Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9640223
R. D. Santos, José Aguilar, E. Puerto
In Machine Learning (ML), there is a lot of research that seek to automate specific processes carried out by data scientists in the generation of knowledge models (predictive, classification, clustering, etc.); however, an open problem is to find mechanisms that allow conferring the ability of self-learning. Thus, a meta-learning mechanism is required to allow ML techniques to self-adapt in order to improve their performance in problem solving, and even in some cases, to induce the learning algorithm itself. In this context, our research defines a meta-learning architecture using Linked Data (LD) for the automatic generation of knowledge models. Specifically, this intelligent architecture is formed by the layers of Knowledge Sources, Meta-Knowledge and Knowledge Modelling, to unify all processes to guarantee a Meta-Learning process. The Knowledge Sources layer is responsible for providing semantic knowledge about the processes of generation of knowledge models; the Meta-Knowledge layer is responsible for controlling the different processes and strategies for the automatic generation of knowledge models; and finally, the Knowledge Modelling layer is responsible for executing ML tasks defined by the Meta-Knowledge layer, among which are the tasks of feature engineering, ML algorithm configuration, model building, among others. Additionally, this article presents a case study to analyze the behavior of the different layers of the architecture, to generate knowledge models. Thus, the main contribution of this research is the definition of a Meta-Learning architecture for ML techniques, which takes advantage of the semantic information described as LD when generating the knowledge models. The preliminary results are very encouraging.
{"title":"A Meta-Learning Architecture based on Linked Data","authors":"R. D. Santos, José Aguilar, E. Puerto","doi":"10.1109/CLEI53233.2021.9640223","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9640223","url":null,"abstract":"In Machine Learning (ML), there is a lot of research that seek to automate specific processes carried out by data scientists in the generation of knowledge models (predictive, classification, clustering, etc.); however, an open problem is to find mechanisms that allow conferring the ability of self-learning. Thus, a meta-learning mechanism is required to allow ML techniques to self-adapt in order to improve their performance in problem solving, and even in some cases, to induce the learning algorithm itself. In this context, our research defines a meta-learning architecture using Linked Data (LD) for the automatic generation of knowledge models. Specifically, this intelligent architecture is formed by the layers of Knowledge Sources, Meta-Knowledge and Knowledge Modelling, to unify all processes to guarantee a Meta-Learning process. The Knowledge Sources layer is responsible for providing semantic knowledge about the processes of generation of knowledge models; the Meta-Knowledge layer is responsible for controlling the different processes and strategies for the automatic generation of knowledge models; and finally, the Knowledge Modelling layer is responsible for executing ML tasks defined by the Meta-Knowledge layer, among which are the tasks of feature engineering, ML algorithm configuration, model building, among others. Additionally, this article presents a case study to analyze the behavior of the different layers of the architecture, to generate knowledge models. Thus, the main contribution of this research is the definition of a Meta-Learning architecture for ML techniques, which takes advantage of the semantic information described as LD when generating the knowledge models. The preliminary results are very encouraging.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"125 25 1","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87824909","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9639963
Luis Silvestre, M. Bastarrica, J. Hurtado, Jacqueline Marín
Hybrid software processes are defined as a combination of practices from traditional and agile methodologies. Most software development companies currently apply this kind of process; however, the appropriate combination of practices is not often clear. To address this issue, we have proposed DynaTail in a previous study, i.e., a method for tailoring a software process and its practices to a particular context to improve a certain characteristic. In this work, we use an MDE approach to formalize all the artifacts involved in DynaTail: processes, context, practices influencing goals, and tailoring transformations. This model-based formalization lays the base for building supporting tools. We illustrate all models with a running example from a Chilean medium-size software development company.
