We present a deterministic model to predict all the phrase boundaries of a syntactic tree, including base constituent boundaries and nested constituent boundaries. The model only uses the word and part-of-speech (POS) information, while general parsers also use the phrase type information. Our model is divided into two stages and finally turned into four classification sub-models. The f-score of our model is comparable to Stanford parser's PCFG model and factored model when tested on Penn Treebank Section 23 using gold-standard POS tags, which shows that phrase boundary identification could be done without phrase labels and could achieve comparable result to Stanford parser.
展开▼
