CN115409535A - A complex product perceptual interaction performance evaluation method based on fusion of multi-source heterogeneous data - Google Patents

Abstract

The invention discloses a complex product perceptual interaction performance evaluation method that integrates multi-source heterogeneous data. The average value establishes an evaluation matrix, calculates the weight of the matrix, and finally establishes a comprehensive evaluation matrix, and judges whether the user evaluation satisfaction degree of each sample in each evaluation dimension is a positive value by comparing with the threshold; the method of the present invention combines both subjective and objective This dimension integrates the three indicators of subjective evaluation, eye movement data and facial expression data. Without affecting the normal operation of the testees, it can accurately and comprehensively obtain the testees' perceptual interactive evaluation of complex products. The design provides a new experimental paradigm and data analysis system, maps out the hidden needs of users, overcomes the uncertainty and ambiguity of traditional methods, and provides an effective reference for product design optimization.

Description

A complex product perceptual interaction performance evaluation method based on fusion of multi-source heterogeneous data

technical field

The invention relates to a complex product perceptual interaction performance evaluation method.

Background technique

Complex products have the characteristics of closed space, dense contacts, and complex function integration, such as the cockpit of airplanes, high-speed rail, submarines, tanks, etc., facing complex dynamic environments, different working scenarios, different operating stages, and the sensibility of users and complex products Interaction is a dynamic evolution process with time series, which is affected by many factors comprehensively, and the interaction mechanism is complex, which makes the interaction process between users and their perceptual interactions particularly complicated. Only relying on single-dimensional data cannot comprehensively and reasonably explain the hidden needs of users .

The patent "A Method for Quantitatively Evaluating Food Consumer Acceptance Using Facial Expression Emotion Recognition and EEG Analysis" (CN113570211A) uses facial expression recognition and EEG analysis methods to evaluate food acceptance, which is currently commonly used for simple products. However, in a complex product environment, the EEG analysis instrument will affect the normal driving operation of the test subjects, resulting in a greatly reduced accuracy of the evaluation results. Therefore, it is necessary to solve the problem of which combination of quantitative tools should be used in a complex product dynamic environment. Perceptual quality to map design requirements is the key to optimizing the appearance design of complex products.

Contents of the invention

Purpose of the invention: The purpose of the invention is to provide a method for comprehensively and accurately evaluating the perceptual interaction between users and complex products by integrating subjective and objective dimensions.

Technical solution: The complex product perceptual interaction performance evaluation method for integrating multi-source heterogeneous data according to the present invention includes the following steps:

(1) Establish a complex product form sample database, and obtain the subjective evaluation data, eye movement experimental data and facial expression experimental data of the subjects in k evaluation dimensions;

(2) Calculate the mean value of subjective evaluation data, mean value of eye movement experiment data and mean value of facial expression experiment data of subjects in each evaluation dimension, and establish subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix;

(3) calculate the weight of described subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix;

(4) Establish a comprehensive evaluation matrix according to the subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix and their weights, and judge whether the user evaluation satisfaction of the product in this evaluation dimension is greater than the threshold according to whether the comprehensive evaluation matrix value is greater than the threshold is a positive value.

Further, the subjective evaluation data in step (1) is Likert scale score; eye movement data is gaze duration and dwell time; facial expression data is emotional efficacy value.

Further, the eye movement evaluation matrix in step (2) is the average value of the reciprocal of the mean reciprocal of the eye movement fixation degree of the subjects in each evaluation dimension, and the eye movement fixation degree is the ratio of the fixation time to the dwell time.

Further, before calculating the weights in step (3), normalization processing is performed on the subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix.

Further, the method for calculating the weight in step (3) is to calculate the reciprocal of the multiple correlation coefficients of the subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix respectively, and then perform normalization processing respectively.

Further, in step (4), if the value of the comprehensive evaluation matrix is greater than the threshold, the user evaluation satisfaction of the product in this evaluation dimension is a positive value; if the value of the comprehensive evaluation matrix is equal to the threshold, the product is in this evaluation dimension. The user evaluation satisfaction of the evaluation dimension is a neutral value; if the value of the comprehensive evaluation matrix is less than the threshold, the user evaluation satisfaction of the product in this evaluation dimension is a negative value.

