Difference between revisions of "Supervised Machine Learning for Fake News Detection"
Adelo Vieira (talk | contribs) (→Summary of Results) (Tag: Visual edit) |
Adelo Vieira (talk | contribs) (→Summary of Results) (Tag: Visual edit) |
||
Line 75: | Line 75: | ||
|[[wikipedia:Thomas_Bayes|Bayes, Thomas]] | |[[wikipedia:Thomas_Bayes|Bayes, Thomas]] | ||
|[https://cran.r-project.org/web/packages/e1071/index.html e1071] | |[https://cran.r-project.org/web/packages/e1071/index.html e1071] | ||
− | |NB | + | |NB |
|[[Establishing an authenticity of sport news by Machine Learning Models#fnd_naivebayes|<math>{\color{blue}76.41%}</math>]] | |[[Establishing an authenticity of sport news by Machine Learning Models#fnd_naivebayes|<math>{\color{blue}76.41%}</math>]] | ||
| | | | ||
Line 88: | Line 88: | ||
|Meyer et al., 2012 | |Meyer et al., 2012 | ||
|We used [[Establishing an authenticity of sport news by Machine Learning Models#The RTextTools package|RTextTools]], which depends on [https://cran.r-project.org/web/packages/e1071/index.html e1071] | |We used [[Establishing an authenticity of sport news by Machine Learning Models#The RTextTools package|RTextTools]], which depends on [https://cran.r-project.org/web/packages/e1071/index.html e1071] | ||
− | |SVM | + | |SVM |
|[[Establishing an authenticity of sport news by Machine Learning Models#fnd_svm|<math>{\color{blue}95.42%}</math>]] | |[[Establishing an authenticity of sport news by Machine Learning Models#fnd_svm|<math>{\color{blue}95.42%}</math>]] | ||
| | | | ||
Line 114: | Line 114: | ||
|Chen & Guestrin, 2016 | |Chen & Guestrin, 2016 | ||
|[[Establishing an authenticity of sport news by Machine Learning Models#The XGBoost package|xgboost]] | |[[Establishing an authenticity of sport news by Machine Learning Models#The XGBoost package|xgboost]] | ||
− | |XGBOOST | + | |XGBOOST |
|[[Establishing an authenticity of sport news by Machine Learning Models#fnd_xgboost|<math>{\color{red}97.42%}</math>]] | |[[Establishing an authenticity of sport news by Machine Learning Models#fnd_xgboost|<math>{\color{red}97.42%}</math>]] | ||
| | | | ||
Line 127: | Line 127: | ||
|Friedman et al., 2010 | |Friedman et al., 2010 | ||
|We used [[Establishing an authenticity of sport news by Machine Learning Models#The RTextTools package|RTextTools]], which depends on [https://cran.r-project.org/web/packages/glmnet/index.html wglmnet] | |We used [[Establishing an authenticity of sport news by Machine Learning Models#The RTextTools package|RTextTools]], which depends on [https://cran.r-project.org/web/packages/glmnet/index.html wglmnet] | ||
− | |GLMNET | + | |GLMNET |
|[[Establishing an authenticity of sport news by Machine Learning Models#fnd_glmnet|<math>{\color{blue}94.58%}</math>]] | |[[Establishing an authenticity of sport news by Machine Learning Models#fnd_glmnet|<math>{\color{blue}94.58%}</math>]] | ||
| | | | ||
Line 140: | Line 140: | ||
|Jurka, 2012 | |Jurka, 2012 | ||
|We used [[Establishing an authenticity of sport news by Machine Learning Models#The RTextTools package|RTextTools]], which depends on [https://cran.r-project.org/web/packages/maxent/index.html maxent] | |We used [[Establishing an authenticity of sport news by Machine Learning Models#The RTextTools package|RTextTools]], which depends on [https://cran.r-project.org/web/packages/maxent/index.html maxent] | ||
− | |MAXENT | + | |MAXENT |
|[[Establishing an authenticity of sport news by Machine Learning Models#fnd_maxent|<math>{\color{blue}96.09%}</math>]] | |[[Establishing an authenticity of sport news by Machine Learning Models#fnd_maxent|<math>{\color{blue}96.09%}</math>]] | ||
| | | |
Revision as of 03:19, 28 April 2019
Contents
Declaration
Acknowledgement
Thanks for Muhammad, Graham and Mark
Abstract
Introduction
Chapter 1
Chapter 2 - Training a Supervised Machine Learning Model
Supervised text Classification for fake news detection Using Machine Learning Models
Procedure
- The Dataset
