Monografias em Ciência da Computação
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, Pontífícia 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 firstname.lastname@example.org
Last update: 18/DECEMBER/2003
CASANOVA, M.A.; LEMOS, M. Optimized buffer management for sequence comparison in molecular biology databases. 21 p. Eng. E-mail: email@example.com
Abstract: Comparing sequences - nucleotide, aminoacid - is one of the basic operations every Molecular Biology database must support. There are two well known sequence comparison algorithms, FAST and BLAST, the latter being more popular (and many variations thereof). We propose in this paper a buffer management strategy to optimize the simultaneous execution of a set P of BLAST processes. The essence of the strategy is to partition P into subsets P1,...,Pn and to allocate a separate set Bi of buffers to each partition Pi. All sequences in the database will then be cycled through the buffers in Bi so that all BLAST processes in Pi will perform their sequence comparison synchronously. We first show that finding the partition that minimizes overall process delay is NP-complete. We then describe a heuristics that is computationally feasible and yet obtains a near optimal solution.
VELOSO, P.A.S. On some meanings of 'generally' and 'rarely'. 11 p. Eng. E-mail: firstname.lastname@example.org
Abstract: We examine some meanings of 'generally' and 'rarely'. The primary motivation is a precise approach to assertions and arguments involving such vague notions, which occur often both in ordinary language and in some branches of science. We focus mainly on the intended meanings of such assertions by analysing some basic intuitions and their underlying presuppositions. This leads to distinguishing various versions according to their behaviour.
VELOSO, P.A.S. On some postulates for 'generally' and 'rarely'. 16 p. Eng. E-mail: email@example.com
Abstract: We examine some postulates for 'generally' and 'rarely'. The primary motivation is a precise approach to assertions and arguments involving such vague notions, which accur often both in ordinary language and in some branches of science. Analysis of some basic intuitions and their underlying presuppositions leads to various distinguishing versions according to their behaviour. We then propose some postulates aiming at characterising the families of subsets corresponding to these notions.
VELOSO, P.A.S. On some logics for qualitative reasoning about 'generally'. 20 p. Eng. E-mail: firstname.lastname@example.org
Abstract: We examine logical systems, with generalised quantifiers, for expressing vague notions, such as 'generally', and reasoning about them. The primary motivation is a precise approach to assertions and arguments involving such vague notions, which occur often both in ordinary language and in some branches of science. Basic intuitions about 'generally' are rendered precise by means of appropriate families of subsets. This gives monotonic and conservative extensions of classical first-order logic, sharing with it various properties. Our logics, though related to default logics, are quite different, belonging to a family of closely connected systems with generalised quantifiers for qualitative reasoning about vague notions.
RENTERIA, R.P.; MILIDIÚ, R.L. APLS: a fast approximate algorithm for partial least-squares. 15 p. Eng. E-mail: email@example.com
Abstract: APLS, an approximate algorithm for solving the Partial Least-Squares problem is presented. When dealing with more than one dependent variable in the standard PLS modeling, one must compute eigenvectors of the mixed independent and dependent variables matrix, thus introducing a time overhead in the algorithm. APLS provides an aproximate method for this calculation with competitive modeling error and improved running time. In our experiments, APLS shows a 40% faster running time when compared to the standard PLS algorithm.
ABELEM, A.J.G.; STANTON, M.A.; RODRIGUEZ, N. QoS fim a fim através da combinação entre serviços integrados e serviços diferenciados. 17 p. Port. E-mail: firstname.lastname@example.org
Abstract: Several alternatives for offering end-to-end QoS in IP internetworks have been proposed over the last few years. Recently, researchers of the IETF ("Internet Engineering Task Force") have proposed a model which combines the integrated services/RSVP (intserv) and differentiated services (diffserv) architectures. In this scheme, the border routers between one region and another play fundamental role, as they are responsible for mapping QoS functionality between the different regions. In this article, an agent for NS simulator (Network Simulator) is presented, which was developed to perform tasks in support of this mapping, which include the choice of a suitable Per-Hop Behavior (PHB), appropriate policing at the border of the diffserv region, and admission control in accordance with the resources available in the diffserv regions.
ROSSETI, I.; ARAGÃO, M.P.; RIBEIRO, C.C.; UCHOA, E.; WERNECK, R.F. New Benchmark instances for the Steiner problem in graphs. 6 p. Eng. E-mail: email@example.com
Abstract: In this work, 50 Steiner problems in graph's instances are proposed. They are based on otimization problems as set covering and code covering. These instances are interesting because, in most of them, a good gap between linear relaxation and primal solution was obtained.
MEDINA, M.T.; RIBEIRO, C.C.; SOARES, L.F.G. Automatic scheduling of hypermedia documents with elastic times. 17 p. Eng. E-mail: firstname.lastname@example.org
Abstract: The problem of automatic scheduling hypermedia documents consists of finding the optimal starting times and durations of objects to be presented, so as to ensure spatial and temporal consistency of the presentation while respecting limits on shrinking and stretching the ideal duration of each object. The combinatorial nature of the minimization of the number of objects whose duration is modified makes it most difficult objective to be tackled by optimization algorithms. We formulated this scheduling problem as a mixed integer programming problem and we report some preliminary investigations and computational results. We propose an original approach to the minimization of the number of objects which are shrinked or stretched. A simple primal heuristic based on variable fixations along the solution of a sequence of linear relaxations of the mixed integer programming formulation is described. Computational experiments on realistic size problems are reported. The effectiveness of the heuristic in finding good approximate solutions within very small processing times makes of it a quite promising approach to be integrated within existing document formatters to perform compile time scheduling or even run time adjustments. We also discuss results obtained by Lagrangean relaxation and propose a dual heuristic using the dual multipliers and modified costs, which consistently improves the solutions found by the primal heuristic.
SILVA, O.R.; GARCIA, A.F.; LUCENA, C.J.P. Achieving system-level and agent-level dependability in multi-agent systems. 15 p. Eng. E-mail: email@example.com
Abstract: Development of multi-agent object-oriented systems is complicated by two main factors: the increasing complexity associated with their system-level and agent-level properties and the need to provide their dependability. This paper explores the recent advances of separation-of-concerns techniques to deal with the increasing complexity in developing dependable multi-agent object-oriented software. We propose a software architecture which: (i) uses aspect-oriented design and programming to handle separately each agency property and facilitate the provision of fault tolerance at the agent-level, and (ii) adopts a reflective associative blackboard architecture in oreder to manipulate separately each property and fault tolerance activities at the system-level. We also demonstrate our multi-agent approach through the Portalware system, a web-based environment for the development of e-commerce portals.
GEROSA, M.A.; FUKS, H.; LUCENA, C.J.P. Uso da categorização e estruturação de mensagens para dinamizar a discussão e reduzir a sobrecarga de informação em cursos via Internet. 17 p. Port.
Abstract: This paper shows how the use of categorization and structuring of messages in asynchronous textual communication tools could be useful in courses delivered via Internet in order to organize the argumentation and guide the participants to reflect about their messages. Although the use of categorization causes an increase in the total number of messages, there is a reduction at the information overload. The use of message categorization in a distance course through the AulaNet environment elaborates on how to define and improve the set of categories.
FUKS, H.; GEROSA, M.A.; LUCENA, C.J.P. The development and application of distance learning courses on the Internet. 14 p. Eng. E-mail: firstname.lastname@example.org
Abstract: This paper presents the methodology, results and difficulties encountered in the development and delivery of a course through the Internet. The course, which is about the application of information technology to education, is being delivered using the AulaNet environment, which is groupware for the creation, participation and maintenance of Web-based courses. A model for group work is provided. This paper also shows how this kind of environment can be used to provide support, and to facilitate and manage cooperative learning.
GEROSA, M.A.; FUKS, H.; LUCENA, C.J.P. Desenvolvimento e aplicação de cursos totalmente a distância na Internet - um estudo de caso. 17 p. Port. E-mail: email@example.com
Abstract: This paper presents the methodology, the results and the difficulties found in the development and delivery of a course through the Internet. The course is about the application of information technology to education. The course is being delivered using the AulaNet environment, a groupware for the creation, participation and maintenance of Web-based courses. This paper also shows how this kind of environment can be used to give support, facilitate and manage cooperative learning.
SILVEIRA, M.S.; BARBOSA, S.D.J. Revelando as affordances do designer via sistemas de help e interjeições de comunicabilidade. 41 p. Eng. E-mail: firstname.lastname@example.org
Abstract: Traditional online help is often function-oriented, unrelated to users' tasks and not situated at users' current context of interaction. This is due mainly to the help development strategy, and to the lack of a help model that takes into account typical users' task flows. In addition to a help model, we also need adequate access structures to the various components of the help system, each providing a different perspective on help content. This paper proposes the integration of a help model with communicability concepts, both based on Semiotic Engineering. Our goal is to provide users with better access and content, which are designed to clarify users' specific doubts, as expressed by users themselves, during interaction.
GARCIA, A.F.; SILVA, V.T.; LUCENA, C.J.P.; MILIDIÚ, R.L. An aspect-oriented design approach for multi-agent systems. 19 p. Eng. E-mail: email@example.com
Abstract: Objects and agents have many similarities, but agents are also driven by beliefs, goals, capabilities, plans, and a number of agency properties such as autonomy, adaptation, interaction, learning and mobility. Moreover, cooperating software agents must incorporate different collaborative capabilities in order to work together in heterogeneous contexts. In practice, a complex application is composed of objects and multiple types of agents, each of them having distinct agency properties and capabilities. An additional difficulty is that these capabilities and a properties typically overlap and interact with each other, and a disciplined scheme to composition is required. This paper dicusses software enginnering approaches for multi-agent systems, and presents a new approach for building multi-agent object-oriented software from early stage of design. This approach (i) describes structured integration of agents into the object model, (ii) incorporates flexible facilities to build different types of software agents, (iii) encourages the separate handling of each property and capability of an agent, (iv) provides explicit support for disciplined and transparent composition of agency properties and capabilities in complex software agents, and (v) allows the production of agent-based software so that it is easy to understand, maintain and reuse. The proposed approach explores the benefits of aspect-based design and programming for the incorporation of agents in object-oriented systems. We also demonstrate our multi-agent approach through the Portalware system, a web-based environment for the development of e-commerce portals.
MILIDIÚ, R.L.; PESSOA, A.A.; LABER, E.S. On-line pipeline transportation of petroleum products with no due dates. 24 p. Eng. E-mail: firstname.lastname@example.org
Abstract: In this paper, we introduce a new model for pipeline transportation petroleum products without due dates. We use a directed multigraph G where arcs represent pipes and nodes represent locations. Moreover, we define an order as a requirement to transport an amount of a given product. Orders are grouped into packages that are received on-line. For this model, we give a necessary and sufficient condition for the existence of a feasible pumping sequence for the next order package. Furthermore, we propose the Batch-to-Pipe Assignment (BPA) algorithm for the case we where all ordered volumes are unitary. This algorithm tests the feasibility condition and, if is satisfied, gives a feasible solution for the next order package in O (r3) time, where r is the number of orders. If the multigraph G is acyclic, then BPA leads to an optimal solution for the on-line problem. In addition, we prove that no algorithm is (1.25 - 0.625/n)-competitive against an adaptive adversary, and no randomized algorithm is (1.125 - 0.3125/n)-competitive against an oblivious adversary when n is the number of nodes in G. Finally, we show that our results can be extended to non-unitary orders.
CORTES, S.C.; LUCENA, C.J.P. Um framework de regras ativas para sistemas de gerência de banco de dados distribuído. 25 p. Port. E-mail: email@example.com
Abstract: Active database management systems make possible the development of advanced database applications, giving support to a semantics that reflects the application behavior through the definition of rules, which specify actions that are automatically triggered by the occurrence of certain events. The use of active rules in a distributed environment allows obtaining more expressiveness, although with a considerable increase of the system complexity. This work proposes a framework for constructing active distributed database management systems that applies the reuse technique (framework) in order to benefit by the specifications which must exist in all systems of this kind, at the same time that allows being configured/extended to specific operational environments, with different rule semantics.
SILVEIRA, M.S.; de SOUZA, C.S. Análise comparativa de sistemas de help online como fonte para avaliação heurística destes sistemas. 42 p. Port. E-mail: firstname.lastname@example.org
Abstract: How can we know precisely who will be accessing a designed computer application, and when? If we don't know this, and thus can't design an application targeted at a specific user group, how can we better help different users to use it in a variety of situations? In order to support users, an application needs to provide them with a help system that not only helps them to do their tasks, but also to understand why the application behaves the way it does. In this work, we have made a comparative analysis of a few office applications' online help systems. Based on this analysis, we propose guidelines that may be used in heuristic help evaluation, in order to better support designers in the creation of more efficient help systems.
GUELL, N.; SCHWABE, D.; BARBOSA, S.D.J. Método de avaliação de usabilidade na Web baseado em modelo e padrões de comportamento. 20 p. Port. E-mail: email@example.com
Abstract: This document proposes a usability evaluation method for Web applications. The proposed method helps designers to diagnose and solve design problems. Our approach is to recognize, during user's interaction with the system, behaviors that are conduct patterns for typical design problems. Initially, behavioral patterns were described, classified and mapped onto classes of design problems. Then, the design problem indicated by each behavioral pattern is diagnosed using the OOHDM methodology. Finally, some design solutions are proposed based on Guidelines and Design Patterns.
RODRIGUES, R.F.; BACHELET, B.; MAHEY, P.; SOARES, L.F.G. Elastic time computation for hypermedia documents. 17 p. Eng. E-mail: firstname.lastname@example.org
Abstract: This paper presents a new elastic time algorithm for adjusting hypermedia document presentation in order to avoid temporal inconsistencies. The algorithm explores the flexibility offered by some hypermedia models in the definition of media object duration and uses this flexibility for choosing the objects to be stretched or shrunk in such a way that the best possible quality of presentation is obtained. Our proposal is based on the "out-of-killer" method for minimum-cost flow problems and it seems to offer more flexibility to model real-time applications in comparison with other previous proposals based on the simplex method.
SEIBEL, L.F.B.; LIFSCHITZ, S. A genome databases framework. 20 p. Eng. E-mail: email@example.com
Abstract: There exists many molecular biology databases, also known as Genome Databases, that need to be integrated together with all related applications and other data sources. This work proposes the use of an object-oriented framework for genome data access and manipulations. The framework approach may be an interesting solution in this context due to the flexibility, reusability and extensibility requirements of this application domain. We will present and discuss the framework in details using class and sequence diagrams that explore both the structural and dynamic parts of the architecture.
MATHIAS FILHO, I.; LUCENA, C.J.P. Uma proposta para a incorporação do modelo de features à linguagem UML. 17 p. Port. E-mail: firstname.lastname@example.org
Abstratc: Feature modeling is one of the most successful techniques in use to promote the reuse of software artifacts as of the initial stages of development. The purpose of this paper is to present a proposal for the incorporation of feature modeling into the UML language, wich has become a standard for the modeling of object-oriented software. The precise and careful specification of this proposal, wich uses formal and informal techniques such as graphic notation, natural language and formal language, will permit the building of tools that provide support to the development of feature models for families of applications, and the instantiation of members of these families based on these models. Moreover, the use of the XMI standard in the representation of the feature models will facilitate their integration with the many CASE tools already existing on the market.
OLIVEIRA, T.C.; MATHIAS FILHO, I.; LUCENA, C.J.P. A framework based approach for workflow software development. 8 p. Eng. E-mail: email@example.com
Abstract: The purpose of Workflow Software is to provide users with an environment in wich they can manipulate and control all the data and tasks involved in achieving a predefined goal (main task). Some features as user-interface, persistency, online guidance and multi-user support, are all-important and should be present in the final software, to obtain a high level of usability. It is also important that the design of the software should take place in an environment that helps the designer to deal with the large amount of definitions obtained during the domain analysis phase and gap between these definitions and the object oriented design model. This paper describes the main features of a tool to support a method that aids the designer to develop workflow software.
MATHIAS FILHO, I.; OLIVEIRA, T.C.; LUCENA, C.J.P. Domain oriented framework construction. 16 p. Eng. E-mail: firstname.lastname@example.org
Abstract: Object-oriented Application Framework is a powerful reuse technique that allows the sharing off requirements, design and code among a set of applications systems instantiated from an original framework. Nevertheless, little attention has been given to the role of an explicit Domain Analysis phase in the framework construction process. This paper describes an approach where the requirements capture for an entire application family takes a central role in the framework development, thus facilitating verification processes and providing reliable documentation that will assist the instantiation step.
OLIVEIRA, T.C.; MATHIAS FILHO, I.; LUCENA, C.J.P. Using XML and frameworks to develop information systems. 11 p. Eng. E-mail: email@example.com
Abstract: To accomplish the software development time and cost constraints this develpoment should take place in an environment that helps the designer to deal with the large amount of concepts obtained during the domain analysis phase and the semantic gap between those concepts and the object oriented desing model, due to their different levels of abstraction. This paper discribes the main features of an environment designed to support the development of IS software based on framework reuse and XML specifications.
MUCHALUAT-SAADE, D.C.; COLCHER, S.; SOARES, L.F.G. Relações de sincronização espaço-temporal hipermídia também merecem status de primeira classe. 18 p. Port. E-mail: firstname.lastname@example.org
Abstract: This paper introduces the concept of hypermidia connector, presenting how spatio-temporal relations may be defined as a specialization of architecture description languages (ADL) connectors. Some features found in ADLs were brought to hypermidia authoring languages, giving first-class status to synchronization relations expressed by links. Among the profits, we can highlight the possibility of reusing a spatio-temporal relation specification in different links with the same behavior and the possibility of defining composite spatio-temporal relations expressed by links, which is an original feature in hypermidia authoring languages.
SILVEIRA, M.S.; DIAS, M.C.; QUENTAL, V.S.D Estudos preliminares para composição de conteúdo de help através de interjeições de comunicabilidade. 46 p. Port. E-mail: email@example.com
Abstract: Accessing help systems by What's this? expressions is easily found in commercial WIMP applications. An improvement to this kind of access is to use communicability utterances, such as Where is?, What now?, and Why should I do this?, for instance. But how can the designer answer these questions? What kind of answer is needed for each one? This work describes studies to help the designer build this kind of answer when he is developing the applications' help systems.
CASTILHO, R.T.L.; GARCIA, A.H. Theses and dissertations presented to the Departamento de Informática of PUC-Rio from 1969 to 2000. 206 p. Eng. E-mail: firstname.lastname@example.org
Abstract: Catalogue encompassing more than 30 years of dissertations and theses from the PUC-Rio Information Technology Department, the first post-graduate Computer Science program in Latin America. It evidences the continued stability of the program in the state-of-the-art of the discipline, is invested of relevant historical value, being a significant example of the evolution of scientific and technological research into the field of Computing in Brazil and is also a record of the many researchers that have populated and continue to do so the academic Computing community in Brazil.
COSTA, R.L.C.; LIFSCHITZ, S. Uma proposta de arquitetura de SGBD baseado em agentes para auto-sintonia de índices. 30 p. Port. E-mail: email@example.com
Abstract: We present an architecture based on software agents for DBMSs' self-tuning. In particular, we are interested in the features of indexes' self-tuning. Thus, an engine is considered that makes possible the automatic selection, creation and exclusion of indexes. A case study is described and reveals adequacy to the agent based architecture specified.
CHAVEZ, C.F.G.; GARCIA, A.F.; LUCENA, C.J.P. From AOP to MDSOC: an experience report. 19 p. Eng. E-mail: firstname.lastname@example.org
Abstract: In this report, we present some insights gathered through the implementation of the Portalware system using the IBM Hyper/J tool, having as a point an aspect-oriented design model and an aspect-oriented implementation (using the Xerox Parc's AspectJ tool). We present a preliminary mapping between the elements of the two approaches and discuss some alternatives for deriving an hyperslice-based solution from the Portalware aspect-oriented model. We describe the selected set of tranformations we have applied, some potential problems and solutions related to the transformation process, and finally, we present resulting Hyper/J solution. One of the motivations behind this development exercise is a pratical assessment of each approach's similarities and differences, strengths and weaknesses to help in the validation of a generic aspect-based design model that can be mapped to the implementation models supplied by AspectJ and Hyper/J.
SILVA, V.T.; LUCENA, C.J.P. Um modelo orientado a objetos para sistemas multi-agentes. 14 p. Port. E-mail: email@example.com
Abstract: In this work, we propose a mapping from agents, and its fundamental properties and mental states, into objects. The proposed objective of the mapping is generate a object-oriented model for multi-agent systems to facilitate the maintainability and reusability of the elements involved in the model. To assure these objectives, a separation of concerns technique namely design patterns was applied to the model. Design patterns provides separation of concerns because they decouple the concerns of interest in classes or in groups of inter-related classes. On the mapping the intra-agents aspects (agent's properties and agent's mental states) have been modeled as well as inter-agents aspects (communication between agents and the environment). The modeling of these aspects and the focus on the utilization of the mental states to model agents makes this approach more appropriate to model cognitive agents (that execute plans based on their goals and beliefs) than reactive agents.
