Monografias em Ciência da Computação

2007

ABSTRACTS

Departmento de Informática 
Pontifícia Universidade Católica do Rio de Janeiro - PUC-Rio
Rio de Janeiro - Brazil


This file contains a list of the technical reports of the Departmento de Informática, Pontifícia Universidade Católica do Janeiro - PUC-Rio, Brazil, which are published in our series Monografias em Ciência da Computação (ISSN 0103-9741), edited by Prof. Carlos Lucena. Please note that the reports not available for download are available in their print format and can be obtained via the e-mail below.
For any questions, requests or suggestions, please contact:
Rosane Castilho bib-di@inf.puc-rio.br

Last update: 30/DECEMBER/2007

INDEX


[MCC01/07]
SOARES, L.F.G.  Fundamentos de sistemas multimídia; Parte 1 - Aquisição, codificação e exibição de dados. 40 p. Port. E-mail: lfgs@inf.puc-rio.br

Abstract: This work comprises the first part of a Multimedia Fundamentals course handouts. It covers data acquisition, coding and exhibition aspects in many representation medias.

[MCC02/07]
CARVALHO, F.G.; RAPOSO, A.B.; GATTASS, M.
  An Approach for enabling the use of immersive virtual reality in desktop hybrid interfaces. 11 p. Eng. E-mail: mgattass@inf.puc-rio.br

Abstract:
Hybrid User Interfaces, which create a heterogeneous environment providing different interaction forms and devices, may be enhanced by exploring more extensively the mixed reality continuum, which ranges from the real world to the complete virtual world, passing by augmented reality and augmented virtuality. Some hybrid interface approaches have been developed making use of the real world or enhanced by augmented reality resources. This work presents an alternative to include immersive virtual reality in hybrid user interfaces in a common desktop setup. In order to enable this inclusion, augmented virtuality was used to enhance the virtual environment with real world information. In this case, that information refers to the physical interaction space available in the users desktop. Some advantages of the use of immersive virtual reality in this context are discussed by means of the analysis of 3D interaction techniques.

[MCC03/07]
PAES, R.B.; LUCENA, C.J.P.; CARVALHO, G.R., COWAN, D.  Event-driven high level model specification of laws in open multi-agent systems . 21 p. Eng. E-mail: lucena@inf.puc-rio.br

Abstract:
The agent development paradigm poses many challenges to software engineering researchers, particularly when the systems are distributed and open. They have little or no control over the actions that agents can perform. Laws are restrictions imposed by a control mechanism to deal with uncertainty and to promote open system dependability. In this paper, we present a high-level event driven conceptual model of laws. XMLaw is an alternative approach to specifying laws in open multi-agent systems that presents high level abstractions and a flexible underlying event-based model. Thus XMLaw allows for flexible composition of the elements from its conceptual model and is flexible enough to accept new elements.

[MCC04/07]
GAZOLA, A.; FURTADO, A.L. Bancos de dados geográficos inteligentes. 17 p. Eng. E-mail: furtado@inf.puc-rio.br

Abstract:
Current research and development of agent-based systems has focused primarily on architectures, protocols, frameworks, messaging infrastructure and community interactions. As intelligent agent-based systems take over operations in the financial community, transportation, manufacturing, utilities, aerospace, and the military, assurances will need to be given to the owners and operators of these systems assuring that these non-deterministic learning systems operate correctly. In this State of the Art report (SotA), we will give an introduction to work presented in the area of testing and debugging distributed systems composed of complex autonomous entities (agents). We will provide pointers to work by large players in the field. We will also discuss the debugging of multi-agents systems. We will explain why this kind of system must be handled differently than less complex systems.

[MCC05/07]
SILVA, B.S.; AURELIANO, V.C.O.; BUENO, A.M.; ARAÚJO, A.C.I.C.; BARBOSA, S.D.J. 
O que profissionais de computação pensam sobre o projeto de software usando o Tablet PC? 20 p. Port. E-mail: simone@inf.puc-rio.br

Abstract: After keyboard and mouse came about, many alternative forms of interaction have been proposed. In particular, one has been gaining attention with the arrival of Tablet PCs: pen-based interaction. This new form of interaction suggests that using a computer can be as natural as writing notes on a piece of paper . In face of this new technological resource, we can pose questions that didn’t make sense before: what can we do with a pen that writes on a computer? What is easier to do in a computer with a pen? What is more difficult? Addressing these questions, this work presents the result of a qualitative research about computer science professionals’ expectations and opinions concerning the use of Tablet PCs with a specific goal: software design. This qualitative research is part of research project Ink-a-Sketch: Combining Model-based and Sketchbased Design of User Interfaces in the Classroom.

