Abstract

Abstract
Developing software agents that play incomplete information games is a demanding task: it is required they incorporate strategies capable of dealing with hidden information and deception and risk management. In Poker, these issues are commonly addressed by estimating opponents' gameplay using a variety of techniques such as Expected Hand Strength (E[HS]) or Hand Potential. In this paper, we propose criteria which can be applied when assessing such techniques, and we have also run benchmark tests which demonstrate their pertinence. We have, however, been faced with a clear gap in terms of the methods' efficiency. While this is not a problem in theoretical models, when implementing such methods in real world applications, they can prove to be painfully slow. In order to address this issue, we propose the Average Rank Strength (ARS) method. It can calculate the strength of a hand of any size through the hand's rank width negligible error, when compared to the original method. Still, the greatest contribution of this method lies in the speed-up factor of about 1000 times over E[HS]. Since most successful agents in the literature use their game abstraction based on E[HS], this breakthrough will significantly contribute towards a much lighter strategy computation, since this routine must be called billions of times. By saving computation time, we believe that future integration of ARS with current game playing algorithms will allow for creating agents with smaller abstraction levels, thus making room for improvement in their overall performance.

Abstract
Traditionally, recommender systems for the web deal with applications that have two dimensions, users and items. Based on access data that relate these dimensions, a recommendation model can be built and used to identify a set of N items that will be of interest to a certain user. In this paper we propose a multidimensional approach, called DaVI (Dimensions as Virtual Items), that consists in inserting contextual and background information as new user-item pairs. The main advantage of this approach is that it can be applied in combination with several existing two-dimensional recommendation algorithms. To evaluate its effectiveness, we used the DaVI approach with two different top-N recommender algorithms, Item-based Collaborative Filtering and Association Rules based, and ran an extensive set of experiments in three different real world data sets. In addition, we have also compared our approach to the previously introduced combined reduction and weight post-filtering approaches. The empirical results strongly indicate that our approach enables the application of existing two-dimensional recommendation algorithms in multidimensional data, exploiting the useful information of these data to improve the predictive ability of top-N recommender systems.

Abstract
In this paper we propose a multilingual method to identify health-related queries and classify them into health categories. Our method uses a consumer health vocabulary and the Unified Medical Language System semantic structure to compute the association degree of a query to medical concepts and categories. This method can be applied in different languages with translated versions of the health vocabulary. To evaluate its efficacy and applicability in two languages we used two manually classified sets of queries, each on a different language. Results are better for the English sample where a distance of 0.38 to the ROC optimal point (0,1) was obtained. This shows some influence of the translation in the method's performance.

Abstract

Abstract
Novelty detection is a useful ability for learning systems, especially in data stream scenarios, where new concepts can appear, known concepts can disappear and concepts can evolve over time. There are several studies in the literature investigating the use of machine learning classification techniques for novelty detection in data streams. However, there is no consensus regarding how to evaluate the performance of these techniques, particular for multiclass problems. In this study, we propose a new evaluation approach for multiclass data streams novelty detection problems. This approach is able to deal with: i) multiclass problems; ii) confusion matrix with a column representing the unknown examples; iii) confusion matrix that increases over time; iv) unsupervised learning, that generates novelties without an association with the problem classes and v) representation of the evaluation measures over time. We evaluate the performance of the proposed approach by known novelty detection algorithms with artificial and real data sets.