; otherwise, the accuracy analysis result is determined to be unqualified profile construction and profile construction optimization is performed.

[0050] The specific process for portrait construction optimization is as follows: The acquired data dimension integrity index deviation and portrait update frequency deviation are input into the linear regression algorithm to adaptively adjust the portrait update frequency. The data dimension integrity index deviation is used to measure the degree of difference between the acquired data dimension integrity index and the preset data dimension integrity index, that is, the difference between the acquired data dimension integrity index and the preset data dimension integrity index. The portrait update frequency deviation is used to measure the degree of difference between the actual portrait update frequency and the preset portrait update frequency, that is, the difference between the actual portrait update frequency and the preset portrait update frequency. If the adaptive adjustment delay time re-acquired within the portrait construction frequency optimization number is not greater than the adaptive adjustment delay time set in the database, and the portrait construction optimization number is monitored by a counter, then the portrait construction optimization is completed and an interference analysis command is sent; otherwise, a portrait construction warning command is sent.

[0051] In this embodiment, the preset data dimension integrity index is represented by extracting the historical data dimension integrity index of each period recorded at the end of the historical profile construction accuracy assessment period in the database, and then summing and averaging the results. This example determines whether the profile construction quality meets the standard by dynamically comparing the data dimension integrity index with the preset threshold. When it is unqualified, an adaptive adjustment mechanism based on the deviation of data dimension integrity and profile update frequency is introduced. Through the dual constraint mechanism of delay duration monitoring and frequency optimization times, the resource waste caused by excessive adjustment is avoided. Finally, closed-loop control is completed while ensuring optimization efficiency. This not only improves the accuracy and response speed of user profile construction, but also ensures the stability and reliability of system operation through the early warning mechanism.

[0052] Furthermore, the interference factor data includes actual commuting distance, actual traffic flow, and actual visibility; actual commuting distance represents the actual distance traveled by a specified user within a specified commuting area from their origin to their destination, obtained through geographic information system (GIS) analysis; actual traffic flow is obtained through monitoring via the traffic management bureau's website or relevant traffic information platforms; and actual visibility distance is obtained through monitoring using a transmission visibility meter.

[0053] In this embodiment, by comprehensively considering three key dimensions, the decision interference of the prediction results is quantitatively assessed. The combination of the three dimensions covers the spatial, temporal and perceptual dimensions of commuting behavior. Through real-time monitoring of multi-dimensional data, abnormal situations in the commuting environment can be detected in a timely manner. Furthermore, through multi-factor cross-validation, more interpretable and guiding quantitative basis is provided for the formulation of adaptive strategies, thereby achieving an accurate characterization of the complexity of the commuting environment and improving the reliability of interference assessment.

[0054] Furthermore, based on the acquired interference factor data, interference analysis is performed on the prediction results of the specified user's travel demand. The specific process is as follows: First, the degree of difference between the actual commuting distance and the corresponding predicted commuting distance is obtained, i.e., the difference between the actual commuting distance and the corresponding predicted commuting distance. Simultaneously, compensation calculations are performed using the actual commuting distance compensation amount in the database to obtain the actual commuting distance coefficient. The specific constraint expression is: In the formula, This represents the actual commuting distance coefficient at the end of the preset interference analysis period. This indicates the compensation amount for the actual commuting distance. This indicates the actual commuting distance at the end of the preset interference analysis period. This indicates the preset actual commuting distance.

[0055] Secondly, the degree of difference between the actual traffic flow and the corresponding predicted traffic flow is obtained, i.e., the difference between the actual traffic flow and the corresponding predicted traffic flow. Simultaneously, compensation calculations are performed using the actual traffic flow compensation amount from the database to obtain the actual traffic flow coefficient. The specific constraint expression is: In the formula, This represents the actual traffic flow coefficient at the end of the preset interference analysis period. This represents the actual traffic flow compensation amount. This represents the actual traffic flow at the end of the preset interference analysis period. This indicates the preset actual traffic flow.

