Research on data visualisation strategies for online information dissemination based on user experience
Published Online: Feb 03, 2025
Received: Sep 23, 2024
Accepted: Dec 31, 2024
DOI: https://doi.org/10.2478/amns-2025-0031
Keywords
© 2025 Xiong Wei, published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
In today’s information age, huge amounts of data are constantly being generated and accumulated. However, simple data cannot directly convey information, and data needs to be transformed into visual forms to better understand and disseminate information. As a powerful tool, data visualisation can present data through charts, graphs and animations to help people understand and disseminate information more intuitively [1–4].
First of all, data visualisation makes complex information easier to understand and digest by presenting data intuitively and graphically. Compared with lengthy text descriptions or cumbersome tables, charts and graphs can convey the core content of data more quickly and directly [5–7]. For example, line graphs and bar charts can clearly show the relationship and trend between data, and pie charts can intuitively show the proportion of each part in the whole. This kind of visualisation helps people to obtain key information more quickly and improves the efficiency of information transfer [8–9].
Secondly, data visualisation helps to refine and summarise the key features and insights of data. By presenting data in a visualised form, we can more easily identify patterns, trends and outliers in the data. For example, scatter plots can help us find correlations between two variables, and box-and-line plots can show the distribution of data and outliers [10–13]. This kind of visualisation allows us to understand the data more deeply and get more profound information from it. In addition, data visualisation can help people to better compare and contrast different datasets, and improve the impact and memorability of information [14–16].
Literature [17] discusses the application of social network analysis and data visualisation techniques in the analysis of information dissemination based on the use of social network analysis methods and the visualisation of huge amounts of data acquired on the Facebook web platform. Literature [18] proposes Uxmood, a tool that provides quantitative and qualitative information with the aim of assisting practitioners in evaluating the feasibility of user experience, which is able to do so by applying sentiment analysis techniques to, for example, video and audio in order to understand the user’s experience. Users were studied and used to assess the efficiency of visual data communication. Literature [19] conducted a visual analysis of international research collaboration patterns using social network analysis to reveal the main characteristics of faculty members in Taiwan’s information management system in terms of international collaboration trends and patterns of research productivity over three decades. Literature [20] deals with the development and implementation of a theoretically supported model for the creation of an information system that enables the dissemination and visualisation of scientific and technological information. The proposed objectives are achieved through social media networking tools and can be effectively controlled in order to continuously improve the system. Literature [21] provides a systematic survey and analysis of information visualisation. It is noted that research on information visualisation has been classified into a taxonomy of major categories such as empirical methods, visualisation frameworks. Future directions are proposed to address the current challenges in the technical field. Literature [22] compares just the landscape of the two fields of information visualisation and data visualisation and identifies the trends in both. The study shows that although information visualisation and data visualisation are developing together, both fields have their developmental characteristics.
In this paper, in order to optimize the data visualization method of network information dissemination, colour symbols and cognitive load experiments were carried out based on the double coding theory to explore the interrelationship between the cognitive effect of visualization on the audience and the level of memory, so as to provide a basis for the evaluation of the visualization effect of network information dissemination. In the colour symbols experiment, the subjects were randomly divided into 21 control groups, of which the effective number was 20, and the final scores were counted using stratified sampling method, and the data were collated and analysed. Cognitive load experiments investigated the task efficiency performance of pie charts, line graphs and bar charts under six types of data analysis tasks, two types of information density, and the efficiency performance of three visualisation charts under three levels of cognitive load for memory reproduction tasks, respectively.
People are inundated with a large amount of information and a wide variety of digital media, which makes audiences less able to receive, select, and process information, which creates information fatigue. At the same time, when the audience expects to share concepts and information, clear visual communication becomes very important. The focus of information design is now on how information can be communicated and understood by users easily. As a result, the multidisciplinary design field of “information visualisation” has emerged. From the perspective of visualisation, “visualisation” is the creation of a mental model or mental image of something. Therefore, visualisation is based on people’s cognitive activities and the process of re-comprehending information, and it emphasises the transformation of complex, invisible things into visible ones [23].
This paper focuses on information visualisation that represents abstract or complex concepts and subtle connections between information and interaction techniques. Data visualisation for web-based information dissemination is an interdisciplinary design field that aims to study the visual presentation of large-scale non-numerical information resources, as well as the use of techniques and methods in graphic imaging to help people understand and analyse data. As one of the closely related branches of information design, information visualisation embodies the main characteristics of information design but also possesses its own unique and advanced way of expression, and its main role is to help users extract valuable information and relationships between them in abstract data and events.
