Information Retrieval Evaluation Experiments
Year: 2024
Type: article
Abstract: Information retrieval (IR) analysis has come a long way since 1953 because to innovative studies such as the MEDLARS project and the Cranfield trials. These showed the inverse link between recall and precision and provided critical performance measures including recall, precision, and dropout. They were unable to offer an explanation for the behavior of retrieval systems. Spark Jones identified five important areas in his early research on both manual and automatic indexing: (1) language comparisons for indexing, (2) comprehensiveness and specificity of indexing, (3) search strategies, (4) output ranking, term weighting, and relevance feedback, and (5) cost-effectiveness. This expanded the scope of IR evaluation above indexing. Despite developments, the majority of studies were limited in their actual relevance by their reliance on controlled laboratory environments and document replacements such as abstracts. The ongoing discussions on evaluation techniques, indicators of success, and relevance evaluations imply that IR evaluation remains a growing field that has to be further developed for practical applications.
Author: Karmakar, Riya
Institution: Jadavpur University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6949564118
FWCI:
Cites:
Cited by:
Related to: