Deconstructing search tasks in interactive information retrieval: A systematic review of task dimensions and predictors

摘要

Increasingly, users are interacting with information retrieval (IR) systems with the goal of addressing complex, ill-defined search tasks. As a result, the interest in deconstructing and predicting search tasks has grown among IR researchers. Given the multidimensionality of search tasks, researchers usually focus on one or multiple dimensions and study the associations between implicit task dimensions and observable predictors. Synthesizing the knowledge of tasks and predictors learned from individual user studies can clarify the progresses we have made as a research community and provide an intellectual benchmark for further explorations of task-based search interactions. This article presents an overview of 76 task-based interactive IR (IIR) studies published between 2000 and 2020 and systematically coded the papers using features such as task dimensions (definitions and operationalizations), task-predictor associations, as well as task prediction models. Results include 1) data illustrating the growth and interdisciplinarity of IIR studies; 2) a comprehensive typology of task dimensions along with the associated measures; 3) a summary of the statistically significant correlations between task dimensions and predictors; 4) a list of the task dimensions being predicted, ground truth labels, and the feature sets employed. In addition, our bibliography of IIR works can be used by students and junior researchers who are new to the area. The results of our review can facilitate the growth of knowledge in IIR community and serve as the basis for future research on new modalities of user-task interactions.