José Everardo Bessa Maia | Universidade Estadual do Ceara (original) (raw)
Papers by José Everardo Bessa Maia
International Journal of Engineering, Nov 24, 2021
International Journal of Computer Applications
A great challenge in Information Retrieval Systems (IRS) is to extract the information intention ... more A great challenge in Information Retrieval Systems (IRS) is to extract the information intention of the user from a command line interface query, so it can recover relevant documents. This problem gets worse in Question-Answering Systems (QAS) in a Closed Domain, for in this scenario, there's a higher divergence between the open language available for the user to elaborate questions and the limited vocabulary in the document collection available in the system (which is usually small). This work proposes and evaluates a system of Query Expansion (QE) for a closed domain QAS based on the semantic similarity between terms of the Word Net and a previously built semantic model using the system's knowledge base. The tests are made by answering questions about the two closed collections of documents showed this method is effective in improving performance of the Closed Domain QAS.
Anais do XI Computer on the Beach - COTB '20
In the Cooperative Targets Observation (CTO) problem, a groupof observing agents with limited vis... more In the Cooperative Targets Observation (CTO) problem, a groupof observing agents with limited vision should be commanded inorder to keep the observation of multiple target agents in motion inorder to maximize the average number of targets observed duringthe period under consideration. Targets are cooperative in the sensethat they do not run away from observers. This article describes andevaluates the application of a modified Hill-Climbing algorithm tothe CTO problem when the movement of the targets is restricted to aplanar graph. It is argued howthis newconfiguration of the problemcan be representative of a practical situation. The performance ofthe proposed algorithm surpasses that of two other algorithmswidely applied to the problem that are classic Hill-Climbing andK-Means.
Anais do XI Computer on the Beach - COTB '20
This paper presents the architecture and operation of a HistoricalNewspaper Page Image Topic Navi... more This paper presents the architecture and operation of a HistoricalNewspaper Page Image Topic Navigation System designed tofacilitate the access and use of social and historical research tothe historical newspaper collection. The system consists of fourmodules which are: Text Subimage Segmentation, Text Extractionand Preprocessing, Topic Network Extraction, and Document Viewingand Retrieval Interface. The algorithmic and technological approachesof each module are described and the initial test resultsare presented.
Anais do Encontro Nacional de Inteligência Artificial e Computacional (ENIAC)
In this work, a modeling and algorithm based on multiagent reinforcement learning is developed fo... more In this work, a modeling and algorithm based on multiagent reinforcement learning is developed for the problem of elevator group dispatch. The main advantage is that, along with the function approximation, this multi-agent solution leads to reduction of the state space, allowing complex states to be addressed with a synthesizing evaluation function. Each elevator is considered an agent that have to decide about two actions: answer or ignore the new call. With some iterations, the agents learn the weights of an evaluation function which approximate the state-action value function. The performance of solution (average waiting time - AWT), shown varying the traffic pattern, flow of people, number of elevators and number of floors, is comparable to other current proposals reported in the literature.
Proceeding XIII Brazilian Congress on Computational Inteligence
Journal of Intelligent Computing
Anais do XV Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2018)
This work proposes and evaluates an approach to query expansion in Information Retrieval based on... more This work proposes and evaluates an approach to query expansion in Information Retrieval based on Local Context Analysis using a Distributional Semantic Representation. In general, the approach performed better compared to that of query expansion using non-distributional, local or global techniques, running over datasets of different application domains.
Anais do XV Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2018)
This article presents Luppar, an Information Retrieval tool for closed collections of documents w... more This article presents Luppar, an Information Retrieval tool for closed collections of documents which uses a local distributional semantic model associated to each corpus. The system performs automatic query expansion using a combination of distributional semantic model and local context analysis and supports relevancy feedback. The performance of the system was evaluated in databases of different domains and presented results equal to or higher than those published in the literature.
Anais do XV Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2018)
This paper presents an approach that uses topic models based on LDA to represent documents in tex... more This paper presents an approach that uses topic models based on LDA to represent documents in text categorization problems. The document representation is achieved through the cosine similarity between document embeddings and embeddings of topic words, creating a Bag-of-Topics (BoT) variant. The performance of this approach is compared against those of two other representations: BoW (Bag-of-Words) and Topic Model, both based on standard tf-idf. Also, to reveal the effect of the classifier, we compared the performance of the nonlinear classifier SVM against that of the linear classifier Naive Bayes, taken as baseline. To evaluate the approach we use two bases, one multi-label (RCV-1) and another single-label (20 Newsgroup). The model presents significant results with low dimensionality when compared to the state of the art.
