We study in this paper how explicit user feedback can be used by a case-based reasoning system to improve the quality of its retrieval phase. More specifically, we explore how both ranking feedback and relevance feedback can be exploited to modify the weights of case features. We propose some options to cope with each type of feedback. We also evaluate, in an interactive setting, their impact on a travel scenario where some user provides feedback on a series of queries. Our results indicate that the combined weight-learning scheme proposed in this paper succeeds, on average, to assign more weights to the features used to formulate relevance and ranking feedback.

In cooperation with the Association for the Advancement of Artificial Intelligence (AAAI), the twenty-first International Conference on Case-Based Reasoning (ICCBR), the premier international meeting on research and applications in Case-Based Reasoning (CBR), was held in July 2013 in Saratoga Springs, NY. ICCBR is the annual meeting of the CBR community and the leading conference on this topic. This year ICCBR featured the Industry Day, the fifth annual Doctoral Consortium and three workshops.

Springs, NY. ICCBR is the annual meeting of the CBR community and the ICCBR also featured a workshop program consisting of three workshops. The main conference track featured 16 research paper presentations, nine posters, and two invited speakers. The papers and posters reflected the state of the art of case-based reasoning, dealing both with open problems at the core of CBR (especially in similarity assessment, case adaptation, and case-based maintenance), as well as trending applications of CBR (especially recommender systems and computer games) and the intersections of CBR with other areas such as multiagent systems. The first invited speaker, Igor Jurisica from the Ontario Cancer Institute and the University of Toronto, spoke about how to scale up case-based reasoning for "big data" applications. The Case-Based Reasoning in Health Sciences workshop, organized by Isabelle Bichindaritz, Cindy Marling, and Stefania Montani, and the EXPPORT workshop (Experience Reuse: Provenance, Process-Orientation and Traces), organized by David Leake, Béatrice Fuchs, Juan A. Recio Garcia, and Stefania Montani, were held jointly and dealt with how to deal with data represented CDPHP, was the local chair; William E. University, and Jonathan Rubin, from Registration information is available at www.aaai.org/Symposia/ the Palo Alto Research Center, were the Spring/ sss14.php.

Case-Bsed Reasoning (CBR) is a recent theory for problem-solving and learning in computers and people.Broadly construed it is the process of solving new problems based on the solution of similar past problems. In the present paper we introduce an absorbing Markov chain on the main steps of the CBR process.In this way we succeed in obtaining the probabilities for the above process to be in a certain step at a certain phase of the solution of the corresponding problem, and a measure for the efficiency of a CBR system. Examples are given to illustrate our results.

The condensed nearest neighbor (CNN) algorithm is a heuristic for reducing the number of prototypical points stored by a nearest neighbor classifier, while keeping the classification rule given by the reduced prototypical set consistent with the full set. I present an upper bound on the number of prototypical points accumulated by CNN. The bound originates in a bound on the number of times the decision rule is updated during training in the multiclass perceptron algorithm, and thus is independent of training set size.

In the panoply of pattern classification techniques, few enjoy the intuitive appeal and simplicity of the nearest neighbor rule: given a set of samples in some metric domain space whose value under some function is known, we estimate the function anywhere in the domain by giving the value of the nearest sample per the metric. More generally, one may use the modal value of the m nearest samples, where m is a fixed positive integer (although m=1 is known to be admissible in the sense that no larger value is asymptotically superior in terms of prediction error). The nearest neighbor rule is nonparametric and extremely general, requiring in principle only that the domain be a metric space. The classic paper on the technique, proving convergence under independent, identically-distributed (iid) sampling, is due to Cover and Hart (1967). Because taking samples is costly, there has been much research in recent years on selective sampling, in which each sample is selected from a pool of candidates ranked by a heuristic; the heuristic tries to guess which candidate would be the most "informative" sample. Lindenbaum et al. (2004) apply selective sampling to the nearest neighbor rule, but their approach sacrifices the austere generality of Cover and Hart; furthermore, their heuristic algorithm is complex and computationally expensive. Here we report recent results that enable selective sampling in the original Cover-Hart setting. Our results pose three selection heuristics and prove that their nearest neighbor rule predictions converge to the true pattern. Two of the algorithms are computationally cheap, with complexity growing linearly in the number of samples. We believe that these results constitute an important advance in the art.

Nowadays, Nearest Neighbor Search becomes more and more important when facing the challenge of big data. Traditionally, to solve this problem, researchers mainly focus on building effective data structures such as hierarchical k-means tree or using hashing methods to accelerate the query process. In this paper, we propose a novel unified approximate nearest neighbor search scheme to combine the advantages of both the effective data structure and the fast Hamming distance computation in hashing methods. In this way, the searching procedure can be further accelerated. Computational complexity analysis and extensive experiments have demonstrated the effectiveness of our proposed scheme.

In this paper we offer a potential solution to the cold-start problem in group recommender systems. To do so, we use information about previous group recommendation events and copy ratings from a user who played a similar role in some previous group event. We show that copying in this way, i.e. conditioned on groups, is superior to copying nothing and also superior to copying ratings from the most similar user known to the system.

This paper proposes an approach for the adaptation of spatial or temporal cases in a case-based reasoning system. Qualitative algebras are used as spatial and temporal knowledge representation languages. The intuition behind this adaptation approach is to apply a substitution and then repair potential inconsistencies, thanks to belief revision on qualitative algebras. A temporal example from the cooking domain is given.

Building on recent research on preference handling in artificial intelligence and related fields, our goal is to develop a coherent and generic methodological framework for case-based reasoning (CBR) on the basis of formal concepts and methods for knowledge representation and reasoning with preferences. A preference-based approach to CBR appears to be appealing for several reasons, notably because case-based experiences naturally lend themselves to representations in terms of preference or order relations. Moreover, the flexibility and expressiveness of a preference-based formalism well accommodate the uncertain and approximate nature of case-based problem solving. In this paper, we outline the basic ideas of preference-based CBR and sketch a formal framework for realizing these ideas.