Special group-oriented visual guidance type interactive interface customization design method

Abstract

The invention relates to the technical field of man-machine interaction, and discloses a special group-oriented visual guidance type interactive interface customization design method, which comprises the following operation steps of S1, multi-modal perception data acquisition: acquiring multi-dimensional physiological and behavior data of a user through gamification tasks (such as finding different, memory cards and color matching); s2, constructing a dynamic user capability model, and integrating perception data and an interaction behavior baseline (such as a task completion path, time consumption and an error rate); according to the visual guidance type interactive interface customization design method oriented to the special group, through multi-mode perception data collection (such as physiological and behavior data collection through a gamification task) and dynamic user ability model construction, 'obstacle type-use scene-ability short board 'of the special group is accurately described, a digital user ability file is formed, and the user ability file is displayed in a user interface. The problems of fuzzy user portraits and insufficient pertinence in existing design are solved, and interface customization has clear data support.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, specifically to a method for custom designing visually guided interactive interfaces for specific groups. Background Technology

[0002] The existing user interface design is based on the physiological, cognitive, and behavioral characteristics of the general user group. It constructs the interface through standardized visual elements (such as buttons of universal size and conventional contrast color schemes) and interaction logic (such as unified form filling process and common navigation layout). It can meet the usage needs of most mainstream users without special disabilities in daily scenarios (such as shopping, information browsing, and basic office work), has a wide user coverage, does not require complex customization for a single segment group, and can achieve efficient user interaction in general scenarios.

[0003] Currently, most user interface designs are based on the physiological and cognitive abilities of ordinary users, and generally suffer from the problem of "excessive general-purpose design but insufficient targeted adaptation," making it difficult to meet the usage needs of special groups (such as visually impaired, cognitively impaired, and motorly impaired individuals, as well as the elderly and children). Specifically, existing technologies have the following core pain points: 1. Limited data collection: Relying solely on basic questionnaires or interviews, there is a lack of multi-dimensional data collection on the physiological (such as vision and motor ability) and behavioral (such as operation path and error pattern) data of special groups, making it impossible to accurately characterize their ability boundaries. 2. Homogeneous guidance solutions: Visual guidance elements (such as buttons and path prompts) use fixed templates and are not customized to address the specific needs of different groups (such as simplifying information for people with cognitive impairments and enhancing contrast for people with visual impairments), resulting in a high rate of accidental touches and difficulty in completing tasks. 3. Lack of iteration mechanism: After the interface is launched, there is no continuous tracking of user behavior and data optimization, which makes it unable to adapt to changes in user capabilities, upgrades in scenario requirements or technology updates, resulting in a long-term decline in adaptability. 4. Insufficient multimodal integration: It is mainly based on visual interaction, lacking or unable to personalize voice and tactile feedback, and has poor compatibility with assistive devices (such as screen readers and blind keyboards), making it difficult to match the interaction preferences of special groups. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a customized design method for visually guided interactive interfaces for special groups. It boasts advantages such as multi-dimensional accurate user modeling, dynamic generation of personalized visual guidance, multi-modal interaction compatibility and adaptation, and data-driven closed-loop iteration. This solves the problems in existing technologies, such as "vague user profiles leading to insufficient guidance targeting, limited adaptation scope due to single interaction methods, and long-term decline in user experience due to a lack of continuous optimization mechanisms." It achieves a shift from "general design" to "customized adaptation," effectively lowering the digital interaction threshold for special groups and improving their usage efficiency and satisfaction.

[0005] (II) Technical Solution To achieve the aforementioned goals of accurate multi-dimensional user modeling, dynamic generation of personalized visual guidance, compatibility and adaptation of multi-modal interactions, and data-driven closed-loop iteration, this invention provides the following technical solution: a customized design method for visually guided interactive interfaces for specific groups, comprising the following steps: Step S1: Multimodal perception data collection, collecting multi-dimensional physiological and behavioral data of users through gamified tasks (such as spot the difference, memory cards, color matching); Step S2: Construct a dynamic user capability model, integrate perception data and interaction behavior baselines (such as task completion path, time consumption, and error rate), and form a digital user capability profile; Step S3: Segment and Profile Tagging. Segment user groups by "barrier type - usage scenario - capability gap", and build a three-dimensional profile for each group, labeling them with "interaction pain points" and "guidance needs". Step S4: Construct a task intent graph, draw a task tree for typical scenarios (such as shopping, medical appointment), and clarify the logical relationships between subtasks; Step S5: Real-time intent inference. Based on the user's real-time behavior (such as dwell area and scrolling behavior) and capability model, a lightweight machine learning algorithm is used to infer the current intent. Step S6: Develop a visually guided atomic component library and establish a configurable library of guiding elements, including focus highlighting, path guidance, semantic enhancement, and multimodal feedback (visual + voice + vibration). Step S7: Rule-driven interface assembly, based on the "IF-THEN" rule engine, matches user models, intents and components; Step S8: Multimodal interaction integration, supporting visual, voice and tactile multimodal feedback, which can be turned on and off independently, and is compatible with auxiliary devices such as screen readers and blind keyboards to ensure compatibility and low latency; Step S9: Multi-dimensional user verification, assemble a user (70% of the target group) + expert team, conduct task scenario testing, and evaluate task success rate, time consumption, and satisfaction (Likert 5-level scale). Step S10: A / B testing and rule optimization. Different guidance schemes are randomly applied to the same intent, the effects are recorded, and the preferred scheme is reused with weighting. Failed interactions trigger rule adjustments. Step S11: User behavior data tracking and analysis, collecting click-through rate, accidental touch rate, frequency of use of guidance path, etc., and identifying design shortcomings, such as redesigning if the symbol accidental touch rate is high. Step S12: Closed-loop iteration mechanism, retest after each round of optimization, until the task success rate is >90% and the satisfaction score is ≥4 points (out of 5). Step S13: Annual upgrade and feedback channel. Based on standard updates, technological developments, and user feedback, the design is upgraded annually, and a user request channel is opened to support the generation of personalized guide templates (such as "memory auxiliary path").

