Staring at Tables: Exploring Conceptual Data Modeling as a Rich Collaborative Activity

Paper Title

Staring at Tables: Exploring Conceptual Data Modeling as a Rich Collaborative Activity

Publication Info

• Topic area: Collaborative conceptual data modeling and its dynamics in human-data interaction.
• Keywords: Conceptual data modeling, collaboration, sensemaking, human-data interaction, data literacy, mental models, negotiation, storytelling, tool design, data visualization.

Background and Problem

• Problem / challenge: Conceptual data modeling (CDM) is a critical but understudied activity, especially in collaborative contexts. Existing research often focuses on individual modeling or simplified datasets, neglecting the situated, iterative, and communicative nature of collaborative CDM.
• Significance: Understanding collaborative CDM is crucial as it shapes how data is structured, interpreted, and used across professional and everyday contexts, impacting data literacy and decision-making.
• Motivation and related work: Prior work has explored individual data modeling, data-centric sensemaking, and collaborative data work but lacks empirical accounts of how people collaboratively construct and align conceptual models. This study addresses this gap by examining how pairs of participants negotiate and create shared conceptual data models.

Solution

• Proposed approach: A collaborative sensemaking study where 22 participants (11 pairs) worked together to construct conceptual data models from a structured dataset, using digital tools and think-aloud protocols.
• Novelty:
  1. Empirical insights into how pairs collaboratively construct conceptual data models, focusing on dialogue, tool use, and expertise.
  2. Identification of collaboration practices, challenges, and dynamics in reaching common ground during CDM.
  3. Narrative participant profiles illustrating differences in collaborative modeling processes.
• Procedure and key techniques:
  • Participants explored a denormalized dataset on long-distance walking routes.
  • They used spreadsheet tools for data exploration and diagramming tools for visualizing models.
  • Sessions were recorded (audio and screen), transcribed, and analyzed using thematic and visual methods.
  • Coding focused on conceptual modeling activities, collaboration practices, and tool use.

Results

• Concrete findings:
  • Participants relied on hypotheses, assumptions, and prior knowledge (e.g., analogies, stories) to navigate the task.
  • Milestones in modeling included identifying duplicates, relationships, and structural patterns.
  • Affirmation and compounding insights were key collaborative mechanisms.
  • Tool use varied by experience, with advanced users employing strategic filtering and visualization early, while novices relied on trial-and-error and external searches.
• Advantage over baselines:
  • Demonstrated how collaboration enhances CDM by surfacing tacit mental models, enabling negotiation, and fostering alignment.
  • Highlighted the role of storytelling and metacognition in overcoming task complexity.
• Experiments / evaluation:
  • Conducted in Austria and the Netherlands across four labs.
  • Participants represented diverse expertise levels (basic to advanced) and used tools like Google Sheets, Miro, and draw.io.
  • Analysis included coding transcripts, visualizing activity timelines, and examining co-occurring codes for modeling and collaboration.
• Limitations and future work:
  • Limited generalizability due to participant demographics (mostly students and academics) and short study duration.
  • Variability in task engagement and tool familiarity influenced outcomes.
  • Future work could explore group dynamics beyond pairs, extend task duration, and develop tools for collaborative CDM.

Summary

This study investigates collaborative conceptual data modeling (CDM) as a socially and communicatively rich activity. By analyzing how 11 pairs of participants constructed shared models, the research identifies key dynamics such as affirmation, storytelling, and negotiation. Results show that collaboration enhances CDM by aligning mental models and enabling iterative refinement, with implications for tool design and data literacy education. Future tools should support dialogue, visualization, and iterative exploration to facilitate collaborative sensemaking in data modeling tasks.