Метки тем

We introduce DocSCAN, a completely unsu- pervised text classification approach built on the Semantic Clustering by Adopting Nearest- Neighbors algorithm. For each document, we obtain semantically informative vectors from a …...