Generating a grammar for statistical training
Richard Anthony Sharman, F. Jelinek, R. L. Mercer · 1990
Parsing sentences of a Natural Language(NL) is an essential requirement for a variety of NL applications, and has been extensively studied. In particular, the sort of tasks which it would be desirable to do, include the ability to tag each word with its part-of-speech; to delineate with brackets, and label with a category name, each syntactic phrase; and to be able to adapt to different types of source material. Despite some 30 years of active research performing these tasks with a high degree of accuracy on unrestricted text is still an unsolved problem.