Adjustable interactive interface design method and system fusing eye movement information of a group

Abstract

The application discloses a method and system for adjustable interactive interface design by fusing eye movement information of a group, which first collects multiple eye movement characteristic values of individuals in the process of interaction with an interactive interface; collects the eye movement characteristic values of the individuals and demographic information to represent individual characteristic sets; obtains the preference degrees of the individuals for different interface design elements to form a preference set; collects the eye movement characteristic values associated with different interface design elements of the individuals, and combines individual weight values to form a relationship network; aggregates the data of the individuals to form a group database with a configurable individual weight distribution; establishes a group decision model, calculates the recommended degree of each interface design element according to the eye movement characteristic values and demographic information of the users, and sorts the attraction of the interface design elements to the users according to the recommended degree. The application can adapt to the changes of different user behavior modes and preferences, and improve user experience and decision efficiency.

Description

Technical Field

[0001] This invention relates to the field of interface design technology, specifically to a method and system for dynamically optimizing and iteratively designing interactive interfaces by integrating eye-tracking technology with group decision-making data. Background Technology

[0002] In recent years, with the rapid development of human-computer interaction technology, the design of user interfaces not only needs to meet users' basic operational needs but also needs to achieve a more efficient user experience, especially in applications under complex decision-making environments, such as multi-user collaborative systems and intelligent data analysis platforms. Against this backdrop, eye-tracking technology, as an intuitive and effective means of capturing user behavior, has gained widespread attention due to its ability to reflect users' attention distribution, operational preferences, and cognitive load in real time. Meanwhile, group decision-making, as an important form of collective intelligence, has been widely applied in fields such as business analysis, medical diagnosis, and education and training. How to combine real-time behavioral data of individual users with group eye-tracking decision-making data has become an important topic in the field of user interface design.

[0003] Traditional user interface design methods often rely on static design schemes, typically using pre-defined logical rules to meet user needs. However, this approach often exhibits insufficient adaptability and delayed feedback when faced with dynamically changing user behavior or multi-user decision-making scenarios. For example, some collaborative tools based on fixed interface designs cannot adjust their interface layout in a timely manner to adapt to different users' attention distributions or decision preferences, thereby reducing decision-making efficiency and user satisfaction.

[0004] While existing research has yielded some progress in eye-tracking technology and group decision theory, combining the two still faces numerous challenges. On one hand, eye-tracking data is complex and requires high real-time performance, necessitating efficient algorithms to support dynamic analysis and model updates. On the other hand, group decision-making involves behavioral interactions and information sharing among multiple users, requiring a balance between coordination and adaptability in interface design. Furthermore, current technologies largely remain at the data acquisition and analysis stage, lacking iterative optimization mechanisms geared towards practical applications. Summary of the Invention

[0005] Purpose of the invention: In view of the problems of poor adaptability, slow response and difficulty in meeting the needs of multi-user collaboration in complex scenarios in the traditional interactive interface design of the above-mentioned existing technologies, the purpose of this invention is to provide an adjustable interactive interface design method and system that integrates group eye-tracking information, which can dynamically adjust the interface layout, functional modules and element design to adapt to changes in different user behavior patterns and preferences, thereby improving user experience and decision-making efficiency.

[0006] Technical solution: To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for designing an adjustable interactive interface that integrates group eye-tracking information, comprising the following steps:

[0008] The eye movement signals of individuals interacting with the interactive interface were collected to assist in the establishment of the group decision-making model, and multiple eye movement feature values ​​were obtained after preprocessing.

[0009] Construct an individual's eye movement feature set {p uj ,u∈{1,2,…,N},j∈{1,2,…,M}}, where M is the number of eye-tracking feature values ​​and N is the number of individuals used to establish the assisted group decision model;

[0010] Construct a set of demographic information {D ul ,u∈{1,2,…,N},l∈{1,2,…,L}}, where L is the number of demographic information categories;

[0011] Obtain the degree of preference of individual u for different interface design elements f, and form a preference set {S}. uf ,u∈

[0012] {1,2,…,N},f∈{1,2,…,H}}; where H is the number of interface design element types;

[0013] Collect individual scores for each eye-tracking feature value j associated with different interface design elements f. And combined with the individual weight value a u Forming a relationship network with adjustable weights {d fj ,f∈{1,2,…,H},j∈{1,2,…,M}};

[0014] By aggregating individual data, a configurable group database with individual weight distribution is formed; the group database includes a set of eye-tracking feature values, a set of preferences, and a set of demographic information.