[0056] Next, the degree of difference between the actual visibility and the corresponding predicted visibility value is obtained, i.e., the difference between the actual visibility and the corresponding predicted visibility value. Simultaneously, a compensation calculation is performed using the actual visibility compensation amount from the database to obtain the actual visibility coefficient. The specific constraint expression is: In the formula, This represents the actual visibility coefficient at the end of the preset interference analysis period. This indicates the actual visibility compensation amount. This indicates the actual visibility at the end of the preset interference analysis period. This indicates the preset actual visibility.

[0057] Finally, the obtained actual commuting distance coefficient, actual traffic flow coefficient, and actual visibility coefficient are coupled to obtain a decision interference index. This index measures the degree of difference between the obtained index and a preset index, i.e., the difference between the obtained and preset indexes. The specific constraint expression is: In the formula, This indicates the decision interference index at the end of the preset interference analysis period.

[0058] In this embodiment, the database pre-stores preset compensation values ​​that are deeply correlated with decision interference indicators. These compensation values ​​establish clear mapping rules between actual commuting distance, actual traffic flow, and actual visibility. It is important to emphasize that these mappings are not randomly generated but carefully designed based on business logic and system characteristics. They can be presented as a one-to-one correspondence or a many-to-one association. For example, when conducting interference analysis, the system can directly substitute the real-time collected actual commuting distance, actual traffic flow, and actual visibility into this preset mapping relationship, thereby quickly and accurately locating and obtaining the actual commuting distance compensation value, actual traffic flow compensation value, and actual visibility compensation value that are appropriate for the current actual commuting distance, actual traffic flow, and actual visibility.

[0059] Crucially, to ensure the consistency of the assessment process and the comparability of the results, this example explicitly stipulates that the value ranges of the actual commuting distance compensation, actual traffic flow compensation, and actual visibility compensation are all strictly limited to the range of 0 to 1, and requires that the sum of these three compensation amounts must be equal to 1.

[0060] It's important to understand that when actual traffic flow increases, road congestion worsens. Although the straight-line commute distance remains unchanged, detours caused by congestion increase the actual commute distance. With reduced visibility, drivers need to maintain larger following distances and lower speeds, further reducing road capacity and thus decreasing actual traffic flow. By integrating the three core elements of actual commute distance, traffic flow, and visibility, this method quantifies spatial, temporal, and environmental disturbances in the commuting environment. This not only overcomes the limitations of single-indicator analysis but also captures nonlinear disturbances that are difficult for traditional models to identify through a multi-factor coupling mechanism, significantly improving the comprehensiveness, dynamism, and explanatory power of the disturbance analysis in the prediction results.

[0061] Furthermore, the specific process for determining whether to perform decision interference optimization based on the interference analysis results is as follows: if the obtained decision interference index does not exceed the preset decision interference index in the database, the interference analysis result is judged as qualified and the interference analysis of the prediction result is completed; if the obtained decision interference index exceeds the preset actual collected decision interference index in the database, the interference analysis result is judged as unqualified, and a decision interference instruction is sent to the designated user and decision interference optimization is performed.

[0062] The specific process for optimizing decision interference is as follows: First, obtain the interaction duration of the decision interference command and determine if it is greater than the preset interaction duration of the interference command in the database. If the interaction duration is greater than the preset duration, obtain a decrease in the visibility sampling frequency. This decrease represents the result of inputting the obtained decision interference index deviation and the second visibility sampling frequency deviation into the hybrid fuzzy PID (Proportional-Integral-Derivative) control algorithm (a type of fuzzy control algorithm). Second, if the interaction duration is less than the preset duration, obtain an increase in the visibility sampling frequency. This increase represents the result of inputting the obtained decision interference index deviation and the first visibility sampling frequency deviation into the hybrid fuzzy PID control algorithm. The process involves obtaining the decision interference... The index deviation and visibility sampling frequency deviation are jointly input into the hybrid fuzzy PID control algorithm, which adaptively performs fuzzy control calculations. The decision interference deviation is dynamically bound to the sampling frequency deviation to automatically adjust the visibility sampling frequency and improve the accuracy of commuter decision predictions. After each visibility sampling frequency adjustment, the fuzzy control calculation delay time is obtained. If the obtained fuzzy control calculation delay time is not greater than the fuzzy control calculation delay time set in the database, it indicates that the current visibility sampling frequency adjustment has not introduced a significant load. Otherwise, the reduction in visibility sampling frequency is mapped in the database based on the obtained sensor load deviation and the re-obtained decision interference index deviation. If the re-obtained decision interference index within the number of visibility sampling frequency reductions does not exceed the preset actual collected decision interference index in the database, and the number of visibility sampling frequency reductions is monitored by a counter, then decision interference optimization is completed and interference analysis of the prediction results is finished. Otherwise, a decision warning command is sent.

