Data-driven user experience optimization method

Abstract

The invention discloses a data-driven user experience optimization method, which is characterized in that real-time UI / UX adjustment can greatly improve the satisfaction degree of a user and reduce operation difficulty or function omission caused by unreasonable interface design, personalized user experience is beneficial to enhancing the viscosity of the user to a platform, long-term use and conversion are promoted, and the user experience is improved. The system can automatically recognize user behaviors and demand changes, so that manual intervention and optimization cost are reduced, and the overall working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of data optimization, and in particular to a data-driven user experience optimization method. Background Art

[0002] Many websites and applications collect subjective feedback from users through questionnaires, user interviews, customer service support, etc., and then adjust their UIs according to these feedbacks. There are also some more professional websites that collect users' behavioral data, such as click streams, dwell times, and page browsing paths, to adjust UI designs. The current low-code platforms perform UI adjustments to pages through rapid configuration.

[0003] Most UI / UX designs are still based on fixed static interface layouts and element configurations. Once released, the design usually does not automatically adjust in real time according to user behavior, and at most can only perform corresponding adaptations according to different screen resolutions. The prior art has greatly reduced the mental burden of user operations, allowing everyone to use the same set of templates to reduce the learning cost, but it cannot vary from person to person.

[0004] Designers usually need to adjust the interface based on periodic user data analysis, and this kind of adjustment often requires a development cycle and re-testing by users. Because these feedbacks are usually periodic, there is often a long lag, and at the same time, it cannot meet the different experiences of different users.

[0005] Most of the current website push is based on user content push, marking the content that users frequently visit, and pushing relevant content when browsing next time. The page personalization is limited to some color themes and fixed title menu layout methods.

[0006] In summary, a data-driven user experience optimization method is needed to solve the deficiencies in the prior art. Summary of the Invention

[0007] In view of the deficiencies of the prior art, the present invention provides a data-driven user experience optimization method, aiming to solve the above problems.

[0008] To achieve the above object, the present invention provides the following technical solution: A data-driven user experience optimization method, including the following steps: Step S1: Data collection, collecting various user behavior data in real time; Step S2: Feature extraction, extracting page features by using natural language processing or image processing; Step S3: Algorithm selection and analysis, selecting different algorithms for analysis according to different user preferences, and using reinforcement learning to adjust the strategy of the recommendation system to generate a suitable page template; Step S4: According to the analysis results, the system automatically generates an optimization plan; Step S5: According to the changes in user behavior, through a real-time feedback mechanism, respond to the changes in user behavior in real time.

[0009] Optionally, the user behavior data in step S1 includes but is not limited to font style, color parameters, layout, theme color, page structure, content frequently clicked by the user, residence time for viewing a certain module, and input content.

[0010] Optionally, the page features extracted in step S2 include visual features, text features, and color features; Visual feature extraction: Extract the visual features of the pre-layout through image processing technology; Text feature extraction: Extract the characteristics of the text content on the page, and process the text using One-Hot Encoding or Word2Vec technology; Color feature extraction: Extract color information using color histograms or RGB space methods.

[0011] Optionally, the algorithms in step S3 include K-means algorithm, Collaborative Filtering algorithm, and deep learning model algorithm; The steps of the K-means algorithm are as follows: Step A1: Initialization, randomly select K center points as the clustering centers; Step A2: Assignment, assign each page template to the cluster corresponding to the center point closest to it; Step A3: Update, update the center point of each cluster to the mean value of all page templates in the current cluster; Step A4: Repeat steps A2 and A3 until the clustering centers reach the set threshold.

[0012] Optionally, the steps of the Collaborative Filtering algorithm are as follows: Step B1: Calculate template similarity, calculate the similarity between templates using the cosine similarity and / or Pearson correlation coefficient method; Step B2: Recommend templates, according to the similarity matrix, recommend other templates similar to the templates liked by the user.

[0013] Optionally, the steps of the deep learning model algorithm are as follows: Step C1: Data preprocessing, adjust the page template image to a fixed size and standardize the image pixel values.