[0015] Establish a group decision-making model to express the relationship between eye-tracking feature values ​​and interactive interface design elements;

[0016] When a user uses a group decision model, multiple eye-tracking feature values ​​are obtained after the user completes the interactive task, as well as the user's demographic information;

[0017] The recommendation level of each interface design element is calculated based on the user's eye movement characteristics and demographic information, and the attractiveness of the interface design elements to the user is ranked according to the recommendation level.

[0018] Furthermore, the formula for calculating the degree of recommendation is:

[0019]

[0020] Among them, G vf S represents the degree to which user v recommends interface design element f. vf This indicates the degree of preference user v has for interface design element f. This represents the score given by user v for each eye-tracking feature value j associated with different interface design elements f, max(d fj ) represents all d fj The maximum value in, max(S) uf ) represents all S uf The maximum value in; the threshold b is d fj The mean, E(d) fj >b) is an indicator function.

[0021] Furthermore, the weighted network of relationships d fj The calculation formula is: in, d ul Let d represent the l-th population statistical characteristic value of individual u. vl w represents the l-th population statistical feature value of user v. l The importance weight represents the l-th feature.

[0022] Furthermore, the group decision-making model includes a domain ontology model and an inference model. The domain ontology model defines processed eye-tracking feature values, the relationship between an individual's different interface design elements and their associated eye-tracking feature values, different interface design elements, and the individual's preference for interface design elements. The inference model is used to complete the recommendation of interface design elements based on the domain ontology model. Combining the domain ontology model and the user's weighting of demographic information, the recommendation degree of each interface design element is obtained. The interface design elements are then re-ranked according to the magnitude of the recommendation degree to obtain the group decision-making result of the model.

[0023] Furthermore, the method also includes:

[0024] Collect user feedback data on the interface design;

[0025] Determine whether the user's eye-tracking features and feedback data have been included in the group decision model. If not, update the model.

[0026] Furthermore, eye movement feature values ​​are obtained after preprocessing the eye movement signals, specifically including:

[0027] The acquired eye movement signal data is denoised, including calculating the distance between consecutive eye movement points, identifying and removing outliers; smoothing the coordinates of consecutive eye movement points; and filtering using a three-point bilateral convolutional filtering method.

[0028] Wavelet filtering is applied to the denoised eye movement signal data, and the wavelet coefficients after removing noise are reconstructed to restore the denoised eye movement signal data.

[0029] Eye movement features are extracted from multiple interactive tasks, including features in the temporal, spatial, and physiological dimensions of eye movement signals. The temporal features include one or more of the following: fixation count, fixation duration, saccade count, and eye movement speed. The spatial features include one or more of the following: fixation path and fixation region. The physiological features include one or more of the following: pupil diameter, blink count, and fixation entropy.

[0030] Furthermore, regarding eye-tracking signal data

[0031] {(X (1) ,Y (1) ),(X (2) ,Y (2) ),(X (3) ,Y (3) )…(X (n) ,Y (n) The three-point bilateral convolutional filtering method is expressed as follows:

[0032] X (i+1) =α1*X (i) +α2*X (i+1) +α3*X (i+2)

[0033]

[0034] X (1) =β1*X (1) +β2*X (2)

[0035] X (n) =β1*X (n) +β2*X (n-1)

[0036] Where X (i) and Y (i) These are the x and y coordinates of the i-th eye movement point, respectively; n is the total number of eye movement points, and L... i L is the distance between the i-th eye movement point and the (i+1)-th eye movement point. i+1 α1, α2, α3 are the distances between the (i+1)th and (i+2)th eye movement points, and α1, α2, α3 are the convolution kernel coefficients.

[0037] β1 and β2 are weighting coefficients.

[0038] Secondly, the present invention provides an adjustable interactive interface design system that integrates group eye-tracking information, comprising:

[0039] The group decision-making model construction module is used to collect eye-tracking signals of individuals interacting with the interface to assist in the establishment of the group decision-making model. After preprocessing, multiple eye-tracking feature values ​​are obtained, forming a set of individual eye-tracking feature values ​​{p uj ,u∈{1,2,…,N},j∈{1,2,…,M}}, where M is the number of eye-tracking feature types and N is the number of individuals for the assisted group decision-making model; construct a demographic information set {D ul ,u∈

[0040] {1,2,…,N},l∈{1,2,…,L}}, where L is the number of demographic information categories; obtain the degree of preference of individual u for different interface design elements f, forming a preference set {S}. uf ,u∈{1,2,…,N},f∈