{"title":"Formalizing the Goal-directed and Context-based Software Process Tailoring Method","authors":"Luis Silvestre, M. Bastarrica, J. Hurtado, Jacqueline Marín","doi":"10.1109/CLEI53233.2021.9639963","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9639963","url":null,"abstract":"Hybrid software processes are defined as a combination of practices from traditional and agile methodologies. Most software development companies currently apply this kind of process; however, the appropriate combination of practices is not often clear. To address this issue, we have proposed DynaTail in a previous study, i.e., a method for tailoring a software process and its practices to a particular context to improve a certain characteristic. In this work, we use an MDE approach to formalize all the artifacts involved in DynaTail: processes, context, practices influencing goals, and tailoring transformations. This model-based formalization lays the base for building supporting tools. We illustrate all models with a running example from a Chilean medium-size software development company.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"46 1","pages":"1-9"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86397514","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9639899
Daniel Arturo Casal Amat, Carlos Buil Aranda, Carlos Valle-Vidal
The SPARQL query language is the standard for querying RDF data and has been implemented in a wide variety of engines. These engines support hundreds of public endpoints on the Web which receive thousands of queries daily. In many cases these endpoints struggle when evaluating complex queries or when they receive too many of them concurrently. They struggle mostly since some of these queries need large amounts of resources to be processed. All these engines have an internal query optimizer that proposes a supposedly optimal query execution plan, however this is a hard task since there may be thousands of possible query plans to consider and the optimizer may not chose the best one. Herein we propose the use of machine learning techniques to help in finding the best query plan for a given query fast, and thus improve the SPARQL servers' performance. We base such optimization in modeling SPARQL queries based on their complexity, operators used within the queries and data accessed, among others. In this work we propose the use of Dense Neural Networks to improve such SPARQL query processing times. Herein we present the general architecture of a neural network for optimizing SPARQL queries and the results over a synthetic benchmark and real world queries. We show that the use of Dense Neural Networks improve the performance of the Nu-SVR approach in about 50% in performance. We also contribute to the community with a dataset of 19,000 queries.
{"title":"A Neural Networks Approach to SPARQL Query Performance Prediction","authors":"Daniel Arturo Casal Amat, Carlos Buil Aranda, Carlos Valle-Vidal","doi":"10.1109/CLEI53233.2021.9639899","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9639899","url":null,"abstract":"The SPARQL query language is the standard for querying RDF data and has been implemented in a wide variety of engines. These engines support hundreds of public endpoints on the Web which receive thousands of queries daily. In many cases these endpoints struggle when evaluating complex queries or when they receive too many of them concurrently. They struggle mostly since some of these queries need large amounts of resources to be processed. All these engines have an internal query optimizer that proposes a supposedly optimal query execution plan, however this is a hard task since there may be thousands of possible query plans to consider and the optimizer may not chose the best one. Herein we propose the use of machine learning techniques to help in finding the best query plan for a given query fast, and thus improve the SPARQL servers' performance. We base such optimization in modeling SPARQL queries based on their complexity, operators used within the queries and data accessed, among others. In this work we propose the use of Dense Neural Networks to improve such SPARQL query processing times. Herein we present the general architecture of a neural network for optimizing SPARQL queries and the results over a synthetic benchmark and real world queries. We show that the use of Dense Neural Networks improve the performance of the Nu-SVR approach in about 50% in performance. We also contribute to the community with a dataset of 19,000 queries.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"1 1","pages":"1-9"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88984934","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9639968
Vinicius dos Santos, P. R. Silva, Erica Ferreira, K. Felizardo, W. Watanabe, Arnaldo Cândido Júnior, G. V. Meinerz, S. Aluísio, N. Vijaykumar
Context: For thousands of years humans have been using natural language to register their knowledge on important information to enable its access to future generations. With internet, a large amount of textual data is produced and shared on a daily basis. So, scientists started to research techniques for efficiently process knowledge stored in textual format. In this context, Natural Language Processing (NLP) became a popular area studying linguistic phenomena and using computational methods to process texts in natural language. In particular, Open Information Extraction (Open IE) was proposed to gather information from plain text. Despite the advances in this area, it is still necessary to map details about how these approaches were proposed to support the community while creating more efficient Open IE systems. Objective: In this paper, we identify, in the literature, the main characteristics of proposed Open IE approaches. Method: First, we extended the search performed in a systematic mapping previously published by using backward snowballing and a manual search. Next, we updated the electronic database search including ACL Anthology. Finally, 159 studies proposing Open IE approaches were considered for data extraction. Results: Data analysis showed a significant increase in the number of studies published about Open IE in the last years. In addition, we provide important details about how these techniques were proposed (e.g., data sets used and output evaluation techniques). Results indicate that researchers started to adopt neural networks to perform Open IE instead of using conventional supervised learning techniques. Conclusion: Recent advances in Artificial Intelligence and neural networks techniques allowed scientists to have a new perspective on how to perform efficient textual data management. Therefore, Open IE approaches gained much attention as they can help in many contexts, especially in knowledge management tasks.
