Language understanding and subsequential transducer learning

摘要

The application of the Onward Subsequential Transducer InferenceAlgorithm (OSTIA) recently introduced by J. Oncina et al. (1993) to(pseudo-) natural language understanding is considered. For thispurpose, a task proposed by J.A. Feldman et al. (1990), as a touchstonefor comparing the capabilities of language learning systems has beenadopted and three increasingly difficult semantic coding schemes havebeen defined for this task. In all cases the OSTIA was consistentlyproved able to learn very compact and accurate transducers fromrelatively small training sets of input-output examples of thetask