Further, the complex product form sample library is a complex product interior form picture.

Further, the threshold in step (4) is selected from the normalized result of Likert scale scores.

Beneficial effects: the advantages of the present invention compared with the prior art are: (1) for the characteristics of complex products, the subjective evaluation method and the physiological measuring instrument adopted in the present invention are neither disturbing nor intrusive, and avoid causing harm to the user's operation. Unnecessary interference does not affect the real feedback of the perceptual interaction between the user and the product; (2) the multi-source heterogeneous data index selected by the present invention can fully reflect the comprehensive performance of the perceptual interaction experience between the user and the complex product, and fully exert its own characteristics At the same time, they complement each other, are more comprehensive and accurate, provide a new experimental paradigm and data analysis system for complex product design, map out the hidden needs of users, overcome the uncertainty and ambiguity of traditional methods, and provide Product design optimization provides an effective reference; (3) The comprehensive performance evaluation method of perceptual interaction based on the fusion of subjective evaluation, eye movement, and facial expression data can calculate a specific value, the interval is between 0-1, and the Likert quantity is selected The median value in the normalized results of the table is used as the threshold, which can quickly and intuitively judge the perceptual interaction performance satisfaction of this evaluation dimension. For complex products with cumbersome interactive operations, users need to implement strict operating regulations, and also need to respond to emergencies Make a scientific and reasonable emergency response, so users of complex products need to maintain rational thinking as much as possible, while avoiding fatigue and misuse. The neutral threshold is more suitable as the evaluation threshold for perceptual interaction of complex products.

Description of drawings

Fig. 1 is a flow chart of the method of the present invention.

Detailed ways

The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

As shown in Figure 1, the complex product perceptual interaction performance evaluation method includes the following steps:

(1) Input the experimental data of subjective evaluation, eye movement and facial expression, and establish a data sample database.

Let n represent the number of subjects, m represent the number of product samples, and k represent the number of evaluation dimensions. Among them, n, m, and k all take positive integers. The subjective evaluation data is the Likert scale score, the eye movement data is the fixation time and dwell time, and the facial expression data is the emotional efficacy value.

(2) Construct the evaluation matrix of each modal data

(2-1) Establish subjective evaluation matrix

According to the measured values of the subjective evaluation data of n subjects in the k-th evaluation dimension, the subjective evaluation data element set matrix Q _k is established:

表示第n名被试者对第m个产品样本在第k个评价维度的李克特量表得分。本实施例中选择5级李克特量表，分值设为(-2,-1,0,1,2)，更能表征用户评价的负向值、中性值和正向值。in,

Indicates the Likert scale score of the nth subject on the mth product sample in the kth evaluation dimension. In this embodiment, a 5-level Likert scale is selected, and the scores are set to (-2, -1, 0, 1, 2), which can better represent the negative value, neutral value and positive value of user evaluation.

作为主观评价评分，得到以主观评价数据解释的单一产品样本在第k个评价维度的评价整体趋势：Calculate the mean value of the Likert scale scores of n subjects on the kth evaluation dimension for each product sample

As a subjective evaluation score, the overall trend of the evaluation of a single product sample in the k-th evaluation dimension explained by the subjective evaluation data is obtained:

Establish a matrix based on the subjective evaluation scores of m product samples in the k-th evaluation dimension, and obtain the subjective evaluation matrix S _k :

(2-2) Establish eye movement evaluation matrix

According to the measured eye movement data of n subjects in the k-th evaluation dimension, the eye movement data element set matrix T _k is established:

分别表示第n名被试者对第m个产品样本在第k个评价维度的注视时长与停留时长。计算其比值得到眼动注视程度

in,

Respectively represent the fixation time and dwell time of the nth subject on the mth product sample in the kth evaluation dimension. Calculate the ratio to get the degree of eye movement fixation

