Eva Armengol

samap: An user-oriented adaptive system for planning tourist visits

In this paper, we present samap, whose goal is to build a software tool to help different people visit different cities. This tool integrates modules that dynamically capture user models, determine lists of activities that can provide more utility to a user given the past experience of the system with similar users, and generates plans that can be executed by the user. This system is intended to work in portable devices (mobile phones, PDAs, etc.,) with internet connection. In this paper, we describe the architecture, the knowledge model that is shared among components using an ontology, and the three components of the tool: user module, case-based module and planning module.

Integrating Induction in a Case-Based Reasoner

This paper focuses on two key issues in building case-based reasoners (CBRs). The first issue is the knowledge engineering phase needed for CBRs as well as knowledge-based systems (KBS); the second issue is the integration of different methods of learning into CBRs. We show that we can use a knowledge modelling framework for the description and implementation of CBR systems; in particular we show how we used it in developing a CBR in the domain of protein purification. In order to encompass CBR (and learning in general) our knowledge modelling framework extends the usual frameworks with the notion of memory. Including memory we provide the capability for storing and retrieving episodes of problem solving, the basis of case-based reasoning and learning. We show here that this framework, and the supporting language NOOS, allows furthermore to integrate other learning methods as needed. Specifically, we show how a method for the induction of class prototypes can be implemented and integrated with case-based methods in an uniform framework.

Similarity Assessment for Relational CBR

Reasoning and learning from cases are based on the concept of similarity often estimated by a distance. This paper presents LAUD, a distance measure that can be used to estimate similarity among relational cases. This measure is adequate for domains where cases are best represented by relations among entities. An experimental evaluation of the accuracy of LAUD is presented for the task of classifying marine sponges.

On the Importance of Similitude: An Entropy-Based Assessment

Assessing the similarity of structured representation of cases in a natural and powerful way is an open issue in case-based reasoning (CBR). In this paper we use the notion of similitude terms, a symbolic representation of structural similarity proposed in an earlier paper. We argue that the issue to be addressed is estimating the relevance of similitude terms with regard to the task at hand, and then we propose a way of using the Case Base to estimate the relevance of similitude terms called the discriminating base. Two specific measures based on Shannon entropy are proposed to assess this relevance: I, the importance of a similitude term, and G, the similitude-based class evidence that estimates class aggregate importance. We show an application of I in the system SPIN for marine sponges identification. A longer version of this paper applies G to two standard Machine Learning datasets for classification tasks.

A knowledge level model of knowledge-based reasoning

We propose to analyze CBR systems at knowledge level following the Components of Expertise methodology. This methodology has been used for design and construction of KBS applications. We have applied it to analyze learning methods of existing systems at knowledge level. As example we develop the knowledge level analysis of CHEF. Then a common task structure of CBR systems is explained. We claim that this sort of analysis can be a first step to integrate different learning methods into case-based reasoning systems.

Remembering similitude terms in CBR

In concept learning, inductive techniques perform a global approximation to the target concept. Instead, lazy learning techniques use local approximations to form an implicit global approximation of the target concept. In this paper we present C-LID, a lazy learning technique that uses LID for generating local approximations to the target concept. LID generates local approximations in the form of similitude terms (symbolic descriptions of what is shared by 2 or more cases). C-LID caches and reuses the similitude terms generated in past cases to improve the problem solving of future problems. The outcome of C-LID (and LID) is assessed with experiments on the Toxicology dataset.

Explanation-based learning: a knowledge level analysis

This paper presents an overview of explanation-based learning (EBL) where the descriptions of EBL methods are realized at the knowledge level. Knowledge level description is a proposal that emphasizes the knowledge content and usage and abstracts away implementation details. Analyzing six different EBL methods at the knowledge level it can be shown what is their shared structure and illuminates their differences in a common framework of description. To describe EBL we use the Components of Expertise (Steels 1991), a knowledge level methodology currently used to describe, analyze, and support the development of expert systems.