The increase in Internet technology and usage has allowed people to share and exchange information more often, and at the same time, people are not only the recipients of information but also the creators of information, so there must exist a way of information dissemination that can cross the boundaries of language and culture. The creation of data visualisation provides people with a more interesting way of obtaining information, and when people use it as a means of communication, it not only changes the meaning of the language of information dissemination and design but also broadens the possibilities of traditional charts and annotations, such as pie charts, line graphs, and bar charts.
Visual translation refers to the transformation of abstract information into readable visual language symbols using visual descriptions (often called symbolic coding). Visual transformation not only relies on inspiration to dictate creativity, but also a creative process with logical thinking and the ability to deduce, which takes over preliminary information collection and information architecture.
In the social communication process of human beings, information is the unity of symbols and meanings. Without visual symbols, information cannot be attached, making it difficult to convey. Therefore, on the basis of realising data and information collection and information architecture, it is necessary to visually transform the information content, i.e., to realise the process of visualisation. The information will be presented to the audience in a clear, intuitive visual presentation, so that they can understand the complex, multidimensional data information, and the intrinsic relationship between the data and information and the audience itself.
Appropriate visual transformation and expression of information, i.e., information visualisation, is one of the effective ways to make information conveyance and total realisation. The information must be attached to the appropriate carrier in order to effectively communicate, so it is necessary to understand the character and characteristics of the elements of visual construction in information visualisation, and effectively select, arrange and organise the visual elements in information design, so as to form an effective visual carrier of information. In this process, the designer is more likely to link the apparent appeal of the visual language with the deep symbolic context and semantics and, at the same time, pay attention to the relationship between objective and subjective information presentation. The expression of information visualisation design is rich and varied from plane to three-dimensional, from static to dynamic, etc., but the visual elements of constructing information mainly contain the symbol language that makes the information more clear and concise, the graphic language that makes the information more charming, the colour language that makes the information more symbolic and appealing, and the process and annotation that makes the information reading more fluent and the semantics more prominent. And the symbol is to carry and convey information. It is a representation of the characteristics of things, a simplified means of understanding things, and the main source of thought. People achieve mutual understanding and communication through the symbol system. Symbols are carriers of information, and they refer to known things related to the characteristics of the object they refer to and the information. There is a relationship between representation and being represented, and understanding and being understood. Symbols can represent both conceptual and physical things, and this representational ability of symbols is jointly given and recognized by the designer and the audience of the encoded information. Interpretation of symbols in information visualisation is also the process of digging out the meaning of the information and the interaction between the audience and the symbols. When the audience interprets the symbols with their own culture and experience, it also includes a personalized understanding of the symbols. The context of any symbol can only be understood within a certain range, and the audience can only accept symbols that conform to a specific context.
The visual translation of the direct message is the understanding and communication of the explicit message. For the designer, it is especially necessary to be in the direction of the message to be conveyed within the constraints, to guide his intuition through his subconscious, and to analyse a large amount of comparative data in order to obtain the most direct and clear way of conveying the message.
In literary writing, we can often see the author’s metaphorical representation of the concepts to be elaborated, but in the text, we can analyse the concepts to be elaborated through certain keywords or contextual semantics. In information visualisation design, this visual transformation (metaphorical representation) of non-direct information design is more about hiding part of the information in the visual language in terms of historical aspects or compositional elements, but this hiding is not just throwing away the information, as in the case of a tree, we can see the various postures of a tree on the ground, but we can also explore down through the trunk to find the roots and the growth veins.
Graphic symbols account for the largest proportion of visual representation symbols in information visualisation, and are both important visual symbols for information visualisation communication design and important carriers and media for information communication. ➀ Graphic symbols include body graphics, simple symbols, charts and words. Main body graphic symbols Graphic symbols are descriptions or portraits of objective objects with similarity and vividness [24]. Graphic understanding is that regardless of language and nationality, it is a universal language that transcends the language barrier between countries and nations, and it is a form of guided language that can directly communicate with the viewer’s thoughts. In information visualisation, “graph” includes both pictures and graphics. The image material obtained through photographic means is a picture, generally involving real people’s information view will be used in the form of a picture, the picture characterises the character itself, and graphical symbols are used most often, so this paper focuses on graphical symbols. Through a variety of graphics software, we can obtain graphic symbols that represent the visual representation of the difference between text and language. Graphics can be reproduced in large quantities through a variety of means and is the main visual representation symbols in the design of information visualisation. The main graphic symbols are closely related to the theme of information visualization, and it can be said that they represent the mapping of the information theme. The main graphic symbols are divided into figurative graphics, abstract graphics, and semi-abstract graphic symbols. Subsidiary icon symbols Subsidiary icon symbols refer to the picture in order to avoid clutter, with a unified image applied to the view, as far as possible to replace the unnecessary appearance of the text so that the information view is easier to understand and communicate. Subsidiary icon symbols mainly include a variety of different forms of pointing symbols, connecting symbols and icons, in the design of the main graphic subject to the style and form, indicating symbols with a sense of flow and direction, such as arrows, with the flow of the line and so on. Accessory icon symbols in information visualisation not only meet functional needs but also have a certain effect on visual psychology.
People’s awareness of colour is a general reaction of sensation, perception or abstract thinking. Colour is an intermediary for expressing thoughts and conveying emotions. Different colours represent different attributes and visual representations, which will directly affect the audience’s ability to store information from vision to sensation to thinking, so colour is also called a symbol. “Colour is dependent on form”. Colors and graphic symbols are combined with visual representations to ensure the harmony of the picture and create a good visual experience for disseminating information to the audience [25].
“Form” is used to refer to the inner structure or law of something. In this paper, “formal symbol” refers to a symbolic way of arranging different visual symbols according to the audience’s visual experience, which provides a reasonable and operable basis for the design of a specific visual representation of information visualisation. The arrangement and combination law of formal symbols is the common visual psychology formed by the audience based on the accumulation of natural order, which has abstractness, stability, and inheritance. For example, the “dispersive” arrangement form described below has the common visual psychology of “cohesion” and “concentration”, which is derived from the phenomenon of dispersal of the sun’s rays in the natural world. Phenomenon.
Dual Coding Theory (DCT): DCT suggests that the human cognitive system consists of two separate but interconnected systems for processing and storing information: verbal coding, which is responsible for processing verbal information, and non-verbal coding, which is responsible for processing non-verbal objects and events (especially imaginative information.) DCT assumes that all of a person’s personal experiences are capable of being mentally reproduced in the human brain and that these experiences can be expressed verbally, forming verbal knowledge; other experiences are non-verbal, forming knowledge about the world, which requires the use of imagination. Experiences that can be expressed verbally form verbal knowledge, which constitutes the overall meaning and structure of language and is the responsibility of verbal coding, while other experiences that are non-verbal form knowledge about the world, which needs to be represented and processed by the imagination and is the responsibility of non-verbal coding. So the five human senses thus establish an orthogonal relationship with the dual coding system, i.e., verbal coding forms a one-to-one correspondence with the pronunciation, shape, meaning, and perception of words, and non-verbal coding forms a correspondence with visual, auditory, tactile, gustatory, and olfactory information in addition to words [26].
The representational units of verbal and non-verbal coding are different, with verbal coding corresponding to lexical elements and non-verbal coding corresponding to image elements. A lexical element is the coded representational unit of any information perceived in the form of language and consists of various verbal codes of varying sizes. A lexical element is a unit of information acquired in a holistic form. For example, “it rains cat and dog” and “cat and dog” a lexical elements. Lexical elements operate sequentially and hierarchically so that when we write silently, if we are given the previous word, the later word can be quickly associated, and if we are given the later word, it takes longer to write the previous word. Elements are any coded representational units of things, events, and situations perceived in a non-verbal form. They vary in size, and different elements are related to each other mainly through combinatorial decomposition. For example, a school consists of a playground, a school building, an administrative building, a cafeteria, etc., and a student’s face consists of elements such as eyes, nose, ears, face, and mouth. One way in which the elephant elements operate with each other is clustering, wholeness and continuity. For example, a song can recall the whole song as soon as the opening accompaniment is reminded. So once a certain pictogram is activated, the overall scene is presented in a clustered panorama. Another way hieroglyphs operate is through similar hooks.
According to the double coding theory above, in order to verify whether the cognition and memory of information elements in data visualisation charts are affected by non-verbal units such as symbolic markers and visual enhancement, 2 information communication data visualisation experiments are conducted to explore the effects of colour symbols and cognitive load on the user experience of information communication, respectively.
The experimental materials were designed uniformly, and the chart information included the chart title and overall theme or morphology, and the charts were not ambiguous due to missing data. The significance (awareness, recall) of the charts in terms of symbolic marking and visual enhancement (colour scheme, dynamic effect, chart type) was quantified using colour scheme, symbolic marking, dynamic effect, and chart type as the four indicators and supplemented by post-experimental interviews on the impact of the different variables on the affective level. The data obtained from the experiments can represent whether the research participants are internalising the presented information or graphic, which can be used to infer which infographics with which attributes have a better communication effect in the audience group.
Accordingly, the following experimental hypothesis is proposed:
Hypothesis H1: Contrasting colour matching charts will enhance cognitive and memory effects compared to approximate colour matching charts.
Hypothesis H2: Irrelevant labelling reduces the cognitive and memory effects of infographics in the audience, and conversely relevant labelling enhances the cognitive and memory effects.
An experimental design with randomised block groups was conducted. In this experiment, the subjects were randomly divided into 21 control groups, of which the effective number was 20, and the final scores were counted using stratified sampling, and the data were organised and analysed.
For marker types, each row is varied into three different marker types based on the same chart. These are the no marker type, the relevant marker type, and the irrelevant marker type. Finally, the simplest chart is defined as the no-marker type chart.
In terms of colour matching, the same chart is varied into two different types of colour matching based on the same chart, which are divided into monochrome type (approximate colour matching) and colour type (contrasting colour matching). In the choice of approximate colour matching, blue (RGB value 88,145,198) is chosen as the main colour, and the brightness and saturation indexes are increased and decreased by a difference of 10 so as to ensure the differentiation and recognizability of the approximate colour matching. In the choice of contrasting colour matching to red, green and blue primary colours as the main colour, respectively, select the colour in the hue ring with a difference of more than 90 ° as the contrasting colour palette for the choice of colour matching scheme. It is guaranteed that the subjects’ vision can identify the intensity of color contrast and the perception of color temperature.
After excluding invalid samples (dependent variable value = null), a total of 120 charts were recorded for awareness and recall data. To exclude the interference of extreme values, the M-estimates in the exploratory analyses were viewed. After weighting the M-estimators, the Huber method was adopted in this study to determine the mean cognitive and memory score values for the different gender samples, and the final results are shown in Table 1. The results show that the mean cognitive score values of males are not significantly different from the mean cognitive score values of females, and the mean memory score values of males are slightly lower than the mean memory score values of females. a. The weighted constant is 1.339, b. The weighted constant is 4.685, c. The weighted constants are 1.700, 3.400, and 8.500, and d. The weighted constant is 1.340*pi.
The total score value is analyzed by m estimation
| Gender | Huber m estimates a | Tugki double right b | Hampel m estimation c | Andrew wave d | |
|---|---|---|---|---|---|
| Awareness | Male | 9.42 | 9.47 | 9.53 | 9.55 |
| Female | 9.58 | 9.61 | 9.65 | 9.7 | |
| Memory degree | Male | 7.08 | 6.99 | 6.98 | 6.96 |
| Female | 7.7 | 7.64 | 7.69 | 7.69 |
Scores were tallied according to the type of chart (6 points for each chart in each group), and the experimental data as a whole were analysed by correlation tests and linear regression using the statistical software for the social sciences, SPSS. One-way analysis of variance (ANOVA) and independent samples t-test were performed on the characteristics of the four independent variables individually. The experimental hypotheses were verified based on the final data analysis results and further elaborated with relevant theories. The correlation test was performed before linear regression was first performed on the overall data. Table 2 displays the outcomes of the Pearson correlation analysis.
Pearson correlation analysis
| Awareness | Memory degree | ||
|---|---|---|---|
| Sign mark | Correlation coefficient | 0.427** | 0.368** |
| p | 0 | 0 | |
| Color collocation | Correlation coefficient | ** | 0.166 |
| p | 0 | 0.167 | |
| Dynamic effect | Correlation coefficient | -0.235* | -0.098 |
| p | -0.015 | 0.477 | |
| Chart type | Correlation coefficient | -0.055 | -0.096 |
| p | 0.671 | 0.406 | |
The value of the correlation coefficient between cognition and symbolic marking is 0.427 and shows significance at the 0.01 level, thus indicating that there is a significant positive correlation between cognition and symbolic marking. Similarly, there is a significant positive correlation between awareness, color scheme, and dynamic effect. The value of the correlation coefficient between cognition and chart type is -0.055, which is close to 0, and the p-value is 0.671>0.05, thus indicating that there is no correlation between cognition and chart type. The correlation coefficient value between memorability and symbolic marking is 0.368 and shows significance at the 0.01 level, thus indicating that there is a significant positive correlation between these two variables. There is no significant correlation between memorability and the color scheme, dynamic effect, and chart type.
The test results showed that the variables involved in this experiment were correlated with cognition and memory. Therefore, linear regression analysis was performed. The results of the linear regression analysis with the dependent variable being cognition are shown in Table 3, D-W value: 2.029, *p<0.05**p<0.01.