[MCC06/07]
COSTA, R.G.L.; BASTOS, P.M.A.; SILVESTRE, B.O.; RODRIGUEZ, N.R. Combining programming paradigms in asynchronous systems. 8 p. Eng. E-mail: noemi@inf.puc-rio.br

Abstract: Event-based systems are becoming more and more popular. However, event orientation imposes a state-machine view of the program which is not always natural or easy for the programmer. In this work we discuss how programming language features can facilitate the construction and integration of different high-level programming abstractions, allowing the programmer to view the code in the way that is most appropriate to each situation, even inside one single application.

[MCC07/07]
FURTADO, A.L., CASANOVA, M.A., BARBOSA, S.D.J., BREITMAN, K.K. Plot mining as an aid to characterization and planning. 16 p. Eng. E-mail: furtado@inf.puc-rio.br

Abstract: In parallel with data mining, plot mining is presented as a valuable way to gain knowledge on management information systems. A compact plot organization and methods for plot collection and handling are described. Plots are then shown to be a significant element in the characterization of the behaviour of agents, as well as a helpful resource for planning. Similarity and analogy are employed to extend the results, both to other cases within the domain involved and to other domains.

[MCC08/07]
CUNHA, L.M. Semantic Web application framework. 156 p. Eng. E-mail: bib-di@inf.puc-rio.br.

Abstract: Documents have been the main vehicle of the Web until some years ago.  With the advent of Web applications, data stored in organizations' databases or legacy systems has been made available to users. However, very often, the exchange of data between those applications themselves or between them and "end-users applications" were not possible since they used different formats for the information representation. The development of standards and the use of eXtensible Markup Language (XML) solved parts of the problem. That was a syntactic solution and it works for several cases, e.g., schema interoperability in Business-to-Business e-commerce scenarios. Nevertheless, the lack of semantics on these data prevented applications to take more advantage of them. The idea behind the Semantic Web is to define explicity the semantics of data available on Web. Therefore, we expect another step forward where applications, being them corporative or for end-users, will "understand" the meaning of the data available on the Web. Once those applications can understand it, they will be able to help users to take advantage of this "data driven" Web and to perform their daily tasks easily. This report proposes a framework for the development of Semantic Web applications. Considering the scenario described in the previous paragraph, the number of possible applications that can be developed is almost infinite. For this reason, we restricted ourselves to examine the solutions that aim to solve the problem presented at the Semantic Web Challenge; and to propose a framework that represent those solutions. The challenge is concerned in demonstrating how Semantic Web techniques can provide valuable or attractive applications to end users. Our main concern was then to demonstrate and help a developer to achieve that value addition or attractiveness, through Semantic Web techniques, in a Software Engineering approach using frameworks.

[MCC09/07]
DUARTE, J.C.; MILIDIU, R.L. Machine learning algorithms for portuguese named entity recognition. 10 p. Eng. E-mail: milidiu@inf.puc-rio.br

Abstract: Named Entity Recognition (NER) is an important task in Natural Language Processing. It provides key features that help on more elaborated document management and information extraction tasks. In this monograph, we propose seven machine learning approaches that use HMM, TBL and SVM to solve Portuguese NER. The performance of each modeling approach is empirically evaluated. The SVM-based extractor shows a 88.11% F-score, which is our best observed value, slightly better than TBL. This is very competitive when compared to state-of-the-art extractors for similar Portuguese NER problems. Our HMM has reasonable precision and recall and does not require any additional expert knowledge. This is an advantage for our HMM over the other approaches. The experimental results suggest that Machine Learning can be useful in Portuguese NER. They also indicate that HMM, TBL and SVM perform well in this natural language processing task.

