Understanding Relevance Judgment in Users' Information Retrieval Behaviour: A Study Based on Library Logs and the MURM

摘要

A new structure that uses latent variables from library logs to indirectly measure the factors in behavior of relevance judgment in Multidimensional User Relevance Model (MURM) is put forward in our research. Our study aims to design quantitative variables to describe the MURM factors for the academic search, especially better interpreting some cognitive concepts of MURM in academic search tasks. The improvement in ranking performance is used as a criterion in our experiment to evaluate the quality of our designed measurement. The results show that our designed variables are effective in measuring the dimensions of MURM, which can be well used in relevance judgment.