The linear regression analysis of recognition is (n= 120)
| Constant | Color collocation | Sign mark | Dynamic effect | Chart type | ||
|---|---|---|---|---|---|---|
| Nonnormal coefficient | B | 3.413 | 1.333 | 0.663 | 0.295 | -0.099 |
| S.E. | 0.32 | 0.18 | 0.104 | 0.3 | 0.135 | |
| Normal factor | Beta | -- | 0.491 | 0.488 | 0.052 | -0.07 |
| t | 10.632 | 6.593 | 5.707 | 1.003 | -0.715 | |
| P | -0.000** | -0.000** | -0.000** | 0.379 | 0.54 | |
| VIF | -- | 1.103 | 1.335 | 1.391 | 1.077 | |
| R2 | 0.387 | |||||
| Adjust R2 | 0.364 | |||||
| F | F | (4,127)=17.856,p=0.000 | ||||
The R-square value of the model is 0.387, implying that colour scheme, symbol marking, dynamic effects, and diagram-type phenotype explain 38.7% of the variation in perception. The F-test of the model was found to pass the F-test (F=17.856, p=0.000<0.05), indicating that the model is valid. The D-W value is around 2, thus indicating that there is no autocorrelation in the model and no correlation between the sample data, which suggests that the model is better. The regression coefficient value of colour matching is 1.333 (t=6.593, p=0.000<0.01), and the regression coefficient value of symbolic marking are 0.663 (t=5.707, p=0.000<0.01), which means that the factors of colour matching and symbolic marking will have a significant positive influence on the relationship of perception.
The results of the linear regression analysis with the dependent variable being memory are shown in Table 4, D-W value: 2.050, *p<0.05**p<0.01.
The linear regression analysis of memory (n= 120)
| Constant | Color collocation | Sign mark | Dynamic effect | Chart type | ||
|---|---|---|---|---|---|---|
| Nonnormal coefficient | B | 3.576 | 0.48 | 0.674 | 0.534 | -0.156 |
| S.E. | 0.418 | 0.235 | 0.144 | 0.357 | 0.173 | |
| Normal factor | Beta | -- | 0.13 | 0.498 | 0.185 | -0.094 |
| t | 9.277 | 1.986 | 4.836 | 1.725 | -0.946 | |
| P | 0.000** | 0.076 | 0.000** | 0.115 | 0.398 | |
| VIF | -- | 1.064 | 1.356 | 1.386 | 1.03 | |
| R2 | 0.181 | |||||
| Adjust R2 | 0.152 | |||||
| F | F | (4,123)=6.134,p=0.000 | ||||
A linear regression analysis was conducted with memory as the dependent variable, and the table shows that the model R-squared value is 0.181, implying that the independent variable explains 18.1% of the variation in memory. Since this experiment measured the long-term memory effect, the graphical attributes had less effect on the recall process of the subjects compared to the cognitive degree, which may be due to the fact that the semantic model of long-term memory also includes processes with individual differences such as storage, retrieval and production. The regression coefficient value for colour matching was 0.48 (t=1.986, p=0.076>0.05), implying that colour matching does not have an influential relationship with recall. For specific analysis, the value of the regression coefficient of symbol marking is 0.674 (t=4.836, p=0.000<0.01), which means that symbol marking produces a significant positive influence relationship on memorability.
In conclusion, hypotheses 1 and 2 are valid. Coloured charts have a significant positive effect on awareness and recall, and charts with relevant markers have a significant positive effect on awareness and recall.
The results of the effect of the color scheme of the information data visualisation on cognition and memory are shown in Table 5 when the chart type is not taken into account. Using independent samples t-tests to study the effect of color scheme on cognition, all of the different color scheme samples showed consistency (p>0.05) and no difference in memory. However, the different colour matching samples showed consistency (p<0.05) and no difference in recognition. The mean value of 3.73 for monochrome samples is significantly lower than the mean value of 4.95 for colour samples.
Test analysis of color matching variables
| Color collocation | t | P | ||
|---|---|---|---|---|
| 0.0(n=42) | 1.0(n=78) | |||
| Awareness | 3.73 | 4.95 | -5.237 | 0.000** |
| Memory degree | 3.65 | 4.09 | -1.921 | 0.056 |
Figure 1 shows the score statistics of colour matching on different chart types. (a) and (b) show the results of cognitive and memorability analyses, respectively. Cognition is significantly higher when the colour scheme is coloured in all chart types of infographics, and there is no significant difference in memorability. The effect of chart type on the results was not significant. It is worth noting that the mean values of scores for monochrome and coloured charts were similar for the line charts in terms of memorability presentation, indicating that the variable of colour scheme had less impact on the memorability of line charts. Much of this phenomenon stems from the fact that the smaller area of the folded line graph shape does not significantly accentuate the colour presentation, especially since weaker colour intensities are difficult to store effectively in the brain during long-term memory. Similarly, it can be observed that the bar chart with the largest color block area can have its memory level significantly affected by color.
Color collocation is counted in different chart types
In dual coding theory, the interaction between the linguistic and non-linguistic systems enables the audience to understand and remember the content more quickly. In this experiment, colour as a “pixel” in the non-linguistic system enhances the referential connection of the multimodal diagrams, and this referential link can speed up the audience’s search for information.
