Revisiting Fast and Slow Thinking in Case-Based Reasoning
A dichotomous Case-Based Reasoning (CBR) model is one in which two kinds of reasoning mechanisms are employed; these may be for realizing fast and slow problem-solving as demanded by the nature of the incoming query. Such dichotomous operation is inspired by Daniel Kahneman’s seminal work on the two modes of thinking observed in humans. In this paper, we present the following three directions of refinement for a dichotomous CBR model: selection of attributes for a fast thinking model based on parsimonious CBR, switching from fast to slow thinking based on constraints derived from domain knowledge and arriving at a complexity measure for evaluating dichotomous models. For all the three improvements identified, we discuss the results on real-world data sets and empirically analyse the effectiveness of the same.
Towards Richer Realizations of Holographic CBR
Holographic Case-Based Reasoning is a framework developed to build cognitively appealing case-based reasoners with proactive and interconnected cases. Improved realizations of the Holographic CBR framework are developed using the principles of dynamic memory proposed by Roger Schank and tested on their cognitive appeal, efficiency, and solution quality compared to other relevant systems.
Holographic Case-Based Reasoning
In this paper, we present a novel extension of CBR that allows cases to be more proactive at problem solving, by enriching case representations and facilitating richer interconnectedness between cases. We empirically study the improvements resulting from a holographic realization on experimental datasets. In addition to making CBR more cognitively appealing, the idea has the potential to lend itself as an elegant general CBR formalism of which diverse realizations of CBR can be viewed as instances.