Optimizing Building Information Modeling System Design Elements for Enhanced User Experience: A Multidimensional Approach

3.1 BIM task scenario analysis

3.1.1 Division of BIM system contextual factors

The division of contexts is primarily determined by the user's state and nature of engagement. This study, grounded in contextual theory, delves into the user-BIM platform interaction from the standpoint of UX. Consequently, the comprehensive contextual influencing factors encountered during user interaction with the system platform are categorized into four contextual factors: user, time, environment, and task. This division serves as the analytical framework for investigating UX issues in the context of the BIM system platform. The specific contextual classifications and division of contextual factors are illustrated in Figure 1.

Contextualized experience design strategies necessitate a departure from user contexts, involving the comprehensive analysis of internal factors such as emotions, cognitive levels, and motivations while also considering the multifaceted factors within the human-computer interaction process. By judiciously harnessing positive elements, users can more seamlessly, joyfully, and efficiently complete tasks. Users' perceptions of cognition, behavior, and emotional attitudes manifest diverse emotional changes throughout the same interaction process. Perceiving users' psychological attributes through external contexts is pivotal.

1.png

Figure 1. Contextual framework of BIM systems

Task scenario analysis encompasses the task a user is currently engaged in, along with associated behavioral activities and events, constituting a vital aspect of contextual analysis. Task scenarios encompass factors like task complexity, flow, the context during task execution, task objectives, and time duration for task completion. Only by meticulously considering all facets, from task initiation to culmination, can system architecture be optimized, enhancing user efficiency.

Environmental context within the BIM system platform primarily pertains to the equipment environment. In comparison, regarding environmental factors, the BIM system encompasses the physical environment in which users exist, including temporal and spatial components. This also incorporates various scenes within the system, ranging from virtual scenarios within railway construction processes to human-machine interaction interfaces within different subsystems.

Time context analysis, guided by cognitive load theory, correlates cognitive load with the time users employ. User behavior analysis, using time duration as a metric, reveals insights into user behavior. The duration of time partly indicates user efficiency during task execution and their emotional shifts. Information overload from data can heighten cognitive load, leading to a dip in user emotions.

Effectively integrating user, task, environmental, and time contextual factors can facilitate the development of a BIM system platform with enhanced UX, nurturing harmonious human-computer interactions, and bolstering user satisfaction.

3.1.2 Purpose and methodology of contextual research

Conducting contextual research analysis constitutes a pivotal facet within the user-centric design process. Contextual research, entailing the analysis of diverse situational variances, subjective experiential perceptions, and interactional behaviors of users, serves as a means of unearthing authentic user requisites. This information, subsequently, serves as the foundational basis for design, facilitating the provision of personalized services to users. In this paper, contextual theory is introduced into the realm of BIM system platform design, enabling the examination of the relationship between various situational factors and user demands. Furthermore, this approach allows for the synthesis of behavioral characteristics observed during the interaction between users and the BIM system platform. The research objectives are delineated as follows:

1. To elucidate the fundamental requirements of users utilizing the BIM platform.

2. To comprehend the issues encountered by users when employing existing platforms, thereby aggregating pain points.

3. To discern whether opportunities for optimization exist through an analysis of user interactions with the utilized BIM platform.

This paper primarily unfolds along three dimensions: questionnaire surveys, user interviews, and behavioral observation methods, as outlined in Table 1, encapsulating the particulars of the research.

(1) Questionnaire Survey

A questionnaire survey serves as a preparatory exercise for in-depth interviews with target users. It aids in gaining an initial understanding of user expectations when using the platform, marking the preliminary stage of research focused on comprehensive analysis and discovery of user requirements pertinent to the BIM platform. The survey designed in this paper predominantly employs online questionnaire distribution as the primary mode of data collection. The distributed survey pertains to aspects of UX related to platform design. From a quantitative perspective, it gathers data on user demographics, motivations for using the BIM platform, usage patterns, and learning behaviors. This preliminary survey aids in ascertaining the fundamental usage requirements of users regarding the BIM platform. Subsequently, the data from the survey is organized and analyzed to provide empirical support for user interviews and subsequent design directions.