At the cognitive-psychological level, a one-way analysis of variance (specifically Welch ANOVA) was used to investigate the variability of symbolic labelling on a total of 2 items of cognition and memory without taking into account the type of chart, and the results are shown in Table 6.
The Welch variance analysis result of the symbol tag variable
| Color collocation | Welch F | P | |||
|---|---|---|---|---|---|
| 0.0(n=42) | 1.0(n=78) | 2.0(n=42) | |||
| Awareness | 4.28 | 3.64 | 5.42 | 39.58 | 0.000** |
| Memory degree | 3.97 | 2.58 | 5.19 | 85.363 | 0.000** |
Where p indicates that the different symbol labelling samples all show significance (p<0.05) for cognition and memory, meaning that the different symbol labelling samples have differences in different values for cognition and memory. Among them, the symbolic marking for cognition showed a 0.01 level of significance (Welch F=39.58, p=0.000), as well as the specific comparison of the differences, can be seen. There is a more obvious difference between the group mean scores comparison results for “2.0>0.0; 2.0>1.0”. Symbol labelling showed a 0.01 level of significance for memorability (Welch F=85.363, p=0.000), and a specific comparison of the differences can be seen. There are more obvious differences in the group mean scores compared to the results of “0.0>1.0; 2.0>0.0; 2.0>1.0”.
Figure 2 shows the performance statistics after considering the chart type factor, and (a) and (b) show the results of the cognitive and memory analyses, respectively. It was found that the variability of perception and recall of pie charts, line graphs, and bar charts was similar to the overall trend, and they all showed that perception and recall were significantly worse in charts with irrelevant markers, while charts with relevant markers had better perception and recall than both charts without markers and charts with irrelevant markers.
The notation is marked by the score statistics on different chart types
Dual coding theory emphasises the importance of graphics for knowledge learning due to the fact that the structural features of visual memory can be processed as a whole unit by both units in a parallel way, but only sequentially if there is only one language. At the same time, additionally, dual coding theory proposes conceptual hook erection, i.e., stimulus specificity plays a key role in image retrieval and recall. Thus, graphic symbols are faster and more complete thought constructs than language.
This section aims to explore, through behavioural experiments and data analysis, the efficiency of different visual diagrams for memory recall tasks under different cognitive loads.
The experimental independent variables remained to be the three visual chart types, which are pie charts, line graphs, and bar charts. Meanwhile, the other independent variable, the number of information, was chosen to be 2, 4, and 8 item values, which is because according to Miller’s study, 4 item values are the general working memory capacity, and 8 item values reach the upper limit of human visual working memory. The names of the items are denoted by the letters A-H. In order to avoid an exercise effect, the data values of each chart were randomly sampled between 1 and 10.
The three dependent variables in the experiment were task error, task completion time, and NASA-TLX task load index. The task error was defined as the subject’s mean error, which is the average of the absolute value of the difference between the reference value and the subject’s filled-in value for each item in the visualisation chart. For example, if the reference values are 5, 2, 5, and 3, and if the subject fills in the values of 5, 4, 6, and 3, the mean error is calculated as (0+2+1+0)/4=0.75. Task completion time is defined as the time interval from the appearance of the answer page to the completion of the user’s filling in the answer page. The NASA-TLX Scale is the subjective evaluation scale of mental load that is currently being used more often. The scale has six entries: mental demands, physical demands, time frame demands, self-performance, effort, and frustration. Each entry is scored from 0 to 20 from left to right, indicating a low to high task load. Self-performance is rated from “perfect” to “failure” from left to right, i.e., the lower the score, the more perfect the self-performance and the lower the task load. The scores of the six items were summed and averaged to obtain the total mental load score.
Table 7 shows a two-by-two pairwise comparison of task errors. There was no significant difference between pie charts, line graphs, and bar graphs at a number of information of 2. At a number of information of 4, there was a significant difference between pie charts and line graphs (p<0.01), bar graphs (p<0.05), and a significant difference between line graphs and bar graphs (p<0.05); at a number of information of 8, there was a significant difference between pie charts and line graphs (p<0.01) and bar graphs (p<0.01).
