MARVIN: A Web-Based System for Representing, Retrieving, and Visualizing Analogies

Analogies are essential in human cognition, reasoning, learning, communication, and problem solving. They can have a profound and broad effect on how we view and understand our world. In this paper we discuss the design, implementation, and evaluation of MARVIN (Markup for Analogy Representation and Visualization for the InterNet), a Web-based system for representing, retrieving, and visualizing human-conceived analogies that provides a medium and a common language for analogy practitioners to share their analogies. We developed a compact XML content model for analogy expressions for use in Web-based environments, and show that the model is capable of representing a wide range of human-conceived analogies. We demonstrate, using XSLT, several example methods for visualizing analogy expressions that use our model. We demonstrate methods for storing and retrieving such expressions and for ranking the retrieved expressions. We designed and implemented MARVIN to demonstrate these methods. A formative evaluation of the MARVIN system found that its visualization and retrieval capabilities are of value to analogy authors and end users. A performance test showed that MARVIN's analogy retrieval is scalable to large analogy archives.