PINHEIRO, S.; VELHO, L.; GATTASS, M. A virtual memory system for real-time visualization of multi-resolution 2D objects. 8 p. Eng. E-mail: firstname.lastname@example.org
Abstract: We describe a predictive memory management system that allows the visualization of 2D graphic objects in real time. This system is based on a virtual memory model. We show how the page loading system predicts and fetches data before it is accessed by the application and how the page replacement system decides which data should be removed from memory. We demonstrate the effectiveness of the system for real-time visualization of large panoramas.
MILIDIÚ,R.L.; PESSOA, A.A. Complexidade de problemas de planejamento de distribuição em oleodutos. 13 p. Port. E-mail: email@example.com
Abstract: Petroleum derivates are typically transported through pipelines in Brazil, and also in other countries. Pipeline networks can be represented by a graph where nodes represent locations and arcs represent oil pipeline segments. A natural approach to solve the distribution problem - PDO - that arises in oil pipeline networks is to develop efficient algorithms for simplified versions of this problem. In this paper, we consider two simplified versions of PDO:PDOS-2, a decision problem with only two distinct products, and PDOP-3, a decision with only three distinct products and a planar graph. We point out that both PDOS-2 and PDOP-3 include several additional simplifying assumptions. Nevertheless, we prove that both problems are NP-hard. We also consider two optimization problems respectively based on PDOS-2 and PDOP-3. In both cases, an optimal solution minimizes the makespan of the pumping operations. For both problems, we prove that there is no(2 - e)-approximate algorithm, for any e>0, unless P=NP.
CORTES, M.; FONTOURA, M.; LUCENA, C. Using refactoring and unification rules to assist framework evolution. 9 p. Eng. E-mail: firstname.lastname@example.org
Abstract: Although object-oriented software development has experienced the benefits of using frameworks, a thorough understanding of how to change them to meet evolving requirement needs is still object of research. Therefore framework development is very expensive, not only because of the intrinsic difficulty related to capturing the domain theory, but also because of the lack of appropriate methods and techniques to support the evolution and redesign of the framework architecture. This paper proposes the use of refactoring and unification rules to assist framework evolution. The approach is illustrated through the JUnit testing framework.
MOURA, A.L.; URURAHY, C.; CERQUEIRA, R.; RODRIGUEZ, N. Dynamic support for nonfunctional requirements in distributed applications. 17 p. Eng. E-mail: email@example.com
Abstract: This work presents an environment that allows distributed applications to adapt dynamically to nonfunctional properties of their components. This environment, based on the programming language Lua and on CORBA, allows applications to select dynamically the components that best suit their requirements, to verify whether the system is satisfying these requirements, and react, when appropriate, to variations in the nonfunctional properties of the services in use. The paper also describes an example application based on the proposed environment.
CORTES, S.C.; LUCENA, C.J.P. Um framework para construção de sistemas de banco de dados móvel com regras ativas. 17 p. Port. E-mail: firstname.lastname@example.org
Abstract: This study presents an overview of the DBS-M-AR framework, with emphasis on its components and hot spots. This is an object-oriented framework for the construction of Mobile Database Management Systems with Active Rules. A detailed discussion of its instantiation is also included.
MONTENEGRO, A.A.; RIBEIRO, C.C; ARAGÃO, M.V.S.P. Um algoritmo randomizado para o problema de determinação do corte s-t mínimo em grafos direcionados. 42 p. Port. E-mail: email@example.com
Abstract: The minimum cut problem in a direct graph consists basically in finding a partitioning of the set of vertices of the graph in two subsets, in such a way that the number (or the sum of the weights) of the edges that the number (or the sum of the wheights) of the edges that connect the partitions is minimized. An important variation of this problem is the s-t minimum cut problem, in which we impose the restriction that each candidadte cut separate two special vertices s and t of the input set. As far as we are know, no eficient and simple randomized algorithm has been proposed to solve this problem. In this work we investigate the applicability of randomization techniques to the s-t minimum, proposing an algorithm and analising its complexity.