[0006] Furthermore, the gamified task design described in step S1 needs to be adapted to the cognitive and physiological characteristics of special groups. For example, for visually impaired groups, the color matching task is replaced with "texture + contrast matching"; for cognitively impaired groups, the number of cards in the memory card task is controlled to 3-5 sets, and the content of the cards uses concrete images of daily life scenes (such as "water cup" and "key") to avoid abstract symbols. At the same time, the task duration can be dynamically adjusted according to the user's initial operation speed to ensure the effectiveness of the data collection process and the user experience.

[0007] Furthermore, the baseline of interactive behavior described in step S2 needs to be divided into benchmark values ​​according to the dual dimensions of "group type - task scenario". For example, in the "medical appointment" scenario, the baseline for task completion time for the elderly with cognitive decline is set to 1.8-2 times that of the general adult group, and the baseline for error rate is set to 1.5 times that of the general adult group. At the same time, the dynamic user ability profile needs to be updated in real time. The update trigger conditions include the user's three consecutive task operation deviations exceeding the baseline by ±30%, the user actively reporting changes in ability, or automatic feature re-extraction after every five cumulative interactive behavior data.

[0008] Furthermore, the “disability type-usage scenario-capability gap” subdivision dimension mentioned in step S3 must cover at least 6 core special groups (visual impairment, hearing impairment, cognitive impairment, limb movement impairment, elderly group, and children group), and the three-dimensional profile of each group includes 3 layers of core information: basic physiological characteristics (such as “remaining visual acuity” and “whether visual aids are used” for visually impaired groups), interactive behavior characteristics (such as “operable area range” and “single operation duration” for limb movement impairment groups), and demand priority (such as “fun guidance priority > efficiency priority” for children group). The labeling must adopt the “pain point-demand-solution direction” association format (such as “pain point: high button accidental touch rate → demand: operation area differentiation → solution direction: increase button spacing + tactile feedback”).

[0009] Furthermore, the lightweight machine learning algorithm described in step S5 must meet the requirements of low computational power dependence (supporting local operation on the terminal side, with a response latency of <100ms). The preferred algorithm model is a decision tree or a Naive Bayes model. Input features include real-time user behavior data (dwell time, operation trajectory, click frequency), historical behavior data (average operation path in similar scenarios), and current scenario features (page complexity, number of core operation buttons). The intent inference result must output "intent type - confidence level". When the confidence level is <70%, an auxiliary confirmation mechanism is triggered (such as a voice inquiry "Do you need help to complete the current step?") to avoid misjudgment leading to guidance deviation.

[0010] Furthermore, the visual guidance atomic component library described in step S6 must meet three major requirements: "configurability, scalability, and compatibility." The configuration dimensions include color (supporting high contrast mode, contrast ratio ≥ 4.5:1), size (button / icon size can be adjusted according to the user's ability model, ranging from 16px to 48px), and feedback intensity (vibration feedback frequency adjustable from 200Hz to 500Hz, and voice feedback speech rate adjustable from 0.8 to 1.5 times). The component library must reserve an extension interface to support the addition of "emotion-aware guidance components" (such as triggering soft tones and soothing voices when the user is anxious during operation). At the same time, all components must pass the WCAG 2.1AA level accessibility standard certification to ensure compatibility with mainstream assistive devices (such as the NVDA screen reader and the Focus40 blind keyboard).

[0011] Furthermore, the multi-dimensional user verification described in step S9 must adopt a "stratified sampling + scenario reproduction" model. The target group sampling must cover different ability levels within the group (e.g., the visually impaired group must include mild, moderate, and severe sub-samples), with a sample size of no less than 50 people. The test scenarios must include three core scenarios (daily high-frequency scenarios such as "shopping payment", key demand scenarios such as "medical appointment", and complex operation scenarios such as "form filling"). In addition to task success rate, time consumption, and satisfaction, the evaluation indicators must include "guidance dependence" (the proportion of users' independent operation) and "error recovery rate" (the proportion of users who recover from misoperation through guidance). Moreover, the expert team must include at least two experts in the field of accessibility design and one representative of special groups to ensure the objectivity of the evaluation.