背景:几千年来，人类一直在使用自然语言来记录他们对重要信息的了解，以便后代能够获得这些信息。有了互联网，每天都会产生和共享大量的文本数据。因此，科学家们开始研究有效处理以文本形式存储的知识的技术。在这种背景下，自然语言处理(NLP)成为研究语言现象和使用计算方法处理自然语言文本的热门领域。特别地，Open Information Extraction (Open IE)被提出从纯文本中收集信息。尽管在这一领域取得了进展，但仍然有必要详细描述这些方法是如何被提出来支持社区的，同时创建更高效的开放IE系统。目的:在本文中，我们在文献中确定了提出的开放IE方法的主要特征。方法:首先，我们通过向后滚雪球和手动搜索扩展了以前发布的系统映射中执行的搜索。接下来，我们更新了包括ACL Anthology在内的电子数据库搜索。最后，159项提出Open IE方法的研究被考虑用于数据提取。结果:数据分析显示，在过去几年中，关于Open IE的研究发表的数量显著增加。此外，我们还提供了关于这些技术是如何提出的重要细节(例如，使用的数据集和输出评估技术)。结果表明，研究人员开始采用神经网络来执行Open IE，而不是使用传统的监督学习技术。结论:人工智能和神经网络技术的最新进展使科学家们对如何执行有效的文本数据管理有了新的视角。因此，开放IE方法获得了很多关注，因为它们可以在很多情况下提供帮助，特别是在知识管理任务中。
{"title":"Using Open Information Extraction to Extract Relations: An Extended Systematic Mapping","authors":"Vinicius dos Santos, P. R. Silva, Erica Ferreira, K. Felizardo, W. Watanabe, Arnaldo Cândido Júnior, G. V. Meinerz, S. Aluísio, N. Vijaykumar","doi":"10.1109/CLEI53233.2021.9639968","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9639968","url":null,"abstract":"Context: For thousands of years humans have been using natural language to register their knowledge on important information to enable its access to future generations. With internet, a large amount of textual data is produced and shared on a daily basis. So, scientists started to research techniques for efficiently process knowledge stored in textual format. In this context, Natural Language Processing (NLP) became a popular area studying linguistic phenomena and using computational methods to process texts in natural language. In particular, Open Information Extraction (Open IE) was proposed to gather information from plain text. Despite the advances in this area, it is still necessary to map details about how these approaches were proposed to support the community while creating more efficient Open IE systems. Objective: In this paper, we identify, in the literature, the main characteristics of proposed Open IE approaches. Method: First, we extended the search performed in a systematic mapping previously published by using backward snowballing and a manual search. Next, we updated the electronic database search including ACL Anthology. Finally, 159 studies proposing Open IE approaches were considered for data extraction. Results: Data analysis showed a significant increase in the number of studies published about Open IE in the last years. In addition, we provide important details about how these techniques were proposed (e.g., data sets used and output evaluation techniques). Results indicate that researchers started to adopt neural networks to perform Open IE instead of using conventional supervised learning techniques. Conclusion: Recent advances in Artificial Intelligence and neural networks techniques allowed scientists to have a new perspective on how to perform efficient textual data management. Therefore, Open IE approaches gained much attention as they can help in many contexts, especially in knowledge management tasks.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"128 1","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76398961","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9640014
Bruno Duarte, L. C. Oliveira, Marcelo Teixeira, Marco A. C. Barbosa
The Traveling Salesman Problem with Draft Limits is a combinatorial optimization problem that consists in calculating routes to be taken by cargo ships without violating draft limits restrictions, so reducing transportation costs. Finding the best route solution, using exact computation, is a problem whose complexity grows exponentially with the number of routes and, therefore, is unfeasible for practical cases. Approximations to the best solution, computed using heuristis and metaheuristics, appear as promising and feasible alternatives to address this problem with reasonable accuracy. This paper exploits two metaheuristics, Genetic and Memetic Algorithms, under the perspective of Evolutionary Algorithms, to address the problem at hand. After they are implemented and applied over a route planning map, their effectiveness are compared against each other and also against the literature. Results suggest that the method based on Memetic Algorithm is slightly better (5.28% average error) in comparison with the Genetic-based approach (12.96%), which is shown to be competitive with respect to the literature.
