An Ordinal Multi-class Classification Method for Readability Assessment of Chinese Documents

摘要

Readability assessment is worthwhile in recommending suitable documents for the readers. In this paper, we propose an Ordinal Multi-class Classification with Voting (OMCV) method for estimating the reading levels of Chinese documents. Based on current achievements of natural language processing, we also design five groups of text features to explore the peculiarities of Chinese. We collect the Chinese primary school language textbook dataset, and conduct experiments to demonstrate the effectiveness of both the method and the features. Experimental results show that our method has potential in improving the performance of the state-of-the-art classification and regression models, and the designed features are valuable in readability assessment of Chinese documents.