A task completion framework to support single-interaction IR research
Authors:
Abstract
Purpose: Within Information Retrieval (IR), most research is directed towards multi-document retrieval and a multi-interaction IR (miIR) user scenario. There are few, if any, IR conceptual models supporting minimal or single-interaction IR (siIR) user scenarios, however the need for siIR systems is growing rapidly. This paper takes the first steps towards constructing a task-oriented conceptual model and experimental framework to support siIR research.
Approach: A first principles approach is employed to develop a task-oriented conceptual model, called Bridging Information Retrieval (BIR). This model is contrasted with the concept of Relevance, a central factor within IR research.
Findings: BIR introduces the central concept of Bridging Information (BI) as the objective of IR systems. BI is the additional information a user requires to complete a task, beyond their innate knowledge. The relationship between BI and relevance is determined.
Research Limitations: The theoretical basis of BIR is derived axiomatically, however the resulting system evaluation model is speculative.
Practical Implications: The proposed operational framework offers researchers a systematic approach to designing and evaluating siIR systems.
Originality: This work contributes a novel task-oriented IR conceptual model and evaluation framework, both centred around the concept of BI for siIR. It also contributes a novel search task classification method.