(2) In-Depth User Interviews

In-depth user interviews refer to a communicative exchange between an interviewer and interviewee. Throughout this interaction, the interviewer judiciously manages and schedules the conversation, posing questions to elicit the interviewee's thoughts on the subject matter. The objective of these interviews is to obtain answers to specific questions. In this paper, the interviews primarily take the form of one-on-one, face-to-face interactions, ensuring that respondents can provide in-depth responses to questions. The interview questions follow a semi-structured format, allowing for spontaneous inquiries within the scope of the prepared questions. The interviews are conducted with typical users and, building upon the findings of the questionnaire survey, delve deeper into users' platform-specific needs. They aim to unearth more profound pain points, elucidate user expectations, and uncover latent demand points. Information about the interviewees and an overview of the research are presented in Tables 2 and 3, respectively.

The in-depth user interview profiles detail each respondent's experience with the existing BIM system platform, the problems they encountered, and their expectations for using the BIM system platform.

Table 1. Scenario research and analysis

Table 2. Respondent information of BIM management platform

Table 3. Overview of some interviewer research

(3) User Behavior Analysis

User behavior analysis is instrumental in reconstructing the authentic process of user-product interaction. Only with a comprehensive understanding of user behavior patterns and operational processes can issues within the user's operational journey be identified. By observing users' interactions within the management platform, a detailed analysis of issues encountered during the user's operational process can be conducted. This analysis is crucial for gaining insights into the practical challenges users face while utilizing the platform.

3.1.3 Contextual research analysis

Traditional UX often relies on singular evaluations, seldom considering multidimensional research for assessing UX. Through conversations with users of the BIM system platform, it has been observed that the UX can vary significantly based on different contexts, even within the same platform. Following a comprehensive investigation of the target users of the BIM platform, this research integrates questionnaire surveys, user interview content, and behavioral observations into contextually relevant categories. Information is then synthesized and analyzed to unearth the demands of BIM platform users during their usage.

(1) User Contextual Analysis

Based on contextual research, the acquired user contextual information is summarized and categorized, elucidating the issues that users encounter during their usage. Drawing from Maslow's Hierarchy of Needs theory, an organization and summarization of target users' needs are conducted, classifying them into five categories: cognitive capabilities, error avoidance, emotional connection, recognition and affirmation, and self-worth, as outlined in Table 4.

Table 4. Target user requirements analysis

Table 5. Scenario requirements summary table

(2) Task Contextual Analysis

Building upon the insights gained from user interviews and behavioral observations, user task information is classified and summarized. The primary task activities of users are found to differ based on various stakeholders and professions, resulting in different task processes. Depending on user types, different contextual interaction processes emerge.

(3) Temporal Contextual Analysis

Temporal contextual analysis is twofold. Firstly, it involves an examination of the duration of platform usage. Longer usage periods lead to increased cognitive load and reduced efficiency for users. Thus, interface adjustments are necessary to modulate the cognitive load for users effectively. Secondly, it considers how user behaviors vary across different time dimensions. Early on, users require an understanding of basic platform operations and procedures. During usage, streamlining operational processes enhances user engagement. Following task completion, reinforcing system interaction feedback provides users with emotional resonance.

(4) Environmental Contextual Analysis

Research conducted during the early stages pertaining to the BIM system platform and device characteristics is categorized and summarized. It is noted that environmental context can significantly impact user emotions. A quiet and conducive environment tends to enhance user efficiency. In an internal context, where users engage in virtual interactions with the system, immersion is crucial. Enhancing the UX in a digital construction environment can involve both visual and auditory design considerations. Differentiating between PC and mobile platforms is essential; PC platforms emphasize task completion, while mobile platforms focus on user engagement.

3.1.4 Summary of user requirements for the BIM system platform

Building upon the contextual research conducted for the BIM system platform, a comprehensive understanding of user requirements within the context of BIM system usage has been derived. This understanding is based on contextual analyses encompassing users, tasks, time, and environment. The resulting user requirement model, as obtained from these analyses, serves as a foundational framework for subsequent design strategies. A detailed summary of these requirements is presented in Table 5.

