Analogical Reuse Research Articles

Reward-predictive representations generalize across tasks in reinforcement learning.

In computer science, reinforcement learning is a powerful framework with which artificial agents can learn to maximize their performance for any given Markov decision process (MDP). Advances over the last decade, in combination with deep neural networks, have enjoyed performance advantages over humans in many difficult task settings. However, such frameworks perform far less favorably when evaluated in their ability to generalize or transfer representations across different tasks. Existing algorithms that facilitate transfer typically are limited to cases in which the transition function or the optimal policy is portable to new contexts, but achieving "deep transfer" characteristic of human behavior has been elusive. Such transfer typically requires discovery of abstractions that permit analogical reuse of previously learned representations to superficially distinct tasks. Here, we demonstrate that abstractions that minimize error in predictions of reward outcomes generalize across tasks with different transition and reward functions. Such reward-predictive representations compress the state space of a task into a lower dimensional representation by combining states that are equivalent in terms of both the transition and reward functions. Because only state equivalences are considered, the resulting state representation is not tied to the transition and reward functions themselves and thus generalizes across tasks with different reward and transition functions. These results contrast with those using abstractions that myopically maximize reward in any given MDP and motivate further experiments in humans and animals to investigate if neural and cognitive systems involved in state representation perform abstractions that facilitate such equivalence relations.

Elaborating analogies from conceptual models

This article defines and analyzes a computational model of similarity which detects analogies between objects based on conceptual descriptions of them, constructed from classification, generalization relations, and attributes. Analogies are detected (elaborated by functions which measure conceptual distances between objects with respect to these semantic modeling abstractions. The model is domain independent and operational upon objects described in nonuniform ways. It does not require any special forms of knowledge for identifying analogies and distinguishes the importance of distinct object elements. Also, it has a polynomial complexity. Due to these characteristics, it may be used in complex tasks involving intra- or interdomain analogical reasoning. So far the similarity model has been applied in the domain of software engineering. First, to support the specification of software requirements by analogical reuse and second, to enable the integration of requirements specifications, generated by the multiple agents involved in information system development. Details of these applications can be found in cited references. Also, we have conducted an empirical evaluation of: (i) the consistency of the estimates generated by the model against human intuition about similarity and (ii) its recall performance in tasks of analogical retrieval, the results of which are presented in this article.

Elaborating analogies from conceptual models

This article defines and analyzes a computational model of similarity which detects analogies between objects based on conceptual descriptions of them, constructed from classification, generalization relations, and attributes. Analogies are detected (elaborated) by functions which measure conceptual distances between objects with respect to these semantic modeling abstractions. The model is domain independent and operational upon objects described in nonuniform ways. It does not require any special forms of knowledge for identifying analogies and distinguishes the importance of distinct object elements. Also, it has a polynomial complexity. Due to these characteristics, it may be used in complex tasks involving intra- or interdomain analogical reasoning. So far the similarity model has been applied in the domain of software engineering. First, to support the specification of software requirements by analogical reuse and second, to enable the integration of requirements specifications, generated by the multiple agents involved in information system development. Details of these applications can be found in cited references. Also, we have conducted an empirical evaluation of: (i) the consistency of the estimates generated by the model against human intuition about similarity and (ii) its recall performance in tasks of analogical retrieval, the results of which are presented in this article. © 1996 John Wiley & Sons, Inc.

Analogical reuse of requirements specifications: A computational model

Specifications of requirements for new software systems can be revised, refined, or completed in reference to specifications of requirements for existing similar systems. Although realized as a form of analogical problem solving, specification by reuse is not adequately supported by available computational models for detecting analogies. This is chiefly due to the following reasons: (1) It is assumed that specifications are expressed according to the same specification model and in a uniform representation scheme. (2) Additional information is needed for the detection of analogies, which is not contained in the specifications. (3) Performance scales poorly with the complexity of specifications. This article presents a computational model for detecting analogies, which addresses these issues to a certain extent. The application of the model in the specification of requirements by analogical reuse is demonstrated through an example, and its sensitivity to the representation of specifications is discussed. F...

An approach to the classification of domain models in support of analogical reuse

This paper presents an approach to classify domain models in order to facilitate reuse through analogy. Domain analysis plays a critical role for systematic reuse, but domain analysis is difficult to perform, especially for new application areas. Analogical approach to reuse can support the domain analysis process by providing software products in a different but analogous domain. In order to achieve this goal, domain models need to be classified. This paper proposes a classification method for domain models. The method is an integration of the enumerative hierarchy and faceted scheme. The classification approach can help the domain analyst to locate an analogous domain to perform the modeling and analysis process. Moreover, the approach is more flexible and more descriptive than conventional classification methods.

Analogical software reuse: Empirical investigations of analogy-based reuse and software engineering practices

This paper presents three empirical studies of software engineering behaviour during systems analytic and analogical reuse tasks. The motivation behind these studies was to inform the design of support tools for the early stages of software development. A first study investigated inexperienced software engineers during a systems analytic task and revealed that they encountered considerable difficulties. The second and third studies investigated analogical software reuse as one means of overcoming these difficulties. The second study involved an experimental investigation of the effectiveness of specification reuse on analytic performance. Reuse proved beneficial, although it appeared to lead to mental laziness manifest as specification copying. In contrast, a third study investigated analogical specification reuse by expert software engineers, to determine how successful analogical comprehension and reuse may best be achieved. Implications of findings from all three studies for the design of support tools during software development are reported.