PAnoramic Learning Map (PALM)

• PAnoramic Learning Map (PALM) is a curriculum-level learning analytics dashboard that integrates cross-course data, individual learning evidence, and historical trends into a digital curriculum map.
• It employs a GIS-inspired, multilayer visualization that overlays course relationships, learner engagement, and grade outcomes to enable detailed drill-down and spatial navigation.
• Empirical evaluations show PALM significantly improves learning attitudes and perceived behavioral control, outperforming traditional systems in usability, understanding, and visual appeal.

Searching arXiv for the specified PALM paper and related acronym usage. arXiv search query: "(Ozaki et al., 24 Jul 2025) PALM PAnoramic Learning Map" PAnoramic Learning Map (PALM) is a curriculum-level learning analytics dashboard that embeds learning analytics into a curriculum map so that learners and educators can inspect cross-course relationships, individual learning evidence, historical cohort patterns, and grades within a unified visual interface. Proposed as a response to the predominance of single-course and short-term analytics, PALM is framed as a hub that connects information otherwise scattered across LMSs, syllabi, grade systems, and traditional curriculum maps, with the explicit aim of supporting self-regulated learning through planning and reflection at university scale (Ozaki et al., 24 Jul 2025).

1. Definition and rationale

PALM was introduced to address a structural limitation in conventional learning analytics: much prior work focuses on one course at a time, emphasizes short-term learning data, and does not adequately represent cross-course relationships or the long-term trajectory of learning. The underlying premise is that university learning is curricular rather than isolated, so prior knowledge acquired in one course may affect performance in later courses, yet conventional dashboards do not readily expose those dependencies (Ozaki et al., 24 Jul 2025).

The system therefore combines two levels of representation that are usually separated. At the macro level, it uses the curriculum map as the organizing structure. At the micro level, it overlays course-specific and learner-specific analytics evidence. In the paper’s terminology, PALM is both a “digitalized curriculum map” and an integration framework for curriculum-level learning analytics, rather than a visualization layer attached to a single course.

This positioning distinguishes PALM from curriculum maps in educational management that remain abstract for learners. Traditional maps show curricular structure, but not the learning evidence needed for planning, monitoring, and reflection. PALM was proposed to bridge precisely that gap by making accumulated learning evidence interpretable in relation to program structure.

2. Conceptual foundations

PALM is explicitly inspired by geographic information systems (GIS). In the GIS analogy, a base map is overlaid with multiple information layers; PALM adopts the same logic by treating the curriculum map as the base map and rendering learning analytics data as layers on top of it (Ozaki et al., 24 Jul 2025).

This layered design serves two conceptual purposes. First, it preserves the curriculum’s global structure while allowing users to inspect particular dimensions of learning activity. Second, it makes fragmented educational data commensurable by aligning them spatially on the same curriculum surface. The paper characterizes this as a panoramic view of learning: a wide-angle representation across the curriculum that still supports detail through hover-based drill-down and layer control.

The multilayer model also organizes educational information across several curriculum levels:

• Curriculum level: overall program structure, course positions, and inter-course relationships.
• Course level: course blocks with outcomes, relevance lines, and comparative indicators.
• Learner level: individual engagement data.
• Cohort / historical level: aggregate data from past learners.
• Outcome level: grades and performance-related evidence.

A plausible implication is that PALM redefines learning analytics as curriculum-aware infrastructure rather than as a local monitoring interface. That interpretation is consistent with the authors’ claim that PALM moves beyond conventional learning analytics toward a comprehensive and scalable approach.

3. System architecture and layered visualization

PALM’s interface is organized into five layers or components. The base display area presents all courses in the selected curriculum as course blocks in a two-dimensional layout resembling a traditional curriculum map; at Kyushu University, this could mean around 180 courses for a faculty or program. The map supports drag-and-drop navigation, mouse-wheel zooming, and spatial browsing (Ozaki et al., 24 Jul 2025).