Proceedings of the 30th Annual ACM Symposium on Applied Computing - SAC '15, 2015
Anais do Encontro Nacional de Inteligência Artificial e Computacional (ENIAC)
This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor... more This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor Networks (WSNs). The energy savings are obtained with the sensor nodes, no longer sensing and transmitting measurements. In this simple strategy, in each epoch each sensor node transmits its measurement with probability p, and does not transmit with probability (1 p), does not correlate with that of any other sensor node. The task at the sink node, which is to estimate the sensor field at non-sensed points, is solved using fuzzy inference to impute the non-transmitted data followed by regression or interpolation of the sensed scalar field. In this, Nadaraya-Watson regression, regression with Fuzzy Inference and Radial Base Functions Interpolation are compared. The compromise curve between the value of p and the accuracy of the sensor field estimation measured by root mean square error (RMSE) is investigated. When compared to a published linear prediction strategy of the literature, th...
Anais do Encontro Nacional de Inteligência Artificial e Computacional (ENIAC)
This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor... more This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor Networks (WSNs). The energy savings are obtained with the sensor nodes, no longer sensing and transmitting measurements. In this simple strategy, in each epoch each sensor node transmits its measurement with probability p, and does not transmit with probability (1 p), does not correlate with that of any other sensor node. The task at the sink node, which is to estimate the sensor field at non-sensed points, is solved using fuzzy inference to impute the non-transmitted data followed by regression or interpolation of the sensed scalar field. In this, Nadaraya-Watson regression, regression with Fuzzy Inference and Radial Base Functions Interpolation are compared. The compromise curve between the value of p and the accuracy of the sensor field estimation measured by root mean square error (RMSE) is investigated. When compared to a published linear prediction strategy of the literature, th...
Resumo. Muitas pesquisas em redes de sensores sem fio (RSSF) têm sido desenvolvidas nos últimos a... more Resumo. Muitas pesquisas em redes de sensores sem fio (RSSF) têm sido desenvolvidas nos últimos anos, com foco na economia de energia dos nós sensores. Para alcançar este objetivo, tais pesquisas utilizam como estratégia a redução de dados enviados na rede. Neste trabalho, é proposta uma estratégia eficiente de agregação e predição de dados em RSSF, com este objetivo, aumentando a vida útil da RSSF. Além do uso da agregação em rede, a proposta apresentada neste artigo introduz o conceito de predição em rede, realizada no processamento de consultas sobre RSSFs. A predição proposta utiliza um modelo de regressão linear, com base nos dados coletados de um único sensor ou com base nos dados de um subconjunto de sensores. Ela é executada de forma distribuída nos diversos sensores de uma RSSF. Os resultados experimentais mostram que a estratégia proposta pode reduzir significativamente o consumo de energia em RSSFs. Abstract. Over the past few years, many research works in Wireless Sensor Networks (WSN) have been focusing on node power saving. In order to achieve this goal, the amount of data sent over the node network is usually reduced. In this work, we propose an efficient strategy that aggregates and predicts data in WSN, aiming to reduce the data volume sent over the network and thus maximizing the network lifetime. Besides the widely used in-network aggregation strategy, this work presents the use of in-network prediction, based on query processing on the network data. Our prediction strategy works with a linear regression model, using data acquired from one or several sensor nodes. It is implemented in various sensor nodes distributed in a WSN. Experimental results show that our strategy is able to significantly reduce power consumption in WSN.
In this work, we show an algorithm to analyze the information of the alarms emitted by the compon... more In this work, we show an algorithm to analyze the information of the alarms emitted by the components of an optical network, in the presence of a fault, aiming to find which network component is causing the failure. The algorithm proposed, named Correlated Fault location Algorithm, CFLA, uses the alarms’ correlation in order to reduce the list of suspected components shown to the network operators. The CFLA is sufficiently robust to work with false alarms, lost alarms and components that do not emit alarms when fail. The algorithm is described and simulated. The results of the simulations are compared to those obtained using another algorithm found in the literature.
sbmac.org.br
... Jorge L. Castro e Silva, José Everardo B. Maia Universidade Estadual do Ceará - Departamento ... more ... Jorge L. Castro e Silva, José Everardo B. Maia Universidade Estadual do Ceará - Departamento de Estatítica e Computação 60740-903, Campus do Itaperi, Fortaleza, CE E-mail: jlcs@larces.uece.br, jmaia@larces.uece.br ...