{"title":"A comparison of Genetic and Memetic Algorithms applied to the Traveling Salesman Problem with Draft Limits","authors":"Bruno Duarte, L. C. Oliveira, Marcelo Teixeira, Marco A. C. Barbosa","doi":"10.1109/CLEI53233.2021.9640014","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9640014","url":null,"abstract":"The Traveling Salesman Problem with Draft Limits is a combinatorial optimization problem that consists in calculating routes to be taken by cargo ships without violating draft limits restrictions, so reducing transportation costs. Finding the best route solution, using exact computation, is a problem whose complexity grows exponentially with the number of routes and, therefore, is unfeasible for practical cases. Approximations to the best solution, computed using heuristis and metaheuristics, appear as promising and feasible alternatives to address this problem with reasonable accuracy. This paper exploits two metaheuristics, Genetic and Memetic Algorithms, under the perspective of Evolutionary Algorithms, to address the problem at hand. After they are implemented and applied over a route planning map, their effectiveness are compared against each other and also against the literature. Results suggest that the method based on Memetic Algorithm is slightly better (5.28% average error) in comparison with the Genetic-based approach (12.96%), which is shown to be competitive with respect to the literature.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"62 26","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91462465","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9639989
Alejandro Tejada-Mesias, Irvin Dongo, Yudith Cardinale, Jose Diaz-Amado
In robotics, object detection in images or videos, obtained in real-time from sensors of robots can be used to support the implementation of service robot tasks (e.g., navigation, model its social behavior, recognize objects in a specific domain), usually accomplished in indoor environments. However, traditional deep learning based object detection techniques present limitations in such indoor environments, specifically related to the detection of small objects and the management of high density of multiple objects. Coupled with these limitations, for specific domains (e.g., hospitals, museums), it is important that the robot, apart from detecting objects, extracts and knows information of the targeted objects. Ontologies, as a part of the Semantic Web, are presented as a feasible option to formally represent the information related to the objects of a particular domain. In this context, this work proposes an object detection and recognition process based on a Deep Learning algorithm, object descriptors, and an ontology. ODROM, an Object Detection and Recognition algorithm supported by Ontologies and applied to Museums, is an implementation to validate the proposal. Experiments show that the usage of ontologies is a good way of desambiguating the detection, obtained with a and $mathbf{mAP}{@}0.5=0.88$ and a $mathbf{mAP}{@}[0.5:0.95]=61%$.