The higher layers encode course relationships, learner engagement, past learners’ engagement, and grades. Hovering over a course block reveals detailed information for each learning data element in a popup.

Layer/component	Representation	Function
Layer 0: Curriculum Map Display Area	Course blocks in a two-dimensional curriculum layout	Base map for navigation and browsing
Layer 1: Course Relevance Lines	Blue lines; thickness indicates relationship strength	Visualizes inter-course relatedness
Layer 2: Individual Learning Engagement	Left half of course block shaded in blue	Shows learner’s own engagement
Layer 3: Past Learners’ Engagement	Right half of course block shaded in orange	Shows historical engagement pattern
Layer 4: Letter Grade	Pin-shaped markers; switchable among letter grade, grade point, or no display	Displays outcome information

Course relevance lines are computed from syllabus text. Specifically, course overview and lecture plan text are vectorized using TF-IDF, and course similarity is then computed using cosine similarity. In operational terms, courses with more similar syllabus descriptions are drawn as more strongly connected. The paper presents this as a way to convey prerequisite-like or topic-related structure visually (Ozaki et al., 24 Jul 2025).

Individual learning engagement is represented as the average of attendance rate, quiz score, and assignment submission. Historical engagement uses the same block structure for previous students, enabling direct comparison between an individual learner’s behavior and the pattern of past learners. Grade information is superimposed separately through pin-shaped markers, and the hover interaction functions as the system’s drill-down mechanism.

The primary data sources are LMS data for attendance, quizzes, assignment submissions, and related behavioral traces; syllabus text for relevance estimation; grade systems for grade display; and traditional curriculum maps or course catalogs for curricular structure. The paper does not provide a full algorithmic data engineering pipeline, but it specifies the conceptual sequence: extract course and student data from existing systems, process and aggregate them, align them with curriculum blocks, and visualize them as layered map elements.

4. Evaluation design and analytical methods

The system evaluation addressed two research questions: whether PALM influences students’ attitudes and intentions regarding self-regulated learning behaviors, especially study planning and reflection, and how PALM compares with existing systems as a dashboard (Ozaki et al., 24 Jul 2025).

The study involved 29 participants: undergraduate students from Electrical Engineering and Computer Science at Kyushu University and master’s students who had graduated from the same department. The procedure had three steps: pre-questionnaire, use of PALM, and post-questionnaire. The full survey was completed within two weeks and had ethical approval.

For effectiveness evaluation, the study used the Theory of Planned Behavior (TPB) with 16 items on a 7-point Likert scale. The four TPB factors were intention, attitude, subjective norm, and perceived behavioral control. Responses were averaged per factor for each participant and compared before and after using the system.

For system evaluation, PALM was compared against an existing-system baseline comprising LMS, syllabus system, grade inquiry systems, traditional curriculum maps in spreadsheet format, and course catalogs. The survey used 28 items derived from the LADS framework, covering visual attraction, usability, understanding level, perceived usefulness, and behavioral changes. Qualitative feedback was also collected regarding the intuitiveness of course relevance lines, impressions of the curriculum map and course blocks, engagement visualizations, and requested additional features.

The statistical analysis used the Shapiro-Wilk test for normality, the paired t-test for pre/post comparisons, and the Wilcoxon signed-rank test when normality was violated. All analyses were carried out using scipy.stats in Python. For paired comparisons, the paper reports the effect size $d_D$ as a standardized mean difference for paired data, $d_D = \overline{D}/s_D$, with $\overline{D}$ the mean of the paired differences and $s_D$ their standard deviation.

5. Empirical findings

The TPB analysis showed statistically significant increases in all four factors after using PALM. Reported means and effect sizes were as follows: intention increased from 5.1 to 5.7 with $d_D = 0.77$; attitude from 5.0 to 5.6 with $d_D = 0.80$; subjective norm from 3.9 to 4.4 with $d_D = 0.58$; and behavioral control from 4.1 to 5.3 with $d_D = 1.21$ (Ozaki et al., 24 Jul 2025).

