A Pr/T-Net formalism for HPSG parsing

This work presents a Pr/T-Net (predicate/transition-net) model for HPSG (head-driven phrase structure grammar) parsing. The model consists of two parts. The first part analyzes the structure of the input string. The second part processes the attribute relations among the components of the input string and establishes the corresponding attributed structure tree. An application example is examined and the result shows that the Pr/T-Net model can do HPSG parsing successfully. This Pr/T-Net formalism overcomes the limitation of some other currently available formalism such as pushdown automata and BFSA, which are not powerful enough to process the complex attribute relations of HPSG.