以作为眼动评分，得到以眼动数据解释的单一产品样本在第k个评价维度的评价整体趋势：For the negative proportional relationship between the eye movement score and the emotional value, the reciprocal of the degree of eye movement fixation is used for calculation, and the mean value of the eye movement fixation degree of each product sample in the k-th evaluation dimension of n subjects is obtained

Taking it as the eye movement score, the overall trend of the evaluation of a single product sample in the k-th evaluation dimension explained by eye movement data is obtained:

Establish a matrix based on the average eye movement scores of m product samples in the k-th evaluation dimension, and obtain the eye movement evaluation matrix E _k :

(2-3) Establish facial expression evaluation matrix

According to the measured values of the facial expression evaluation data of n subjects in the k-th evaluation dimension, the facial expression evaluation data element set matrix C _k is established:

表示第n名被试者对第m个产品样本在第k个评价维度的面部表情效价值。in,

Indicates the facial expression efficacy value of the nth subject on the mth product sample in the kth evaluation dimension.

作为面部表情评分，得到以面部表情数据解释的单一产品样本在第k个评价维度的评价整体趋势：Calculate the mean value of the facial expression effect value of n subjects for each product sample in the kth evaluation dimension

As the facial expression score, the overall trend of the evaluation of a single product sample in the k-th evaluation dimension explained by the facial expression data is obtained:

Establish a matrix based on the average facial expression scores of m product samples in the k-th evaluation dimension, and obtain the facial expression evaluation matrix F _k :

(3) Matrix data normalization processing

Perform linear transformation on the data in each evaluation matrix, so that the evaluation result value is mapped to [0,1].

Substitute S _k , E _k , F _k into the above formula to obtain S _k ^* , E _k ^* , F _k ^* .

(4) Calculate the weight of each evaluation matrix

Calculate the multiple correlation coefficient of each evaluation matrix, the formula is:

为参数均值，

为待估参数估计值。对R_ik取倒数并进行归一化处理得到单一模态各指标权重w_ik：in

is the parameter mean,

is the estimated value of the parameter to be estimated. Take the reciprocal of R _ik and perform normalization processing to obtain the weight w _ik of each index in a single mode:

Substitute the data in the matrices S _k ^* , E _k ^* , F _k ^* into the above formula to get w _sk , w _ek , w _fk .

(5) Construct a comprehensive evaluation matrix and output evaluation values

Synthesize the normalized subjective evaluation matrix S _k ^* , eye movement evaluation matrix E _k ^* , facial expression evaluation matrix F _k ^* , and the weights w _sk , w _ek , w _fk corresponding to the matrices in turn to construct a single evaluation dimension product Comprehensive evaluation matrix Z _k :

Then the product comprehensive evaluation matrix Z is:

Z=[Z ₁ ,Z ₂ ,…,Z _k ]

Z is an m×k matrix, which represents the evaluation value of each product sample in k evaluation dimensions during the product evaluation process.

(6) Analysis and evaluation results

After normalizing the scores of the 5-level Likert scale, in this embodiment, the median value of 0.5 indicates that the evaluation result is a neutral value, greater than 0.5 indicates that the evaluation result is a positive value, and less than 0.5 indicates that the evaluation result is negative Because the matrix of subjective evaluation, eye movement evaluation and facial expression evaluation has been normalized, the normalized result of Likert scale can be used as the threshold to represent the subjective evaluation, eye movement evaluation and facial expression evaluation. satisfaction. In actual use, other thresholds can be selected according to the level of the Likert scale or the type of product.

For different products, users have different expectations for perceptual interaction, and the selection of thresholds for judging perceptual interaction performance is also different. For example, in consumer products, such as household items, small appliances, 3C products, etc., product homogeneity is common, so users’ emotional interaction with products is expected to be a very typical positive emotion, hoping that the product can give people a pleasant experience , so the selection of the threshold will be more biased towards positive sentiment according to the specific product. As for complex products, because of their cumbersome and complex interactive operations, strict implementation of operating regulations is required, and scientific and reasonable emergency response to emergencies is also required. Then, users of complex products need to maintain rational thinking as much as possible, while avoiding fatigue and misuse. The threshold of perceptual interaction should be neutral.

