WO2021147292A1 - Product design scheme decision-making method combining electroencephalogram and eye movement with user similarity - Google Patents

Abstract

A product design scheme decision-making method combining electroencephalogram and eye movement with user similarity. The method comprises three parallel steps: (1) collecting original electroencephalogram (EEG) signals and removing artifacts and noises, extracting a β waveband of EEG about an emotion awakening degree and an α waveband of EEG about an emotional valence, and extracting potential changes of event-related potentials ERPs during decision maker scheme selection; (2) collecting an attention degree obtained on the basis of eye movement behavior data; and (3) obtaining a scheme selection time duration of the decision maker to obtain a scheme preference degree. The three data is combined to obtain implicit decision-making behavior data, and explicit decision-making data is obtained according to the selections of the decision makers; an association degree between the decision-making behavior data of different decision makers is calculated to obtain the user similarity. According to the method, explicit decision-making behaviors and implicit decision-making behaviors are collected, a correlation between the decision makers and schemes is established and a recommendation order is calculated, and an optimal scheme set based on a user similarity relationship is formed.

Description

Product design scheme decision-making method combining EEG and eye movement combined with user similarity Technical field

The invention belongs to the technical field of product design decision-making methods, and specifically relates to a product design scheme decision-making method that combines brain electricity and eye movement with user similarity.

Background technique

In an environment where market competition is intensifying and product life cycles are shortening, companies must concise design and development processes to gain competitive advantage, and quickly discover the optimal solution of design schemes through scientific, standardized, objective and effective design decision-making methods. In the process of product design and decision-making, complexity and uncertainty are an important factor for long decision-making time and low efficiency. Design decision-making is a multi-objective decision-making problem oriented to multi-attribute groups. Multi-attribute emphasizes the gathering and agreement of opinions, and multi-objective emphasizes that the optimization of schemes requires multi-dimensional factors. Under the background that the problem of information overload is becoming more and more obvious, the recommendation system is widely used as an important means of information filtering technology. It searches for the relevant interest domains between different users in the massive data to construct the similarity function between users. This method can also solve the product design decision-making problem based on small sample data and solve the multi-attribute problem of decision-makers in the decision-making process. The decision-making behavior in the product development process is largely based on the overall judgment of the plan. The EEG band frequency and human eye gaze feedback can be used to analyze and evaluate the pros and cons of the product design, and the decision makers can be found based on human cognitive evaluation and perceptual integrity The eye movement data during the decision-making process is based on the EEG data obtained from the different mental states and emotional fluctuations of people facing different solutions in the decision-making process, combined with the aforementioned user similarity to find the common choice of decision-makers, and solve the group-oriented decision-making process Multi-objective decision-making problem in.

Summary of the invention

Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a product design decision-making method that combines EEG and eye movements with user similarity, aiming to use user similarity and physiological measurement data to obtain human " "Explicit decision-making behavior" and "recessive decision-making behavior", the combination of the two provide a new method of design decision-making.

Technical solution: In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is:

A product design decision-making method that combines EEG and eye movement with user similarity. Through similarity calculation, data processing and correlation analysis techniques, two "explicit decision-making behaviors" and "recessive decision-making behaviors" are extracted Behavior elements, the behaviors of the two attributes have different weight coefficients, and the "decision maker-plan" matrix is calculated. The method includes the following steps:

(1) Enter a variety of design plans that meet the target product, establish a target product design sample library, and the system displays the plans covering the full sample in groups in the form of pictures;

(2) The decision-maker clicks on the picture to select each group of options displayed, the system records the single selection time and the total selection time, and the speed of the decision-maker’s choice of the option indicates the decision-maker’s preference for the option P;

(3) Decision makers wear eye trackers and EEG at the same time during the selection process, and record eye movement data and EEG data;

(4) Obtaining explicit decision-making behaviors based on the decision-maker’s choice of options, and at the same time obtaining implicit decision-making behaviors based on their eye movement data, EEG data, and the speed of program selection, and establishing a behavior matrix between decision-makers and options;

(5) Calculate the user similarity between decision makers according to the behavior matrix, and obtain the priority ranking of the schemes. The scheme decision matrix D=user similarity×behavior matrix.

The target product design sample library in the step (1), the total number of samples in the sample library S is n, S _n = {s1, s2,..., sj}, s _j represents the j-th plan; in the process of plan decision-making Recommend schemes covering the entire sample. The schemes are randomly grouped and displayed on the system interface in the form of pictures. The number of scheme samples displayed on the interface each time is r, and r generally takes 3-5.

