Korean Erroneous Sentence Classification With Integrated Eojeol Embedding

This paper attempts to analyze the Korean sentence classification system. Sentence is task of classifying an input based on predefined categories. However, spelling or space error contained in causes problems morphological analysis and tokenization. proposes a novel approach Integrated Eojeol (Korean syntactic word separated by space) Embedding reduce effect poorly analyzed morphemes classification. The also two noise insertion methods that further improve performance. Our evaluation results indicate applying proposed existing classifiers, accuracy erroneous sentences increased 8% 15%.