[0014] Step C2: Convolutional layer, extract the local features of the image through multiple convolutional layers.

[0015] Step C3: Pooling layer, reducing the feature dimension and retaining important information.

[0016] Step C4: Fully connected layer, classifying the extracted features through the fully connected layer and outputting a template recommendation; Step C5: Adopting reinforcement learning to adjust the strategy of the recommendation system to generate the most suitable page template.

[0017] Optionally, the optimization solutions in step S4 include optimization based on sentiment analysis, in-depth customization optimization based on user portraits, and intelligent adjustment optimization based on AI polishing; Optimization based on sentiment analysis is carried out in the following way: Step D1: Collect user comments and feedback text information to obtain user behavior data.

[0018] Step D2: Use a sentiment analysis model to judge the sentiment tendency of the text information to obtain the sentiment analysis result; Step D3: Combine the sentiment analysis result with the user behavior data to generate a comprehensive data set.

[0019] Step D4: Input the comprehensive data set and generate an optimization solution recommendation through algorithm logic or a machine learning model.

[0020] Optionally, the in-depth customization optimization based on user portraits is carried out in the following way: Step E1: Collect and integrate multi-dimensional information of users to build user portraits; Step E2: Segment users according to the user portraits to form different user groups; Step E3: For each user group, combine its behavior data and user portraits to generate personalized optimization solution recommendations; Step E4: The system pushes the personalized optimization solutions to the corresponding user groups to achieve in-depth customization.

[0021] Optionally, the intelligent adjustment optimization based on AI polishing is carried out in the following way: Step F1: The user makes custom adjustments according to the needs; Step F2: The system conducts intelligent analysis on the adjusted page; Step F3: According to the analysis result, the system automatically adjusts the parameters of the page elements; Step F4: The user views the effect of the adjusted page and makes further adjustments or confirms and saves according to the needs.

[0022] Advantages of the present invention: 1. In the present invention, real-time UI / UX adjustment can greatly improve user satisfaction, reduce operation difficulties or functional omissions caused by unreasonable interface design. The personalized user experience helps enhance user stickiness to the platform, promotes long-term use and conversion. The system can automatically identify changes in user behavior and needs, thereby reducing manual intervention and optimization costs, and improving overall work efficiency; 2. In the present invention, by collecting various behavior data of users and using natural language processing or image processing technology to extract page features, this method can more accurately understand user preferences. Combining different algorithms (such as K-means, collaborative filtering, and deep learning models) for analysis, personalized page templates and optimization schemes can be generated, thereby enhancing user satisfaction and experience. By adopting reinforcement learning to adjust the strategy of the recommendation system and responding to changes in user behavior through a real-time feedback mechanism, the system can dynamically adapt to changes in user needs and provide content and services that are more in line with the user's current interests and preferences; 3. In the present invention, it is not limited to a single type of user behavior data, but includes various visual and text characteristics such as font styles, color parameters, layouts, etc., as well as multi-dimensional optimization methods such as intelligent adjustment based on sentiment analysis, in-depth customization of user portraits, and AI polishing, comprehensively improving the quality of the user experience. Through automated data analysis and algorithm application, the need for manual intervention is reduced, and the efficiency and accuracy of the optimization process are improved. For example, using a convolutional neural network to extract image features can effectively identify and classify page elements; while the collaborative filtering algorithm can provide recommendations more in line with the user's preferences by calculating template similarities. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram of a method flow of the present invention.

[0024] Figure 2 It is a schematic diagram of a processing flow of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0026] As Figure 1 and Figure 2 shown, a data-driven user experience optimization method includes the following: The system will first collect real-time user behavior data through various means. Use machine learning algorithms to analyze the user's behavior patterns and identify the user's needs, preferences, pain points, and potential churn points. The specific steps are as follows: Step 1: Data collection: font style, color parameters, layout (such as navigation bar, sidebar, main content area, etc.), theme color, page structure (such as grid layout, card layout, etc.), content frequently clicked by the user, viewing time of a certain module, input content, etc.