[0041] {1,2,…,H}}; where H is the number of interface design elements; collect the scores of individual u for each eye-tracking feature value j associated with different interface design elements f. And combined with the individual weight value a u Forming a relationship network with adjustable weights {d fj f∈{1,2,…,H},j∈{1,2,…,M}}; By aggregating individual data, a group database with configurable individual weight distribution is formed; the group database includes a set of eye-tracking feature values, a set of demographic information, and a set of preferences; a group decision model is established to express the relationship between eye-tracking feature values ​​and interactive interface design elements;

[0042] The human-computer interaction module is used to interact with the user, obtain multiple eye-tracking feature values ​​after the user completes the interaction task, as well as the user's demographic information;

[0043] The recommendation ranking module is used to calculate the recommendation level of each interface design element based on the user's eye movement feature value and demographic information, and to rank the interface design elements according to the recommendation level to the user's attractiveness.

[0044] Furthermore, the system also includes an automatic iterative update module, used to collect user feedback data on the interactive interface design; determine whether the user's eye-tracking feature values ​​and feedback data have been included in the group decision model; if not, update the model.

[0045] Thirdly, the present invention provides a computer program product, including a computer program / instruction, wherein when the computer program / instruction is executed by a processor, it implements the steps of the adjustable interactive interface design method that integrates group eye-tracking information.

[0046] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0047] 1) The interface design method proposed in this invention is not only based on objective user eye-tracking data, but also integrates a model derived from group eye-tracking behavior and preferences. This allows for the acquisition of personalized interface designs for each user based on their eye-tracking characteristic values. Furthermore, the model can be updated in real-time based on user input data, optimizing the accuracy of interface design-related information recommendations.

[0048] 2) In this invention, users can adjust the weights and modify the content of the group eye-tracking model to match a more personalized recommendation method. By adjusting the model's parameters, a more accurate recommendation experience that better matches individual preferences can be provided.

[0049] 3) Based on the model framework of this invention, multi-dimensional eye movement features can be adjusted and integrated in real time according to different interaction scenarios, and dynamically adjusted according to specific interaction scenarios and task requirements, which improves the adaptability of eye movement data. With the support of multi-dimensional eye movement data, the interface design can respond to user behavior more accurately and optimize the response speed and accuracy of the decision support system.

[0050] 4) This invention utilizes eye-tracking data, which accurately reflects users' focus and operational paths on the interface, helping designers identify issues related to the visibility and layout of important information. By analyzing user gaze duration and scanning trajectories, problems such as unclear or complex interfaces can be identified, allowing for optimization of information structure and interaction logic. Simultaneously, eye-tracking data provides objective feedback based on actual user behavior, supporting the scientific basis of design improvements and reducing the subjective biases of traditional methods. Combined with real-time analysis, eye-tracking data can also be used to customize personalized designs for different user groups, significantly improving interface operational efficiency and user satisfaction.

[0051] 5) This invention proposes an interactive interface design method and system that integrates eye-tracking and group decision-making. Users can input eye-tracking data to provide feedback on their individual focus areas and operation paths within the interaction module, while simultaneously inputting preference and evaluation data to assist the group decision-making model in optimizing the interface design. This invention allows users to dynamically adjust the weights of eye-tracking data and group preference data in interface design optimization based on actual needs, thereby achieving a design scheme that combines personalization with group commonality, improving the scientific nature of interface design and the adaptability of user experience.

[0052] 6) This invention enhances the intelligence and adaptability of the interface, and is widely applicable to smart terminals, multi-user collaboration platforms and complex data visualization scenarios, which can promote technological progress in the field of human-computer interaction. Attached Figure Description

[0053] Figure 1 A diagram illustrating the analytical process of designing an interactive interface that integrates eye-tracking and group decision-making.

[0054] Figure 2 A schematic diagram of the domain ontology model designed to assist the interface.

[0055] Figure 3 This is a diagram illustrating an application example used by a user. Detailed Implementation

[0056] The technical solution of the present invention will now be clearly and completely described in conjunction with the accompanying drawings and specific embodiments.

