# semeval2014task9

**Repository Path**: plusyou13/semeval2014task9

## Basic Information

- **Project Name**: semeval2014task9
- **Description**:  UniGe Project - External participation in task 9 of Semeval 2014 : Sentiment Analysis in Twitter
- **Primary Language**: Unknown
- **License**: GPL-3.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-04-17
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Semeval 2014 task 9 : Sentiment Analysis in Twitter
*External participation in task 9 of Semeval 2014*

## First step into the Machine Learning

My first project of machine learning. My goal was to improve the results of the shared task.

The purpose of the shared task is to classify in a supervised manner a set of tweets into three classes: neutral, positive and negative.

To do it I had to do at least 3 steps: a baseline, implement by myself a neuron (perceptron), and finally propose improvements.

### Baseline

Personally I propose for my baseline to be bastard on a set of positive and negative word to classify them.

### Perceptron

From the training set I implement a perceptron from scratch.

### Improvement: Average Perceptron & Actif learning

To improve my results I implement an average perceptron. My data is evaluated by three neurons before classifying a tweet. I tried a dozen hypotheses and combinations of hypotheses here to correctly bias my machine learning.
I propose competing systems to make active learning.

## More infos

Read the [report](https://github.com/poggioenzo/semeval2014task9/blob/master/Rapport-MELS.pdf).