[0063] Sensor load deviation is used to quantify the degree of deviation between the sensor's output characteristics after a load is applied and its output characteristics when unloaded; decision interference index deviation is used to measure the degree of difference between the acquired decision interference index and the preset decision interference index, that is, the difference between the acquired decision interference index and the preset decision interference index; visibility acquisition frequency deviation includes a first visibility acquisition frequency deviation and a second visibility acquisition frequency deviation. The first visibility acquisition frequency deviation represents the difference between the visibility acquisition frequency set value and the current visibility acquisition frequency, and the second visibility acquisition frequency set value represents the difference between the current visibility acquisition frequency and the visibility acquisition frequency set value.

[0064] In this embodiment, the preset decision interference index is represented by extracting the historical decision interference indexes of each period recorded at the end of the historical decision interference assessment period in the database, and then summing and averaging them. This example constructs a triple environmental variable to accurately quantify commuting decision interference under foggy weather events, and innovatively introduces a dynamic visibility acquisition frequency adjustment mechanism. It uses a hybrid fuzzy PID control algorithm to fuzzily control and associate the decision interference deviation with the sampling frequency deviation. Through multi-source data coupling, the spatiotemporal resolution of interference analysis is improved, and the robustness and resource utilization efficiency of the commuting decision system under extreme weather conditions are significantly enhanced.

[0065] like Figure 6The diagram shown is a structural schematic of a personalized prediction system for residents' travel needs based on user profiles, provided in an embodiment of this application. This system includes: a data collection effect analysis module, an accuracy analysis module, and an interference analysis module. The data collection effect analysis module analyzes the data collection process of a specified user information data using acquired data collection effect data, obtaining a data collection effect analysis result. This result measures the quantitative degree to which the data collection effect data contributes to the collection of the specified user information. The accuracy analysis module determines whether data collection optimization should be performed based on the data collection effect analysis result, and simultaneously performs an accuracy analysis on the construction process of a specified user profile based on acquired profile construction data, obtaining an accuracy analysis result. Data collection optimization indicates… The comprehensiveness of extracted user information data is improved by adjusting correlation analysis and data collection frequency. Accuracy analysis measures the accuracy of analysis of corresponding text and voice data modalities during the construction of user profiles. The interference analysis module determines whether to optimize profile construction based on the accuracy analysis results. It also performs interference analysis on the prediction results of user travel demand based on the acquired interference factor data, and determines whether to optimize decision interference based on the results. Profile construction optimization means improving the accuracy of user profile generation by adjusting the profile update frequency. Interference analysis measures the degree of interference of interference factor data on the prediction results of user travel demand. Decision interference optimization means improving the reliability of user travel decision prediction by adjusting the visibility collection frequency.

[0066] In summary, this application embodiment analyzes the collection effect of the specified user information data collection process using the acquired collection effect data, determines whether to optimize data collection based on the collection effect analysis results, determines whether to optimize profile construction based on the accuracy analysis results, and finally performs interference analysis on the prediction results of the specified user's travel demand based on the acquired interference factor data, and determines whether to optimize decision interference based on the interference analysis results. This achieves the effect of improving the accuracy of residents' commuting decision prediction under foggy weather events, and effectively solves the problem of low accuracy of corresponding commuting decision prediction in the user profile analysis process under foggy weather events.

