Short Text Classification Using kNN Based on Distance Function

Abstract: In the present day circumstances nowadays, the scope of short text such as Twitter messages, blogs, chat massages, book and movie summaries, forum, news feeds, and customer review is increasing very drastically. These applications pose a tremendous challenge to the text classifications due to sparseness of the relevant data & lack of similarity between the words. Short text classification is nothing but a process of assigning various input short texts to one or more target categories based on its contents.
Here we compared various algorithms such as Support Vector Machine, Naive Bayes, K-Nearest Neighbor, etc... . Based on this comparison I have selected knn for this. This paper includes various methods to reduce processing time and give good accuracy for testing instances.

Keywords: Short Text, Nearest Neighbor, Euclidean distance, Manhattan distance, Mahalanobis distance, Cosine similarity, k-fold cross validation , Holdout, Loocv, Repeated random sub-sampling