International Journal of Engineering, Nov 24, 2021
International Journal of Computer Applications
A great challenge in Information Retrieval Systems (IRS) is to extract the information intention ... more A great challenge in Information Retrieval Systems (IRS) is to extract the information intention of the user from a command line interface query, so it can recover relevant documents. This problem gets worse in Question-Answering Systems (QAS) in a Closed Domain, for in this scenario, there's a higher divergence between the open language available for the user to elaborate questions and the limited vocabulary in the document collection available in the system (which is usually small). This work proposes and evaluates a system of Query Expansion (QE) for a closed domain QAS based on the semantic similarity between terms of the Word Net and a previously built semantic model using the system's knowledge base. The tests are made by answering questions about the two closed collections of documents showed this method is effective in improving performance of the Closed Domain QAS.
Anais do XI Computer on the Beach - COTB '20
In the Cooperative Targets Observation (CTO) problem, a groupof observing agents with limited vis... more In the Cooperative Targets Observation (CTO) problem, a groupof observing agents with limited vision should be commanded inorder to keep the observation of multiple target agents in motion inorder to maximize the average number of targets observed duringthe period under consideration. Targets are cooperative in the sensethat they do not run away from observers. This article describes andevaluates the application of a modified Hill-Climbing algorithm tothe CTO problem when the movement of the targets is restricted to aplanar graph. It is argued howthis newconfiguration of the problemcan be representative of a practical situation. The performance ofthe proposed algorithm surpasses that of two other algorithmswidely applied to the problem that are classic Hill-Climbing andK-Means.
Anais do XI Computer on the Beach - COTB '20
This paper presents the architecture and operation of a HistoricalNewspaper Page Image Topic Navi... more This paper presents the architecture and operation of a HistoricalNewspaper Page Image Topic Navigation System designed tofacilitate the access and use of social and historical research tothe historical newspaper collection. The system consists of fourmodules which are: Text Subimage Segmentation, Text Extractionand Preprocessing, Topic Network Extraction, and Document Viewingand Retrieval Interface. The algorithmic and technological approachesof each module are described and the initial test resultsare presented.
Anais do Encontro Nacional de Inteligência Artificial e Computacional (ENIAC)
In this work, a modeling and algorithm based on multiagent reinforcement learning is developed fo... more In this work, a modeling and algorithm based on multiagent reinforcement learning is developed for the problem of elevator group dispatch. The main advantage is that, along with the function approximation, this multi-agent solution leads to reduction of the state space, allowing complex states to be addressed with a synthesizing evaluation function. Each elevator is considered an agent that have to decide about two actions: answer or ignore the new call. With some iterations, the agents learn the weights of an evaluation function which approximate the state-action value function. The performance of solution (average waiting time - AWT), shown varying the traffic pattern, flow of people, number of elevators and number of floors, is comparable to other current proposals reported in the literature.
Proceeding XIII Brazilian Congress on Computational Inteligence
Journal of Intelligent Computing
Anais do XV Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2018)
This work proposes and evaluates an approach to query expansion in Information Retrieval based on... more This work proposes and evaluates an approach to query expansion in Information Retrieval based on Local Context Analysis using a Distributional Semantic Representation. In general, the approach performed better compared to that of query expansion using non-distributional, local or global techniques, running over datasets of different application domains.
Anais do XV Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2018)
This article presents Luppar, an Information Retrieval tool for closed collections of documents w... more This article presents Luppar, an Information Retrieval tool for closed collections of documents which uses a local distributional semantic model associated to each corpus. The system performs automatic query expansion using a combination of distributional semantic model and local context analysis and supports relevancy feedback. The performance of the system was evaluated in databases of different domains and presented results equal to or higher than those published in the literature.
Anais do XV Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2018)
This paper presents an approach that uses topic models based on LDA to represent documents in tex... more This paper presents an approach that uses topic models based on LDA to represent documents in text categorization problems. The document representation is achieved through the cosine similarity between document embeddings and embeddings of topic words, creating a Bag-of-Topics (BoT) variant. The performance of this approach is compared against those of two other representations: BoW (Bag-of-Words) and Topic Model, both based on standard tf-idf. Also, to reveal the effect of the classifier, we compared the performance of the nonlinear classifier SVM against that of the linear classifier Naive Bayes, taken as baseline. To evaluate the approach we use two bases, one multi-label (RCV-1) and another single-label (20 Newsgroup). The model presents significant results with low dimensionality when compared to the state of the art.