[0067] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0069] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0070] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0071] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0072] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A personalized prediction method for residents' travel needs based on user profiles, characterized in that, Includes the following steps: Step 1: Analyze the collection effect of the specified user information data collection process using the acquired collection effect data to obtain the collection effect analysis results. The collection effect analysis is used to measure the quantitative degree of the collection effect data on the collection effect of the specified user information. Step 2: Based on the results of the data collection effect analysis, determine whether to optimize the data collection. At the same time, based on the acquired profile construction data, perform an accuracy analysis on the construction process of the specified user profile to obtain the accuracy analysis results. The data collection optimization means improving the comprehensiveness of the extraction of specified user information data by adjusting the correlation analysis and adjusting the data collection frequency. The accuracy analysis is used to measure the analysis accuracy of the corresponding text data modal and voice data modal during the construction of the specified user profile. Step 3: Based on the accuracy analysis results, determine whether to optimize the user profile construction. Simultaneously, based on the acquired interference factor data, perform interference analysis on the prediction results of the specified user's travel demand to obtain interference analysis results. Based on the interference analysis results, determine whether to optimize decision interference. The user profile construction optimization refers to improving the accuracy of user profile generation by adjusting the profile update frequency. The interference analysis is used to measure the degree of interference of interference factor data on the user's travel demand prediction results. The decision interference optimization refers to improving the reliability of user travel decision prediction by adjusting the visibility collection frequency. Interference factor data include actual commuting distance, actual traffic flow, and actual visibility; actual commuting distance represents the actual distance traveled by a specified user within a specified commuting area from their origin to their destination, obtained through geographic information system (GIS) analysis; actual traffic flow is obtained through monitoring via the traffic management bureau's website or relevant traffic information platforms; actual visibility distance is obtained through monitoring using a transmission-type visibility meter. Based on the acquired interference factor data, interference analysis is performed on the prediction results of the travel demand of a specified user. The specific process is as follows: First, the degree of difference between the actual commuting distance and the corresponding predicted commuting distance is obtained, that is, the difference between the actual commuting distance and the corresponding predicted commuting distance. At the same time, compensation calculation is performed by combining the actual commuting distance compensation amount in the database to obtain the actual commuting distance coefficient. The specific constraint expression is: ， In the formula, This represents the actual commuting distance coefficient at the end of the preset interference analysis period. This indicates the compensation amount for the actual commuting distance. This indicates the actual commuting distance at the end of the preset interference analysis period. This indicates the preset actual commuting distance; The method involves obtaining the degree of difference between the actual traffic flow and the corresponding predicted traffic flow, i.e., the difference between the actual traffic flow and the predicted traffic flow. Simultaneously, it combines this difference with the actual traffic flow compensation amount from the database to perform compensation calculations, resulting in the actual traffic flow coefficient. The specific constraint expression is: ， In the formula, This represents the actual traffic flow coefficient at the end of the preset interference analysis period. This represents the actual traffic flow compensation amount. This represents the actual traffic flow at the end of the preset interference analysis period. This indicates the preset actual traffic flow; The difference between the actual visibility and the corresponding predicted visibility value is obtained, i.e., the difference between the actual visibility and the corresponding predicted visibility value. Simultaneously, compensation calculations are performed using the actual visibility compensation amount from the database to obtain the actual visibility coefficient. The specific constraint expression is: ， In the formula, This represents the actual visibility coefficient at the end of the preset interference analysis period. This indicates the actual visibility compensation amount. This indicates the actual visibility at the end of the preset interference analysis period. This indicates the preset actual visibility. Finally, the obtained actual commuting distance coefficient, actual traffic flow coefficient, and actual visibility coefficient are coupled to obtain a decision interference index. This index measures the degree of difference between the obtained index and a preset index, i.e., the difference between the obtained and preset indexes. The specific constraint expression is: In the formula, This indicates the decision interference index at the end of the preset interference analysis period.

2. The personalized prediction method for residents' travel needs based on user profiles as described in claim 1, characterized in that, The data collected includes social platform information acquisition values, destination extraction values, and image feature extraction values. The social platform information acquisition value is used to quantify the number of valid information successfully collected from specified user information data at the end of the preset data collection period. The destination extraction value is used to quantify the number of destination keywords extracted from the valid destination information successfully collected from the specified user information data at the end of the preset data collection period. The image feature extraction value is used to quantify the extraction efficiency of valid features of specified user information data at the end of a preset data collection period.

