<p>Visualizing multidimensional spatial data is an essential visual analysis strategy, it helps us interpret and communicate how different variables correlate to geographical information. In this study, we proposed an abstract contextual visualization that encodes data on the boundaries of spatial distributions and developed a new algorithm, AuroraMap. AuroraMap projects the spatial data to the boundaries of the distributions and color-encodes the densities continuously. We further conducted the user experiments, and the results show users can detect the relative locations and scopes of the clusters. Furthermore, users can quantitatively determine the peak value of each cluster’s density. The method provides three contributions: (1) freeing up and saving the graphical visualization space; (2) assisting the users to quantitatively estimate the clusters inside distributions; (3) facilitating the visual comparisons for multiple and multivariate spatial distributions. In the end, we demonstrated two applications with real-world religious infrastructural data by AuroraMap to visualize geospatial data within complex boundaries and compare multiple variables in one graph.</p><p> </p>

Enabling human understanding of high-dimensional (HD) data is critical for scientific research but highly challenging. To deal with large datasets, probabilistic-based non-linear DR models, like UMAP and t-SNE, lead the performance on reducing the high dimensionality. However, considering the trade-off between global and local structure preservation and the randomness initialized for computation, applying non-linear models in different parameter settings to unknown high-dimensional structure data may return different 2D visual forms. Much critical neighborhood relationship may be falsely imposed, and uncertainty may be introduced into the low-dimensional embedding visualizations, so-called distortion. In this work, a survey has been conducted to illustrate the most state-of-the-art layout enrichment works for interpreting dimensionality reduction methods and results. Responding to the lack of visual interpretation techniques to probabilistic-based DR methods, we propose a visualization technique called ManiGraph, which facilitates users to explore multi-view 2D embeddings via mesoscopic structure graphs. A dynamic mesoscopic structure first subsets HD data by a hexagonal grid in visual space from non-linear embedding (e.g., UMAP). Then, it measures the regional adapted trustworthiness/continuity and visualizes the restored missing and highlighted false connections between subsets from high-dimensional space to the low-dimensional in a node-linkage manner. The visualization helps users understand and interpret the distortion from both visualization and model stages. We further demonstrate the user cases tested on intuitive 3D toy datasets, fashion-MNIST, and single-cell RNA sequencing with domain experts in unsupervised scenarios. This work will potentially benefit the data science community, from toolkit users to DR algorithm developers.<br>

<p>External representations (ERs) used in science education are multimodal ensembles consisting of design elements to convey educational meanings to the audience. As an example of a dynamic ER, an animation presenting its content features (i.e., scientific concepts) via varying the feature’s depiction over time. A production team invited the dissertation author to inspect their creation of a biochemistry animation about the role of the actin cytoskeleton in cell motility and the animation’s implication on learning. To address this, the author developed a four-step methodology entitled the Multimodal Variation Analysis of Dynamic External Representations (MVADER) that deconstructs the animation’s content and design to inspect how each content feature is conveyed via the animation’s design elements.</p><p><br></p><p> </p><p>This dissertation research investigated the actin animation’s educational value and the MVADER’s utility in animation evaluation. The research design was guided by descriptive case study methodology and an integrated framework consisting of the variation theory, multimodal analysis, and visual analytics. As stated above, the animation was analyzed using MVADER. The development of the actin animation and the content features the production team members intended to convey via the animation were studied by analyzing the communication records between the members, observing the team meetings, and interviewing the members individually. Furthermore, students’ learning experiences from watching the animation were examined via semi-structured interviews coupled with post- storyboarding. Moreover, the instructions of MVADER and its applications in studying the actin animation were reviewed to determine the MVADER’s usefulness as an animation evaluation tool.</p><p><br></p><p> </p><p>Findings of this research indicate that the three educators in the production team intended the actin animation to convey forty-three content features to the undergraduate biology students. At least 50% of the student who participated in this thesis learned thirty-five of these forty-three (> 80%) features. Evidence suggests that the animation’s effectiveness to convey its features was associated with the features’ depiction time, the number of identified design elements applied to depict the features, and the features’ variation of depiction over time.</p><p><br></p><p>Additionally, one-third of the student participants made similar mistakes regarding two content features after watching the actin animation: the F-actin elongation and the F-actin crosslink structure in lamellipodia. The analysis reveals the animation’s potential design flaws that might have contributed to these common misconceptions. Furthermore, two disruptors to the creation process and the educational value of the actin animation were identified: the vagueness of the learning goals and the designer’s placement of the animation’s beauty over its reach to the learning goals. The vagueness of the learning goals hampered the narration scripting process. On the other hand, the designer’s prioritization of the animation’s aesthetic led to the inclusion of a “beauty shot” in the animation that caused students’ confusion.</p><p><br></p><p> </p><p>MVADER was used to examine the content, design, and their relationships in the actin animation at multiple aspects and granularities. The result of MVADER was compared with the students’ learning outcomes from watching the animation to identify the characteristics of content’s depiction that were constructive and disruptive to learning. These findings led to several practical recommendations to teach using the actin animation and create educational ERs.</p><p><br></p><p> </p><p>To conclude, this dissertation discloses the connections between the creation process, the content and design, and the educational implication of a biochemistry animation. It also introduces MVADER as a novel ER analysis tool to the education research and visualization communities. MVADER can be applied in various formats of static and dynamic ERs and beyond the disciplines of biology and chemistry.</p>