[0027] Step 2: Feature extraction: Use natural language processing (NLP) or image processing to extract page visual features from existing templates: Visual features: Extract the visual features of the layout through image processing techniques (such as extracting design elements in the page through CNN).

[0028] Text features: Extract the characteristics of the text content of the page, including font size, color, etc., and process the text using techniques such as One-Hot Encoding or Word2Vec.

[0029] Color features: Extract color information using methods such as color histograms or RGB spaces.

[0030] Step 3: Algorithm selection: K-means: Classify different page templates through a clustering algorithm to find similar page designs. Pages can be clustered based on parameters such as the visual features and layout style of the template, so as to recommend a template algorithm similar to the user's historical preferences. Steps: · Initialization: Randomly select K center points as the clustering centers.

[0031] · Assignment: Assign each page template to the cluster corresponding to the center point closest to it.

[0032] · Update: Update the center point of each cluster to the mean of all page templates in the current cluster.

[0033] · Repeat steps 2 and 3 until the clustering centers no longer change or change very little.

[0034] Collaborative Filtering: Collaborative filtering recommends similar templates based on the user's historical preferences. This method predicts the templates that the user may like based on the similarity between the user and the templates. It is divided into user-based collaborative filtering and item-based collaborative filtering.

[0035] Steps: · Calculate template similarity: Use methods such as cosine similarity and Pearson correlation coefficient to calculate the similarity between templates.

[0036] · Recommended templates: Based on the similarity matrix, recommend other templates similar to the templates that the user likes.

[0037] Deep learning model: Convolutional neural network (CNN) is very suitable for extracting features from images and can be used to analyze the visual elements of page templates. By training a CNN model, we can automatically extract layout and design style features from page images, thus realizing template recommendation.

[0038] Steps: · Data preprocessing: Resize the page template images to a fixed size and standardize the image pixel values.

[0039] · Convolutional layer: Extract local features of the image through multiple convolutional layers, such as color, shape, layout, etc.

[0040] · Pooling layer: Reduce the feature dimension and retain important information.

[0041] · Fully connected layer: Classify the extracted features through the fully connected layer and output a template recommendation.

[0042] Use reinforcement learning to gradually adjust the strategy of the recommendation system, and then generate the most suitable page template.

[0043] According to the analysis results, the system can automatically generate optimization plan recommendations 1) Optimization recommendation based on sentiment analysis: When generating an optimization plan, the system not only considers the user's behavior data (such as clicks, dwell time, etc.), but also introduces sentiment analysis technology to analyze the sentiment tendency (positive, negative, neutral) in text information such as user comments and feedback through natural language processing (NLP). According to the user's sentiment feedback, the system can more accurately identify the user's pain points and needs, and thus generate an optimization plan that is closer to the user's expectations.

[0044] Implementation steps: · Collect text information such as user comments and feedback.

[0045] · Use a sentiment analysis model (such as a deep learning model based on LSTM, BERT, etc.) to judge the sentiment tendency of the text information.

[0046] · Combine the sentiment analysis results with the behavior data and use them together as the input for generating the optimization plan.

[0047] · The system generates more accurate optimization plan recommendations based on the comprehensive input.

[0048] 2) Deep customization based on user profiles: When generating optimization solutions, the system not only considers the current user's behavior data but also combines the user profile for deep customization. The user profile includes multi-dimensional information such as the user's age, gender, region, occupation, interests, etc. By deeply mining the user profile, the system can understand the user more comprehensively and thus generate more personalized optimization solutions.

[0049] Implementation steps: · Collect and integrate the user's multi-dimensional information to build a user profile.

[0050] · Segment the users according to the user profile to form different user groups.

[0051] · For each user group, combine its behavior data and user profile to generate personalized optimization solution recommendations.

[0052] · The system pushes the personalized optimization solutions to the corresponding user groups to achieve deep customization.

