Explanation-based learning and finite state transducers: Applications to parsing lexicalized tree adjoining grammars

B. Srinivas (University of Pennsylvania)

Explanation-based Learning techniques have been applied in NLP for speeding up parsing in limited domains. In [1], we showed that combining the LTAG representation with the EBL technique provides a novel method for parsing based on a FST mechanism and demonstrated a speedup in parsing times on the ATIS corpus. In this paper, we extend that approach to account for cases of ``long distance extractions'' and show that the same FST mechanism can be used for these sentences as well.


PS version (6 pages, 160k)

PDF version (6 pages, 152k)