[MCC10/07]
LOBATO, C.A.; GARCIA, A.F.; ROMANOVSKY, A.; LUCENA, C.J.P. An aspect-oriented software architecture for code mobility. 52 p. Eng. E-mail: lucena@inf.puc-rio.br

Abstract: Mobile agents have come forward as a technique for tackling the complexity of open distributed applications. However, the pervasive nature of code mobility implies that it cannot be modularized using only object-oriented (OO) concepts. In fact, developers frequently evidence the presence of mobility tangling and scattering in their modules. Despite these problems, they usually rely on OO application programming interfaces (APIs) offered by the mobility platforms. Such API-oriented designs suffer a number of architectural restrictions and there is a pressing need for empowering developers with an architecture supporting a flexible incorporation of code mobility in the agent applications. This work presents an aspect-oriented software architecture, called ArchM, ensuring a clean modularization, a more straightfoward introduction, and an improved variability of code mobility in mobile agent systems. It addresses OO APIs' restrictions and is independent on specific platforms and applications. An ArchM implementation provides solutions to fine-grained problems related to mobility tangling and scattering in the implementation level. The usefulness and usability of ArchM has been assessed within the context of two case studies, and through its composition with two mobility platforms.

[MCC11/07]
SANTOS, C.N.; MILIDIÚ, R.L. Probabilistic classifications with TBL. 14 p. Eng. E-mail: milidiu@inf.puc-rio.br

Abstract: The classifiers produced by the Transformation Based error-driven Learning (TBL) algorithm do not produce uncertainty measures by default. Nevertheless, there are situations like active and semi-supervised learning where the application requires both the sample's classification and the classification confidence. In this paper, we present a novel method which enables a TBL classifier to generate a probability distribution over the class labels. To assess the quality of this probability distribution, we carry out four experiments: cross entropy, perplexity, rejection curve and active learning. These experiments allows us to compare our method with another one proposed in the literature, the TBLDT. Our method, despite being simple and straightforward, outperforms TBLDT in all four experiments.

[MCC12/07]
SILVA, B.S.; BARBOSA, S.D.J. Designing human-computer interaction with MoLIC diagrams - a practical guide. 45 p. Eng. E-mail: simone@inf.puc-rio.br

Abstract: This document describes the 2nd edition of MoLIC in the form of a designer's manual. First, we briefly introduce MoLIC's theoretical foundation, the semiotic engineering theory of Human-Computer Interaction. This introduction is necessary for an efficient use of MoLIC. We then describe how to define the user-system interaction according to the 2nd edition of MoLIC, presenting a number of examples to illustrate various usages of the notation.

[MCC13/07]
CHAVEZ, C.F.G.; GARCIA, A.F.; BATISTA, T.V.; RASHID, A.; SANT'ANNA, C.N. Composing architectural aspects based on style semantics. 23 p. Eng. E-mail: bib-di@inf.puc-rio.br

Abstract: An architectural style can be regarded as a paradigm of architectural modularity that encompasses syntactic descriptions, semantic models and constraints over them. In this paper, we analyze the influence exerted by architectural styles over the nature of architectural aspects. We propose style-based join point models that expose high-level join points based on style semantics, with the goal of enhancing composition of architectural aspects in hybrid software architectures. We present a real-life case study involving several styles to demonstrate the expressiveness of the style-oriented join point model. We also assess the scalability of our proposal in the presence of four stylistic composition categories, namely hierarchy, specialization, conjunction, and overlapping.
Finally, we discuss the interplay of the style-based architecture composition model and the other conventional models.

[MCC14/07]
GATTI, M.A.; LUCENA, C.J.P. An agent-based approach for building biological systems: improving the software engineering for complex and adaptative multi-agent systems. 18 p. Eng. E-mail: lucena@inf.puc-rio.br

Abstract: Biological systems are typically open, distributed and complex systems, comprised of multiple interacting autonomic elements that exhibit emergent behavior. Design and development of such systems is a non-trivial task that by definition requires specific software engineering approaches, including the use of specialized modeling techniques. Multi-agent systems are a relatively new way to address complex systems. Usually the idea is that the agents used are rather simple, and the complexity and adaptability of such a system are modeled by the interaction between these agents. This paper starts situating the reader in the biological systems context. Then it describes how multi-agent systems can fulfill their needs of modeling and simulating. To exemplify, we briefly describe five applications with different targets and approaches in this area. Finally, we present the research challenges for developing software engineering of multi-agent systems capable of implementing the behavior of complex and adaptative systems such as a biological system or any other with those characteristics.

