ERIC Number: ED107151
Record Type: RIE
Publication Date: 1975-Mar-21
Reference Count: 0
Detecting Syntactic Ambiguity: Three Augmented Transition Network Techniques.
Herman, L. Russell, Jr.
When a grammar is expressed in augmented transition network (ATN) form, the problem of detecting syntactic ambuguity reduces to finding all possible paths through the ATNs. Each successfully terminating path through the ATN generates an acceptable parsing of the input string. Two ATN forms, minimal-node and pseudo-tree, are described along with the conventions for traversing each. The two forms are compared in regard to efficient use of computer time and space and in regard to appropriateness for each of the three path-finding techniques. Three techniques are discussed for finding all acceptable paths through ATNs. The techniques are "Backtracking,""Simultaneous Parallel Analysis," and "Amputate And Re-enter." Relative merits of the three techniques are discussed in terms of computer execution time, required data storage, programmer time, and amenability of the program to modification. A rudimentary ATN-based parser for English has been written in SPITBOL to test the implementation of these techniques. (Author)
Publication Type: Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Southeastern Conference on Linguistics (SECOL) (13th, Vanderbilt University, March 1975)