{"title":"ODROM: Object Detection and Recognition supported by Ontologies and applied to Museums","authors":"Alejandro Tejada-Mesias, Irvin Dongo, Yudith Cardinale, Jose Diaz-Amado","doi":"10.1109/CLEI53233.2021.9639989","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9639989","url":null,"abstract":"In robotics, object detection in images or videos, obtained in real-time from sensors of robots can be used to support the implementation of service robot tasks (e.g., navigation, model its social behavior, recognize objects in a specific domain), usually accomplished in indoor environments. However, traditional deep learning based object detection techniques present limitations in such indoor environments, specifically related to the detection of small objects and the management of high density of multiple objects. Coupled with these limitations, for specific domains (e.g., hospitals, museums), it is important that the robot, apart from detecting objects, extracts and knows information of the targeted objects. Ontologies, as a part of the Semantic Web, are presented as a feasible option to formally represent the information related to the objects of a particular domain. In this context, this work proposes an object detection and recognition process based on a Deep Learning algorithm, object descriptors, and an ontology. ODROM, an Object Detection and Recognition algorithm supported by Ontologies and applied to Museums, is an implementation to validate the proposal. Experiments show that the usage of ontologies is a good way of desambiguating the detection, obtained with a and $mathbf{mAP}{@}0.5=0.88$ and a $mathbf{mAP}{@}[0.5:0.95]=61%$.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"87 1","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85730784","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9639970
Eduardo Coronel, B. Barán, P. Gardel
In the present work the problem of Optimal Placement of remote controlled Switches in an Electric Power Distribution Systems is analyzed with a Multi-objective approach. The selected objectives consider economic, technical, operational and social aspects. Real data from Paraguayan East Region is used. For the evaluation of solutions, an evaluating function based on the Monte Carlo method is implemented, which estimates the defined indices through the simulation of failures in the network, whose probabilities are obtained using the Reliability Block Diagram method. Six different multi-objective algorithms were implemented and compared using ONVGR and Hypervolume metrics, having the Multi-objective Ant Colony Optimization algorithms the best performance.
{"title":"Optimal Placement of Remote Controlled Switches in Electric Power Distribution Systems with a Multi-Objective Approach","authors":"Eduardo Coronel, B. Barán, P. Gardel","doi":"10.1109/CLEI53233.2021.9639970","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9639970","url":null,"abstract":"In the present work the problem of Optimal Placement of remote controlled Switches in an Electric Power Distribution Systems is analyzed with a Multi-objective approach. The selected objectives consider economic, technical, operational and social aspects. Real data from Paraguayan East Region is used. For the evaluation of solutions, an evaluating function based on the Monte Carlo method is implemented, which estimates the defined indices through the simulation of failures in the network, whose probabilities are obtained using the Reliability Block Diagram method. Six different multi-objective algorithms were implemented and compared using ONVGR and Hypervolume metrics, having the Multi-objective Ant Colony Optimization algorithms the best performance.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"70 11","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91505032","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9640128
O. Herrera, Patrica Mejías, Alejandra Cid
The practice of teaching to train the professionals of the future is increasingly demanding. Many of the trainers in Higher Education Institutions do not have a training in pedagogy, so they do their best to achieve good results in their students. Teachers base their methodology on repeating patterns learned, training received, formal and informal support from pedagogical support units in universities, among other elements. This article presents a platform that supports the design of pedagogical strategies, which is based on three pillars. First, planning based on pedagogical criteria (Bloom's taxonomy and conversational framework). Second, a design that arises from assessment, understood as an always formative activity. And third, collaboration both in the design of the activities and in sharing these designs with others. This platform has been used for designs in the computing area with positive evaluations from the participating teachers. Teachers from other disciplines have also joined, confirming the usefulness and transversality of the platform. A direct impact on students with more solid and relevant learning is expected.
{"title":"Rethinking the design of learning modules: An assessment centered strategy with ICT support","authors":"O. Herrera, Patrica Mejías, Alejandra Cid","doi":"10.1109/CLEI53233.2021.9640128","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9640128","url":null,"abstract":"The practice of teaching to train the professionals of the future is increasingly demanding. Many of the trainers in Higher Education Institutions do not have a training in pedagogy, so they do their best to achieve good results in their students. Teachers base their methodology on repeating patterns learned, training received, formal and informal support from pedagogical support units in universities, among other elements. This article presents a platform that supports the design of pedagogical strategies, which is based on three pillars. First, planning based on pedagogical criteria (Bloom's taxonomy and conversational framework). Second, a design that arises from assessment, understood as an always formative activity. And third, collaboration both in the design of the activities and in sharing these designs with others. This platform has been used for designs in the computing area with positive evaluations from the participating teachers. Teachers from other disciplines have also joined, confirming the usefulness and transversality of the platform. A direct impact on students with more solid and relevant learning is expected.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"26 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84803699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9640190
Orlando Montenegro, O. S. Pabón, Raúl Ernesto Gutiérrez de Piñerez Reyes
Online product reviews are becoming common and are being used more frequently by consumers to choose the most competitive products. Negation detection is a crucial task for information extraction from product review texts because negation can change the meaning of opinions given by consumers about products or services. Although several approaches have been proposed for negation detection in product reviews, research efforts have concentrated mainly on English. This paper describes a transformer-based approach for detecting negation in product reviews written in Spanish. This approach takes advantage of transfer learning techniques and uses a BERT-based model to perform negation detection. Performed tests using the SFU corpus for Spanish, showed an F1 score of 95.4% in the cue detection task and 91.5% in the scope resolution task. Our finding suggests that our BERT-based approach is feasible to perform negation detection in Spanish.
