Interruptable Transition Networks

A specialized transition network mechanism, the interruptable transition network (ITN) is used to perform the last of three stages in a multiprocessor syntactic parser. This approach can be seen as an exercise in implementing a parsing procedure of the active chart parser family. Most of the ATN parser implementations use the left-to-right top-down chronological backtracking control structure (cf. Bates, 1978 for discussion). The control strategies of the active chart type permit a blend of bottom-up and top-down parsing at the expense of time and space overhead (cf. Kaplan, 1973). The environment in which the interruptable transition network (ITN) has been implemented is not similar to 6hat of a typical ATN model. Nor is it a straightforward implementation of an active chart. ITN is responsible for one stage in a multiprocessor parsing technique described in Lozinskii & Nirenburg, (1982a and b), where parsing is performed in essentially the bottom-up fashion in parallel by a set of relatively small and "dumb" processing units running identical software. The process involves three stages: (a) producing the candidate strings of preterminal category symbols; (b) determining the positions in this string at which higher-level constituents start and (c) determining the closing boundaries of these constituents. Each of the processors allocated to the first stage obtains the set of all syntactic readings of one word in the input string. Using a table grammar, the processors then choose a subset of the word's readings to ensure compatibility with similar subsets generated by this processor's right and left neighbor. Stage 2 uses the results of stage 1 and a different tabular grammar to establish the left ("opening") boundaries for composite sentence constituents, such as NP or PP. The output of this stage assumes the form of a string of triads llabel x M), where lah~l belongs to the vocabulary of constituent types. In our implementation this set includes S, NP, VP, PP, NP& (the "virtual" NP), Del (the delimiter), etc. X and M are the left and the right indices of the boundaries of these constituents in the input string. They mark the points at which parentheses are to be opened (x) and closed (y) in the tree representation. The values x and y relate to positions of words in the initial input string. For example, the sentence (i) will be processed at stage 2 into the string (2). The '?' in (2) stand for unknown coordinates y. (i) The very big brick building that sits 1 2 3 4 5 6 7 on the hill belongs to the university. 8 9 i0 ii 12 13 14 (2) (s 1 ?)(np 1 ?)(s 6 ?)(np& 6 6) (vp 7 ?)(pp 8 ?)(np 9 ?)(vp ii ?) (pp 12 ?)(np 13 ?) It is at this point that the interruptable transition network starts its work of finding the unknown boundary coordinates and thus determining the upper levels of the parse tree. An input string ~ triads long will be allocated n identical processors. Initially the chunk of every participating processor will be one triad long. After these processors finish with their chunks (either succeeding or failing to find the missing coordinate) a "change of levels" interrupt occurs: the size of the chunks is doubled and the number of active processors halved. These latter continue the scanning of the I TN from the point they were interrupted taking as input what was formerly the chunk of their right neighbor. Note that all constituents already closed in that chunk are transparent to the current processor and already closed in that chunk are transparent to the current processor and are not rescanned. The number of active processors steadily reduces during parsing. The choice of processors that are to remain active is made with the help of the Pyramid protocol (cf. Uozinskii & Nirenburg, 1982). The processors released