Catálogo de publicaciones - libros

The discovery of new and potentially meaningful relationships between concepts in the biomedical literature has attracted the attention of a lot of researchers in text mining. The main motivation is found in the increasing availability of the biomedical literature which makes it difficult for researchers in biomedicine to keep up with research progresses without the help of automatic knowledge discovery techniques. More than 14 million abstracts of this literature are contained in the Medline collection and are available online. In this paper we present the application of an association rule mining method to Medline abstracts in order to detect associations between concepts as indication of the existence of a biomedical relation among them. The discovery process fully exploits the MeSH (Medical Subject Headings) taxonomy, that is, a set of hierarchically related biomedical terms which permits to express associations at different levels of abstraction (generalized association rules). We report experimental results on a collection of abstracts obtained by querying Medline on a specific disease and we show the effectiveness of some filtering and browsing techniques designed to manage the huge amount of generalized associations that may be generated on real data.

In the context of the Pangea project at IBM, we needed to design an information extraction module in order to extract some information from datasheets. Contrary to several information extraction systems based on some machine learning techniques that need some linguistic parsing of the documents, we propose an hybrid approach based on association rules mining and decision tree learning that does not require any linguistic processing. The system may be parameterized in various ways that influence the efficiency of the information extraction rules we discovered. The experiments show the system does not need a large training set to perform well.

In this paper we propose an ontology-supported technique to preprocess the remark fields in real-life customer servicing databases in order to discover useful information to help re-categorize misclassified service records owing to human ignorance or bad design of problem categorization. This process restores the database into one with more meaningful data in each record, which facilitates subsequent data analysis. Our experience in applying the technique to a real-life database shows a substantial quality improvement can be obtained in mining association rules from the database.

In this paper, a multi-objective genetic algorithm is proposed to deal with a real-world fuzzy job shop scheduling problem. Fuzzy sets are used to model uncertain due dates and processing times of jobs. The objectives considered are average tardiness and the number of tardy jobs. Fuzzy sets are used to represent satisfaction grades for the objectives taking into consideration the preferences of the decision maker. A genetic algorithm is developed to search for the solution with maximum satisfaction grades for the objectives. The developed algorithm is tested on real-world data from a printing company. The experiments include different aggregation operators for combining the objectives.

This paper describes the application of a genetic algorithm to nearest-neighbour based imputation of sample data into a census data dataset. The genetic algorithm optimises the selection and weights of variables used for measuring distance. The results show that the measure of fit can be improved by selecting imputation variables using a genetic algorithm. The percentage of variance explained in the goal variables increases compared to a simple selection of imputation variables. This quantitative approach to the selection of imputation variables does not deny the importance of expertise. Human expertise is still essential in defining the optional set of imputation variables.

This paper describes a Genetic Algorithm approach to solve a task scheduling problem at a small size manufacturing company. The operational solution must fulfill two basic requirements: low cost and usability. The proposal was implemented and results obtained with the system lead to better results compared to previous and non-computerized solutions.

In this paper, an efficient genetic algorithm (EGA) for TSK-type neural fuzzy identifier (TNFI) is proposed for solving identification problem. For the proposed EGA method, the better chromosomes will be initially generated while the better mutation points will be determined for performing efficient mutation. The adjustable parameters of a TNFI model are coded as real number components and are searched by EGA method. The advantages of the proposed learning algorithm are that, first, it converges quickly and the obtained fuzzy rules are more precise. Secondly, the proposed EGA method only takes a few population sizes.

In this paper, we propose an overall architecture for hardware implementation of genetic algorithms. The proposed architecture is independent of such specifics. It implements the fitness computation using a neural networks.

The Multi-Vehicle routing problem (MVRP) in real time is a graph modification problem. In order to solve this kind of problems, alternative approaches have been investigated. Evolutionary Algorithms (EAs) have presented relevant results. However, these methodologies require special encoding to achieve proper performance when large graphs are considered. We propose a representation based on NDE [Delbem et al., (2004a); Delbem et al., (2004b)] for directed graphs. An EA using the proposed encoding was developed and evaluated for the MVRP.

This paper proposes two novel methods to optimize quantitative association rules. We utilize a multi-objective Genetic Algorithm (GA) in the process. One of the methods deals with partial optimal, and the other method investigates complete optimal. Experimental results on Letter Recognition Database from UCI Machine Learning Repository demonstrate the effectiveness and applicability of the proposed approaches.