The two pairs of the task error are compared
| Mean difference | 95% confidence interval | ||||||
|---|---|---|---|---|---|---|---|
| Information | Alarm mode(I) | Alarm mode(J) | (I-J) | S.E. | Sig. | Low | Up |
| 2 term | Pie | Line | -0.034 | 0.109 | 0.995 | -0.283 | 0.237 |
| Bar | -0.021 | 0.109 | 1.001 | -0.342 | 0.202 | ||
| Line | Pie | 0.04 | 0.109 | 1.049 | -0.209 | 0.313 | |
| Bar | -0.017 | 0.109 | 1.035 | -0.322 | 0.25 | ||
| Bar | Pie | 0.045 | 0.109 | 0.976 | -0.187 | 0.338 | |
| Line | 0 | 0.109 | 1.052 | -0.211 | 0.314 | ||
| 4 term | Pie | Line | -0.54 | 0.109 | -0.028 | -0.863 | -0.298 |
| Bar | -0.26 | 0.109 | 0.043 | -0.526 | -0.009 | ||
| Line | Pie | 0.56 | 0.109 | -0.036 | 0.335 | 0.824 | |
| Bar | -0.316 | 0.109 | 0.003 | 0.028 | 0.535 | ||
| Bar | Pie | 0.286 | 0.109 | 0.047 | -0.013 | 0.58 | |
| Line | 0.323 | 0.109 | 0.047 | -0.562 | 0.005 | ||
| 8 term | Pie | Line | 0.349 | 0.109 | -0.015 | 0.069 | 0.61 |
| Bar | 0.736 | 0.109 | 0.028 | 0.415 | 0.945 | ||
| Line | Pie | -0.374 | 0.109 | -0.035 | -0.632 | -0.076 | |
| Bar | 0.344 | 0.109 | -0.007 | 0.101 | 0.596 | ||
| Bar | Pie | -0.699 | 0.109 | 0 | -0.979 | -0.438 | |
| Line | -0.385 | 0.109 | -0.035 | -0.619 | -0.1 | ||
In order to further investigate the task error comparison of different visualisation chart types, descriptive analysis was used to statistic the mean values of the task errors of the three visualisation chart types under the three numbers of information. Figure 3 shows the statistical results.
The mean of the task error
The results show that when the number of information is 2, task errors are displayed in descending order in pie charts, line graphs, and bar graphs. When the number of information is 4, the task errors are displayed in pie charts, bar graphs, and line graphs in descending order. At a number of information of 8, the task errors were in descending order, bar graphs, line graphs, and pie charts. Combined with the results of the simple effects analysis, it was found that when the number of information was 2, there was no significant difference in the errors caused by different chart types; when the number of information was 4, the errors caused by pie charts were significantly smaller than those caused by line graphs and bar charts, and the errors caused by bar charts were significantly smaller than those caused by line graphs; and when the number of information was 8, the errors caused by pie charts were significantly higher than those caused by line graphs and bar charts, and the errors caused by bar charts were significantly smaller than those caused by line graphs.
Table 8 shows a two-by-two pairwise comparison of task completion times. At an information count of 2, there is no significant difference between pie charts, line graphs, and bar graphs. At a number of information of 4, there was a significant difference between the pie charts and the line graphs (p<0.01) and bar graphs (p<0.05), respectively, and no significant difference between the line graphs and bar graphs (p=0.985); at a number of information of 8, there was a significant difference between the pie charts and the line graphs (p<0.01) and bar graphs (p<0.01), respectively, and a significant difference between the line graphs and bar graphs (p<0.05) are significantly different from each other.
Two pairs of comparison between the time of the task
| Mean difference | 95% confidence interval | ||||||
|---|---|---|---|---|---|---|---|
| Information | Alarm mode(I) | Alarm mode(J) | (I-J) | S.E. | Sig. | Low | Up |
| 2 term | Pie | Line | 1.528 | 1.783 | 0.837 | -2.973 | 6.076 |
| Bar | 1.059 | 1.783 | 0.968 | -3.48 | 5.58 | ||
| Line | Pie | -1.557 | 1.783 | 0.819 | -6.046 | 2.968 | |
| Bar | -0.481 | 1.783 | 1.066 | -5.003 | 4.055 | ||
| Bar | Pie | -1.039 | 1.783 | 0.98 | -5.565 | 3.456 | |
| Line | 0.495 | 1.783 | 1.054 | -4.046 | 5.013 | ||
| 4 term | Pie | Line | 5.282 | 1.783 | 0.007 | 0.745 | 9.813 |
| Bar | 4.679 | 1.783 | 0.04 | 0.137 | 9.205 | ||
| Line | Pie | -5.302 | 1.783 | 0.053 | -9.81 | -0.732 | |
| Bar | -0.559 | 1.783 | 1.018 | -5.1 | 3.958 | ||
| Bar | Pie | -4.699 | 1.783 | 0.034 | -9.233 | -0.153 | |
| Line | 0.57 | 1.783 | 1.052 | -3.965 | 5.13 | ||
| 8 term | Pie | Line | 13.634 | 1.783 | 0.024 | 9.105 | 18.175 |
| Bar | 8.224 | 1.783 | -0.021 | 3.698 | 12.746 | ||
| Line | Pie | -13.63 | 1.783 | -0.012 | -18.159 | -9.12 | |
| Bar | -5.408 | 1.783 | 0.02 | -9.994 | -0.884 | ||
| Bar | Pie | -8.239 | 1.783 | -0.003 | -12.753 | -3.677 | |
| Line | 5.44 | 1.783 | 0.036 | 0.887 | 9.987 | ||
Descriptive analysis was used to count the mean values of task completion times for the three visualisation chart types for the three numbers of information. Figure 4 shows the statistical results. The results show that at the low, medium, and high levels of information, the task completion time is a line chart, bar chart, and pie chart in descending order. Combined with the results of simple effects analysis, it was found that there was no significant difference in the task completion time of different chart types when the number of information was 2. When the number of information was 4, the task completion time of pie charts was significantly greater than that of line graphs and bar charts; when the number of information was 8, the task completion time of pie charts was significantly greater than that of line graphs and bar charts, and the task completion time of bar charts was significantly greater than that of line graphs.
