Sentiment Analysis is to build machine learning models that can determine the tone (positive, negative, neutral) of the text (e.g., movie reviews, tweets).
It is one of the most important and standard tasks in NLP. However, Arabic sentiment analysis has not been studied at a level as high as other languages, e.g., English, Chinese, French. One of the key factors is the lack of high-quality and large-scale training data.
In this challenge, we release 55K annotated tweets for Arabic sentiment analysis. This dataset will also be extremely useful for building Arabic language models, which are essential for supporting other Arabic NLP tasks, such as Question&Answering, Machine Translation, and Text Generation.
More information about the dataset can be found in a paper at https://arxiv.org/abs/2011.00578


## Important Notes:
- To be eligible to participate in this competition, please register via the competition [web page](https://wti.kaust.edu.sa/solve/Arabic-Sentiment-Analysis-Challenge) and download the dataset from there.
- Follow carefully the instructions on this page and the evaluation page.
- The deadline for this completion is May 31, 2021 (11:59 PM UTC). Any further deadline extension will be announced on the event website.
- Winners will be announced 15 days after the deadline.

## Data Pre-processing
The labels of the training and testing files should be converted (mapped) into integer values.
- Convert the label values (sentiment column) from string to integers following this: 
Positive→  1, Negative →  -1, Neutral →  0
- In python, you can do something like:  
`df.sentiment= df.sentiment.map({'Positive’:1,  'Negative':-1, 'Neutral':0})`