- Splitting the data into Train and Test data
- Cleaning the data
- Building the Document-Term Matrix
- Model Building
- Cross validation
- Making predictions from the model created and displaying a Confusion matrix
Results
Summary of Results
Algorithms | Author | Package | Keyword | Accuracy | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Kaggle Fake News Dataset | Fake News Detector Dataset | Gofaaas Fake News Dataset | Using the KFN* Model to make predictions over the Gofaaas Dataset | Using the FND** Model to make predictions over the Gofaaas Dataset | |||||||
Train:70%
Test: 30% |
Cross validation | Train:70%
Test: 30% |
Cross validation | Train:70%
Test: 30% |
Cross validation | ||||||
Naive Bayes | Bayes, Thomas | e1071 | NB | ||||||||
Support vector machine | Meyer et al., 2012 | We used RTextTools, which depends on e1071 | SVM | ||||||||
Random forest | Liawand Wiener, 2002 | randomForest | RF | ||||||||
Extreme Gradient Boosting | Chen & Guestrin, 2016 | xgboost | XGBOOST | ||||||||
General linearized models | Friedman et al., 2010 | We used RTextTools, which depends on wglmnet | GLMNET | ||||||||
Maximum entropy | Jurka, 2012 | We used RTextTools, which depends on maxent | MAXENT | ||||||||
* KFN (Kaggle Fake News Model): Model created using 70% (train data) of the Kaggle Fake News Dataset.
** FND (Fake News Detector Model) : Model created using 70% (train data) of the Fake News Detector Dataset. |
Evaluation of Results
We evaluate our approach in different settings. First, weperform cross-validation on our noisy training set; second,and more importantly, we train models on the training setand validate them against a manually created gold standard.17Moreover, we evaluate two variants, i.e., including and exclud-ing user features. [smb:home/adelo/1-system/1-disco_local/1-mis_archivos/1-pe/1-ciencia/1-computacion/2-data_analysis-machine_learning/gofaaaz-machine_learning/5-References/7-Weakly_supervised_searning_for_fake_news_detection_on_twitter.pdf]
Datasets used
Kaggle Fake News Dataset
https://www.kaggle.com/c/fake-news/data
Distribution of the data:
The distribution of Stance
classes in train_stances.csv
is as follows:
rows | unrelated | discuss | agree | disagree |
---|---|---|---|---|
49972 | 0.73131 | 0.17828 | 0.0736012 | 0.0168094 |
Fake News Detector Dataset
Gofaaas Fake News Dataset
Algorithms
Naive Bayes
Naïve Bayes is based on the Bayesian theorem, there in order to understand Naïve Bayes it is important to first understand the Bayesian theorem.
Bayesian theorem is a mathematical formula for determining conditional probability which is the probability of something, happening given that something else has already occurred.
- P(c|x) is the posterior probability of class (target) given predictor (attribute).
- P(c) is the prior probability of class.
- P(x|c) is the likelihood which is the probability of predictor given class.
- P(x) is the prior probability of predictor.
Prior probability, in Bayesian statistical inference, is the probability of an event before new data is collected.
Posterior probability is the revised probability of an event occurring after taking into consideration new information.
In statistical terms, the posterior probability is the probability of event A occurring given that event B has occurred.
Support vector machine
Random forest
Extreme Gradient Boosting
The XGBoost R package
XGBoost - Extreme Gradient Boosting:
- https://xgboost.readthedocs.io/en/latest/
- https://cran.r-project.org/web/packages/xgboost/index.html
The RTextTools package
RTextTools - A Supervised Learning Package for Text Classification:
- https://journal.r-project.org/archive/2013/RJ-2013-001/RJ-2013-001.pdf
- http://www.rtexttools.com/
- https://cran.r-project.org/web/packages/RTextTools/index.html
Chapter 3 - The Gofaaas-Fake News Detector R Package
Installation
Functions
Chapter 4 - Gofaas Web App
A way to interact, test and display the model results
Conclusion