Inductive Logic Programming (ILP) is a form of machine learning. It is an approach to machine learning where relations are induced from examples. The relations take the form of predicate logic. In ILP logic is used as the hypothesis language and the primar
Inductive Logic Programming for Classification of Interrogative Sentences and Frame Grammars over a Restricted Domain
Cameron A. Hughes
Youngstown State University
CTEST LaboratoriesTracey HughesYoungstown State Universitytdhughes@ysu.eduCTEST Laboratoriesctestlabs@ctestlabs.org
Abstract
Inductive Logic Programming (ILP) is a form of machine learning. It is an approach to machine learning where relations are induced from examples. The relations take the form of predicate logic. In ILP logic is used as the hypothesis language and the primary artifact of learning is a set of predicate logic formulas that constitute a logic program. Because the primary artifact of learning in ILP is a logic program, ILP is often described as the intersection between machine learning and logic programming.
One of the primary advantages of using ILP for machine learning is that ILP provides a very general way of specifying apriori knowledge or domain knowledge. In addition to providing a convenient method of specifying background knowledge, ILP focuses more on the learning of relations as opposed to attribute-value learning. The support for recursive structures in an ILP context is of particular utility in learning the relations. In this paper we demonstrate a technique for using ILP to learn sentence frame grammars to be used in interrogative sentence processing. A sentence frame is one of the simplest techniques in natural language processing for capturing grammatical description. The grammatical description that we focus on in this paper is the interrogative sentence.
1. Introduction
Inductive Logic Programming (ILP) is a method of machine learning. It is a technique in machine learning for inducing relations from examples. The relations take the form of predicate logic. The predicate logic is often represented as Horn clauses. In ILP, predicate logic is used as the hypothesis language and the result of learning is also represented as logical formulae in predicate logic. This makes logic programming languages such as Prolog and Mercury particularly suitable for implementing ILP methods. ILP methods present positive examples of relations and negative examples to the learner. The learner is the program that will induce the unknown relations. The positive and negative examples are presented to the learner and the learner must produce a relation or set of relations that capture the positive examples and exclude the negative examples. If the hypothesis (artifact) that the learner produces covers the positive examples the learner is said to be complete. If the artifact that the learner produces does not cover the negative examples the learner is said to be consistent. One of the primary goals of an ILP method is to be complete and consistent. We can now define the basic form of the ILP problem:
In this paper we are interested in finding a logic program P that learns the appropriate sentence frame
暂无评价 12页 免费 Inductive Logic Programm... 暂无评价 7页 免费... Generalized quantifiers in declarative and interrogative sentences ABSTRACT: In...
Inductive Logic Programm... 暂无评价 7页 免费 ... English Grammar Classification of Sentences, 楚军...(interrogative sentences): general questions (yes/...
Inference推理 and logical analysis of sentences ?...Deductive logic 演绎逻辑 vs. Inductive logic The ...(ii) interrogative 表示疑问的: What do you ...
to artificial intelligence, to computer programming...particular to the general 3 (called inductive ...interrogative sentence, 疑问句,不是命题. 8 ...
to artificial intelligence, to computer programming...particular to the general 3 (called inductive ...interrogative sentence, 疑问句,不是命题. 8 ...
热门文档