Governor: Turning Open Government Data Portals into Interactive Databases

Interactive Data VisualizationKnowledge Management & Team AwarenessData Scientists & AnalystsGovernment Officials & Civil ServantsHCI Researchers

Title of the Paper

Governor: Turning Open Government Data Portals into Interactive Databases

Paper Information

  • Subject Area: Data Management and Open Government Data
  • Keywords: Open Data, Data Integration, Databases, User Interface, Data Visualization, Data Provenance

Research Background and Problem Statement

  • Problems or Challenges:

    • Current Open Government Data Portals (OGDPs) have limited functionality, primarily offering search capabilities for dataset titles and descriptions. Users must download files and use third-party tools to analyze the data.
    • Non-programmer users face difficulties in efficiently finding and utilizing open data, and the data integration process is cumbersome.
    • Existing OGDPs fail to meet advanced user needs for data integration and data provenance, requiring users to independently manage the sources of raw data, which is error-prone and challenging.
  • Significance of the Problem:

    • Open data aims to enhance government transparency, enabling citizens, researchers, and journalists to analyze data and monitor national policies and societal performance.
    • Addressing the limitations of OGDPs could provide significant societal value, such as identifying inefficiencies in government operations and uncovering corruption.
  • Research Motivation and Related Work:

    • Existing OGDP search functions rely on metadata accuracy, and poor metadata quality can hinder users from discovering relevant data.
    • Dataset search engines (e.g., Google Dataset Search) and existing interactive data integration systems (e.g., Voyager, Ronin) fail to meet the non-programming needs of OGDP users and lack support for union, join operations, and data provenance.
    • No system currently provides non-technical users with database management system (DBMS) functionalities to effectively unlock the potential of open data integration.

Solution

  • Proposed Method or Solution:

    • Introduced a browser-based open-source system named Governor, which models OGDPs as interactive databases.
    • By extending existing OGDP functionalities, Governor supports record-level search within tables and direct data table previews without requiring file downloads.
    • Provides intelligent recommendations, identifying and suggesting unionable and joinable tables for data integration.
    • Offers a color-coded visualization interface, enabling users to interactively manage data integration sources and trace back to the original tables in OGDPs.
  • Innovations:

    • Introduced core DBMS functionalities to OGDPs, such as record-level search, data integration, and provenance management.
    • Designed an interactive system with a low technical barrier, addressing the pain points of non-programmer users.
    • Provided clear visualization and provenance features, allowing users to quickly understand the composition and sources of integrated tables.
  • Implementation Steps and Key Technologies:

    1. Search Functionality:
      • Offers two modes: metadata search and record-level search.
      • Handles search requests using MongoDB and Elasticsearch.
    2. Data Integration:
      • Extends table rows via union operations or adds new columns via join operations.
      • Performs interactive SQL processing of integrated tables in the browser using DuckDB-Wasm.
    3. Data Provenance and Color Coding:
      • Uses color coding to indicate the sources of merged or joined data in integrated tables.
      • Allows users to trace value origins back to source columns and original tables via a visual interface.
    4. System Architecture:
      • Fetches OGDP metadata and table data via the CKAN API.
      • Compresses data using the Parquet format to reduce transmission load.
      • Supports integration with various government data portals.

Research Outcomes

  • Specific Achievements:

    • Designed and implemented the Governor system, equipping OGDPs with interactive DBMS functionalities.
    • Evaluated the system through two user studies, demonstrating its efficiency and user-friendliness in data search and integration tasks.
  • Advantages over Existing Solutions:

    • Compared to existing tools like Google Dataset Search, Auctus, and TODS, Governor supports deep record-level searches and direct multi-table integration.
    • Automated integration recommendations significantly reduce the time and error rates associated with manual downloading, organizing, and merging of tables.
    • The visual data provenance feature greatly improves the efficiency of data auditing and verification.
  • Experimental or Evaluation Results:

    • In a controlled task and an open-ended task user study:
      • Most users successfully completed the tasks with good performance in terms of time and accuracy.
      • NASA TLX questionnaire results indicated low workload perception and high task success rates when using Governor.
      • Users generally found Governor easy to learn, intuitive, and powerful, particularly in table integration and data provenance functionalities.
    • Compared to existing tools, Governor provided a superior user experience, especially for non-programmer users, who showed high acceptance of the system.
  • Limitations and Future Directions:

    • Limited filtering and data cleaning functionalities; users expressed a need for additional features like column-based filtering and data aggregation.
    • Current data integration is restricted to recommended tables; future work could allow users to freely select tables for integration.
    • Academic validation did not include direct baseline comparisons (e.g., traditional OGDP tools); further research in real-world scenarios is needed.

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/papers/chi/96383/2023

AdRecommended

Learn AI Coding at CodeNow

open_in_newOpen DOI Link
DOI: https://doi.org/10.1145/3544548.3580868
At a Glance

Paper Snapshot

fact_check
dataset
Source
CHI
calendar_month
Year
2023
emoji_events
Award
No award tagged
group
Authors
4 authors
sell
Subtopics
Interactive Data Visualization, Knowledge Management & Team Awareness
work
Professions
Data Scientists & Analysts, Government Officials & Civil Servants, HCI Researchers
article
Content Status
Full text indexed
hub
Related Papers
1 related papers