Analyzing and categorizing the sentiment of tweets.
Main goal: compare different approaches using different data sets to see which combination succeeds in associating the right sentiment with each tweet.
Secondary goal: create a firefox extension that associate at each tweet a sentiment value converted in an emoji:
| Value | Emoji |
|---|---|
| 0 | |
| 1 | 😀 |
-
[238.8MB] ~ 1,600,000 tweets
Every tweet has been annotated (0 = negative, 2 = neutral, 4 = positive), this dataset contains 6 fields/column:
-
target: polarity of the tweet (0 = negative, 2 = neutral, 4 = positive)
-
ID: tweet ID
-
date: date of the tweet
-
flag: the query
-
user: the user that tweeted
-
text: the text of the tweet
-
Allows to predict a binary outcome using the sigmoid function which outputs a probability between 0 (Negative) and 1 (Positive)
Allows to plot labelled data as points in a multi-dimensional space, there will be present a decision boundary that divides the data points in Positive and Negative
Use the dataset features to create Positive / Negative questions and continually split the dataset until you isolate all data points belonging to each class
A meta estimator that fits a number of decision tree classifiers on various sub-samples of the dataset and uses averaging to improve the predictive accuracy and control over-fitting
-
Precision
-
Recall
-
Accuracy
-
F1-Score