[0012] Furthermore, the annual upgrade described in step S13 must follow the "demand priority ranking" mechanism. The priority of upgrade content is determined based on the following criteria: frequency of user feedback (≥20 feedbacks for the same demand are preferred), changes in industry standards (e.g., synchronous adaptation is required after the release of the new national standard for accessibility), and technical feasibility (the cost of implementing new technologies is ≤30% of the current system maintenance budget). Personalized guide template generation must support user-defined core parameters (e.g., "memory auxiliary path" can be customized with "step prompt interval" and "prompt method combination"). After the template is generated, it must undergo small-scale verification (sample size ≥10 people, task success rate ≥85%) before it can be made available for use. At the same time, a template sharing channel should be opened to allow users to share templates with good adaptation effects.

[0013] (III) Beneficial Effects Compared with existing technologies, this invention provides a customized design method for visually guided interactive interfaces for specific groups, which has the following beneficial effects: 1. This visually guided interactive interface customization design method for special groups accurately portrays the "disability type-use scenario-capability shortcomings" of special groups through multimodal perception data collection (such as collecting physiological and behavioral data through gamified tasks) and dynamic user capability model construction, forming a digital user capability profile. This solves the problems of vague user profiles and insufficient targeting in existing designs, and provides clear data support for interface customization.

[0014] 2. This customized design method for visually guided interactive interfaces for special groups achieves precise matching between the interface guidance scheme and the user's real-time needs and capabilities through real-time intent inference (a lightweight algorithm based on user behavior and capability models) and rule-driven interface assembly, combined with a configurable visual guidance atomic component library. This avoids the problem of ineffective guidance caused by fixed templates and improves the operational efficiency of special groups.

[0015] 3. This customized design method for visually guided interactive interfaces for special groups integrates multimodal interaction (supporting independent switching of visual, voice, and tactile feedback) and adapts to assistive devices (compatible with screen readers and blind keyboards) to meet the interaction preferences of special groups with different types of disabilities, break the limitations of single visual interaction, ensure the compatibility and low latency of the interaction process, and lower the threshold for use.

[0016] 4. This customized design method for visually guided interactive interfaces for special groups achieves continuous design optimization through multi-dimensional user verification (target group + expert team testing), A / B testing optimization and closed-loop iteration mechanism (until the task success rate is >90% and the satisfaction score is ≥4 points), coupled with annual upgrades and user feedback channels. This solves the problems of lack of iteration and long-term decline in adaptability of existing designs, and ensures that the interface meets the changing needs of special groups in the long term. Attached Figure Description

[0017] Figure 1 Simplified flowchart of the visually guided interactive interface customization design method of the present invention; Figure 2 This is a flowchart of the user perception modeling and precise demand profiling process of the present invention; Figure 3 This is a flowchart illustrating the intent understanding and task graph construction process of this invention; Figure 4 This is a flowchart illustrating the dynamic interface generation and multimodal guidance process of this invention. Figure 5 This is a flowchart of the system verification and data-driven optimization process for this invention. Figure 6 This is a flowchart illustrating the long-term iteration and ecosystem co-construction process of this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1-6 A method for customizing visually guided interactive interfaces for specific groups, comprising the following steps: Step S1: Multimodal perception data collection, collecting multi-dimensional physiological and behavioral data of users through gamified tasks (such as spot the difference, memory cards, color matching); Step S2: Construct a dynamic user capability model, integrate perception data and interaction behavior baselines (such as task completion path, time consumption, and error rate), and form a digital user capability profile; Step S3: Segment and Profile Tagging. Segment user groups by "barrier type - usage scenario - capability gap", and build a three-dimensional profile for each group, labeling them with "interaction pain points" and "guidance needs". Step S4: Construct a task intent graph, draw a task tree for typical scenarios (such as shopping, medical appointment), and clarify the logical relationships between subtasks; Step S5: Real-time intent inference. Based on the user's real-time behavior (such as dwell area and scrolling behavior) and capability model, a lightweight machine learning algorithm is used to infer the current intent. Step S6: Develop a visually guided atomic component library and establish a configurable library of guiding elements, including focus highlighting, path guidance, semantic enhancement, and multimodal feedback (visual + voice + vibration). Step S7: Rule-driven interface assembly, based on the "IF-THEN" rule engine, matches user models, intents and components; Step S8: Multimodal interaction integration, supporting visual, voice and tactile multimodal feedback, which can be turned on and off independently, and is compatible with auxiliary devices such as screen readers and blind keyboards to ensure compatibility and low latency; Step S9: Multi-dimensional user verification, assemble a user (70% of the target group) + expert team, conduct task scenario testing, and evaluate task success rate, time consumption, and satisfaction (Likert 5-level scale). Step S10: A / B testing and rule optimization. Different guidance schemes are randomly applied to the same intent, the effects are recorded, and the preferred scheme is reused with weighting. Failed interactions trigger rule adjustments. Step S11: User behavior data tracking and analysis, collecting click-through rate, accidental touch rate, frequency of use of guidance path, etc., and identifying design shortcomings, such as redesigning if the symbol accidental touch rate is high. Step S12: Closed-loop iteration mechanism, retest after each round of optimization, until the task success rate is >90% and the satisfaction score is ≥4 points (out of 5). Step S13: Annual upgrade and feedback channel. Based on standard updates, technological developments, and user feedback, the design is upgraded annually, and a user request channel is opened to support the generation of personalized guide templates (such as "memory auxiliary path").