3.2 Obtaining BIM system user requirements based on KANO-QFD

Building upon the user requirements summarized earlier using the contextual theory, design strategies are proposed based on contextual factors: users, tasks, environment, and time. A KANO-QFD model is employed to categorize and prioritize user requirements and transform them into design elements, thereby deriving relevant design strategies.

3.2.1 Requirement acquisition process

By combining the KANO model with QFD, the model's functionality can be effectively enhanced and refined, thereby improving its reliability. The application of the KANO-QFD model not only enables the precise analysis of user requirements but also helps in determining the correct service direction. This guides product design to align with user requirements, ensuring user satisfaction. In this paper, improvements to the BIM-based system platform are approached from the perspective of contextual analysis and user requirements. The process is illustrated in Figure 2.

2.png

Figure 2. User Requirements analysis flowchart

3.2.2 Clustering user requirement attributes based on the KANO model

In user requirement analysis, the KANO model is a crucial method for categorizing and prioritizing user requirements. Grounded in the analysis of the impact on user satisfaction, Professor Kano has classified product quality attributes into five categories: Must-be Quality (M), One-dimensional Quality (O), Attractive Quality (A), Indifferent Quality (I), and Reverse Quality (R). This classification is based on the relationship between different types of quality characteristics and user satisfaction evaluations, as depicted in Figure 3.

3.png

Figure 3. KANO model

Combining the design requirements summarized in the previous section with user feedback, demand keywords were identified and selected. A KANO survey questionnaire was designed, and user evaluations were recorded for each requirement. Data were collected and analyzed, primarily through online KANO questionnaire surveys, resulting in a total of 96 valid responses. Statistical analysis was performed using SPSS, yielding a reliability coefficient of 0.827 for this research questionnaire. The analysis of demand evaluations is presented in Table 6.

3.2.3 Design objectives based on the QFD model

Utilizing the user requirements, functionalities are categorized based on the contextual factors of users, tasks, time, and environment. Specifically, 6 user requirements have been selected for further analysis, as presented in Table 7. These requirements are categorized within the dimensions of user context, task context, time context, and environmental context.

CS represents customer satisfaction, while CD represents customer dissatisfaction. Calculating CS and CD values helps reflect the average impact of each quality attribute on customer satisfaction. The results are presented in Table 8.

A relationship matrix between user requirements and design elements, indicating strong correlations, weak correlations, or no correlations, was generated based on user requirements. The QFD methodology was employed to calculate and construct the Quality House, as shown in Table 9. This matrix serves as a valuable tool for mapping the relationships between user needs and design features, facilitating the prioritization of design efforts.

Based on the output results from the Quality House, the prioritization of design requirements has been established. Combining these prioritized design requirements with the existing ones, the following design objectives have been determined: 1. Optimize Interface Layout Design; 2. Develop Color Scheme Design; 3. Refine Operational Workflow Design.

These objectives will guide the design efforts to ensure that user needs and expectations are met effectively.

Table 6. Statistical analysis of KANO questionnaire results

Table 7. Design elements and design requirements table

Table 8. The relationship between quality of experience and the impact of user satisfaction

Table 9. Relationship matrix of need elements and functional requirements

Note: ◇ indicates strong correlation, □ indicates medium correlation, and △ indicates weak correlation.

3.3 BIM system interface layout optimization based on user requirements

Step 1: Establish interface layout design principles. These principles should be user-centric, incorporating principles from cognitive load, ergonomics, and aesthetics. These principles are summarized and transformed into objective functions.

Step 2: Build the interface layout optimization model and employ the SSA to optimize the objective functions.

Step 3: Evaluate the usability of the proposed layout. Introduce eye-tracking measurements for usability assessment.

Table 10. Coding of the 12 layout elements and their importance and frequency of use

Table 11. 12 correlations between elements to be laid out

Table 12. Calculated result coordinates

4.png

Figure 4. Low fidelity map after layout optimization

The system platform interface is selected as the research subject. Initially, the information elements on the platform's homepage are encoded. These elements include: Digital Factory, Material Management, Quality Inspection, Work Valuation, Completion Data, Change Design, Credit Evaluation, Fund Management, Contract Management, Work Processes, Time Display, and System Name - totaling twelve elements for layout. Experts are invited to score decision elements, and elements are standardized, as shown in Table 10. 12 correlations between elements to be laid out as shown in Table 11.