[MCC15/07]
DUARTE, J.C.; MILIDIU, R.L. Generalized boosting learning. 6 p. Eng. E-mail: milidiu@inf.puc-rio.br

Abstract: Boosting is a machine learning technique to combine several weak algorithms and improve their accuracies. In each iteration, the algorithm changes the weights of the examples building a different classifier. A simple final voting scheme among each classifier defines the classification of a new instance. The most common used algorithm based on boosting is AdaBoos, which starts with an uniform distribution for the examples. Unfortunately, there is no guarantee that this is the best choice for a initial distribution. We propose a boosting
approach to model this issue, in which one can set any initial weight distribution. Besides that, we can also introduce a cost function to charge errors in most representative examples. For instance, if examples come in a sequential order, with human intervention, the earlier learned examples are more representative. In this kind of environment, one can use a skewed initial distribution like Zipf or geometric. We show the necessary changes in the original algorithm to accommodate the choice of any initial weight distribution.

[MCC16/07]
DUARTE, A.R.; HAEUSLER, E.H.; RIBEIRO, C.C.; URRUTIA, S. Referee assignment in sports leagues. 16 p. Eng. E-mail: hermann@inf.puc-rio.br

Abstract: Optimization in sports is a field of increasing interest. Combinatorial optimization techniques have been applied e.g. to game scheduling and playoff elimination. A common problem usually found in sports management is the assignment of referees to games already scheduled. There are a number of rules and objectives that should be taken into account when referees are assigned to games. We address a simplified version of a referee assignment problem common to many amateur leagues of sports such as soccer, and basketball. The problem is formulated by integer programming and its decision version is proved to be NP-complete. To tackle real-life large instances of the referee assignment problem, we propose a three-phase heuristic approach based on a constructive procedure, a repair heuristic to make solutions feasible, and a local search heuristic to improve feasible solutions. Numerical results on realistic instances are presented and discussed.

[MCC17/07]
SOUSA, D.X.; LIFSCHITZ, S. A avaliação do e-value para execução do BLAST sobre bases de dados fragmentadas. 15 p. Port. E-mail: sergio@inf.puc-rio.br

Abstract: The BLAST tool is widely used in bioinformatics research projects, helping with the biological sequences comparisons and homology search. Stand-alone executions, involving a single input sequence (query) and a relatively small database, no performance problems arise. However, with large databases, there is a need for execution approaches that achieve better execution times. An alternative would be parallel computing environments, and in most cases a distributed database allocation through the available machines. Particulary, when the sequence database is fragmented, correctness must be guaranteed in order to get output results that are consistent with the equivalent serial execution. This fundamental problem occurs due to the statistics used by BLAST in its heuristics, such as the e-value. Computations are directly related to the size of the database. When the latter is fragmented, the results obtained may not infer the same semantics as when BLAST is run in the conventional (serial) way. This paper aims at discussing the BLAST parallel evaluation on fragmented databases. Our goal is to support developers and users with respect to their decisions regarding configuration and parameters for BLAST execution in these situations.

[MCC18/07]
FURTADO, A.L. Applying analogy for the generation of entity-relationship schemas. 11 p. Eng. E-mail: furtado@inf.puc-rio.br

Abstract: To support the generation of database schemas of information systems, starting from analogous predefined schemas, a five-step process is described. It involves generic and blended spaces, whose utilization is essential to achieve the passage from the source space into the target space in such a way that differences and conflicts can be detected, and, whenever possible, conciliated. The convenience to work with multiple source schemas to cover distinct aspects of a target schema, as well the possibility of creating schemas at the generic and blended spaces, are briefly considered.

[MCC19/07]
SOUZA, C.; LABER, E.S.; VALENTIM, C.; CARDOSO, E. A polite policy for revisiting web pages. 10 p. Eng. E-mail: laber@inf.puc-rio.br

Abstract: We propose and analyze an efficient scheduling policy for revisiting web pages so as to keep a local repository as up to date as possible. The main contribution of our policy is the fact that it takes into account the politeness constraint in order to compute the revisiting times. Our experiments suggest that the performance of our policy is very close to the optimal one and that it is significantly better than one obtained using a reasonable variation of the policy proposed in a former work.

