| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.14 MB | Adobe PDF |
Abstract(s)
Ao longo dos anos, impulsionada pelo avanço tecnológico e pela constante inovação
de serviços, a sociedade tem intensificado substancialmente o uso das redes sociais,
alcançando níveis de dependência (SNS, 2020).
A rede social X (antigo Twitter) é uma rede social que permite aos seus utilizadores
partilhar as suas opiniões. Esta é uma das redes sociais mais utilizadas atualmente
(Martin, 2023) e, quando surgiu, era permitido um máximo de 140 caracteres por
tweet, valor que foi aumentado para 280 em 2017. Em Portugal, 2,5 milhões de
pessoas usam a rede social X e, mundialmente, esta rede social conta com mais de 330
milhões de utilizadores, o que origina em média a criação de 5 787 tweets por segundo
em todo o mundo, o que equivale a quase 500 milhões diariamente (Bacelar, 2022).
Estes tweets geram uma quantidade significativa de dados, dados estes que estão a ser
cada vez mais importantes na atualidade, pois é possível retirar informações relevantes
e/ou ser capaz de identificar perfis. Isto é alcançado através da tecnologia e da
descoberta de novas técnicas e algoritmos, sendo estas as principais razões para o
desenvolvimento deste projeto.
O projeto teve os seguintes objetivos:
• Tratar Dados de forma a conseguir obter o melhor resultado possível.
• Determinar tópicos para conseguir perceber os assuntos.
• Classificar os tweets de modo a determinar se os utilizadores são a favor,
contra ou neutros relativamente aos tópicos discutidos nas redes sociais.
• Recomendar conteúdos com base no sentimento detetado anteriormente, tendo
em conta o tópico abordado.
Com o intuito de esclarecer os objetivos deste estudo, propõe-se o seguinte exemplo
ilustrativo:
Se o Presidente da República for a entidade selecionada, procede-se, assim, à recolha
de tweets que façam menção ao perfil "@PresidenteDaRepublica". Em seguida, serão
realizadas as etapas necessárias para o tratamento adequado desses tweets, seguido
pela identificação dos tópicos mais frequentemente discutidos em relação a essa figura
pública. Isso estabelecerá a base de dados fundamental para a condução deste projeto. Uma vez obtida a base de dados, sólida e coesa, realiza-se a análise de sentimento
com o objetivo de identificar e classificar os perfis de utilizadores em categorias como
"Positivo", "Neutro" ou "Negativo" em relação a diversos temas relacionados com a
entidade. Posteriormente, será implementado um sistema de recomendação para
enviar conteúdos diferenciados a cada grupo de utilizadores previamente identificados.
Esta abordagem permitirá potencialmente influenciar as opiniões das pessoas ou
auxiliá-las na formação de opiniões bem fundamentadas e construtivas,
independentemente das opiniões serem positivas ou negativas a determinados
assuntos.
ABSTRACT: Over the years, propelled by technological advancements and continuous service innovation, society has significantly intensified the use of social media, reaching substantial levels of dependency (SNS, 2020). The social network X (previously known as Twitter) is a social media platform that allows its users to share their opinions. It is one of the most widely used social networks today (Martin, 2023) and, when it was first introduced, it allowed a maximum of 140 characters per tweet, a limit that was increased to 280 characters in 2017. In Portugal, 2.5 million people use X social network, and globally, this social network boasts over 330 million users, resulting in an average of 5,787 tweets per second worldwide (Bacelar, 2022). This leads to the creation of nearly 500 million tweets daily. These data are becoming increasingly significant in contemporary society, as they provide valuable information and the ability to identify profiles within the current society, this is achieved through technology and the discovery of new techniques and algorithms, which are the main reasons for the development of this project. The project had the following objectives: • Clean data in order to achieve the best possible outcome. • Determine topics to understand the subjects present in the data. • Classify tweets in order toto determine whether users are in favor, against, or neutral regarding the topics discussed on social networks. • Recommend content based on the detected sentiment. To clarify the objectives of this study, consider the following example, if the President of the Republic is the selected entity, we will proceed with the collection of tweets mentioning the profile "@PresidenteDaRepublica". Subsequently, the necessary steps will be taken for the proper treatment of these tweets, followed by the identification of the most frequently discussed topics related to this public figure. This will establish the fundamental database for the execution of this project. Once a solid and cohesive database is obtained, we will conduct sentiment analysis with the aim of identifying and classifying user profiles into categories such as "Positive", "Neutral," or "Negative" in relation to various topics related to the president of the republic. Subsequently, a recommendation system will be implemented to deliver differentiated content to each group of previously identified users. This approach will potentially influence people's opinions or assist them in forming more well-founded and constructive opinions, regardless of whether they are positive or negative on certain subjects. Keywords: Sentiment analysis, Topic Identification, Machine Learning, Web Scrapping, Natural Language Processing, Recommender Systems, Python, X Social Network.
ABSTRACT: Over the years, propelled by technological advancements and continuous service innovation, society has significantly intensified the use of social media, reaching substantial levels of dependency (SNS, 2020). The social network X (previously known as Twitter) is a social media platform that allows its users to share their opinions. It is one of the most widely used social networks today (Martin, 2023) and, when it was first introduced, it allowed a maximum of 140 characters per tweet, a limit that was increased to 280 characters in 2017. In Portugal, 2.5 million people use X social network, and globally, this social network boasts over 330 million users, resulting in an average of 5,787 tweets per second worldwide (Bacelar, 2022). This leads to the creation of nearly 500 million tweets daily. These data are becoming increasingly significant in contemporary society, as they provide valuable information and the ability to identify profiles within the current society, this is achieved through technology and the discovery of new techniques and algorithms, which are the main reasons for the development of this project. The project had the following objectives: • Clean data in order to achieve the best possible outcome. • Determine topics to understand the subjects present in the data. • Classify tweets in order toto determine whether users are in favor, against, or neutral regarding the topics discussed on social networks. • Recommend content based on the detected sentiment. To clarify the objectives of this study, consider the following example, if the President of the Republic is the selected entity, we will proceed with the collection of tweets mentioning the profile "@PresidenteDaRepublica". Subsequently, the necessary steps will be taken for the proper treatment of these tweets, followed by the identification of the most frequently discussed topics related to this public figure. This will establish the fundamental database for the execution of this project. Once a solid and cohesive database is obtained, we will conduct sentiment analysis with the aim of identifying and classifying user profiles into categories such as "Positive", "Neutral," or "Negative" in relation to various topics related to the president of the republic. Subsequently, a recommendation system will be implemented to deliver differentiated content to each group of previously identified users. This approach will potentially influence people's opinions or assist them in forming more well-founded and constructive opinions, regardless of whether they are positive or negative on certain subjects. Keywords: Sentiment analysis, Topic Identification, Machine Learning, Web Scrapping, Natural Language Processing, Recommender Systems, Python, X Social Network.
Description
Keywords
Análise de Sentimento Identificação de Tópicos Machine Learning Web Scrapping Natural Language Processing Recommender Systems Python Rede Social X