Must be an integer.

The total number of decision makers U in the step (2) is q, U _q = {u1, u2,..., ui}, u _i represents the i-th decision maker. In each group of options, the decision maker will record the selected option as 1, and the non-selected option as 0, and establish an explicit decision matrix Ex(u _i , s _j ) between people and options. The total number of schemes marked as 1 for each decision maker

In the step (2), _{the time required for a given decision maker u i to} select option j is t _cij , that is, a single group selection time, and the total selection time is T _ci ; The preference function P(u _i ,s _j ):

In the step (3), the eye movement data of the decision maker needs to be collected, saccade and gaze are collected. The method of recording eye movement data is to take the ratio of the fixation duration of a single plan to the total fixation duration of a single plan To measure the degree of attention of the decision maker to the plan, the same is true for the video frequency. Single-plan fixation time t _eij , single-plan fixation time T _eij , single-plan fixation video rate f _eij , and single-plan fixation video rate _Feij . This method only calculates the single-plan fixation time and the largest single-plan fixation rate among a set of plans, that is, extracts the maximum single-plan fixation duration and the largest single-plan fixation rate, so the number of recorded plans is

Establish the eye movement decision matrix E(u _i , s _j ) of the person and the project:

In the step (3), obtaining the EEG data of the decision maker requires artifact removal and digital filtering of the collected original EEG signals. The method of recording EEG data is: collecting the original EEG signals and removing the original EEG signals. , EMG, including artifacts, use digital filtering to remove noise, extract the EEG β-band about emotional arousal and the α-band about emotional valence, and extract the potential changes of event-related potential ERPs when the decision maker chooses the plan. dt refers to the differential of time t as the independent variable, f _α and f _β refer to the frequency values of the β-band of EEG regarding emotional arousal and the α-band of emotional valence, respectively. The EEG decision matrix B(u _i , s _j ):

The explicit decision-making behavior in the step (4) is expressed as an explicit decision-making matrix Ex(u _i , s _j ), and the implicit decision-making behavior Im(u _i , s _j ):

Im(u _i ,s _j )=(E,B,P)

Get the behavior matrix C:

C(u _i ,s _j )=(Ex(u _i ,s _j ), Ex(u _i ,s _j ), Ex(u _i ,s _j ))+Im(u _i ,s _j )

In the calculation of user similarity in the step (5), given the decision maker u ₁ and the decision maker u ₂ , calculate the user similarity Sim according to the behavior matrix C, and let C(u ₁ ) indicate that the decision maker u ₁ has had positive feedback _{Let C(u 2} ) be the set of the first 10 plans for which the decision maker u ₂ has positive feedback, then the user similarity Sim:

Finally, get the plan decision matrix D:

D=Sim×C.

Beneficial effects: The present invention has the following characteristics:

(1) Design decision-making methods based on user similarity, use the results of program screening to simplify the tedious product analysis process, and provide innovative methods and decision-making support for the design and decision-making of enterprises and groups;

(2) In the present invention, the implicit selection obtained by combining the display selection of the decision maker with the physiological measurement data improves the accuracy of decision-making on the basis of cognition;

(3) The present invention reduces the subjective influence of decision-makers, and can accurately explore the design direction with the combination of quantitative and qualitative decision-making mode.

Description of the drawings

Figure 1 is the general flow chart of product decision-making;

Figure 2 is a detailed flow chart of the method;

Figure 3 is the homepage of the product decision-making platform;

Figure 4 is the start interface of product decision-making;

Figure 5 shows the decision data display interface;

Figure 6 shows the output interface of product decision-making results.

Detailed ways

The invention discloses a product design scheme decision-making method that combines brain electricity and eye movement with user similarity, and solves the problem of low reliability of scheme selection results caused by human cognitive ambiguity and subjective influence in the process of design decision-making . The new method for design plan decision-making proposed by the present invention is suitable for the evaluation and testing stage of the design plan, and is to use the plan set in the design sample library for further screening processing. Starting from the principle of screening, this method is suitable for the design of consumer products. The method consists of three parallel steps. The first branch step: collect original EEG signals, remove artifacts such as oculogram and electromyography, use digital filtering to remove noise, and extract EEG β-bands on emotional arousal and emotional valence And extract the potential changes of event-related potential ERPs when the decision maker’s plan is selected; the second branch step collects saccades, gaze and gaze behavior data; the third branch acquires the time of the decision maker’s plan selection Get the degree of preference. Combine the data of these three steps to obtain implicit decision-making behavior data, obtain explicit decision-making data according to the choice of decision makers, and calculate the correlation between decision-making behavior data of different decision makers to obtain user similarity. This method collects explicit decision-making behaviors and implicit decision-making behaviors, establishes the correlation relationship between the decision maker and the plan, calculates the recommendation order of the plan based on this correlation relationship, and forms a set of preferred plans based on the user similarity relationship.