[0020] In the implementation of the case, the gamification task design in step S1 needs to be adapted to the cognitive and physiological characteristics of special groups. For example, for visually impaired groups, the color matching task is replaced with "texture + contrast matching"; for cognitively impaired groups, the number of cards in the memory card task is controlled to 3-5 sets, and the content of the cards uses concrete images of life scenes (such as "water cup" and "key") to avoid abstract symbols. At the same time, the task duration can be dynamically adjusted according to the user's initial operation speed to ensure the effectiveness of the data collection process and the user experience.

[0021] The design logic of gamified tasks should follow the principle of "obstacle compensation - ability stimulation". That is, for the functional defects of special groups, tasks should be constructed in a way that replaces senses or simplifies cognition. At the same time, the fun of the game (such as points rewards and level unlocks) should be used to maintain user engagement. For example, for people with physical movement disorders, fine-clicking tasks can be adjusted into a combination of "area touch + voice confirmation" tasks, which reduces the requirements for movement precision and enhances operation perception through voice feedback. For children, task scenarios can be integrated with cartoon characters and story-based plots, turning the data collection process into a game segment of "helping animated characters complete an adventure". By designing gamified tasks that are adapted to the cognitive and physiological characteristics of special groups, we can obtain more realistic and comprehensive user behavior data (because users are more likely to naturally demonstrate their operating habits and ability boundaries in a gamified context). On the other hand, it also increases the willingness of special groups to participate in data collection, solving the problems of low user cooperation and poor data validity in traditional surveys, and providing high-quality raw data support for the subsequent construction of dynamic user ability models.

[0022] In the implementation of the case, the baseline of interaction behavior in step S2 needs to be divided into benchmark values ​​according to the two dimensions of "group type - task scenario". For example, in the "medical appointment" scenario, the baseline for task completion time for the elderly with cognitive decline is set to 1.8-2 times that of the general adult group, and the baseline for error rate is set to 1.5 times that of the general adult group. At the same time, the dynamic user ability profile needs to be updated in real time. The update trigger conditions include the user's three consecutive task operation deviations exceeding the baseline by ±30%, the user actively reporting changes in ability, or the automatic triggering of feature re-extraction after every 5 cumulative interaction behavior data.

[0023] The determination of the dual-dimensional benchmark values ​​for "group type - task scenario" relies on large-sample studies of special group behaviors and evaluations by industry experts. Taking the "medical appointment" scenario as an example, the benchmarks for task completion time and error rate for the elderly with cognitive decline are determined by comprehensively considering a large amount of actual operational data of this group (covering different degrees of cognitive decline) and combining the judgments of experts in geriatrics and human-computer interaction on the "correlation between cognitive function and interaction efficiency". The real-time update mechanism of the dynamic user ability profile adopts a "triggered condition layering + feature extraction algorithm adaptive" mode. Different triggering conditions correspond to different feature extraction priorities and algorithm parameters (for example, when users actively report changes in their abilities, deep feature extraction will be prioritized for the ability dimensions related to the feedback). By dividing benchmark values ​​into two dimensions—"group type" and "task scenario"—interaction behavior assessments become more scenario-specific and accurate for different groups, avoiding the use of a single benchmark to measure the interaction performance of all groups and scenarios. This allows for precise identification of changes in user capabilities. Real-time updated dynamic user capability profiles can promptly capture changes in user interaction characteristics during long-term use, scenario switching, or capability fluctuations, providing a real-time basis for dynamic adjustments to interface customization designs. This ensures that the interactive interface always matches the user's current capability status, improving the efficiency and experience of special groups using the interactive interface.

[0024] In the implementation of the case, the detailed dimensions of "disability type - usage scenario - capability gap" in step S3 must cover at least 6 core special groups (visual impairment, hearing impairment, cognitive impairment, limb movement impairment, elderly group, and children group), and the three-dimensional profile of each group includes 3 layers of core information: basic physiological characteristics (such as "remaining visual acuity" and "whether visual aids are used" for visually impaired groups), interactive behavior characteristics (such as "operable area range" and "duration of single operation" for limb movement impairment groups), and demand priority (such as "fun guidance priority > efficiency priority" for children group). The labeling must adopt the "pain point - demand - solution direction" association format (such as "pain point: high button accident rate → demand: operation area differentiation → solution direction: increase button spacing + tactile feedback").