Task completion time mean
Two-by-two comparisons of chart types were made using the LSD test. The data results are shown in Table 9. The comparison shows that there is a significant difference between pie charts and line charts (p<0.01), between pie charts and bar charts (p<0.05), and no significant difference between bar charts and line charts (p=0.649).
The two pairs of the task load index are compared
| Mean difference | 95% confidence interval | |||||
|---|---|---|---|---|---|---|
| Alarm mode(I) | Alarm mode(J) | (I-J) | S.E. | Sig. | Lower limit | Upper limit |
| Pie chart | Line diagram | 1.555 | 0.473 | 0.014 | 0.546 | 2.577 |
| Bar chart | -0.282 | 0.473 | -0.001 | 0.264 | 2.303 | |
| Line diagram | Pie chart | -1.574 | 0.473 | -0.013 | -2.574 | -0.58 |
| Bar chart | 0.246 | 0.473 | 0.649 | -1.268 | -0.273 | |
| Bar chart | Pie chart | -1.314 | 0.473 | 0.012 | -2.342 | 5.477 |
| Line diagram | 0.259 | 0.473 | 0.616 | -0.738 | 1.297 | |
Descriptive analysis was used to count the mean values of the task load index for the three visualisation chart types for the three numbers of information. Figure 5 shows the results of the analysis. The results show that when the number of information is 2, the task load index is represented by a bar chart, line chart, and pie chart in descending order. When the amount of information is 4 and 8, the task load index is displayed in the order of line graph, bar graph, and pie chart, starting from small to large. Combining the results of the two-by-two comparison, it was found that the cognitive load induced by pie charts was significantly greater than that of line and bar charts.
The mean of the task load index
Based on the above theoretical analysis and experimental verification results, the following information visualisation strategies are proposed:
When visual coding is performed between multiple pages, the mapping relationship between data dimensions and visual coding should be consistent and coherent. The user’s cognitive efficiency increases with the closer the mapping relationship between the data constructed on the previous page and the visual coding to the next page. The principle of consistency in multi-page visual linguistic representation can be divided into the consistency of visual coding representation and the consistency of user operations between pages. Due to the complexity and abstractness of data information, the successive presentation of multi-page visual language representations greatly increases the cognitive complexity of users compared to the parallel presentation of single pages. In order to ensure that users more easily understand the visual language representations, it is necessary to maintain consistency in the basic visual language representations such as shape, colour and spatial position of the dimensional coding.
Since working memory is only a system used by users to temporarily store and process data and information, only a small portion of the received information will be cognised and temporarily remembered by the user, which requires that the visual, linguistic representation of the visualisation of information and data be carried out in such a way as to minimise the dimensions, quantity, complexity, etc. of the content stored by the user, and to facilitate the user’s identification and retention of important information.
The information contained in information visualization is usually abstracted high-dimensional multivariate data, and the visual language presentation of visualization also faces problems with complex visual coding and complex memory. For multi-page visual language representation, we should try to optimise the data information and visual coding elements, pay attention to the user’s visual cognitive model and experience, and choose the visual coding elements that conform to people’s mental model instead of pursuing the visual presentation effect of the visual presentation and ignoring the user’s reception effect of the information in the reading process. Pay attention to avoiding obscure visual language representations and uncommon terminology that may affect the transmission of information, resulting in users’ inability to understand and receive it.
This paper takes the dual coding theory in cognitive psychology as the starting point and analyzes the effect of color symbols and cognitive load on the process of information visualization through experiments. In this way, it provides a basis for evaluating the communication effect of infographics in network information communication. It has been found that contrasting color schemes and symbols with meaningful associations can help the audience understand and remember infographics. The error induced by pie charts is significantly higher than that induced by line and bar charts, and the error induced by bar charts is significantly smaller than that induced by line charts. The cognitive load that pie charts cause is significantly higher than that of line and bar charts. Therefore, strategies for displaying information and visualizing data are proposed for reference.