3. The personalized prediction method for residents' travel needs based on user profiles as described in claim 2, characterized in that, The process of analyzing the collection effect of the specified user information data by acquiring the collection effect data is as follows: The degree of difference between the social platform information acquisition value and the preset social platform information acquisition value in the database is obtained, and the compensation calculation is performed in combination with the social platform information acquisition compensation amount in the database to obtain the social platform information acquisition coefficient. The degree of difference between the destination extraction value and the preset destination extraction value in the database is obtained, and a compensation calculation is performed in combination with the destination extraction value compensation amount in the database to obtain the destination extraction coefficient. The degree of difference between the extracted image feature values ​​and the preset image feature extraction values ​​in the database is obtained, and the compensation calculation is performed in combination with the compensation amount of the extracted image feature values ​​in the database to obtain the image feature extraction coefficients; The acquisition coefficients of social platform information, destination extraction coefficients, and image feature extraction coefficients are coupled to obtain the acquisition effect index, which is used to measure the degree of difference between the acquired acquisition effect index and the preset acquisition effect index.

4. The personalized prediction method for residents' travel demand based on user profiles as described in claim 3, characterized in that, The specific process for determining whether to optimize data collection based on the analysis results of the collection effect is as follows: If the acquired collection effect index does not exceed the preset collection effect index in the database, the collection effect analysis result will be judged as qualified data collection and an accuracy analysis command will be sent; otherwise, the collection effect analysis result will be judged as unqualified data collection and voice information extraction optimization will be performed. The specific process for optimizing voice information extraction is as follows: The acquired acquisition effect index deviation and the speech information extraction efficiency deviation are jointly input into the SHAP value attribution calculation for multimodal feature fusion. The acquisition effect index deviation is used to measure the degree of difference between the acquired acquisition effect index and the preset acquisition effect index, and the speech information extraction efficiency deviation is used to measure the degree of difference between the acquired speech information extraction efficiency and the preset speech information extraction efficiency. After optimizing speech information extraction, the multimodal fusion delay time is obtained. If the obtained multimodal fusion delay time is not greater than the multimodal fusion delay time set in the database, it indicates that the current speech information extraction optimization is effective. Otherwise, the reduction in data acquisition frequency is mapped in the database based on the obtained CPU memory usage rate and the deviation of the re-acquired acquisition effect index. If the multimodal fusion delay time reacquired within the number of times the data acquisition frequency is reduced is not greater than the multimodal fusion delay time set in the database, then the voice information extraction optimization is completed and an accuracy analysis command is sent; otherwise, an acquisition effect warning command is sent.

5. The personalized prediction method for residents' travel needs based on user profiles as described in claim 1, characterized in that, The profile construction data includes text information coverage values ​​and voice information coverage values; The text information coverage value is used to quantify the availability of valid text data successfully obtained from text data at the end of the preset construction period for a specified user profile; The voice information coverage value is used to quantify the contribution of the valid user intent successfully obtained from the voice data at the end of the preset construction period for a specified user profile. The accuracy analysis of the process of constructing a specified user profile based on the acquired profile construction data is as follows: The data dimensional integrity index is obtained by geometrically averaging the acquired text information coverage value and voice information coverage value. The data dimension integrity metric is used to reflect the quantitative degree of accuracy of the profile construction data in constructing a specified user profile.

6. The personalized prediction method for residents' travel needs based on user profiles as described in claim 5, characterized in that, The specific process for determining whether to optimize profile construction based on accuracy analysis results is as follows: If the data dimension integrity index obtained exceeds the preset data dimension integrity index in the database, the accuracy analysis result will be judged as unqualified profile construction and an interference analysis command will be sent; otherwise, the accuracy analysis result will be judged as unqualified profile construction and profile construction optimization will be performed. The specific process for portrait construction and optimization is as follows: The obtained data dimension integrity index deviation and the image update frequency deviation are jointly input into the linear regression algorithm to adaptively adjust the image update frequency. The data dimension integrity index deviation is used to measure the degree of difference between the obtained data dimension integrity index and the preset data dimension integrity index, and the image update frequency deviation is used to measure the degree of difference between the actual image update frequency and the preset image update frequency. If the adaptive adjustment delay time re-acquired within the number of optimization attempts for profile building is not greater than the adaptive adjustment delay time set in the database, then profile building optimization is completed and an interference analysis command is sent; otherwise, a profile building warning command is sent.