MONTENEGRO, A.A.; CARVALHO, P.C.P. Algoritmos de fluxo máximo para o problema de correspondência em visão estéreo. 55 p. Port. E-mail: firstname.lastname@example.org
Abstract: The epipolar constraint is one of the most important properties used to solve the stereo correspondence problem. This constraint makes possible the solution of the problem for each pair of epipolar lines separetly. The main drawback with this approach is that it does not exploit the coeherence that exists among the epipolar lines, yielding some artifacts in the final solution. In this work we investigate some recently proposed methods based on network flow algorithms that are capable of taking into consideration the coherence among epipolar lines.
MONTENEGRO, A.A.; GATTASS, M. Determinação de superfícies de profundidade através de visão estéreo. 48 p. Port. E-mail: email@example.com
Abstract: In this work we present the basic theoretical and computational aspects of stereo vision. Our main focus is in the geometry involved in the process as well as in the way the problem can be modelled. Some traditional methods, as the correlation based and the minimum path based methods, are presented in detail as exemples of how the problem can be solved computationally.
CRUZ, S.O.; LUCENA, C.J.P.; RANGEL, J.L.M. Identificando objetos através de pronomes. 26 p. Port. E-mail: firstname.lastname@example.org
Abstract: The need to know the object identification to request it's sevices is a restriction in the construction of object oriented system. To reduce this restriction we have proposed a form of generic reference to an object named pronoum. Pronoum allows the programmer to specify how to send a message without knowing the destination object by name. With pronouns, the visibility of an object increases without breaking the rule of encapsulation. With this feature decoupled server classes may be constructed and used in different contexts. Some pronouns are defined in this work. Their usefulness in the development of object oriented systems is illustrated through examples. An implementation approach based on open languages is suggested.
GARCIA, A.F.; CHAVEZ, C., SILVA, V.T.; LUCENA, C.J.P. Developing multi-agent software: an aspect-based approach and a pattern-based approach. 16 p. Eng. E-mail: email@example.com
Abstract: This paper proposes and compares two software engineering approaches for multi-agent systems.The first proposal explores the benefits of aspect-based design and programming and the second one is based on design patterns. Both approaches have the following goals: (i) describe structured integration of agents into the object model, (ii) incorporate flexible facilities to build different types of software agents, (iii) encourage the separate handling of each property and capability of an agent, (iv) provide explicit support for disciplined and transparent composition of agency properties and capabilities in complex software agents, and (v) allow the production of agent-based software so that it is easy to understand, maintain and reuse. We also demonstrate our proposals through the Portalware system, a web-based environment for the development of e-commerce portals.
CUNHA, L.M.; GARCIA, A.F.; ASSIS, R.L.; FUKS, H.; LUCENA, C.J.P. O "acompanhamento da participação" e avaliação no ambiente AulaNet. 8p. Port. E-mail: firstname.lastname@example.org
Abstract: Student evaluation in Web-based courses has to be part of the process of learning, i.e., it should not be dealt as an isolated issue. In this context, this paper presents evaluation concepts and their application in the AulaNet environment, presenting a mechanism for assessing the participation of the students. The mechanism goal is to provide information about the student evolution while performing learning activities. This information can be visualized by means of the follow-up report.
MOURA, M.A.M.; ENDLER, M. An infrastructure for management of fault tolerant CORBA applications. 21 p. Eng. E-mail: email@example.com
Abstract: In the last years CORBA has become a de facto standard for the development of object-oriented distributed applications. In spite of some important features such as support for interoperability and heterogeneity, only very recently CORBA specification, with the adoption of the Fault Tolerant CORBA standard (FT-CORBA), addressed issues concerning object replication and reliable communication between objects, which are key components to provide fault tolerance for distributed object systems. Since mid-90's several research projects have proposed and explored some approaches to incorporate fault tolerance mechanisms in CORBA but most of the systems developed offer a rather basic support for managing groups of object replicas. Our work builds upon one of these systems, called OGS. We created an extended OGS (called OGS+) and on the top of it we developed Juggler, a distributed services that provides means for flexible and automatic management of CORBA replicated objects through the deployment of a set of high-level interfaces based on the ones standardized by FT-CORBA.