[MCC20/07]
NUNES, C.P.B.; STAA, A.v. Processos de software open source. 20 p. Port. E-mail: arndt@inf.puc-rio.br

Abstract: The open source software development movement has increased significantly, and this refers to the success of some projects, due to their quality and trust-worthiness. In these projects these attributes are achieved through an meaningful characteristic of the open source, which is the active participation of the users. In this paper, details of the open source software development process and the main activities of its development are presented. Some experimental studies about the open source software process maturity, as well as details of some projects, as Mozilla, Apache, and Linux, are also discussed. Finally, a development process is proposed in this paper.

[MCC21/07]
GATTI, M.A.C.; STAA, A.v.; LUCENA, C.J.P. AUML-BP: a basic agent oriented software development process model using AUML. 25 p. Eng. E-mail: arndt@inf.puc-rio.br

Abstract: Agent-Oriented Software Engineering (AOSE) has emerged as the discipline devoted to the engineering of complex software systems based on the multi-agent systems paradigm. Research in the field of AOSE includes the identification and development of both conceptual tools (e.g., formal modeling) and practical tools (e.g., agent-based infrastructures) to support software engineers and programmers in the analysis, design and development of multi-agent systems. Among others, a great deal of effort in the AOSE area focuses on the definition of methodologies to guide the process of developing multi-agent systems. As the AOSE methodologies have been proposed so far, they are not enough for practical agent software development without a clear understanding of the software development process model that should underlie the methodology. In order to have a good process and sucessfully finish the project, it is necessary to explicitly adopt either general methods and methodologies, or specifically suitable ones. In this context, this paper proposes AUML-BP (AUML Basic Process), a basic agent oriented software development process model using AUML.

[MCC22/07]
VITERBO FILHO, J.; ENDLER, M.; RODRIGUES, V.J.S. Discovering services with restricted location scope in ubiquitous environments. 11 p. Eng. E-mail: endler@inf.puc-rio.br

Abstract: In ubiquitous computing systems, the mobility of users and their devices results in recurring disconnections and reconnections with different networks, and the corresponding dynamic change of the network and domain-specific resources and services accessible from the user's device. On the other hand, some services are available to be used only by users that are located in a well defined region. In this highly dynamic and heterogeneous scenario, applications must be capable of discovering the appropriate instances of the required services in each visited network or region. In order to support such spontaneous interaction, we propose a discovery service based on the notion of a (geographic) location scope. This discovery service is one of the MoCA architecture, a middleware that supports the development and deployment of location-aware ubiquitous applications.

[MCC23/07]
SILVESTRE, B.O.; RODRIGUEZ, N.R.; BRIOT, J-P. Flexibility for synchronization abstractions in distributed programming. 9 p. Eng. E-mail: noemi@inf.puc-rio.br

Abstract: Most proposals for synchronization mechanisms in concurrent and distributed systems have privileged the idea of a single abstraction that can solve all synchronization issues. However, in the context of loosely coupled wide-area applications, it is interesting to have available different synchronization mechanisms, so that the developer may chose the most suitable mechanism for each part of his application. Besides, it is important that
synchronization facilities guarantee the desired constraints while imposing as little blocking as possible. In this work, we discuss how the use of dynamic programming language can help us deal with this problem, proving support for building different building blocks and abstractions. We base our discussion on the Lua programming language, and show how we can use it to implement different intra and inter-object synchronization mechanisms proposed in the literature
.

[MCC24/07]
KARLSSON, B.; CIARLINI, A.E.M.; FEIJÓ, B.; FURTADO, A.L. Applying the plan-recognition/ plan-generation paradigm to interactive storytelling: the LOGTELL case study. 18 p. Eng. E-mail: bruno@inf.puc-rio.br.

Abstract: A key issue in interactive storytelling is how to generate stories which are, at the same time, interesting, and coherent. On the one hand, it is desirable to provide means for the user to intervene in the story. But, on the other hand, it is necessary to guarantee that user intervention will not introduce events has violate the rules of the intended genre. This paper describes the usage of a plan recognition/ plan generation paradigm in LOGTELL, a logic-based tool for the interactive generation and dramatization of stories. We focus on the specification of a formal logic model for events and characters' behaviour and on how the tool helps the interactive composition of plots through the adaptation of fully or partial generated plots. Based on the model, the user can interact with the tool at various levels, obtaining a variety of stories agreeable to individual tastes, within the imposed coherence requirements. The system alternates stages of goal inference, planning, plan recognition, user intervention and 3D visualization. Our experiments have shown that the system can be used not only for entertainment purposes but also, more generally, to help in the creation and adaptation of stories in conformity with a specified genre.

