Politecnico di Torino (logo)

Assessing Public Engagement in Climate Change from the Eye of Twitter

Soha Torki

Assessing Public Engagement in Climate Change from the Eye of Twitter.

Rel. Silvia Anna Chiusano, Hamed Haddadi. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2018

PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (2MB) | Preview

Climate change which refers to the shift in weather patterns and global temperatures is one of the main issues across the world today. According to scientists, the extensive amount of greenhouse gases emissions has caused a tremendous increase in global temperatures in recent years in such a way that August 2017 was the second warmest August in 137 years of record-keeping.Twitter as one of the most popular social networking websites, has become a common platform for climate change conversations. The climate change debates on Twitter is generally categorized into two types of opinion. One group believe that climate change is happening and actions should be taken to fight it and protect the earth. While the second group do not believe in climate change. This group that is known as climate change deniers or skeptics, claim that climate change is not happening or if it is, this is not caused by the human activities and there is no need to take actions.In this thesis a dataset of tweets about climate change is created using Twitter API in order to analyze the public opinion on climate change by means of machine learning techniques as an approach for sentiment analysis. The tweets are collected using popular climate change hashtags and are labeled as positive and negative which correspond to two opinion about climate change discussed earlier. The positive group are those who believe in climate change while the negative group support the climate change denial. Then the classification algorithms are applied on the dataset. The experiments show that Support Vector Machine classifier with $0.97$ accuracy and Logistic Regression classification with $0.95$ accuracy using the unigrams, sentiment lexicon, word embeddings and Twitter features have the best performance metrics.Analysis on the dataset show that more than 70\% of the tweets are positive and from the negative tweets more than 85\% are from the United States mainly in Georgia, Kentucky, Texas, Ohio and Arizona. This is while according to the Emission Database for Global Atmospheric Research (EDGAR) and EPA, the U.S is the second biggest $CO_2$ emitters in the world and Texas has the most amount of annual $CO_2$ emission in the U.S. it is obvious that climate change deniers are the biggest $CO_2$ emitters who want to stop the regulations on their activities.As future developments on this work, deep learning techniques could be used for classification. The analysis could also be performed on other social networks and on bigger data set in order to make wider assessment.

Relators: Silvia Anna Chiusano, Hamed Haddadi
Academic year: 2017/18
Publication type: Electronic
Number of Pages: 106
Corso di laurea: Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering)
Classe di laurea: New organization > Master science > LM-32 - COMPUTER SYSTEMS ENGINEERING
Aziende collaboratrici: UNSPECIFIED
URI: http://webthesis.biblio.polito.it/id/eprint/8020
Modify record (reserved for operators) Modify record (reserved for operators)