The Parallel Expert Parser (PEP): a thoroughly revised descendent of the Word Expert Parser (WEP)

~H>st z'act in this paper we present PEP (the Parallel Expert Parser~ Devos 198'7), a radically revised descendant of WEP (the Word Expert Parser, Small 1980). WEP's idea of linguistic entities as interacting processes has been retained, but its adherence to the word as the only entity has been rejected. Experts exist at different levels, communicate through rigidly defined protocols and are now fully designed to run in parallel. A prototype of PEP is implemented in Flat Concurrent Prolog ar{d runs in a Logix environment. of information is more complex tl~an a node in a connectionlst model (it may be a rule, for instance)~ but in which one attempts to keep the parallel, computation involving the items of information mere under control than can be done in a connectionist.model. (For examples of coarse-grain parallel NLU, see Hirakawa 1983 or Matsumoto 3.987). The research we present here is of the latter type of parallel NI.U. A potentially parallel NLU system (the Word Expert Parser, Small 1980) has been drastically revised so as to allow a truly parallel implementation (viz. in Flat ConCurrent Prolog, using the Logix environment (Silverman et al. 1986)); we call the resulting system the Parallel Expert Parser (PEP, Devos 1987). Io Introduction Work on parallel natural language understanding (NLU) is only starting to emerge. (This even holds for work on any kind of parallel AI (see e.g. Kowalik 1988)). In general, there seem to be two kinds of approaches to parallel NLU. On the one hand, there is what we call fine-grain parallelism; on the other hand, there is coarse-grain parallelism. With fine-grain parallel NLU we refer basically to the connectionist approach and its decendants. Connectionist models feature huge networks of small nodes of information; computation is represented by fluctuations of the activation levels of nodes and by (parallel) transmission of excitation and inhibition along connections. 1986) o With coarse-grain parallel NL~ we refer to a more modest kind, in which the smallest item 2. The WoEd Expmrt Parser (WEP) b~iefly described The Word Expert Parser (WEP, Small 1980) is a natural language understanding program in the AI tradition of semantic parsing (see also Hirst 1983v Hahn 1986, Cottrell 1985, Adriaens 1986a/b for WEP-inspired or-related work). The organization of the model differs strongly from that of a "classical" NLU system. Rather than having a number of components of rules that are applied (serially) to linguistic input by a …