[MCC25/07]
RADEMAKER, A.; AMARAL, F.N.; HAEUSLER, E.H. A sequent calculus for ALC. 14 p. 18 p. Eng. E-mail: hermann@inf.puc-rio.br

Abstract: Description Logics is a family of formalisms used to represent knowledge of a domain. In contrast with others knowledge representation systems, Description Logic are equipped with a formal, logic-based semantics. Knowledge representation systems based on description logics provide various inference capabilities that deduce implicit knowledge from the explicity represented knowledge. We present a sequent calculus for ALC, a basic Description Logic. The first motivation for developing such system is the extraction of computational content of ALC proofs. The present calculus is an intermediate step towards a Natural Deduction System for ALC.

[MCC26/07]
MAGALHAES, J.A.P.; STAA, A.v.; LUCENA, C.J.P. Evaluating the recovery oriented apprach through the systematic development of real complex applications. 19 p. Eng. E-mail: arndt@inf.puc-rio.br

Abstract: Recovery oriented software is built with the perspective that hardware or software failures and operation mistakes are facts to be coped with, since they are problems that cannot be fully solved while developing real complex applications. Consequently, any software will always have a non-zero chance of failure. Some of these failures may be caused by defects that may be removed or encapsulated. From the point of view of removing or encapsulating defects, a failure is considered to be trivial, when i) the required effort to identify and eliminate or encapsulate the causing defect is small, ii) the risk of making mistakes in these steps is also small, and iii) the consequences of the failure are tolerable. It is highly important to design systems in such a way that most (ideally all) of the failures are trivial. Such systems are called "debuggable systems". In this work, we present the results of systematic applying techniques that focus in creating debuggable software for real embedded applications.

[MCC27/07]
LABER, E.S.; MOLINARO, M.S. Improved approximations for the hotlink assignment problem and binary searching on tree. 38 p. Eng. E-mail: laber@inf.puc-rio.br

Abstract: Let G = (V,E) be a directed acyclic graph representing a web site, where nodes correspond to pages and arcs to hyperlinks. In this context, hotlinks are defined as shortcuts (new arcs) added to web pages of G in order to reduce the time spent by users to reach their desired information. In this paper, we consider the problem where G is a rooted directed tree and the goal is minimizing the expected time spent by users by assigning at most k hotlinks to each node. For the most studied version of this problem where at most one hotlink can be assigned from each node, we prove the existence of an FPTAS, improving upon the constant factor algorithm recently obtained in by Jacobs and Wads. In addition, we develop the first constant factor approximation algorithm for the most general version where k hotlinks can be assigned from each node. Finally, we give an evidence that the technique developed to obtain this last result can be of independent interest since, based on it, we develop a linear time algorithm that provides the first constant approximation for the average case version of the problem of binary searching in trees, a natural generalization of the classical problem of searching in lists with different access probabilities.

[MCC28/07]
SILVA, D.S.P.; ARAGÃO; M.V.S.P. An approximation algorithm for permutation flowshop scheduling problem via Erdos-Szekeres theorem extensions. 8 p. Eng. E-mail: poggi@inf.puc-rio.br

Abstract: This work presents a deterministic approximation algorithm for Permutation Flowshop Scheduling Problem (PFS) with performance ratio 2V2n+m and time complexity (nm+n log n), where n is the number of jobs and m is the number of machines of an instance. This result consists in the first known deterministic approximation algorithm for PFS in which performance ratio is not a linear factor of n or m. In the case that n=O(m) this is the best approximation algorithm already obtained for PFS. A novel technique for performance guarantee analysis of PFS solutions is developed, exploring its correlation with Weighted Monotone Subsequence Problems.