[0025] The construction of the subdivision of "disability type - usage scenario - capability gap" needs to be based on interdisciplinary research results and actual scenario needs. For visually impaired people, "remaining visual acuity" can be determined through professional vision test data or users' daily visual assistance behaviors (such as whether they rely on high-contrast displays), while "whether they use visual aids" is related to the type of device (such as electronic visual aids, magnifying glasses) and device usage habits (such as usage time and preferred parameters). The collection of interactive behavior characteristics should rely on technologies such as eye tracking and motion capture to accurately record the operating area range (such as the screen area accessible to the hand and touch accuracy) and the duration of a single operation (such as clicking a button) for people with limb movement disorders. (Time from initial touch to release of the button); Determining the priority of needs requires combining user interviews, behavioral observations, and group commonality research. For example, in educational apps, fun guidance can increase children's willingness to continue using the app, so it takes priority over efficiency. In functional scenarios such as medical appointments, the elderly have a higher priority for operational efficiency. The "pain point-need-solution direction" association format is derived from analyzing a large number of cases of problems encountered by special groups using existing interfaces (such as the problem of accidental button touches, which is caused by small button spacing and lack of tactile feedback, leading to accidental touches by users with physical or visual impairments). This is a logical chain from problem to need to solution. By creating a comprehensive and accurate three-dimensional profile covering multiple core special groups and containing multi-layered core information, we can fully and accurately present the interaction needs and capability boundaries of these special groups in different scenarios. This provides detailed and targeted basis for the subsequent customized design of the interactive interface, avoiding the problem of insufficient adaptability due to an incomplete understanding of user characteristics during the design process. The "pain point-need-solution direction" labeling format allows designers to more clearly deduce design needs and directions from the actual problems of users, accelerating the interface customization design process, improving the effectiveness of design solutions, and ensuring that the final designed interface can effectively solve the interaction pain points of special groups.

[0026] In the implementation of this case, the lightweight machine learning algorithm in step S5 needs to meet the requirements of low computational power dependence (supporting local operation on the terminal side, with a response latency of <100ms). The preferred algorithm model is a decision tree or a Naive Bayes model. Input features include real-time user behavior data (dwell time, operation trajectory, click frequency), historical behavior data (average operation path in similar scenarios), and current scenario features (page complexity, number of core operation buttons). The intent inference result needs to output "intent type - confidence level". When the confidence level is <70%, an auxiliary confirmation mechanism is triggered (such as a voice inquiry "Do you need help to complete the current step?") to avoid misjudgment leading to guidance deviation.

[0027] Among them, the lightweight machine learning algorithms chosen are decision trees or Naive Bayes models because these two types of models have the characteristics of low computational complexity and fast training and inference speed, making them very suitable for running locally on the terminal side without relying on the powerful computing resources of the cloud. They can effectively control the response latency to within 100ms. The selection of input features is a key factor in comprehensively considering user interaction. Real-time user behavior data (dwell time, operation trajectory, click frequency) can reflect the immediate intent of the current operation, historical behavior data (average operation path in similar scenarios) can reflect user habits and preferences, and current scenario features (page complexity, number of core operation buttons) can help judge the impact of the scenario on user intent. The intent inference result outputs "intent type - confidence level", and triggers an auxiliary confirmation mechanism when the confidence level is <70%. This is to reduce the risk of misjudgment while ensuring the efficiency of intent inference. The auxiliary confirmation mechanism of voice inquiry can further clarify the intent through interaction with the user, avoiding incorrect guidance due to misjudgment, and bringing a worse user experience to special groups. By employing lightweight machine learning algorithms that meet the requirements of low computational power dependence and low latency, user intentions can be inferred in real time on the terminal without waiting for cloud data transmission and processing, ensuring smooth interaction. This is especially important for groups with physical and cognitive impairments, avoiding operation interruptions or errors due to latency. Combining multi-dimensional input features with auxiliary confirmation mechanisms for intention inference can more accurately grasp user operation intentions, thereby generating visual guidance solutions that better meet user needs. This improves the efficiency and accuracy of interaction between special groups and the interface, allowing visual guidance to truly assist special groups in their operations.

[0028] In the implementation of this case, the visual guidance atomic component library in step S6 must meet three major requirements: "configurability, scalability, and compatibility." The configuration dimensions include color (supporting high contrast mode, contrast ratio ≥ 4.5:1), size (button / icon size can be adjusted according to the user's ability model, ranging from 16px to 48px), and feedback intensity (vibration feedback frequency adjustable from 200Hz to 500Hz, and voice feedback speech rate adjustable from 0.8 to 1.5 times). The component library must reserve an extension interface to support the addition of "emotion-aware guidance components" (such as triggering soft tones and soothing voices when the user is anxious during operation). At the same time, all components must pass the WCAG 2.1AA level accessibility standard certification to ensure compatibility with mainstream assistive devices (such as the NVDA screen reader and the Focus40 keyboard for the blind).