When the evaluation value is greater than 0.5, it means that the user evaluation satisfaction of the product in this dimension is a positive value, and the closer the value is to 1, the higher the satisfaction;

When the evaluation value is equal to 0.5, it means that the user evaluation satisfaction of the product in this evaluation dimension is a neutral value;

When the evaluation value is less than 0.5, it means that the user evaluation satisfaction of the product in this evaluation dimension is a negative value, and the closer the value is to 0, the lower the satisfaction is.

The method of the present invention is verified by specific experimental data below.

Taking the interior form of the cockpit of a business jet as an example, the perceptual interactive performance evaluation of multi-source heterogeneous data is carried out.

(1) After screening and image processing, determine 10 business jet cockpit interior shape pictures as a complex product sample library, that is, m=10; invite 29 male pilots as subjects, that is, n=29; defined in this embodiment The three perceptual interaction subjective evaluation dimensions are the figurative linear features of the cockpit interior, the functional effect corresponding to the cockpit shape, and the emotional experience in the perceptual interaction between the pilot and the cockpit interior shape, corresponding to k=1, k=2, The case of k=3.

(2) Record the pilot's subjective evaluation, eye movement and facial expression experimental data in each perceptual interaction dimension, and establish a data sample database. The subjective evaluation data is the 5-level Likert scale score, the eye movement data is the fixation time and dwell time, and the facial expression data is the emotional efficacy value.

(3) Construction of each modal data evaluation matrix

(3-1) Establish subjective evaluation matrix

作为主观评价评分，再由

建立主观评价矩阵S_k：According to the measured value of the subjective evaluation data of 29 subjects in the k-th evaluation dimension, the subjective evaluation data element set matrix Q _k is established, and the 29 subjects’ evaluation of each product sample in the k-th evaluation dimension is calculated by Q _k The mean Likert scale score for

As a subjective evaluation score, and then by

Establish subjective evaluation matrix S _k :

(3-2) Establish eye movement evaluation matrix

再由

建立眼动评价矩阵E_k：According to the measured value of the eye movement data of 29 subjects in the k-th evaluation dimension, the eye-movement data element set matrix T _k is established, and the 29 subjects’ evaluation of each product sample in the k-th evaluation dimension is obtained from T _k mean eye fixation

Then by

Establish eye movement evaluation matrix E _k :

(3-3) Establish facial expression evaluation matrix

作为面部表情评分，由

建立面部表情评价矩阵F_k：According to the measured value of the facial expression evaluation data of 29 subjects in the k-th evaluation dimension, the facial expression evaluation data element set matrix C _k is established, and the evaluation of each product sample by the 29 subjects in the k-th is calculated by C _k Dimension facial expression efficacy value mean

as a facial expression score, given by

Establish facial expression evaluation matrix F _k :

(1) Normalize the data in each evaluation matrix, so that the evaluation result value is mapped to [0,1]. S _k ^* , E _k ^* , F _k ^* are obtained.

(2) Calculate the weights w _sk , w _ek , and w _fk of each matrix.

w _s1 =36.04% w _e1 =36.63 w _f1 =27.33

w _s2 = 41.11% w _e2 = 28.32 w _f2 = 30.57

w _s3 =27.25% w _e3 =30.92 w _f3 =41.83

(3) Construct a comprehensive evaluation matrix Z _k of business jet samples with a single evaluation dimension:

Then the perceptual interaction comprehensive performance evaluation matrix Z of the sample library is:

When the evaluation value is greater than 0.5, it means that the user evaluation satisfaction of the business jet in this evaluation dimension is a positive value, and the closer the value is to 1, the higher the satisfaction is.

When the evaluation value is equal to 0.5, it means that the user evaluation satisfaction of the business jet in this evaluation dimension is a neutral value;

When the evaluation value is less than 0.5, it means that the user evaluation satisfaction of the business jet in this evaluation dimension is a negative value. And the closer the value is to 0, the lower the satisfaction.

