A carregar...
Publicação
Sentiment analysis: the case of twitch chat - Mining user feedback from livestream chats
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 3.47 MB | Adobe PDF |
Orientador(es)
Resumo(s)
In a world where users often share their thoughts and opinions through online communication
channels, applications that can tap into these channels as to extract consumer feedback have
become increasingly valuable. Traditional marketing research techniques such as interviews or
surveys offer results that pale in comparison to sentiment analysis applications that can extract
organic feedback from an extremely large selection, with very little resources and in real-time.
This thesis focuses on proposing and developing one of these tools that targets livestreams,
which have, over the years, seen a massive increase in popularity from both a user-base
standpoint as well as brand involvement. We chose the livestreaming platform “Twitch” as the
target of research and developed a sentiment analysis model, using rule-based approaches,
capable of interpreting user chat messages and identifying whether those messages are negative,
positive or neutral. Additionally, an application was developed to better view and analyze the
results of the model. By segmenting our results by product reveal, we also exhibit how the
application allows for the extraction of various insights about the public’s opinion of that
product.
Descrição
Project Work presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Information Systems and Technologies Management
Palavras-chave
Sentiment Analysis Opinion Mining Livestreams Twitch Text Mining