[0029] The "configurable" feature of the visual guidance atomic component library is designed based on the diverse physiological and interactive needs of special groups. The high-contrast color mode references the WCAG (Web Content Accessibility Guidelines) standard to ensure that visually impaired individuals (such as those with low vision) can clearly distinguish interface elements. The dynamic adjustment of button / icon sizes is based on parameters such as "limb movement precision" and "visual recognition distance" in the user ability model. For example, users with limb movement disorders, who have difficulty with fine motor skills, can be fitted with larger buttons, while children, due to their visual development characteristics, require a suitable size range. The adjustable feedback intensity caters to the perceptual sensitivity of different groups; for instance, the elderly may have weaker perception of vibration and voice, so feedback parameters can be increased, and cognitive impairments... If the disabled population is easily disturbed by strong feedback, the parameters can be reduced. The "expandable" expansion interface design adopts a modular architecture, which facilitates the rapid integration of new functions such as "emotion perception guidance components" based on new research results or technological developments. For example, it can collect users' physiological signals (such as heart rate and skin conductance response) through biosensors to determine anxiety state, and then trigger corresponding visual, tactile or voice soothing guidance. In terms of "compatibility", it has passed the WCAG2.1AA level accessibility standard certification, strictly follows the interaction protocol and data format requirements of assistive devices, and ensures seamless connection with devices such as the NVDA screen reader and the Focus40 blind keyboard in terms of command transmission and information feedback, avoiding problems such as operation delays and missing functions. By providing a visual guidance atomic component library that meets the requirements of "configurability, scalability, and compatibility", it can flexibly adapt to the interaction needs of different special groups in different scenarios, and provide rich and accurate visual guidance elements to support the customization of the interactive interface. At the same time, good compatibility ensures that special groups can use the interface smoothly with the help of assistive devices, while scalability allows the component library to continue to evolve with the development of technology and needs, thereby improving the long-term adaptability of the interactive interface to special groups and the user experience.

[0030] In the implementation of the case, the multi-dimensional user verification in step S9 should adopt the "stratified sampling + scenario reproduction" model. The target group sampling should cover different ability levels within the group (e.g., the visually impaired group should include mild, moderate and severe sub-samples), and the sample size should be no less than 50 people. The test scenarios should include 3 types of core scenarios (daily high-frequency scenarios such as "shopping payment", key demand scenarios such as "medical appointment", and complex operation scenarios such as "form filling"). In addition to task success rate, time consumption and satisfaction, the evaluation indicators should also include "guidance dependence" (the proportion of users' independent operation) and "error recovery rate" (the proportion of users who recover from misoperation through guidance). The expert team should include at least 2 experts in the field of accessibility design and 1 representative of special groups to ensure the objectivity of the evaluation.

[0031] The stratified sampling method, employing a "stratified sampling + scenario reproduction" approach, ensures that the samples represent the overall ability distribution of the special group. Taking the visually impaired group as an example, mild, moderate, and severe subsamples correspond to different degrees of visual function retention, comprehensively testing the interface's performance under different visual conditions. A minimum sample size of 50 people is required for statistical analysis, guaranteeing the statistical significance of the validation results. The test scenarios cover three core categories: high-frequency daily scenarios, key needs scenarios, and complex operations. High-frequency daily scenarios (such as "shopping payment") verify the interface's convenience in regular use, while key need scenarios (such as "medical appointment") verify the support for the core needs of the special group. The verification interface demonstrates the effectiveness of guidance in complex operation scenarios (such as "form filling") under high-difficulty tasks. "Guidance dependency" (the percentage of users who operate independently) assesses the ability of special groups to interact independently under visual guidance. "Error recovery rate" (the percentage of users who recover from misoperations through guidance) assesses the role of guidance in correcting errors. These two indicators supplement the shortcomings of traditional task success rate, time consumption, and satisfaction indicators, and more comprehensively reflect the actual effect of visual guidance. The expert team includes experts in the field of accessibility design and representatives of special groups. Accessibility design experts evaluate from the perspective of professional design standards and technology, while representatives of special groups evaluate from the perspective of their own real-world usage experience and needs, ensuring the comprehensiveness and objectivity of the evaluation. By adopting a "stratified sampling + scenario reproduction" model, multi-dimensional evaluation indicators, and multi-dimensional user verification by a professional and diverse expert team, we can comprehensively and realistically test the effectiveness of the interactive interface in special groups, promptly identify design problems and deficiencies, provide accurate basis for subsequent interface optimization, and ensure that the final delivered interactive interface can effectively meet the interaction needs of special groups and improve their usage efficiency and satisfaction.

[0032] In the implementation of the case, the annual upgrade in step S13 must follow the "demand priority ranking" mechanism. The priority of the upgrade content is determined based on the following criteria: frequency of user feedback (≥20 feedbacks for the same demand are preferred), changes in industry standards (such as synchronous adaptation after the release of the new national standard for accessibility), and technical feasibility (the cost of implementing new technologies is ≤30% of the current system maintenance budget). The generation of personalized guidance templates must support user-defined core parameters (such as "memory assistance path" can be customized with "step prompt interval" and "prompt method combination"). After the template is generated, it must be verified on a small scale (sample size ≥10 people, task success rate ≥85%) before it can be made available for use. At the same time, a template sharing channel should be opened to allow users to share templates with good adaptation effects.

