ENEM nas redes sociais: minera??o de textos e clusteriza??o

Submitted by Jos? Henrique Henrique (jose.neves@ufvjm.edu.br) on 2018-07-24T17:34:56Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) leila_maria_silva.pdf: 2106552 bytes, checksum: 53ba37c88f3aa004f2201a85b74fd640 (MD5) === Approved for entry into arch...

Full description

Bibliographic Details
Main Author: Silva, Leila Maria
Other Authors: Guelpeli, Marcus Vin?cius Carvalho
Language:Portuguese
Published: UFVJM 2018
Subjects:
Online Access:http://acervo.ufvjm.edu.br/jspui/handle/1/1776
Description
Summary:Submitted by Jos? Henrique Henrique (jose.neves@ufvjm.edu.br) on 2018-07-24T17:34:56Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) leila_maria_silva.pdf: 2106552 bytes, checksum: 53ba37c88f3aa004f2201a85b74fd640 (MD5) === Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2018-10-04T19:43:35Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) leila_maria_silva.pdf: 2106552 bytes, checksum: 53ba37c88f3aa004f2201a85b74fd640 (MD5) === Made available in DSpace on 2018-10-04T19:43:35Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) leila_maria_silva.pdf: 2106552 bytes, checksum: 53ba37c88f3aa004f2201a85b74fd640 (MD5) Previous issue date: 2017 === A internet ? hoje a maior fonte de informa??o eletr?nica existente. Cresce a cada dia o n?mero de usu?rios da internet, e consequentemente o uso das redes sociais online. S?o muitas as informa??es novas que ficam embutidas nas bases de dados textuais. Por causa da sua natureza din?mica, ou seja, milh?es de p?ginas surgem e desaparecem todos os dias, a tarefa de encontrar informa??es relevantes nessas bases de dados se torna muito dif?cil. As t?cnicas de minera??o de textos para a descoberta de informa??es na web surgiram da necessidade de sanar este problema. O presente trabalho versa sobre a aplica??o de m?todos de minera??o de textos com clusteriza??o na grande quantidade de mensagens sobre o Exame Nacional do Ensino M?dio no ano de 2016 provenientes da rede social Twitter. O foco deste estudo est? na obten??o de grupos de textos, a fim de possibilitar uma visualiza??o resumida e sintetizada dos assuntos mais comentados pelos usu?rios. Para manipula??o dessas bases textuais, o Modelo Cassiopeia foi utilizado empregando seu algoritmo de agrupamento textual que tem como principal finalidade gerar agrupamentos, ou seja, clusters (grupos) de documentos textuais que apresentam algum tipo de similaridade. O Modelo Cassiopeia apresenta um limite de processamento com a quantidade m?xima de 700 tweets. Os tweets passam primeiramente pela fase de limpeza dos textos no pr?-processamento, logo ap?s, a utiliza??o do algoritmo no processamento e por fim, as an?lises dos resultados no p?s-processamento. Os resultados obtidos neste trabalho mostram valores coesos quanto ? similaridade dos documentos dentro de um cluster e entre os clusters, avaliados por medidas de agrupamento textual, proposto pelo Modelo Cassiopeia. Isso demonstra a aplicabilidade dessa proposta para a visualiza??o sintetizada das informa??es mais significativas de um determinado tema, muitas vezes permitindo que a??es sejam antecipadas e impactos sobre a popula??o afetada sejam reduzidos. === Disserta??o (Mestrado Profissional) ? Programa de P?s-Gradua??o em Educa??o, Universidade Federal dos Vales do Jequitinhonha e Mucuri, 2017. === The Internet is today the largest source of existing electronic information. The number of Internet users is increasing daily, and consequently the use of online networks online. There are many new information that is embedded in textual databases. Because of its dynamic nature- that is, millions of pages and other numbers-a task of finding relevant information in those databases becomes very difficult. The techniques of text mining for a discovery of information on the web came from the need to heal this problem. The present work is about an application of methods of text mining with clustering in the large amount of messages on the National High School Exams in the year 2016 issu social network Twitter. The focus of this study is on obtaining groups of texts in order to enable a summary and synthesized publication of the appropriate comments of the users. For manipulation of textual bases, the Cassiopeia Model was used by using its textual grouping algorithm that has as main purpose to generate clusters, that is, clusters of textual documents and executed some kind of similarity. The Cassiopeia Model has a processing limit with a maximum of 700 tweets. The tweets first pass through the phase of cleaning the texts without preprocessing, afterwards, a use of the algorithm without processing and, finally, as analysis of the results without post-processing. The results obtained in this work are more closely related to the similarity of the documents within the cluster and between the clusters, through the measurements of textual grouping, proposed by the Cassiopeia Model. This demonstrates an application for an uninformed publication of the most important information on a given topic, often allowing actions to be anticipated and impacts on an affected population to be reduced.