Marcelo Luis Errecalde

Evaluation of internal validity measures in short-text corpora

Short texts clustering is one of the most difficult tasks in natural language processing due to the low frequencies of the document terms. We are interested in analysing these kind of corpora in order to develop novel techniques that may be used to improve results obtained by classical clustering algorithms. In this paper we are presenting an evaluation of different internal clustering validity measures in order to determine the possible correlation between these measures and that of the F-Measure, a well-known external clustering measure used to calculate the performance of clustering algorithms. We have used several short-text corpora in the experiments carried out. The obtained correlation with a particular set of internal validity measures let us to conclude that some of them may be used to improve the performance of text clustering algorithms.

Measuring the quality of web content using factual information

Nowadays, many decisions are based on information found in the Web. For the most part, the disseminating sources are not certified, and hence an assessment of the quality and credibility of Web content became more important than ever. With factual density we present a simple statistical quality measure that is based on facts extracted from Web content using Open Information Extraction. In a first case study, we use this measure to identify featured/good articles in Wikipedia. We compare the factual density measure with word count, a measure that has successfully been applied to this task in the past. Our evaluation corroborates the good performance of word count in Wikipedia since featured/good articles are often longer than non-featured. However, for articles of similar lengths the word count measure fails while factual density can separate between them with an F-measure of 90.4%. We also investigate the use of relational features for categorizing Wikipedia articles into featured/good versus non-featured ones. If articles have similar lengths, we achieve an F-measure of 86.7% and 84% otherwise.

An efficient Particle Swarm Optimization approach to cluster short texts

Short texts such as evaluations of commercial products, news, FAQs and scientific abstracts are important resources on the Web due to the constant requirements of people to use this on line information in real life. In this context, the clustering of short texts is a significant analysis task and a discrete Particle Swarm Optimization (PSO) algorithm named CLUDIPSO has recently shown a promising performance in this type of problems. CLUDIPSO obtained high quality results with small corpora although, with larger corpora, a significant deterioration of performance was observed. This article presents CLUDIPSO^*, an improved version of CLUDIPSO, which includes a different representation of particles, a more efficient evaluation of the function to be optimized and some modifications in the mutation operator. Experimental results with corpora containing scientific abstracts, news and short legal documents obtained from the Web, show that CLUDIPSO^* is an effective clustering method for short-text corpora of small and medium size.

An application of defeasible logic programming to decision making in a robotic environment

Decision making models for autonomous agents have received increased attention, particularly in the field of intelligent robots. In this paper we will show how a Defeasible Logic Programming approach with an underlying argumentation based semantics, could be applied in a robotic domain for knowledge representation and reasoning about which task to perform next. At this end, we have selected a simple application domain, consisting of a micro-world environment using real and simulated robots for cleaning tasks.

Experiencing minix as a didactical aid for operating systems courses

Minix is a Unix clone Operating Systems to be run on IBM PCs and compatibles, designed by Tanembaum [10] for courses in the area.Accepting the Tanembaums proposal, this document describes the results of some extensions on the internal work of Minix as an exercise on Operating Systems Design and Implementation that attempts to transfer that experience to other groups of interest.The paper intends to be interpreted as a report remarking what kind of work was done having at our disposal an extensively documented copy of the source code of an operating system, taking into account that the developers are undergraduates in Computer Science.Further details on implementations will be available in future publications [1], [4], [6].

An argumentation-based BDI personal assistant

BDI models and argumentation-based approaches are powerful tools that can play a fundamental role in implementing intelligent systems for complex business and industrial problems. Some recent works have started studying the integration of both approaches from a theoretical point of view. However, little effort has been made to analize how these technologies can be effectively integrated in complex systems which require the use of well-known development resources, such as agent development platforms and other programming tools. This work addresses that problem, showing how an argumentation-based reasoning service can be used in different components of a BDI agent implemented with a Jadex platform. All the concepts involved in our proposal are exemplified by designing and implementing a travel assistant agent that allows to observe how BDI and argumentation approaches can be effectively integrated in a working system developed with freely available technologies.

A possibilistic defeasible logic programming approach to argumentation-based decision-making

The development of symbolic approaches to decision-making has become an ever-growing research line in artificial intelligence; argumentation has contributed to that with its unique strengths. Following this trend, this article proposes a general-purpose decision framework based on argumentation. Given a set of alternatives posed to the decision-maker, the framework represents the agents preferences and knowledge by an epistemic component developed using possibilistic defeasible logic programming. The reasons by which a particular alternative is deemed better than another are explicitly considered in the argumentation process involved in warranting information from the epistemic component. The information warranted by the dialectical process is then used in decision rules that implement the agents general decision-making policy. Essentially, decision rules establish patterns of behaviour of the agent specifying under which conditions a set of alternatives will be considered acceptable; moreover, a methodology for programming the agents epistemic component is defined. It is demonstrated that programming the agents epistemic component following this methodology exhibits some interesting properties with respect to the selected alternatives; also, when all the relevant information regarding the agents preferences is specified, its choice behaviour coincides with respect to the optimum preference derived from a rational preference relation.

A general bio-inspired method to improve the short-text clustering task

“Short-text clustering” is a very important research field due to the current tendency for people to use very short documents, e.g. blogs, text-messaging and others. In some recent works, new clustering algorithms have been proposed to deal with this difficult problem and novel bio-inspired methods have reported the best results in this area. In this work, a general bio-inspired method based on the AntTree approach is proposed for this task. It takes as input the results obtained by arbitrary clustering algorithms and refines them in different stages. The proposal shows an interesting improvement in the results obtained with different algorithms on several short-text collections.

An approach to decision making based on dynamic argumentation systems

In this paper we introduce a formalism for single-agent decision making that is based on Dynamic Argumentation Frameworks. The formalism can be used to justify a choice, which is based on the current situation the agent is involved. Taking advantage of the inference mechanism of the argumentation formalism, it is possible to consider preference relations, and conflicts among the available alternatives for that reasoning. With this formalization, given a particular set of evidence, the justified conclusions supported by warranted arguments will be used by the agents decision rules to determine which alternatives will be selected. We also present an algorithm that implements a choice function based on our formalization. Finally, we complete our presentation by introducing formal results that relate the proposed framework with approaches of classical decision theory.

On the Use of Reliable-Negatives Selection Strategies in the PU Learning Approach for Quality Flaws Prediction in Wikipedia

Learning from positive and unlabeled examples (PU learning) has proven to be an effective method in several Web mining applications. In particular, in the 1st International Competition on Quality Flaw Prediction in Wikipedia in 2012, a tailored PU learning approach performed best amongst the competitors. A key feature of that approach is the introduction of sampling strategies within the original PU learning procedure. The paper in hand revisits the winner approach of 2012 and elaborates on neglected aspects in order to provide evidence for the usefulness of sampling in PU learning. In this regard, we propose a modification to this PU learning approach, and we show how the different sampling strategies affect the flaw prediction effectiveness. Our analysis is based on the original evaluation corpus of the 2012-competition on quality flaw prediction. A main outcome is that under the best sampling strategy, our new modified version of PU learning increases in average the flaw prediction effectiveness by 18.31%, when compared against the winning approach of the competition.