[0057] This invention discloses an adjustable interactive interface design method that integrates group eye-tracking information. By capturing user attention and decision preferences in real time and combining it with a proposed group decision-making model, an interactive interface design method adapted to multi-user scenarios is constructed to improve user decision-making efficiency and interface interaction experience. This interactive interface design method can better match users' eye movements and preferences, automatically proposing interactive interfaces that are more user-friendly and improve interaction efficiency. In a specific embodiment, participants with experience in eye-tracking testing and interface design are first invited to assist in building the model. After the model is built, it can be used to determine the user's preference for a certain design element by importing user eye-tracking data and demographic information with different weights. Furthermore, it can determine whether various types of data of new users are included in the model; if not, the model automatically iterates and updates.

[0058] like Figure 1 As shown in the figure, the interactive interface design method integrating eye movement and group decision-making disclosed in this invention mainly includes the following steps: collecting and determining the degree of individual eye movement data preferences and constructing a demographic information set; taking all preference values ​​to form a group database; establishing a domain ontology model containing basic information of the target decision domain; collecting users' eye movement feature values; calculating the recommendation degree of each interface design element and forming a recommendation ranking; collecting user feedback data on the interface design; and determining whether the existing domain ontology model needs to be updated. The specific steps of the adjustable interactive interface design method integrating group eye movement information according to this invention are described in detail below.

[0059] Step S1: Collect eye movement signals of individuals in the assisted group decision-making model during their interaction with the interactive interface, and obtain multiple eye movement feature values ​​after preprocessing. In this step, for a given individual u, collect multiple eye movement feature values ​​such as the number of fixations, fixation time, and pupil diameter, which represent the objective eye movement feedback during the interaction with the interactive interface.

[0060] Step S2: Set the set of eye-tracking feature values ​​{p} from all interactions of N individuals. uj ,u∈{1,2,…,N},j∈{1,2,…,M}}. Where N is the number of individuals established by the assisted group decision model, and M is the number of eye-tracking feature values.

[0061] Step S3: Construct a demographic information set {D} ul ,u∈{1,2,…,N},l∈{1,2,…,L}; where L is the number of demographic information categories;

[0062] Step S4: Obtain the preference characteristics of individual u (u∈{1,2,…,N}) for different interface design elements f, forming a preference set {S} uf ,u∈{1,2,…,N},f∈{1,2,…,H}}. Where H is the number of interface design element types.

[0063] Step S5: Collect the scores of individual u (u∈{1,2,…,N}) for each eye-tracking feature value j associated with different interface design elements f. And combined with the assigned individual weight value a u A network of relationships was then formed. fj ,f∈{1,2,…,H},j∈{1,2,…,M}}.

[0064] Step S6: By aggregating individual data, a group database with configurable individual u (u∈{1,2,…,N}) weight distribution is formed, including a set of eye-tracking feature values, a set of preferences, and a set of demographic information.

[0065] Step S7: Establish a group decision model to express the relationship between eye-tracking feature values ​​and interactive interface design elements.

[0066] Step S8: When user v uses the group decision model, collect eye movement feature values ​​such as the number of fixations, fixation time, number of saccades, saccade time, pupil diameter, and number of blinks after the user completes an interactive task; at the same time, collect user demographic information such as age, gender, and occupation.

[0067] Step S9: Calculate the recommendation level G for each interface design element based on the user's eye-tracking characteristics and demographic information. vf Re-based on G vfThe numerical values ​​rank the attractiveness of interface design elements to users, and the final evaluation result can be obtained by combining the weights of demographic information that users can adjust themselves.

[0068] Step S10: Collect user feedback data on the interface design, including objective and subjective data. Determine whether the user's eye-tracking feature values ​​and feedback data are already included in the current group decision-making model. If not, update the model.

[0069] Therefore, the adjustable interactive interface design method that integrates group eye-tracking information described in this embodiment of the invention collects user eye-tracking data and group preference data, comprehensively analyzes user visual attention points and group decision-making tendencies, and optimizes interface layout and interaction design. The design optimization results can be presented to the user in real time through the feedback module, supporting dynamic adjustments to achieve personalized and universal interface design, and improve the scientific nature and adaptability of user experience.

[0070] For example, in step S1, for individual u, various signal data are recorded using an eye-tracking device. These signals can reflect the individual's attention and preference for a certain stimulus (such as an image, interface, text, etc.). Common eye-tracking features include: fixation count, fixation duration, pupil diameter, etc. In specific implementation, the eye-tracking signal data collected in the user test design interface scenario undergoes a series of processing steps, such as noise reduction. The eye-tracking data to be denoised is recorded as: {(X1,Y1)(X2,Y2)(X3,Y3)…(X…Y1…Y2 ... n ,Y n )}, and calculate the Euclidean distance {L1,L2…L n-1 The distance between adjacent eye movement points is calculated using the following formula:

[0071]

[0072] Where (X) i ,Y i ) and (X i+1 ,Y i+1 (x) represents the position of two consecutive eye movement points in the coordinate space, where X is the abscissa of the eye movement point and Y is the ordinate of the eye movement point.