By following the aforementioned steps, encoding, and employing the SSA for iterative optimization, the layout elements at the 20th iteration are presented in Table 12.

Combining the data, the layout optimization scheme obtained is shown in Figure 4.

3.4 Color scheme prediction based on user requirements

The mapping between image space and color space is not a one-to-one relationship. To make the color scheme align with color image semantics, the semantic analysis method is applied to address the complex mapping issue between color space and image space. The color image assessment design process is as follows:

Step 1: Extract feature colors from the existing platform using the improved GNG clustering algorithm to determine the colors for the color scheme, constructing a color scheme sample library.

Step 2: Gather sensory image adjectives for the BIM system platform and select representative samples and adjectives to construct the image space.

Step 3: Explore the relationship between image vocabulary and color samples through semantic differential methods, building a mapping model. Express sensory demands and expectations for colors through image vocabulary.

Step 4: Utilize gray clustering methods to categorize image vocabulary into three grayscale categories: "lively-neutral-lonely." This is used to predict the overall color scheme tendencies for the human-computer interface.

3.4.1 Establishing feature color groups

In this study, the improved GNG algorithm is used for clustering. Since input images are typically stored in 8-bit grayscale format, representing data within 256 grayscale levels, every color pixel falls within the range of 0-255. When inputting data into the improved GNG neural network, all color pixels are normalized to values between 0 and 1, which corresponds to the range of 0-255. The experimental steps are as follows: Search for color image semantics related to project management system platforms. Collect a large number of interface color design examples from online media. Invite 20 professionals, including teachers and graduate students specializing in design, to conduct preliminary screening of the collected samples. Ultimately, select 30 interface color samples with significant relevance to the target color image semantics for color clustering based on the adaptive GNG algorithm.

Through iterative optimization and the construction of a sample library using extracted feature color RGB values, a total of 90 colors are obtained. These colors are then grouped based on color similarity, resulting in 10 groups, as depicted in Figure 5.

5.png

Figure 5. Characteristic color grouping

3.4.2 Color image semantics analysis and semantic mapping model

Color semantic analysis can be expressed through antonymous forms. According to semantic classification, three common features are extracted for all color semantics in the sample library: evaluative, activity, and potential. The results obtained through semantic induction are shown in Table 13.

Color features can evoke emotions and associations in users. For example, green often conveys environmental consciousness, tranquility, and associations with nature, such as forests. Users can express their expectations and basic needs for colors through the derived image semantics words. Predicting a color scheme that aligns with user color image preferences during the early stages of color design can enhance user satisfaction with the system platform's color scheme.

The image vocabulary involved is obtained through the Semantic Differential method. Initially, 156 sensory words were collected through literature research and user interviews. Through group discussions and evaluations, negative words were removed and similar words were merged. The 18 most frequent image vocabulary words were selected, and antonymous pairs were created for each. The result is 18 sets of vocabulary related to the sensory image of the system platform, as shown in Table 14.

Table 13. Semantic categorization

Table 14. Sensory imagery phrase

Table 15. Color imagery evaluation score statistics

Color features have the ability to evoke emotions and associations in users.

Firstly, select one representative color sample from each of the 10 groups of similar color samples. Designate these selected colors as experimental samples and assign unique identifiers to each sample, Sample_n, n=1,2,3,…,10.

In accordance with the field of Human-Computer Interaction (HCI), we extended invitations to a total of 30 participants possessing backgrounds in design, aged between 25±5 years, with normal or corrected-to-normal vision, Sample_m, n=1,2,3,…,30.

Subsequently, employing the Semantic Differential Method, we constructed a five-point evaluation questionnaire to capture psychological variations across different dimensions. This questionnaire amalgamated ten representative color samples with carefully selected color-associated lexical terms. The color image assessments were positioned on a scale ranging form -2 to 2, with 0 signifying neutrality, -2 indicating discomfort, and 2 representing pleasure.

The experimental procedure involved the random presentation of the ten selected color samples on a computer screen to participants. Each participant was allotted a specific period for observing the color samples and providing semantic evaluations. These evaluations were quantitatively analyzed using the Likert scale.