Proceedings of the 30th Annual ACM Symposium on Applied Computing - SAC '15, 2015
Anais do Encontro Nacional de Inteligência Artificial e Computacional (ENIAC)
This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor... more This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor Networks (WSNs). The energy savings are obtained with the sensor nodes, no longer sensing and transmitting measurements. In this simple strategy, in each epoch each sensor node transmits its measurement with probability p, and does not transmit with probability (1 p), does not correlate with that of any other sensor node. The task at the sink node, which is to estimate the sensor field at non-sensed points, is solved using fuzzy inference to impute the non-transmitted data followed by regression or interpolation of the sensed scalar field. In this, Nadaraya-Watson regression, regression with Fuzzy Inference and Radial Base Functions Interpolation are compared. The compromise curve between the value of p and the accuracy of the sensor field estimation measured by root mean square error (RMSE) is investigated. When compared to a published linear prediction strategy of the literature, th...
Anais do Encontro Nacional de Inteligência Artificial e Computacional (ENIAC)
This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor... more This work presents and evaluates a fuzzy-probabilistic strategy to save energy in Wireless Sensor Networks (WSNs). The energy savings are obtained with the sensor nodes, no longer sensing and transmitting measurements. In this simple strategy, in each epoch each sensor node transmits its measurement with probability p, and does not transmit with probability (1 p), does not correlate with that of any other sensor node. The task at the sink node, which is to estimate the sensor field at non-sensed points, is solved using fuzzy inference to impute the non-transmitted data followed by regression or interpolation of the sensed scalar field. In this, Nadaraya-Watson regression, regression with Fuzzy Inference and Radial Base Functions Interpolation are compared. The compromise curve between the value of p and the accuracy of the sensor field estimation measured by root mean square error (RMSE) is investigated. When compared to a published linear prediction strategy of the literature, th...
Resumo. Muitas pesquisas em redes de sensores sem fio (RSSF) têm sido desenvolvidas nos últimos a... more Resumo. Muitas pesquisas em redes de sensores sem fio (RSSF) têm sido desenvolvidas nos últimos anos, com foco na economia de energia dos nós sensores. Para alcançar este objetivo, tais pesquisas utilizam como estratégia a redução de dados enviados na rede. Neste trabalho, é proposta uma estratégia eficiente de agregação e predição de dados em RSSF, com este objetivo, aumentando a vida útil da RSSF. Além do uso da agregação em rede, a proposta apresentada neste artigo introduz o conceito de predição em rede, realizada no processamento de consultas sobre RSSFs. A predição proposta utiliza um modelo de regressão linear, com base nos dados coletados de um único sensor ou com base nos dados de um subconjunto de sensores. Ela é executada de forma distribuída nos diversos sensores de uma RSSF. Os resultados experimentais mostram que a estratégia proposta pode reduzir significativamente o consumo de energia em RSSFs. Abstract. Over the past few years, many research works in Wireless Sensor Networks (WSN) have been focusing on node power saving. In order to achieve this goal, the amount of data sent over the node network is usually reduced. In this work, we propose an efficient strategy that aggregates and predicts data in WSN, aiming to reduce the data volume sent over the network and thus maximizing the network lifetime. Besides the widely used in-network aggregation strategy, this work presents the use of in-network prediction, based on query processing on the network data. Our prediction strategy works with a linear regression model, using data acquired from one or several sensor nodes. It is implemented in various sensor nodes distributed in a WSN. Experimental results show that our strategy is able to significantly reduce power consumption in WSN.
In this work, we show an algorithm to analyze the information of the alarms emitted by the compon... more In this work, we show an algorithm to analyze the information of the alarms emitted by the components of an optical network, in the presence of a fault, aiming to find which network component is causing the failure. The algorithm proposed, named Correlated Fault location Algorithm, CFLA, uses the alarms’ correlation in order to reduce the list of suspected components shown to the network operators. The CFLA is sufficiently robust to work with false alarms, lost alarms and components that do not emit alarms when fail. The algorithm is described and simulated. The results of the simulations are compared to those obtained using another algorithm found in the literature.
sbmac.org.br
... Jorge L. Castro e Silva, José Everardo B. Maia Universidade Estadual do Ceará - Departamento ... more ... Jorge L. Castro e Silva, José Everardo B. Maia Universidade Estadual do Ceará - Departamento de Estatítica e Computação 60740-903, Campus do Itaperi, Fortaleza, CE E-mail: jlcs@larces.uece.br, jmaia@larces.uece.br ...