[0073] By calculating the distance between consecutive points, noise in eye data can be initially reduced, such as identifying outliers (e.g., sudden jumps due to device errors or user blinking) and processing them (e.g., removal or interpolation). Euclidean distance can also be used to determine whether a user's eye movements are fixation or saccades. A smaller Euclidean distance indicates a smaller range of change between eye movement points, which may indicate fixation; a larger Euclidean distance indicates rapid eye movement, which may indicate saccades.

[0074] The coordinates of continuous points are smoothed, and the moving average method is used to further reduce noise interference. The specific formula is expressed as follows:

[0075]

[0076] Where y i x is the smoothed value of the i-th eye-tracking coordinate. j Let j be the j-th point in the original data, and k be the half-width of the sliding window, which determines the neighborhood range when taking the average. For the i-th point, the average of the k points before it and the k points after it can be calculated, which can quickly process large-scale eye-tracking data.

[0077] Finally, an improved three-point bilateral convolutional filtering method is used to process the eye-tracking data. The filtered eye-tracking data is recorded as {(X (1) ,Y (1) ),(X (2) ,Y (2) ),(X (3) ,Y (3) )…(X (n) ,Y (n) )},

[0078] X (i+1) =α1*X (i) +α2*X (i+1) +α3*X (i+2)

[0079]

[0080] X (1) =β1*X (1) +β2*X (2)

[0081] X (n) =β1*X (n) +β2*X (n-1)

[0082] Where α1, α2, and α3 are the convolution kernel coefficients, with α2 = 0.5; β1 and β2 are the weight coefficients, with β1 = 0.8 and β2 = 0.2.

[0083] Then, wavelet filtering is performed on the denoised eye-tracking data to further extract detailed information from the denoised data, remove residual noise and retain eye-tracking features, and obtain wavelet coefficients composed of detailed noise and eye-tracking features.

[0084] Utilizing the property differences of wavelet coefficients at different scales (noise is mainly concentrated in high-frequency components), the high-frequency detailed coefficients d are analyzed. jA threshold is set, and coefficients smaller than the threshold are considered noise and set to 0. Soft thresholding is used to remove wavelet coefficients from the noise. Compared to hard thresholding, soft thresholding avoids over-smoothing and preserves important details of the eye-tracking signal. The specific method is as follows:

[0085]

[0086] The threshold T can be selected based on the noise level σ, and the specific method is as follows:

[0087]

[0088] Where σ is the standard deviation of the noise and B is the number of data points.

[0089] The wavelet coefficients after noise removal are reconstructed to restore the denoised eye-tracking data.

[0090] The feature value calculation in step S1 can be: extracting eye movement feature values ​​from K tasks. For example, eye movement feature values ​​such as fixation count, fixation duration, saccade count, eye movement velocity, pupil diameter, blink count, fixation entropy, etc., can be calculated based on the temporal, spatial, and physiological dimensions of the eye movement signal.

[0091] The temporal dimension features focus on the temporal variation characteristics of eye movement signals. By analyzing indicators such as the duration, frequency, and speed of fixation and saccades, it reveals the user's attention allocation and information processing characteristics during the task. The spatial dimension features focus on the spatial distribution of eye movement points and are used to analyze information such as the user's fixation path and fixation area. The physiological dimension features extract indicators related to the user's physiological state from the eye movement signals, such as pupil size, fixation entropy, and blinking behavior.

[0092] Calculate each eye movement feature value, taking eye movement velocity and fixation entropy as examples:

[0093] The formula for calculating eye movement speed is as follows:

[0094]

[0095] Where V is the eye movement speed, usually measured in degrees per second; D is the distance the eye moves during one saccade, usually measured in degrees; and t is the time required to complete the saccade, usually measured in seconds.

[0096] The formula for calculating gaze entropy is as follows:

[0097]

[0098] Where n is the number of AOIs partitioned; p i It is the fixation probability of the i-th AOI (the ratio of the number of fixations or the duration of fixation to the total number of fixations or the duration of fixation).

[0099] In step S4, the preference characteristics of individual u (u∈{1,2,…,N}) for interface design elements f (such as color scheme, layout, font size, button style, icon style, etc.) are directly obtained through questionnaires, interviews or user feedback tools, and a preference set is established.

