Vietnamese spelling detection and correction using Bi-gram, Minimum Edit Distance, SoundEx algorithms with some additional heuristics
Phuong H. Nguyen, Thuan D. Ngo, Dung A. Phan, Thu P. T. Dinh, Huỳnh Quyết Thắng · 2008
The spelling checking problem is considered to contain two main phases: the detecting phase and the correcting phase. In this paper, we present a new approach for Vietnamese spelling checking based on Vietnamese characteristics for each phase. Our research approach includes the use of a syllable Bi-gram in combination with parts of speech (POS) to find out suspected syllables. In the correcting phase, we based on minimum edit distance, SoundEx algorithms and some heuristics to build a weight function for assessing suggestion candidates. The training corpus and the test set were collected from e-newspapers.