[0053] 3) Dynamic adjustment and optimization strategy: The system can adjust the optimization strategy in real time according to changes in user behavior. For example, when a user frequently clicks on a certain area but fails to complete the expected operation, the system can automatically identify this behavior pattern and dynamically adjust the layout or interaction method of this area to reduce the user's operation difficulty and improve user satisfaction.

[0054] Implementation steps: · Collect user behavior data in real time and conduct dynamic analysis · When it is recognized that the user behavior pattern has changed, the system adjusts the corresponding optimization solution according to the preset optimization strategy library.

[0055] · The system pushes the adjusted layout or interaction method to the user in real time and collects the user's feedback data.

[0056] · According to the user feedback data, the system continuously optimizes and adjusts the optimization strategy library to achieve continuous iteration and optimization.

[0057] 4) Intelligent adjustment based on AI polishing: After the user's self-defined adjustment, the system provides an AI polishing function that can automatically detect the rationality of the user's adjustment and perform intelligent optimization. For example, when the user's adjustment of the layout causes page elements to overlap or the typesetting to be unappealing, the system can automatically adjust the element positions or spacing to present a more beautiful page effect.

[0058] Implementation steps: · After the user completes the self-defined adjustment, click the AI polishing button · The system performs intelligent analysis on the adjusted page to detect problems such as typesetting and element overlap.

[0059] · Based on the analysis results, the system automatically adjusts parameters such as the position and spacing of page elements to optimize the page effect.

[0060] · Users can view the adjusted page effect and make further adjustments or confirm and save according to their needs.

[0061] The key innovation of this technology lies in its real-time feedback mechanism (point 2 in this article). Different from traditional periodic data analysis and manual design adjustments, this system can respond in real time to changes in user behavior. For example, if a user stays on a certain page for too long, the system can automatically prompt possible operation steps or simplify the interaction process. The system will continuously adjust its optimization strategy according to user feedback (such as click-through rate, operation completion rate, satisfaction score, etc.), forming an adaptive learning mechanism to continuously improve the user experience.

[0062] By customizing a layout style exclusive to each user based on their behavior data and preferences, the system can provide personalized UI / UX designs for different user groups. For example, different interface layouts or functions are shown for new users, old users, users from different regions or devices. In a multi-device environment, the system can automatically adapt to different screen sizes and resolutions (providing corresponding responsive designs for mobile devices for the styles configured on the PC side), providing a seamless cross-device user experience.

[0063] The real-time UI / UX adjustment of the present invention can greatly improve user satisfaction, reduce operation difficulties or functional omissions caused by unreasonable interface design. The personalized user experience helps to enhance user stickiness to the platform, promote long-term use and conversion. The system can automatically identify changes in user behavior and needs, thereby reducing manual intervention and optimization costs and improving overall work efficiency; By collecting various behavior data of users and using natural language processing or image processing technologies to extract page features, this method can more accurately understand user preferences. Combining different algorithms (such as K-means, collaborative filtering, and deep learning models) for analysis, personalized page templates and optimization schemes can be generated, thereby improving user satisfaction and experience. Adopting reinforcement learning to adjust the strategy of the recommendation system and responding to changes in user behavior through a real-time feedback mechanism enables the system to dynamically adapt to changes in user needs and provide content and services that are more in line with the user's current interests and preferences; Not limited to a single type of user behavior data, but including various visual and text features such as font styles, color parameters, layouts, etc., as well as multi-dimensional optimization methods such as intelligent adjustment based on sentiment analysis, in-depth customization of user portraits, and AI polishing, comprehensively improving the quality of the user experience. Through automated data analysis and algorithm application, the need for manual intervention is reduced, and the efficiency and accuracy of the optimization process are improved. For example, using a convolutional neural network to extract image features can effectively identify and classify page elements; while the collaborative filtering algorithm can provide recommendations more in line with the user's preferences by calculating the template similarity.