7. The personalized prediction method for residents' travel demand based on user profiles as described in claim 1, characterized in that... The specific process for determining whether to perform decision interference optimization based on the interference analysis results is as follows: If the obtained decision interference index does not exceed the preset actual collected decision interference index in the database, the interference analysis result will be judged as qualified and the interference analysis of the prediction result will be completed. If the obtained decision interference index exceeds the preset actual decision interference index in the database, the interference analysis result will be judged as unqualified, and a decision interference instruction will be sent to the designated user to optimize the decision interference.

8. The personalized prediction method for residents' travel needs based on user profiles as described in claim 7, characterized in that, The specific process for optimizing decision interference is as follows: Obtain the interaction duration of decision interference instructions and determine whether the obtained interaction duration of decision interference instructions is greater than the preset interaction duration of interference instructions in the database; If the obtained decision interference instruction interaction duration is greater than the preset decision interference instruction interaction duration in the database, then the visibility acquisition frequency reduction value is obtained. The visibility acquisition frequency reduction value represents the result of inputting the obtained decision interference index deviation and the second visibility acquisition frequency deviation into the hybrid fuzzy PID control algorithm. If the interaction time of the acquired decision interference instruction is less than the preset interaction time of the decision interference instruction in the database, then the visibility acquisition frequency increase value is acquired. The visibility acquisition frequency increase value represents the result of inputting the acquired decision interference index deviation and the first visibility acquisition frequency deviation into the hybrid fuzzy PID control algorithm. The obtained decision interference index deviation and visibility acquisition frequency deviation are input into the hybrid fuzzy PID control algorithm to adaptively perform fuzzy control calculations, and the decision interference deviation and sampling frequency deviation are dynamically bound to achieve automatic adjustment of visibility acquisition frequency. After a visibility acquisition frequency adjustment, the fuzzy control calculation delay time is obtained. If the obtained fuzzy control calculation delay time is not greater than the fuzzy control calculation delay time set in the database, it indicates that the current visibility acquisition frequency adjustment has not introduced a significant load. Otherwise, the reduction in visibility acquisition frequency is obtained by mapping the obtained sensor load deviation and the re-acquired decision interference index deviation in the database. If the decision interference index reacquired within the number of times the visibility collection frequency is reduced does not exceed the preset actual collection decision interference index in the database, then the decision interference optimization is completed and the interference analysis of the prediction results is completed; otherwise, a decision warning instruction is sent. The decision interference index deviation is used to measure the degree of difference between the obtained decision interference index and the preset decision interference index.

9. A personalized prediction system for residents' travel needs based on user profiles, used to perform the method described in any one of claims 1-8, characterized in that, include: The system includes a data acquisition effect analysis module, an accuracy analysis module, and an interference analysis module. The data collection effect analysis module is used to analyze the data collection effect of the specified user information data collection process using the acquired data collection effect data, and obtain the data collection effect analysis result. The data collection effect analysis is used to measure the quantitative degree of the data collection effect on the data collection effect of the specified user information. The accuracy analysis module is used to determine whether to optimize data collection based on the results of the data collection effect analysis. At the same time, it performs accuracy analysis on the construction process of the specified user profile based on the acquired profile construction data to obtain the accuracy analysis results. The data collection optimization means improving the comprehensiveness of the extraction of specified user information data by adjusting the correlation analysis and adjusting the data collection frequency. The accuracy analysis is used to measure the analysis accuracy of the corresponding text data modality and voice data modality in the process of constructing the specified user profile. The interference analysis module is used to determine whether to optimize user profile construction based on the accuracy analysis results. Simultaneously, it performs interference analysis on the predicted travel demand of a specified user based on the acquired interference factor data, obtaining interference analysis results. Based on these results, it determines whether to optimize decision interference. User profile construction optimization refers to improving the accuracy of user profile generation by adjusting the profile update frequency. The interference analysis measures the degree of interference from interference factor data on the predicted user travel demand. Decision interference optimization refers to improving the reliability of user travel decision prediction by adjusting the visibility collection frequency.

Images