[MCC29/07]
SANTOS, C.N.; MILIDIÚ, R.L. Entropy guide transformation learning. 11 p. Eng. E-mail: milidiu@inf.puc-rio.br

Abstract: This work presents a new machine learning strategy that combines the feature selection characteristics of Decision Trees (DT) and the robustness of Transformation Based Learning (TBL). The proposed method, Entropy Guided Transformation Learning (ETL), produces transformation rules that are more effective than decision trees and also eliminates the need of a problem domain expert to build TBL templates. We carry out experiments with three computational linguistic tasks: Portuguese noun phrase chunking, English base noun phrase chunking and text chunking. In all three tasks, ETL shows better results than Decision Trees and TBL with hand-crafted templates. ETL also provides a new training strategy that accelerates transformation learning by a factor or five. For Portuguese noun phrase chunking, ETL shows the best reported results for the task. For the other two linguistic tasks, ETL shows state-of-the-art competitive results and maintains the advantages of using a rule based system.

[MCC30/07]
SHAHAM, N.; LABER, E.S. Methods fot the acceleration of "non-local means" noise reduction algorithm. 33 p. Eng. E-mail: laber@inf.puc-rio.br

Abstract: "Non Local Means" is an innovative noise reduction algorithm for images presented by Buades and Morel in 2004. It performs remarkably better than older generation algorithms but has a performance penalty that prevents it
from being used in mainstream consumer application. The objective of this work is to find ways of reducing the time-complexity of the algorithm and enabling its use in main stream image processing applications such as home photography or photo printing centers.

[MCC31/07]
LOAIZA FERNANDEZ, M.E.; RAPOSO, A.B.; GATTASS, M. A novel optical tracking algorithm for point-based projective invariant marker patterns. 11 p. Eng. E-mail: mgattass@tecgraf.inf.puc-rio.br

Abstract: In this monograph, we describe a novel algorithm to group, label, identify and perform optical tracking of marker sets, which are grouped into two specific configurations, and whose projective invariant properties will allow
obtaining a unique identification for each predefined marker pattern. These configurations are formed by 4 collinear and 5 coplanar markers. This unique identification is used to correctly recognize various and different marker patterns inside the same tracking area, in real time. The algorithm only needs image coordinates of markers to perform the identification of marker patterns. For grouping the dispersed markers that appear in the image, the algorithm uses a "device and conquer" strategy to segment the image and give some neighborhood reference among markers.

[MCC32/07]
BIM, S.A.; LEITÃO, C.F.; SOUZA, C.S. The challenge of teaching HCI qualitative methods: a case study on the communicability evaluation method. 19 p. Eng. E-mail: clarisse@inf.puc-rio.br

Abstract: The challenges of teaching qualitative HCI evaluation methods for undergraduate students in computing-related programs are presented here through the description of a qualitative study about Communicability Evaluation Method (CEM), a Semiotic Engineering evaluation method, which is an example of qualitative HCI evaluation method. Different perspectives on CEM, expressed by the methods creators, by teachers, practitioners and students bring some interesting outcomes that are also shared by other qualitative HCI methods. Among, the most relevant results, is the fact that teachers themselves express their difficulty with mastering knowledge paradigms and methods that are not popular in their own professional training.

[MCC33/07]
NUNES, I.O.
Implementação do modelo e da arquitetura BDI. 13 p. Eng. E-mail: bib-di@inf.puc-rio.br

Abstract:
Agents with reasoning are been used to model intelligent behavior in multiagent systems. The architecture proposed by Rao and Geogreff in 1995, which is based in the BDI model (belief-desire-intention), has been used with success in situations where modeling the human reasoning is necessary. Several implementations have emerged for this architecture. Therefore, this work presents and compare four of these implementations: JACK, Jadex, JAM and Jason.

[MCC34/07]
CIARLINI, A.E.M.; CASANOVA, M.A.; FURTADO, A.L.; VELOSO, P.A.S. Treating literary genres as application domains. 42 p. Eng. E-mail: casanova@inf.puc-rio.br

Abstract: Literary genres, of prime relevance to storytelling, can be regarded as a particular kind of application domain. As such, they can be usefully characterized by combining notions drawn from literary theory with well-known models developed for information systems. Once a genre is specified with some rigor in a constructive way, it becomes possible to determine whether a given plot is a legitimate representative of the genre, as well as to generate such plots. This paper presents a conceptual modeling method with this purpose, based on a plan recognition / plan generation paradigm. The method leads to the formulation of static, dynamic and behavioral schemas, expressed in temporal logic, and allows multi-stage interactive plot generation. The paper also describes a prototype tool, developed to support the method, and includes a case study, involving a simple Swords and Dragons genre.