The present invention will be further described below in conjunction with the drawings and embodiments.

A product design decision-making method that combines EEG and eye movement with user similarity. Through similarity calculation, data processing and correlation analysis techniques, two "explicit decision-making behaviors" and "recessive decision-making behaviors" are extracted Behavior elements, the behaviors of the two attributes have different weight coefficients, and the "decision maker-plan" matrix is calculated. The method includes the following steps:

1. After logging in, the operator enters the homepage of the product design plan decision-making system, clicks "New Experiment" to add an experiment name, and adds a design plan that meets the target product to establish a target product design sample library, and the system displays the samples covering all samples in groups Plan, and then set the experimental parameters such as the number of samples displayed in a single group of the interface, the minimum experiment time of a single group, the number of decision makers participating in the experiment, etc., as shown in Figure 3.

2. Under the guidance of the operator, each decision maker checks the "Notes" on the right side of the system homepage, and conducts a pre-experiment to select a set of satisfactory solutions, as shown in Figure 3. After the comprehension, the decision maker wears the eye tracker and EEG at the same time with the assistance of the operator to record the eye movement data and EEG data. The decision maker enters his personal basic information and clicks "start experiment" as shown in Figure 4.

3. The experiment starts at the time, and the decision maker clicks on the picture to select the plan that he thinks meets the decision goal display under the condition that it is greater than or equal to the minimum experimental limit time requirement, and the plan is displayed in groups. The system records the single selection time and the total selection time, and the degree of preference P of the decision maker on the plan is expressed by the speed of the decision maker's choice of the plan. After the last group is displayed, the experiment will automatically end and timekeeping;

4. View the experimental data record as shown in Figure 5. The eye movement data obtained in the experiment corresponds to the EEG data one by one according to time. Obtain explicit decision-making behaviors based on the decision-maker’s selection of options, and at the same time obtain implicit decision-making behaviors based on their eye tracking data, EEG data, and the speed of program selection, and establish a behavior matrix between the decision-maker and the program;

5. Calculate the user similarity between decision makers according to the behavior matrix, and get the priority ranking of the schemes. The scheme decision matrix D = user similarity × behavior matrix. The calculation process is carried out in the background of the system. Figure 6 shows the best decision-making solution group shown in the form of pictures, and the product shown in the leftmost picture is the best decision-making solution for the experiment. The target product design sample library in the step (1), the total number of samples in the sample library S is n, S _n = {s1, s2,..., sj}, where s _j represents the jth plan; in the plan decision process The schemes covering the full sample are recommended in the middle. The schemes are randomly grouped and displayed on the system interface in the form of pictures. The number of scheme samples displayed on the interface each time is r, and r generally takes 3-5.

Must be an integer.

The total number of decision makers U in the step (2) is q, U _q = {u1, u2,..., ui}, u _i represents the i-th decision maker. In each group of options, the decision maker will record the selected option as 1, and the non-selected option as 0, and establish an explicit decision matrix Ex(u _i , s _j ) between people and options. The total number of schemes marked as 1 for each decision maker

In the step (2), _{the time required for a given decision maker u i to} select option j is t _cij , that is, a single group selection time, and the total selection time is T _ci ; The preference function P(u _i ,s _j ):

In the step (3), the eye movement data of the decision maker needs to be collected, saccades and gazes are collected. The method of recording eye movement data is to take the ratio of the gaze duration of a single plan to the total gaze duration of a single set of plans To measure the degree of attention of the decision maker to the plan, the same is true for the video frequency. Single-plan fixation time t _eij , single-scheme total fixation time T _eij , single-scheme fixation video rate f _eij , single-scheme fixation video rate _Feij . This method only calculates the one with the largest single-plan fixation time and single-plan fixation rate in a set of plans, that is, extracts the maximum single-plan fixation duration and the largest single-plan fixation rate, so the number of recorded plans is

Establish the eye movement decision matrix E(u _i , s _j ) of the person and the project:

In the step (3), obtaining the EEG data of the decision maker requires artifact removal and digital filtering of the collected original EEG signals. The method of recording EEG data is: collecting the original EEG signals and removing the original EEG signals. , EMG, including artifacts, use digital filtering to remove noise, extract the EEG β-band about emotional arousal and the α-band about emotional valence, and extract the potential changes of event-related potential ERPs when the decision maker chooses the plan. dt refers to the differential of time t as the independent variable, f _α and f _β refer to the frequency values of the β-band of EEG regarding emotional arousal and the α-band of emotional valence, respectively. The EEG decision matrix B(u _i , s _j ):

The explicit decision-making behavior in the step (4) is expressed as an explicit decision-making matrix Ex(u _i , s _j ), and the implicit decision-making behavior Im(u _i , s _j ):

Im(u _i ,s _j )=(E,B,P)

Get the behavior matrix C:

C(u _i ,s _j )=(Ex(u _i ,s _j ), Ex(u _i ,s _j ), Ex(u _i ,s _j ))+Im(u _i ,s _j )

In the calculation of user similarity in the step (5), given the decision maker u ₁ and the decision maker u ₂ , calculate the user similarity Sim according to the behavior matrix C, let C(u ₁ ) indicate that the decision maker u ₁ has had positive feedback _{Let C(u 2} ) be the set of the first 10 plans for which the decision maker u ₂ has positive feedback, then the user similarity Sim:

Finally, get the plan decision matrix D:

D=Sim×C.

The above are only preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications are also It should be regarded as the protection scope of the present invention.

Claims (6)

1. A design scheme decision-making method that combines EEG and eye movement with user similarity. It is characterized by taking the decision maker’s subjective choice as explicit decision data, and the decision maker’s eye movement data, EEG data and preference degree as implicit decision data. The decision-making data is combined with different decision-making levels to calculate the user similarity to obtain the final decision of the product, including the following steps:

   (1) Establish a sample library of target product design schemes;

   (2) The decision maker selects a sample of the plan, and obtains the explicit decision matrix Ex according to his explicit decision behavior;

   (3) The system records the selection time and total selection time of a single set of samples, and expresses the preference function P in terms of speed;

   The decision maker wears an eye tracker and an EEG device at the same time during the selection process, records eye movement data and EEG data, and obtains the eye movement decision matrix E and the EEG decision matrix B; the implicit behavior matrix Im is: Im = (E , B, P);

   (4) Establish the correlation between the decision maker and the plan sample, that is, the behavior matrix C: C = (Ex, Ex, Ex) + Im;

   (5) Calculate the user similarity Sim between decision makers according to the behavior matrix C;

   (6) The final scheme decision matrix D is the common result of user similarity and behavior matrix: D=Sim×C.
2. The method for designing plan decision-making that combines brain electricity and eye movement combined with user similarity according to claim 1, characterized in that: in step (2), the decision maker chooses each set of plans, and the selected plan is It is recorded as 1, and if it is not selected, it is recorded as 0. An explicit decision matrix Ex(u _i , s _j ) between people and solutions is established, where u _i represents the i-th decision maker and s _j represents the j-th solution.
3. The method for designing decision-making combining EEG and eye movement combined with user similarity according to claim 1, characterized in that: in step (3), the eye movement decision matrix E _ij is:

   

   Single-plan fixation time t _eij , single-scheme total fixation time T _eij , single-scheme fixation video rate f _eij , single-scheme fixation video rate _Feij .
4. The method for designing decision-making combining EEG and eye movement combined with user similarity according to claim 1, characterized in that: in step (3), the EEG decision matrix B _ij is:

   

   The time required for the decision maker u _{i to} select the plan s _{j is t cij} , which is the single group selection time; dt refers to the differential of the time t, and f _α and f _β refer to the β band of EEG regarding emotional arousal and emotion The frequency value of the alpha band of the potency.
5. According to claim 1, a design scheme decision-making method combining EEG and eye movement combined with user similarity is characterized in that: in step (3), the decision-maker's choice of scheme is used to express the decision-maker u _i The preference function P(u _i , s _j ) of the scheme s _j:

   

   Among them, t _cij is the time required for the decision maker u _{i to} select the plan s _j , that is, the single group selection time; T _ci is the total selection time;
6. The method for designing scheme decision-making combining EEG and eye movement combined with user similarity according to claim 1, characterized in that: the user similarity calculation method is: given decision makers u ₁ and decision makers u ₂ , according to The behavior matrix C(u _i , s _j ) calculates the user similarity Sim, let C(u ₁ ) denote the set of the first 10 schemes that the decision-maker u ₁ has had positive feedback, and let C(u ₂ ) be the decision-maker u ₂ has After the first 10 schemes with positive feedback, the user similarity Sim:

   

Images