Catálogo de publicaciones - libros

Qualitative spatial and temporal calculi are usually formulated on a particular level of granularity and with a particular domain of spatial or temporal entities. If the granularity or the domain of an existing calculus doesn’t match the requirements of an application, it is either possible to express all information using the given calculus or to customize the calculus. In this paper we distinguish the possible ways of customizing a spatial and temporal calculus and analyze when and how computational properties can be inherited from the original calculus. We present different algorithms for customizing calculi and proof techniques for analyzing their computational properties. We demonstrate our algorithms and techniques on the Interval Algebra for which we obtain some interesting results and observations. We close our paper with results from an empirical analysis which shows that customizing a calculus can lead to a considerably better reasoning performance than using the non-customized calculus.

In this paper we propose a modified and improved method, called . is a popular feature selection algorithm proposed by Kira and Rendell in 1992. Although compared to many other feature selection methods or its extensions are found to be superior, in this paper we show that it can be further improved. In , in the main loop, a number of instances are randomly selected using simple random sampling (), and for each of these selected instances, the nearest hit and miss are determined, and these are used to assign ranks to the features. fails to represent the whole dataset properly when the sampling ratio is small (i.e., when the data is large), and/or when data is noisy. In we use an efficient method to select instances. The proposed method is based on the idea that a sample has similar distribution to that of the whole. We approximate the data distribution by the frequencies of attribute-values. Experimental comparison with shows that performs significantly better particularly for large and/or noisy domain.

The use of association rule mining carries the attendant challenge of focusing on appropriate data subsets so as to reduce the volume of association rules produced. The intent is to heuristically identify “interesting” rules more efficiently, from less data. This challenge is similar to that of identifying “high-value” attributes within the more general framework of machine learning, where early identification of key attributes can profoundly influence the learning outcome. In developing heuristics for improving the focus of association rule mining, there is also the question of where in the overall process such heuristics are applied. For example, many such focusing methods have been applied the generation of a large number of rules, providing a kind of ranking or filtering. An alternative is to constrain the input data earlier in the data mining process, in an attempt to deploy heuristics in advance, and hope that early resource savings provide similar or even better mining results. In this paper we consider possible improvements to the problem of achieving focus in web mining, by investigating both the articulation and deployment of rule constraints to help attain analysis convergence and reduce computational resource requirements.

In general, document classification research focuses on the automated placement of unseen documents into pre-defined categories. This is regarded as one core technical component of knowledge management systems, because it can support to handle explicit knowledge more systematically and improve knowledge sharing among the users. Document classification in knowledge management systems should support incremental knowledge acquisition and maintenance because of the dynamic knowledge changes involved. We propose the MCRDR document classifier as an incremental and maintainable document classification solution. Even though our system successfully supported personal level document classification, we did not examine its capability as a document classification tool in multi-user based knowledge management contexts. This paper focuses on the analysis of document classification results performed by multiple users. Our analysis reveals that even though the same documents and the classification structure are given to the users, they have very different document classification patterns and different acceptance results for each other’s classification results. Furthermore, our results show that the integration of multiple users’ classification may improve document classification performance in the knowledge management context.

Semantic orientation (SO) for texts is often determined on the basis of the positive or negative polarity, or sentiment, found in the text. Polarity is typically extracted using the positive and negative words in the text, with a particular focus on adjectives, since they convey a high degree of opinion. Not all adjectives are created equal, however. Adjectives found in certain parts of the text, and adjectives that refer to particular aspects of what is being evaluated have more significance for the overall sentiment of the text. To capitalize upon this, we weigh adjectives according to their relevance and create three measures of SO: a baseline SO using all adjectives (no restriction); SO using adjectives found in on-topic sentences as determined by a decision-tree classifier; and SO using adjectives in the nuclei of sentences extracted from a high-level discourse parse of the text. In both cases of restricting adjectives based on relevance, performance is comparable to current results in automated SO extraction. Improvements in the decision classifier and discourse parser will likely cause this result to surpass current benchmarks.

Few years old children lift and manipulate unfamiliar objects more dexterously than today’s robots. Therefore, it has arisen an interest at the artificial intelligence community to look for inspiration on neurophysiological studies to design better models for the robots. In a human dexterous manipulation a crucial event is the detection of incipient slips. Humans detect the incipient slips based on the responses of their tactile mechanoreceptors. In this paper, we propose a method to detect the incipient slips using artificial neural networks that receive as input simulated human afferent responses. This method is strongly inspired on neurophysiological studies of the afferent responses during the human dexterous manipulation. Finite element analysis was used to model two fingers and an object, and simulated experiments using the proposed method were done. To the best of our knowledge, this is the first time that simulated human afferent signals are combined with finite element analysis and artificial neural networks, to detect the incipient slips.

A prerequisite of any successful social robot is the ability to express its internal state and intention to humans in a natural way. Many researchers studied verbal and nonverbal feedback from humanoid robots or humanoid robotic heads but there is little research about the possible feedback mechanisms of non-humanoid and especially miniature robots. The TalkBack experiment is a trial to fill this gap by investigating the effectiveness of using motion cues as a feedback mechanism and comparing it to verbal feedback. The results of the experiment showed that there is no significant difference in the task completion accuracy and time or in the feeling of naturalness between these two modalities and there is a statistically significant improvement when using any of them compared with the no-feedback case. Moreover the subjects selected the motion cues feedback mechanism more frequently as the preferred feedback modality for them.

Visitors to cultural heritage sites are often overwhelmed by the information available in the space they are exploring. The challenge is to find items of relevance in the limited time available. Mobile computer systems can provide guidance and point to relevant information by identifying and recommending content that matches a user’s interests. In this paper we infer implicit ratings from observed viewing times, and outline a collaborative user modelling approach to predict a user’s interests and expected viewing times. We make predictions about viewing themes (item sets) taking into account the visitor’s time limit. Our model based on relative interests with imputed ratings yielded the best performance.

This paper proposes a collaborative filtering method with user-created tags focusing on changes of web content and internet services. Collaborative tagging is employed as an approach in order to grasp and filter users’ preferences for items. In addition, we explore several advantages of collaborative tagging for future searching and information sharing which is used for automatic analysis of user preference and recommendation. We present empirical experiments using real dataset from to demonstrate our algorithm and evaluate performance compared with existing works.

The goal of this paper is to promote a new research area named Computational Imagination, which is defined as the science of modeling human imagination by creating artificial agents with intelligence, emotions and imagination. The new discipline aims to (i) study the interplay between cognition, emotion and imagery, (ii) analyze the way perceptions, emotions, prior knowledge and context influence imagination, and (iii) design agents capable of forming concepts and images of something that is neither perceived nor sensed as real, and of dealing with the reality by using the creative power of their imagination. The paper outlines some of the most challenging and intriguing research questions, and directions for further research.