Learning from examples is an important characteristic feature of intelligence in both natural and artificial intelligent agents. In this paper, we study learnability of term rewriting systems from positive examples alone. We define a class of linear-bounded term rewriting systems that are inferable from positive examples. In linear-bounded term rewriting systems, nesting of defined symbols is allowed in right-hand sides, unlike the class of flat systems considered in Krishna Rao [8]. The class of linear-bounded TRSs is rich enough to include many divide-and-conquer programs like addition, logarithm, tree-count, list-count, split, append, reverse etc.
Cite as: Rao, M.R.K.K. (2006). Learnability of Term Rewrite Systems from Positive Examples. In Proc. Twelfth Computing: The Australasian Theory Symposium (CATS2006), Hobart, Australia. CRPIT, 51. Gudmundsson, J. and Jay, B., Eds. ACS. 133-137.
(from crpit.com)
(local if available)