TheoryOn: Designing a Construct-Based Search Engine to Reduce Information Overload for Behavioral Science Research

Abstract

The accumulated literature base in the behavioral sciences represents a great source of knowledge on human behaviors, and yet the same literature has grown beyond human comprehension. We address this information overload problem by proposing a novel IT artifact --- TheoryOn. Based on the design science paradigm, we identify five design requirements. We first adapt the ontology learning layer cake framework to develop a four-step process --- hypothesis extraction, construct extraction, construct relationship extraction and theory extraction --- to automatically extract integral parts of behavioral theories. We then design four functionalities allowing researchers to quickly access synonymous constructs, construct relationships and theoretically related constructs e.g. antecedents and consequents, as well as integrate related theories. To illustrate the applicability and usefulness, we use a dataset of all the relevant behavioral studies from three top journals in Information Systems and Psychology and conduct an A/B test between the prototype TheoryOn system and the EBSCOhost full-text search engine.