{"title":"A Deep Learning Approach for Negation Detection from Product Reviews written in Spanish","authors":"Orlando Montenegro, O. S. Pabón, Raúl Ernesto Gutiérrez de Piñerez Reyes","doi":"10.1109/CLEI53233.2021.9640190","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9640190","url":null,"abstract":"Online product reviews are becoming common and are being used more frequently by consumers to choose the most competitive products. Negation detection is a crucial task for information extraction from product review texts because negation can change the meaning of opinions given by consumers about products or services. Although several approaches have been proposed for negation detection in product reviews, research efforts have concentrated mainly on English. This paper describes a transformer-based approach for detecting negation in product reviews written in Spanish. This approach takes advantage of transfer learning techniques and uses a BERT-based model to perform negation detection. Performed tests using the SFU corpus for Spanish, showed an F1 score of 95.4% in the cue detection task and 91.5% in the scope resolution task. Our finding suggests that our BERT-based approach is feasible to perform negation detection in Spanish.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"1 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82309310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-25DOI: 10.1109/CLEI53233.2021.9640084
Raúl Marichal, Ernesto Dufrechu, P. Ezzatti
The solution of sparse triangular linear systems is an important building block for a large number of numerical methods used in science and engineering. It is then crucial to count with implementations of this operation that can execute efficiently in the most recent hardware platforms. In the case of GPUs, several methods have been proposed in the last years. These methods belong to two main categories. On the one hand, there are the methods that rely on a previous analysis of the sparse matrix to determine a better execution schedule and, on the other hand, there are methods that decide this scheduling dynamically. The experimental results in the literature are not conclussive in favour of any of these strategies. However, the experimental evaluations usually focus on the use case where many systems have to be solved with the same sparse matrix, where the analysis phase needs to be performed only once and its cost is not important in relation with the total runtime. In this work we are interested in determining which is the best strategy, according to the degree of parallelism of the problem, when only one sytem is to be solved. The experimental evaluation performed on NVIDIA P100 accelerators shows that the self-scheduled routines present important advantages when the degree of parallelism of the problem allows it.
{"title":"Assessing the solution of one sparse triangular linear system on multi-many core platforms","authors":"Raúl Marichal, Ernesto Dufrechu, P. Ezzatti","doi":"10.1109/CLEI53233.2021.9640084","DOIUrl":"https://doi.org/10.1109/CLEI53233.2021.9640084","url":null,"abstract":"The solution of sparse triangular linear systems is an important building block for a large number of numerical methods used in science and engineering. It is then crucial to count with implementations of this operation that can execute efficiently in the most recent hardware platforms. In the case of GPUs, several methods have been proposed in the last years. These methods belong to two main categories. On the one hand, there are the methods that rely on a previous analysis of the sparse matrix to determine a better execution schedule and, on the other hand, there are methods that decide this scheduling dynamically. The experimental results in the literature are not conclussive in favour of any of these strategies. However, the experimental evaluations usually focus on the use case where many systems have to be solved with the same sparse matrix, where the analysis phase needs to be performed only once and its cost is not important in relation with the total runtime. In this work we are interested in determining which is the best strategy, according to the degree of parallelism of the problem, when only one sytem is to be solved. The experimental evaluation performed on NVIDIA P100 accelerators shows that the self-scheduled routines present important advantages when the degree of parallelism of the problem allows it.","PeriodicalId":6803,"journal":{"name":"2021 XLVII Latin American Computing Conference (CLEI)","volume":"51 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80394574","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：