José Luis Álvarez

A New Self-adaptative Crossover Operator for Real-Coded Evolutionary Algorithms

In this paper we propose a new self-adaptative crossover operator for real coded evolutionary algorithms. This operator has the capacity to simulate other real-coded crossover operators dynamically and, therefore, it has the capacity to achieve exploration and exploitation dynamically during the evolutionary process according to the best individuals. In other words, the proposed crossover operator may handle the generational diversity of the population in such a way that it may either generate additional population diversity from the current one, allowing exploration to take effect, or use the diversity previously generated to exploit the better solutions. In order to test the performance of this crossover, we have used a set of test functions and have made a comparative study of the proposed crossover against other classic crossover operators. The analysis of the results allows us to affirm that the proposed operator has a very suitable behavior; although, it should be noted that it offers a better behavior applied to complex search spaces than simple ones.

MAVE: Multilevel wrApper Verification systEm

Wrappers are pieces of software used to extract data from websites and structure them for further application processing. Unfortunately, websites are continuously evolving and structural changes happen with no forewarning, which usually results in wrappers working incorrectly. Thus, wrappers maintenance is necessary for detecting whether wrapper is extracting erroneous data. The solution consists of using verification models to detect whether wrapper output is statistically similar to the output produced by the wrapper itself when it was successfully invoked in the past. Current proposals present some weaknesses, as the data used to build these models are supposed to be homogeneous, independent, or representative enough, or following a single predefined mathematical model. In this paper, we present MAVE, a novel multilevel wrapper verification system that is based on one-class classification techniques to overcome previous weaknesses. The experimental results show that our proposal outperforms accuracy of current solutions.

Applying ant colony hybrid metaheuristics to wrapper verification

First application of Ant Colony Metaheuristic to verify information extracted by web wrappers.New multilevel verification system that improves the results achieved by current techniques.Enumeration of current techniques weakness.Reformulation of wrapper verification problem as a combinational optimization problem.Applying non-parametric testing techniques to ascertain the statistical significance among results. Wrappers are pieces of software used to extract data from websites and structure them for further application processing. Unfortunately, websites are continuously evolving and structural changes happen with no forewarning, which usually results in wrappers working incorrectly. Thus, wrappers maintenance is necessary for detecting whether wrapper is extracting erroneous data. The solution consists of using verification models to detect whether wrapper output is statistically similar to the output produced by the wrapper itself when it was successfully invoked in the past. Current proposals present some weaknesses, as the data used to build these models are supposed to be homogeneous or that the features of this data set can be mapped to an n-dimensional space of independent dimensions when there is a correlation among their features. In this paper, a new verification system based on the Best-Worst Ant System (BWAS) is presented to overcome previous weaknesses. The experimental results show an accuracy improvement of 7.5% over current solutions.

On Relational Learning for Information Extraction

The extraction and integration of data from multiples sources are required in current companies which manage their business process by heterogeneous collaborating applications. However, integrating web applications is an arduous task because they are intended for human consumption and they do not provide APIs to access to their data automatically.Web Information extractors are used for this purpose but, they mostly provide ad-hoc highly domain dependent solutions. In this paper we aim at devising Information Extractors with a FOIL based core algorithm. It is a widely used first order rule learning algorithm since their rules are substantially more expressive and allow to learn complex concepts that cannot be represented in the attribute-value format. Furthermore, we focus on integrating other scoring functions to check if we can improve the rule search guide speeding up the learning process in order to make FOIL tractable in real-world domains such as Web sources.

Gpax: Genetic Parabolic Adaptive Crossover Operator

Abstract In this paper we propose a new crossover operator for real coded evolutionary algorithms that is based on a parabolic probability density function. This density function depends on two real parameters α and β which have the capacity to achieve exploration and exploitation dynamically during the evolutionary process in relation to the best individuals. In other words, the proposed crossover operator is able to handle the generational diversity of the population in such a way that it can either generate additional population diversity, therefore allowing exploration to take effect, or use the diversity previously generated to exploit the better solutions. In order to test the performance of this crossover, we have used a set of test functions and have made a comparative study of the proposed crossover against other classic crossover operators. The analysis of the results allows us to affirm that the proposed operator displays a very suitable behavior, although, it should be noted that it offers a b...

Mining Web Pages Using Features of Rendering HTML Elements in the Web Browser

The Web is the largest repository of useful information available for human users, but it is usual that Web Pages do not provide an API to get access to its information automatically. In order to solve this problem, Information Extractors are developed. We present a new methodology to induce Information Extractors from the Web. It is based on rendering HTML elements in the Web browser. The methodology uses a KDD process to mining a dataset with features of the elements in the Web page. An experimentation over 10 web sites has been made and the results show the effectiveness of the methodology.

Feature Selection to Improve Information Verifier Models in EAI

Reduce maintenance costs of Enterprise Application Integration (EAI) solutions becomes a challenge when we are trying to integrate friendly web applications. This problem can be solved using automated systems that allow to navigate, extract, structure and verify relevant information. The extracted information is characterized using complex models that later are used to check whether the information is valid. In this paper we will demonstrate empirically that feature selection techniques simplify and improve the models obtained for verification of information.

JISBD03 - Feature Selection to Improve Information Verifier Models in EAI

Reduce maintenance costs of Enterprise Application Integration (EAI) solutions becomes a challenge when we are trying to integrate friendly web applications. This problem can be solved using automated systems that allow to navigate, extract, structure and verify relevant information. The extracted information is characterized using complex models that later are used to check whether the information is valid. In this paper we will demonstrate empirically that feature selection techniques simplify and improve the models obtained for verification of information.