[0033] In the "demand priority ranking" mechanism, the frequency of user feedback must be statistically analyzed based on a long-term, multi-channel user feedback collection system to ensure that feedback sources for the same demand are authentic and representative. When the same demand receives ≥20 feedback responses, it indicates that the demand has high prevalence among specific groups and should be given priority. For adaptation to changes in industry standards, a standards tracking and interpretation team should be established to promptly translate the technical requirements and design specifications in new national standards (such as the new accessibility standard) into specific interface upgrade requirements, ensuring that the product meets the latest industry accessibility requirements. Technical feasibility assessment requires a comprehensive consideration of R&D costs, time costs, and the current system maintenance budget. When the cost of implementing a new technology is ≤30% of the current system maintenance budget, the new technology can be introduced under controllable costs, improving product performance. Without placing excessive financial pressure on the existing system operation, the user-defined core parameters of the personalized guidance templates are set around the common interactive assistance needs of special groups. The "step prompt interval" of the "memory assistance path" can be adapted to the memory rhythm of people with cognitive impairments. The "prompt method combination" (such as visual + voice, tactile + visual, etc.) can meet the perceptual preferences of different disability types. The small-scale validation sample size is ≥10 people and the task success rate is ≥85%. This is to verify the effectiveness and applicability of the template to the target group before the template is released on a large scale, and to avoid defective templates affecting the user experience. The opening of the template sharing channel, with the help of experience sharing among users, allows templates with good adaptation effects to spread quickly among special groups, improving the overall user experience of the interactive interface for the group. By following the "demand priority" mechanism for annual upgrades, we can ensure that the upgraded interface focuses on the most urgent and common needs of special groups and the latest industry requirements, while also taking into account technical feasibility, making the upgraded interface more suitable for actual use scenarios. The support for user-defined core parameters and verified personalized guidance templates can provide more targeted interactive guidance for different special groups, while the open template sharing channel can promote the dissemination of high-quality adaptation experience, further improving the efficiency and satisfaction of special groups in using the interface, and achieving continuous optimization and universal sharing of the interface.

[0034] In summary, this customized visually guided interactive interface design method for special groups accurately portrays the "disability type-usage scenario-capability gap" of special groups through multimodal perception data collection (such as collecting physiological and behavioral data through gamified tasks) and dynamic user capability model construction, forming a digital user capability profile. This solves the problems of vague user profiles and insufficient targeting in existing designs, providing clear data support for interface customization. Through real-time intent inference (a lightweight algorithm based on user behavior and capability models) and rule-driven interface assembly, combined with a configurable visual guidance atomic component library, it achieves precise matching between the interface guidance scheme and the user's real-time needs and capabilities, avoiding the problem of ineffective guidance caused by fixed templates, and improving the operational efficiency of special groups.

[0035] Furthermore, by integrating multimodal interaction (supporting independent switching of visual, voice, and haptic feedback) and adapting to assistive devices (compatible with screen readers and blind keyboards), it meets the interaction preferences of special groups with different types of disabilities, breaks the limitations of single visual interaction, ensures the compatibility and low latency of the interaction process, lowers the threshold for use, and achieves continuous design optimization through multi-dimensional user verification (target group + expert team testing), A / B testing optimization and closed-loop iteration mechanism (until the task success rate is >90% and the satisfaction score is ≥4 points), coupled with annual upgrades and user feedback channels. This solves the problems of lack of iteration and long-term decline in adaptability of the existing design, ensuring that the interface keeps up with the changing needs of special groups.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for designing a visual guidance type interactive interface for a specific group, characterized in that: Comprise the following operation steps: Step S1: Multimodal perception data collection, through gamification tasks (such as find different, memory card, color matching) to collect user's multidimensional physiological and behavioral data; Step S2: Constructing a dynamic user ability model, integrating perception data and interaction behavior baseline (such as task completion path, time consumption, error rate), forming a digital user ability profile; Step S3: Subdivide groups and portrait labeling, subdivide user groups according to "obstacle type-use scenario-ability short board", and construct a complete portrait for each group, label "interaction pain points" and "guidance needs" tags; Step S4: Constructing a task intention graph, drawing a task tree for typical scenarios (such as shopping, medical appointment), and clearly defining the logical relationship between subtasks; Step S5: Real-time intention inference, based on user real-time behavior (such as stay area, scrolling behavior) and ability model, using lightweight machine learning algorithm to infer current intention; Step S6: Visual guidance atomic component library development, establishing a configurable guidance element library, including focus highlight, path guidance, semantic reinforcement, multimodal feedback (visual+voice+vibration) and so on; Step S7: Rule-driven interface assembly, based on "IF-THEN" rule engine, matching user model, intention and component; Step S8: Multimodal interaction fusion, supporting visual, voice, tactile multimodal feedback, which can be independently switched, and adapting to screen readers, blind keyboards and other assistive devices, ensuring compatibility and low delay; Step S9: Multidimensional user verification, establishing user (70% of target group) + expert team, carrying out task scenario test, and evaluating task success rate, time consumption, satisfaction (Likert 5-level scale); Step S10: A / B testing and rule tuning, randomly using different guidance schemes for the same intention, recording the effect, and optimizing the scheme for weighted reuse, and triggering rule adjustment for failed interaction; Step S11: User behavior data tracking and analysis, collecting click rate, mis-touch rate, guidance path usage frequency, and identifying design short boards, such as high symbol mis-touch rate for redesign; Step S12: Closed-loop iteration mechanism, retest after each round of optimization until task success rate > 90%, satisfaction ≥ 4 points (5-point system); Step S13: Annual upgrade and feedback channel, according to standard update, technology development, user feedback, design upgrade every year, and open user demand channel, support personalized guidance template generation (such as "memory assistance path").

