Search Techniques for Learning Probabilistic Models of Word Sense Disambiguation, inAAAI Sp.pdf

App ears in the Working Notes of the AAAI Spring Symp osium on Search Techniques for Problem Solving Under Uncertainty and Incomplete Information March Palo Alto CA Search Techniques for Learning Probabilistic Mo dels of Word Sense Disambiguation Ted Pedersen Department of Computer Science California Polytechnic State University San Luis Obisp o CA tp edersecsccalp olyedu Abstract yet general enough to handle the sizeable numb er of events not directly observed in that sample A para The development of automatic natural language un metric form is to o complex if a substantial numb er of derstanding systems remains an elusive goal Given the highly ambiguous nature of the syntax and se parameters have zerovalued estimates this indicates mantics of natural language it is not p ossible to de that the available sample of text simply do es not con velop rulebased approaches to understanding even tain enough information to supp ort the estimates re very limited domains of text The diculty in sp eci quired by the mo del However a parametric form is fying a complete set of rules and their exceptions has led to the rise of probabilistic approaches where mo d to o simple if relevant dep endencies among features are els of natural language are learned from large corp ora not represent