The largest effect was observed for perceived behavioral control. The authors interpret this as indicating that PALM clarifies the links between learning actions and outcomes through the visual presentation of individual learning histories and statistical trends, thereby making learners feel more able to control how they study. In practical terms, the paper associates PALM with improved awareness of one’s own learning history, study planning, and reflection.

The system comparison against existing tools also favored PALM on all LADS dimensions, again with large effect sizes. Visual attraction increased from 3.4 for existing systems to 6.0 for PALM with $d_D = 2.40$; usability from 3.7 to 5.9 with $d_D = 1.84$; understanding level from 3.8 to 6.1 with $d_D = \overline{D}/s_D$0; perceived usefulness from 4.0 to 5.7 with $d_D = \overline{D}/s_D$1; and behavioral changes from 3.7 to 5.3 with $d_D = \overline{D}/s_D$2. The reported pattern indicates that participants judged PALM substantially more attractive, understandable, and useful than the disconnected system ensemble they normally used.

Qualitative responses were favorable but not uniform. Excluding two “I don’t know” responses, 89% of respondents judged the course relevance lines “very intuitive” or “somewhat intuitive.” Positive comments emphasized the visibility of recommended prerequisite courses, relationships already perceived by students from taking courses, improved readability relative to spreadsheet curriculum maps, intuitive zooming, color intensity for comparison, and the ability to link grades to behaviors such as attendance. Critiques concerned excessive blue in the display, unclear indicator meanings, and a desire for grade distribution or percentile information. Some relevance links were perceived as counterintuitive, including a connection between Compiler and Mathematics for Electrical Engineering.

6. Distinctiveness, limitations, and nomenclature

PALM differs from conventional learning analytics dashboards primarily in scope and structural anchoring. Conventional dashboards typically focus on a single course, display current or recent learning activity, and remain disconnected from curriculum design. PALM is instead curriculum-wide, visually links multiple courses, overlays individual and historical data, and embeds analytics within a curriculum map. The authors therefore present it not simply as a monitoring tool for one class but as cross-course learning infrastructure (Ozaki et al., 24 Jul 2025).

A common misconception, explicitly addressed in the paper, is to treat PALM as merely a visually improved dashboard. The authors reject that characterization by emphasizing that PALM is an integration framework: its significance lies in aligning heterogeneous educational information on a common curricular substrate. This suggests that the principal contribution is architectural as much as graphical.

The paper also identifies several limitations. PALM does not yet support dynamic data updates, which complicates large-scale practical deployment. Subjective norm improved less than the other TPB factors, suggesting a need for better peer-level information. Users requested supplementary guidance for interpretation and use, as well as more personalized presentation. The role of instructors in making use of PALM remains unexplored, and long-term effects across broader populations, semesters, and cohorts are not yet established.

Within the broader literature, the acronym “PALM” is not unique. In 2025, arXiv also published “To Label or Not to Label: PALM — A Predictive Model for Evaluating Sample Efficiency in Active Learning Models,” where PALM denotes “Performance Analysis of Active Learning Models,” a parametric model for active learning trajectories rather than a curriculum-level dashboard (Machnio et al., 21 Jul 2025). The shared acronym is nominal rather than conceptual; the two works address different domains, methods, and evaluation problems.

Overall, PAnoramic Learning Map is best understood as a curriculum-aware learning analytics framework that turns the curriculum map into an analytics hub. Its central claim is that learning analytics intended to support self-regulated learning at university scale must represent long-term curricular progression, not only short-term activity within isolated courses. The reported evaluation indicates that such curriculum-level integration can improve awareness of study planning and reflection, especially through increased perceived behavioral control, while also outperforming existing disconnected systems in visual attraction, usability, understanding, usefulness, and reported behavioral change.