Generalizing predicates with string arguments

Abstract

Abstract. The least general generalization (LGG) of strings may cause an overgeneralization in the generalization process of the clauses of predicates with string arguments. We propose a specific generalization (SG) for strings to reduce overgeneralization. SGs of strings are used in the generalization of a set of strings representing the arguments of a set of positive examples of a predicate with string arguments. In order to create a SG of two strings, first, a unique match sequence between these strings is found. A unique match sequence of two strings consists of similarities and differences to represent similar parts and differing parts between those strings. The differences in the unique match sequence are replaced to create a SG of those strings. In the generalization process, a coverage algorithm based on SGs of strings or learning heuristics based on match sequences are used.

Anti-unification for Unranked Terms and Hedges

“…Therefore, there were attempts to compute only special generalizations. Cicekli and Cicekli [11] are interested only in generalizations that contain the longest common subsequence of the original words provided that this longest common subsequence contains all the letters that the given words have in common. Biere [7] considers only ε-free generalizations.…”

“…The algorithm for the latter defined in Cicekli and Cicekli [11] computes not all, but only a specific word generalization. The input words w 1 and w 2 should satisfy the following properties: (i) Their longest common subsequence w should not share a letter with the words w 1 − w and w 2 − w (called the differences), obtained from w 1 and w 2 , respectively, by deleting the subsequence w, and (ii) w 1 − w and w 2 − w should not share a letter.…”

“…The reason of such a combination is that in S the AUPs are between difference words. According to Cicekli and Cicekli [11], they are generalized differently than the original words. They should be decomposed to see whether there are two AUPs that can be generalized with the same variable.…”

Induction of logical relations based on specific generalization of strings

“…The bottomup method Least General Generalization proposed in [2] may cause overgeneralization in the clause generation in the absence of negative examples. In [1], a specific generalization (SG) of two strings is proposed to reduce overgeneralization. To compute SG, unique match sequence, which is a sequence of similarities and differences, is found in the initial step and followed by the generalization by replacing differences with variables.…”

“…Although [1] proposes a heuristic for generalization of strings, it is far from being an ILP system because of lack of background knowledge processing. The purpose of the research that is materialized in this paper was to develop an ILP system based on the specific generalization idea.…”