[0064] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, or improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A data-driven user experience optimization method, characterized in that: The following steps are involved: Step S1: Data collection, real-time collection of various user behavior data; Step S2: feature extraction, using natural language processing or image processing to extract page features; Step S3: Algorithm selection and analysis: different algorithms are selected for analysis according to different user preferences, and reinforcement learning is used to adjust the recommendation system strategy to generate a suitable page template; Step S4: Based on the analysis results, the system automatically generates an optimization plan; Step S5: According to the changes in user behavior, respond to the changes in user behavior in real time through a real-time feedback mechanism.

2. The data-driven user experience optimization method according to claim 1, characterized in that: The user behavior data in step S1 includes, but is not limited to, font style, color parameters, layout, theme color, page structure, user frequently clicked content, viewing time of a module, and input content.

3. The data-driven user experience optimization method according to claim 1, characterized in that: The page features extracted in step S2 include visual features, text features and color features; Visual feature extraction: Visual features arranged in advance through image processing technology; Text feature extraction: Extract features from the text content of the page and process the text using One-Hot Encoding or Word2Vec technology; Color feature extraction: Use color histogram or RGB space method to extract color information.

4. The data-driven user experience optimization method according to claim 1, characterized in that: The algorithms in step S3 include K-means algorithm, Collaborative Filtering algorithm and deep learning model algorithm; The steps of K-means algorithm are as follows: Step A1: Initialization, randomly select K center points as cluster centers; Step A2: Assign each page template to the cluster corresponding to the center point closest to it; Step A3: Update, update the center point of each cluster to the mean of all page templates in the current cluster; Step A4: Repeat steps A2 and A3 until the cluster center reaches the set threshold.

5. The data-driven user experience optimization method according to claim 4, characterized in that: The steps of the Collaborative Filtering algorithm are as follows: Step B1: Calculate template similarity, using cosine similarity and / or Pearson correlation coefficient method to calculate the similarity between templates; Step B2: Recommend templates. Based on the similarity matrix, recommend other templates similar to the templates that the user likes.

6. The data-driven user experience optimization method according to claim 4, characterized in that: The deep learning model algorithm steps are as follows: Step C1: Data preprocessing, adjusting the page template image to a fixed size and standardizing the image pixel value; Step C2: Convolutional layer, extracting local features of the image through multiple convolutional layers; Step C3: Pooling layer, reducing feature dimensions and retaining important information; Step C4: Fully connected layer, classify the extracted features through the fully connected layer and output a template recommendation; Step C5: Use reinforcement learning to adjust the strategy of the recommendation system and generate the most appropriate page template.

7. The data-driven user experience optimization method according to claim 1, characterized in that: The optimization scheme in step S4 includes optimization based on sentiment analysis, deep customization optimization based on user portraits, and intelligent adjustment optimization based on AI polishing; Based on sentiment analysis optimization, through the following methods: Step D1: Collect user comments and feedback text information to obtain user behavior data; Step D2: Use the sentiment analysis model to judge the sentiment tendency of the text information and obtain the sentiment analysis result; Step D3: Combine sentiment analysis results with user behavior data to generate a comprehensive dataset; Step D4: Input the comprehensive data set and generate optimization solution recommendations through algorithm logic or machine learning models.

8. The data-driven user experience optimization method according to claim 7, characterized in that: The deep customization optimization based on user portrait is achieved in the following ways: Step E1: Collect and integrate multi-dimensional information of users to build user portraits; Step E2: Segment users according to user portraits to form different user groups; Step E3: For each user group, generate personalized optimization solution recommendations based on their behavior data and user profiles; Step E4: The system pushes the personalized optimization plan to the corresponding user group to achieve deep customization.

9. The data-driven user experience optimization method according to claim 7, characterized in that: The intelligent adjustment and optimization based on AI polishing is achieved in the following ways: Step F1: The user makes custom adjustments according to needs; Step F2: The system performs intelligent analysis on the adjusted page; Step F3: Based on the analysis results, the system automatically adjusts the parameters of the page elements; Step F4: The user views the adjusted page effect and makes further adjustments or confirms saving as needed.

Images