Weka is a collection of machine learning algorithms for data mining tasks. The algorithms can either be applied directly to a dataset or called from your own Java code. Weka contains tools for data pre-processing, classification, regression, clustering, association rules, and visualization. It is also well-suited for developing new machine learning schemes.
Arff is a data file format used by Weka. Look at http://www.cs.waikato.ac.nz/~ml/weka/arff.html for more information on it. You can find syntax highlighters for arff file in various editors at http://www.cs.waikato.ac.nz/~ml/weka/highlighting/index.html
Source: Weka FAQ