[0100] In step S5, eye-tracking data associated with different interface design elements f for individual u (u∈{1,2,…,N}) are collected to form a relationship network {d fj The expression f∈{1,2,…,H},j∈{1,2,…,M}} further reveals the distribution of users' visual attention and the interdependence between elements, providing strong support for interface design optimization and helping to improve user experience.

[0101] In step S6, the basic aggregation method is used to collect the preferences of individual u for different interface design elements f. In order to more accurately reflect the group preferences, different weights can be assigned to the preference data of different users to reflect the importance or representativeness. Finally, based on the weighted aggregation results, a group database with configurable weight distribution of individual u (u∈{1,2,…,N}) is established.

[0102] In step S7, a domain-specific ontology model is created that can express the relationship between eye-tracking feature values ​​and interactive interface design elements. This model aims to systematically capture the core elements and attributes of interactive interface design, providing a solid foundation for intelligent reasoning. It will include basic elements of interactive interface design, such as color, font, layout, and interactive components, aiming to automate design, optimize interface layout, and meet users' personalized needs.

[0103] The model defines and assigns values ​​to class, relationship, instance, and score. Class represents the processed eye-tracking feature values, with the index denoted by j; relationship represents the relationship between an individual u (u∈{1,2,…,N}) and its associated eye-tracking data, denoted by d, based on weights. fj The symbol represents the interface design element; instance represents different interface design elements, indexed by f; score represents the degree of preference of an individual u (u∈{1,2,…,N}) for the interface design element, denoted by S. uf This information directly contributes to the formation of group decisions.

[0104] In step S8, the collection of eye-tracking feature values ​​is conducted during user participation in the experiment. Users will perform a series of test tasks based on specific interface design elements or product options, assuming they complete K such tasks. Each time a user completes a task, the eye-tracking sensor captures their eye-tracking data. This data is then recorded and stored in the system. Following the process in step S1, the eye-tracking data is preprocessed, and feature values ​​of the eye-tracking signals are calculated for subsequent analysis of the correlation between user preferences for different design elements and the eye-tracking feature values.

[0105] In step S9, the recommendation level of each interface design element is calculated, and a recommendation ranking is formed by combining the user's weighting of demographic information.

[0106] User demographic information is often transformed into a series of quantitative features, such as gender, age, occupation, and educational background. Based on this information, users can adjust the weight allocation of N individuals, for example, increasing the weight of individuals with similar demographic information to their own, thereby improving the accuracy of the recommendation system. In some implementations, individual weights can be determined using the following methods:

[0107] First, define the demographic information set D. ul The demographic information feature vector of each individual u (u∈{1,2,…,N}):

[0108] D u ={d u1 ,d u2 ,...d uL}

[0109] Where L represents a dimension of demographic information (e.g., gender, age, occupation, etc.);

[0110] d ul ,l∈{1,2,…,L} represents the population statistical characteristic value of the l-th individual u.

[0111] Next, define the demographic feature vector of the current user:

[0112] D v ={d v1 ,d v2 ,...d vl}

[0113] Different demographic characteristics may have varying degrees of importance, and users can be allowed to assign weights to each characteristic:

[0114]

[0115] Among them, w lThe importance weight of the l-th feature (defined by the current user).

[0116] Finally, based on similarity S u Adjust the weights of individual u (u∈{1,2,…,N}):

[0117]

[0118] Among them, a u The weights are normalized (ensuring the sum of all weights is 1), representing the degree of influence of individual u (u∈{1,2,…,N}) in group decision-making; N is the total number of people in the group.

[0119] Calculate the weighted d fj The value is calculated using the following formula:

[0120] Construct a group decision-making reasoning model to complete the recommendation of interface design elements based on the above-mentioned domain ontology model. The basic logic is based on the group database p. uj D ul and S uf (u∈{1,2,…,N}), d fj Information from f, combined with Figure 2 The domain ontology model for auxiliary interface design, and the weighting of user demographic information, yield the recommendation level G for each interface design element f. vf Re-based on G vf The numerical values ​​used to rank interface design elements will yield group decision results for a specific interface design element. vf The calculation is expressed as:

[0121]

[0122] Where the threshold b is all d fj The average value, S vf This indicates the degree of preference user v has for interface design element f. This represents the eye-tracking feature values ​​associated with user v for different interface design elements f, max(d fj ) represents all d fj The maximum value in, max(S) uf ) represents all S uf The maximum value in E(d). fj >b) is an indicator function, defined as:

[0123]

[0124] In step S10, after using the interactive interface for a period of time, users' opinions on the interface design and their user experience are collected through tools such as questionnaires and scales, including both objective and subjective data.

