Data Retrieval = Text Retrieval?

Abstract

Due to the comparatively more recent emergence of data retrieval systems than text-based search engines, the former have still yet to attain similar maturity in terms of standards and techniques. Most of the existing solutions for data retrieval are more or less makeshift adaptations of text retrieval systems rather than purpose-built solutions specially designed to cater to the particular peculiarities, subtleties, and unique requirements of research datasets. In this paper we probe into the key differences between text and data retrieval that bear practical relevance to the retrieval question; these differences we demonstrate by evaluating some representative examples of research data repositories as well as presenting findings from previous studies.

Identifying Design Requirements of a User-Centered Research Data Management System

“…Face-to-face interviews; address aims 1 & 2, and is reported in [32] 3. Market appraisal and review; addresses aim 1, and is reported in [17,26] 4. A technical experiment comparing data retrieval (DR) with traditional information retrieval (IR); addresses aims 2 & 3, and is reported in [16] Methods 1 and 2 above (questionnaire and interview) were conducted sequentially, the purpose of the latter being chiefly to probe further into and expand upon some of the findings and hints from the former.…”

“…In computing file sizes, we considered for text retrieval (research publications) only full research papers; and for data retrieval (research datasets), both the dataset itself and any documentation(s) it comes with. Detailed findings and discussions may be found in [16,17].…”

“…Notwithstanding this lacuna, however, various efforts among institutions and research centers towards developing RDM products have resulted in user-needs studies prior to system design and made evident attempts to accommodate those needs [1,[13][14][15]. But, though notable, the designs of these systems still leave much room for improvement [16,17].…”