For example, the user evaluation satisfaction of sample 1 in the two evaluation dimensions of k=2 and k=3 is greater than 0.5, while the satisfaction of k=1 is low, so the subsequent optimization design can be carried out for the evaluation dimension of k=1; 5 The user satisfaction in the three evaluation dimensions is less than 0.5, and the design of the three evaluation dimensions needs to be further optimized; the user satisfaction of sample 8 in the k=2 evaluation dimension is greater than 0.5, and the k=1, k=3 satisfaction Both are low, and the satisfaction degree of k=3 is far less than 0.5, so the subsequent optimization design can be carried out for the two evaluation dimensions of k=1 and k=3, and focus on optimizing the latter.

Claims (10)

1. A complex product perceptual interaction performance evaluation method that integrates multi-source heterogeneous data, characterized in that it comprises the following steps: (1) Establish a complex product form sample database, and obtain the subjective evaluation data, eye movement experimental data and facial expression experimental data of the subjects in k evaluation dimensions; (2) Calculate the mean value of subjective evaluation data, mean value of eye movement experiment data and mean value of facial expression experiment data of subjects in each evaluation dimension, and establish subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix; (3) calculate the weight of described subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix; (4) Establish a comprehensive evaluation matrix according to the subjective evaluation matrix, eye movement evaluation matrix and facial expression evaluation matrix and their weights, and judge whether the user evaluation satisfaction of the product in this evaluation dimension is greater than the threshold according to whether the comprehensive evaluation matrix value is greater than the threshold is a positive value. 2. The complex product perceptual interaction performance evaluation method fused with multi-source heterogeneous data according to claim 1, characterized in that the subjective evaluation data in step (1) is Likert scale scores. 3. The complex product perceptual interaction performance evaluation method fused with multi-source heterogeneous data according to claim 1, characterized in that the eye movement data in step (1) are gaze duration and dwell duration. 4. The complex product perceptual interaction performance evaluation method for merging multi-source heterogeneous data according to claim 1, characterized in that the facial expression data in step (1) is emotional efficacy value. 5. The complex product perceptual interaction performance evaluation method for fusion of multi-source heterogeneous data according to claim 1, characterized in that the eye movement evaluation matrix in step (2) is the eye movement of the subject in each evaluation dimension The average value of the reciprocal of the mean fixation degree, the eye movement fixation degree is the ratio of the fixation duration to the dwell duration. 6. The complex product perceptual interaction performance evaluation method for fusion of multi-source heterogeneous data according to claim 1, characterized in that before step (3) calculates the weight, the subjective evaluation matrix, eye movement evaluation matrix and The facial expression evaluation matrix is normalized. 7. The complex product perceptual interaction performance evaluation method for fusion of multi-source heterogeneous data according to claim 1, characterized in that the method for calculating the weight in step (3) is to calculate the subjective evaluation matrix and eye movement evaluation respectively. matrix and the reciprocal of the complex correlation coefficient of the facial expression evaluation matrix, and then normalized respectively. 8. The complex product perceptual interaction performance evaluation method for merging multi-source heterogeneous data according to claim 1, characterized in that, in step (4), if the value of the comprehensive evaluation matrix is greater than the threshold, the product is in the evaluation dimension The user evaluation satisfaction of the product is a positive value; if the value of the comprehensive evaluation matrix is equal to the threshold, the user evaluation satisfaction of the product in this evaluation dimension is a neutral value; if the value of the comprehensive evaluation matrix is less than the threshold, the product The user evaluation satisfaction in this evaluation dimension is a negative value. 9. The complex product perceptual interaction performance evaluation method that integrates multi-source heterogeneous data according to claim 1, wherein the complex product form sample library is a complex product interior form picture. 10. The complex product perceptual interaction performance evaluation method for fusion of multi-source heterogeneous data according to claim 2, characterized in that the threshold value in step (4) is selected from the normalized results of Likert scale scores .

Images