[0125] Based on the recorded and processed user eye-tracking data and user feedback, the first step is to check whether the new eye-tracking data is covered by existing domain ontology models. This is done by extracting features from the new data and checking whether its distribution matches the training data. The Kullback-Leibler (KL) Divergence method is then used to measure the difference between the two distributions.

[0126]

[0127] KL divergence is an asymmetric method for measuring the difference between two probability distributions P and Q. If P and Q are completely identical, the KL divergence value is 0. The larger the KL divergence value, the greater the deviation of Q from P. Here, P(j) represents the probability of distribution P at eye-tracking feature value j; Q(j) represents the probability of distribution Q at eye-tracking feature value j.

[0128] If the difference is too large, it may indicate that the user's eye movement features are not covered by the existing domain ontology model. In this case, it is necessary to fine-tune the existing model using new data, that is, to incorporate the new data into the original model; otherwise, no adjustment to the model is required.

[0129] Based on the same inventive concept, embodiments of the present invention also disclose an adjustable interactive interface design system that integrates group eye-tracking information, comprising:

[0130] The group decision-making model construction module is used to collect eye-tracking signals of individuals interacting with the interface to assist in the establishment of the group decision-making model. After preprocessing, multiple eye-tracking feature values ​​are obtained, forming a set of individual eye-tracking feature values ​​{p uj ,u∈{1,2,…,N},j∈{1,2,…,M}}; Construct a demographic information set {D ul ,u∈{1,2,…,N},l∈{1,2,…,L}}; obtain the degree of preference of individual u for different interface design elements f, forming a preference set {S} uf ,u∈{1,2,…,N},f∈{1,2,…,H}}; Collect the scores of individual u for each eye-tracking feature value j associated with different interface design elements f. And combined with the individual weight value a u Forming a relationship network with adjustable weights {d fjf∈{1,2,…,H},j∈{1,2,…,M}}; By aggregating individual data, a group database with configurable individual weight distribution is formed; the group database includes a set of eye-tracking feature values, a set of demographic information, and a set of preferences; a group decision model is established to express the relationship between eye-tracking feature values ​​and interactive interface design elements;

[0131] The human-computer interaction module is used to interact with the user, obtain multiple eye-tracking feature values ​​after the user completes the interaction task, as well as the user's demographic information;

[0132] The recommendation ranking module is used to calculate the recommendation level of each interface design element based on the user's eye movement feature value and demographic information, and to rank the interface design elements according to the recommendation level to the user's attractiveness.

[0133] Furthermore, the system also includes an automatic iterative update module, used to collect user feedback data on the interactive interface design; determine whether the user's eye-tracking feature values ​​and feedback data have been included in the group decision model; if not, update the model.

[0134] This invention also discloses a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the adjustable interactive interface design method that integrates group eye-tracking information.

Claims

1. A method for designing an adjustable interactive interface that integrates group eye-tracking information, characterized in that, Includes the following steps: Eye movement signals of individuals interacting with the interactive interface during the establishment of a assisted group decision-making model were collected and preprocessed to obtain multiple eye movement feature values. The preprocessing included filtering using a three-point bilateral convolutional filtering method. For the eye movement signal data... The three-point bilateral convolutional filtering method is expressed as: ； ； ； ； in and Let x and y be the x and y coordinates of the i-th eye movement point, respectively; n is the total number of eye movement points. Let be the distance between the i-th eye movement point and the (i+1)-th eye movement point. Let be the distance between the (i+1)th eye movement point and the (i+2)th eye movement point. These are the convolution kernel coefficients; These are the weighting coefficients; Construct an individual's eye-tracking feature set { }, where M is the number of eye-tracking feature types and N is the number of individuals used to assist in the establishment of the group decision model; Construct a set of demographic information { , where L is the number of demographic information categories; Obtain individual For different interface design elements The degree of preference forms a preference set { }; where H represents the number of interface design elements; Collect individuals For different interface design elements The score of each associated eye movement feature value j And combined with individual weight values Forming a relationship network whose weights can be adjusted by users. ; By aggregating individual data, a configurable group database with individual weight distribution is formed; the group database includes a set of eye-tracking feature values, a set of preferences, and a set of demographic information. Establish a group decision model that includes a domain ontology model and a reasoning model to express the relationship between eye-tracking feature values ​​and interactive interface design elements; When a user uses a group decision model, multiple eye-tracking feature values ​​are obtained after the user completes the interactive task, as well as the user's demographic information; The recommendation level of each interface design element is calculated based on the user's eye movement characteristics and demographic information, and the attractiveness of the interface design elements to the user is ranked according to the recommendation level. The formula for calculating the degree of recommendation is: ； in, Indicates user For interface design elements Recommendation level Indicates user For interface design elements The degree of preference, Indicates user For different interface design elements The score for each associated eye movement feature value j, Indicates all The maximum value in, Indicates all Maximum value in; threshold for The mean, This is an indicator function.

