Usage
  • 30 views
  • 30 downloads

Table Union Search with Preferences

  • Author / Creator
    Mirzaei, Hamed
  • We study the problem of Table Union Search (TUS) in the presence of preferences. Two tables are unionable if their column values are drawn from the same domains. This notion of unionability is too coarse to be effective in down-stream tasks.
    The result of a table search for unionability is often less relevant to the needs of users, and selecting top few is subjective and depends on the follow-up operations.

    This thesis introduces preferences for table unionability, as a way to reduce the search space and focus on rows and columns that are important for the follow-up operations. But, adding preferences introduces a few challenges to the process. Firstly, one may need extra information such as the relationship between webtables which TUS does not consider. Secondly, there is usually additional overload, which can be costly when searching a large set of webtables. We study a few approaches to address these challenges.

    We evaluate the efficiency and effectiveness of preferences on three down-stream tasks, showing that adding preferences significantly improves the performance of these tasks.

  • Subjects / Keywords
  • Graduation date
    Fall 2023
  • Type of Item
    Thesis
  • Degree
    Master of Science
  • DOI
    https://doi.org/10.7939/r3-1fyq-8064
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.