Upon completion of the experiment, participant data encompassing assessments of color semantics were collected. After all participants had evaluated the ten prototypical color samples, the data underwent averaging, yielding the evaluation results as illustrated in Table 15.

3.4.3 Color image prediction based on grey clustering

The mapping relationship between monochromatic colors and color image semantics has been established through evaluations, as described earlier. However, typical color schemes in human-computer interfaces often involve the combination of multiple colors rather than a single color. To address this, this paper employs the Grey Clustering Method for color scheme prediction.

To begin, the set of color samples within the BIM system platform's color schemes is denoted as C=(1,2,…,n), while the collection of affective image vocabulary is represented as Y=(1,2,…,m). The clustering whitening coefficient for the third color scheme with respect to the fourth color image vocabulary is indicated as Coefficient f_ij. Consequently, this forms the color scheme matrix, denoted as Matrix A.

$A=\begin{matrix}   \begin{matrix}   1 & 2 & \cdot \cdot \cdot  & n  \\\end{matrix}  \\   \left[ \begin{matrix}   {{d}_{11}} & {{d}_{22}} & \cdot \cdot \cdot  & {{d}_{1n}}  \\   {{d}_{21}} & {{d}_{22}} & \cdot \cdot \cdot  & {{d}_{2n}}  \\   \cdot \cdot \cdot  & \cdot \cdot \cdot  & \cdot \cdot \cdot  & \cdot \cdot \cdot   \\   {{d}_{m1}} & {{d}_{m2}} & \cdot \cdot \cdot  & {{d}_{mn}}  \\\end{matrix} \right]\begin{matrix}   1  \\   2  \\   \cdot \cdot \cdot   \\   m  \\\end{matrix}  \\\end{matrix}$                     (1)

The calculation formula for determining the whitening weight function $I_j^e\left(f_{i j}\right)$ for color $j$ where color $f_{i j}$ is classified within group $\delta_i^e\left(\delta_i^e \subseteq(0,1]\right)$, and obtaining the clustering coefficient $\delta_i^e$ for the $i$ color scheme with respect to the $e$ semantic set can be expressed as follows:

$\delta_j^e=\sum_{j=1}^q I_j{ }^e\left(f_{i j}\right) g_j{ }^e$                         (2)

In the provided context, where A represents the clustering weight, the calculation formula for $g_j{ }^e$ can be expressed as follows:

$g_j{ }^e=\frac{\varepsilon_j^e}{\sum_{j=1}^q \varepsilon_j^e}$                    (3)

In the equation, $\varepsilon_j^e$ represents the threshold value for color j being a subclass of color e.

From the information provided, we can deduce that the clustering coefficient of the $i$ color scheme with respect to the e grayscale is denoted as $\delta_j^e$. By calculating these clustering coefficients, we obtain a vector $\delta_i=\left(\delta_1, \delta_2, \cdots, \delta_e\right)$ representing the color scheme's clustering effects on grayscale. The maximum value in this vector corresponds to the color scheme's image value.

To assign color scheme, color perception, and semantics, classifications are made based on different clustering indicators and the associated whitening values across e grayscales. This process is used for categorization. In this paper, the Grey Clustering Method is employed to associate the color sample set with grayscale image grayness, determining the gray membership function for whitening. This allows for vector calculations and categorization. A predictive model for color scheme images based on the BIM system platform is established, and a color scheme collection is constructed.

Therefore, assuming a ratio of color areas as are_i, with color area ratio Q=(are₁, are₂, … are_n), and are₁+are₂+…+ are_n, we can establish the membership function for the semantic set described by the adjective "lively-lonely" as depicted in Figure 6. Typically, there are three levels of gray, and the semantic evaluation function is as follows:

6.png

Figure 6. Graph of the affiliation function of a semantic set

From Figure 7, it is evident that the semantic set is associated with two units along the horizontal axis. When converting the membership function of this semantic set into a whitening function, the threshold value A is defined as B. The whitening weight function (C) for color scheme effects on color image semantics is illustrated in the following diagram:

7.png

Figure 7. The whitening weight function of perceptual imagery