2. The adjustable interactive interface design method for integrating group eye-tracking information according to claim 1, characterized in that, Weighted network of relationships The calculation formula is: ;in, , , Representing an individual The Individual demographic characteristic values, Indicates user The Individual demographic characteristic values, Representing the The importance weights of each feature.

3. The adjustable interactive interface design method for integrating group eye-tracking information according to claim 1, characterized in that, The ontology model in the domain defines processed eye movement feature values, the relationship between an individual's different interface design elements and their associated eye movement feature values, different interface design elements, and the degree of individual preference for interface design elements. The inference model is used to recommend interface design elements based on the domain ontology model. It combines the domain ontology model with the user's weighting of demographic information to obtain the recommendation degree of each interface design element. The interface design elements are then reordered according to the magnitude of the recommendation degree to obtain the group decision result of the model.

4. The adjustable interactive interface design method for integrating group eye-tracking information according to claim 1, characterized in that, Also includes: Collect user feedback data on the interface design; Determine whether the user's eye-tracking features and feedback data have been included in the group decision model. If not, update the model.

5. The adjustable interactive interface design method for integrating group eye-tracking information according to claim 1, characterized in that, Eye movement feature values ​​are obtained after preprocessing the eye movement signals, specifically including: The acquired eye movement signal data is denoised, including calculating the distance between consecutive eye movement points, identifying and removing outliers; smoothing the coordinates of consecutive eye movement points; and filtering using a three-point bilateral convolutional filtering method. Wavelet filtering is applied to the denoised eye movement signal data, and the wavelet coefficients after removing noise are reconstructed to restore the denoised eye movement signal data. Eye movement features are extracted from multiple interactive tasks, including features in the temporal, spatial, and physiological dimensions of eye movement signals. The temporal features include one or more of the following: fixation count, fixation duration, saccade count, and eye movement speed. The spatial features include one or more of the following: fixation path and fixation region. The physiological features include one or more of the following: pupil diameter, blink count, and fixation entropy.

6. An adjustable interactive interface design system integrating group eye-tracking information, used to implement the adjustable interactive interface design method integrating group eye-tracking information according to any one of claims 1-5, characterized in that, include: The group decision model construction module is used to collect eye movement signals of individuals interacting with the interactive interface to assist in the establishment of the group decision model, and obtain multiple eye movement feature values ​​after preprocessing. Forming an individual's set of eye movement feature values ​​{ }, where M is the number of eye-tracking feature types and N is the number of individuals used to build the assisted group decision-making model; construct a demographic information set { Where L is the number of demographic information categories; obtaining individual For different interface design elements The degree of preference forms a preference set. Where H represents the number of interface design element types; collect individual For different interface design elements The score of each associated eye movement feature value j And combined with individual weight values Forming a relationship network whose weights can be adjusted by users. By aggregating individual data, a configurable group database with individual weight distribution is formed; the group database includes a set of eye-tracking feature values, a set of demographic information, and a set of preferences; a group decision-making model is established to express the relationship between eye-tracking feature values ​​and interactive interface design elements; The human-computer interaction module is used to interact with the user, obtain multiple eye-tracking feature values ​​after the user completes the interaction task, as well as the user's demographic information; The recommendation ranking module is used to calculate the recommendation level of each interface design element based on the user's eye movement feature value and demographic information, and to rank the interface design elements according to the recommendation level to the user's attractiveness.

7. The adjustable interactive interface design system for integrating group eye-tracking information according to claim 6, characterized in that, Also includes: An automatic iterative update module is used to collect user feedback data on the interface design; Determine whether the user's eye-tracking features and feedback data have been included in the group decision model. If not, update the model.

8. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the adjustable interactive interface design method for fusing group eye-tracking information according to any one of claims 1-5.

Images