2. The method of claim 1, wherein the method further comprises: The gamification task design of step S1 needs to adapt to the cognitive and physiological characteristics of special groups, for example, for visually impaired groups, color matching tasks are replaced by "texture + contrast matching"; for cognitive impaired groups, the number of memory card tasks is controlled at 3-5 groups, and the card content uses concrete life scene images (such as "water cup" "key"), avoiding abstract symbols, and the task duration can be dynamically adjusted according to the user's initial operation speed, to ensure the effectiveness and user experience of the data collection process.

3. The method of claim 1, wherein: The baseline of the interaction behavior in step S2 needs to be divided into two dimensions of "group type-task scenario", for example, in the "medical appointment" scenario, the task completion time baseline of the elderly cognitive decline group is set to 1.8-2 times that of the ordinary adult group, and the error rate baseline is set to 1.5 times that of the ordinary adult group. The dynamic user ability profile needs to be updated in real time, and the update trigger conditions include that the user's continuous 3 times of task operation deviation exceeds the baseline ±30%, the user actively feedbacks the change of ability, or the characteristic re-extraction is automatically triggered after every 5 times of interaction behavior data is accumulated.

4. The method of claim 1, wherein: The "disorder type-use scenario-ability short board" subdivision dimension in step S3 needs to cover at least 6 core special groups (visual impairment, hearing impairment, cognitive impairment, motor impairment, elderly group, and children group), and the stereoscopic portrait of each group contains 3 layers of core information: basic physiological characteristics (such as "remaining visual acuity" and "whether to use visual aid devices" of the visual impairment group), interaction behavior characteristics (such as "operable area range" and "single operation duration" of the motor impairment group), and demand priority (such as "interest guidance priority > efficiency priority" of the children group), and the label annotation needs to use the "pain point-demand-solution direction" association format (such as "pain point: high button mis-touch rate → demand: operation area division → solution direction: increase button spacing + tactile feedback").

5. The method of claim 1, wherein: The lightweight machine learning algorithm in step S5 needs to meet the low algorithm power dependence (supporting local running on the terminal side, response delay <100ms), and the algorithm model is preferably a decision tree or a naive Bayes model, and the input features include real-time behavior data of the user (dwell time, operation trajectory, click frequency), historical behavior data (average operation path in the same scenario), and current scene features (page complexity, number of core operation buttons); the intent inference result needs to output "intent type-confidence", and when the confidence is <70%, the auxiliary confirmation mechanism (such as voice inquiry "whether to need help to complete the current step") is triggered to avoid misjudgment leading to guidance deviation.

6. The method of claim 1, wherein: The visual guidance atomic component library in step S6 needs to meet the three requirements of "configurable-expandable-compatibility": the configuration dimensions include color (supporting high contrast mode, contrast ≥4.5:1), size (button / icon size can be adjusted according to the user ability model, range 16px-48px), and feedback intensity (vibration feedback frequency 200Hz-500Hz adjustable, voice feedback speed 0.8-1.5 times adjustable); the component library needs to reserve an expansion interface to support the addition of "emotion perception guidance components" (such as triggering soft color and soothing voice when the user is anxious); at the same time, all components need to pass the WCAG2.1AA level accessibility standard certification to ensure compatibility with mainstream assistive devices (such as screen reader NVDA and blind keyboard Focus40).

7. The method of claim 1, wherein: The multi-dimensional user verification described in step S9 needs to use the "stratified sampling + scenario reproduction" mode. The target group sampling needs to cover different ability levels within the group (for example, the visually impaired group needs to include mild, moderate, and severe sub-samples), and the sample size should not be less than 50 people. The test scenarios need to include three types of core scenarios (daily high-frequency scenarios such as "shopping payment", key demand scenarios such as "hospital appointment", and complex operation scenarios such as "form filling"), and in addition to the task success rate, time consumption, and satisfaction, the evaluation indicators also need to include "guidance dependency" (the proportion of user self-operation) and "error recovery rate" (the proportion of recovery after error operation). The expert team needs to include at least 2 experts in the field of barrier-free design and 1 representative of special groups to ensure the objectivity of the evaluation.

8. The method of claim 1, wherein: The annual upgrade described in step S13 needs to follow the "demand priority ranking" mechanism. The priority determination of the upgrade content includes: user feedback frequency (the same demand feedback ≥ 20 times is preferred), industry standard changes (such as the need to adapt to new barrier-free national standards after their release), and technical feasibility (the cost of new technology landing is ≤ 30% of the current system maintenance budget). The generation of personalized guidance templates needs to support user-defined core parameters (such as "memory assistance path" which can be customized "step prompt interval time" and "prompt mode combination"), and the template needs to be verified in a small scale (sample size ≥ 10 people, task success rate ≥ 85%) before it can be opened for use. At the same time, a template sharing channel is opened to allow users to share templates with good adaptation effect.

Images