An improved MDL-based compression algorithm for unsupervised word segmentation
The 51st Annual Meeting of the Association for Computational Linguistics - Short Papers (ACL Short Papers 2013)
Sofia, Bulgaria, August 4-9, 2013
In this paper, we analyze the objective function used in regularized compression, a recently proposed MDL-based unsupervised word segmentation algorithm. By exploiting its connection to change in description length, we uncover a novel lower-bound approximate to the original objective. The proposed new objective improves the baseline regularized compressor, achieving comparable performance as the other state-of-the-art methods.
Conference Manager (V2.61.0 - Rev. 2792M)