Simple inductive learning algorithms assume that all attribute values are available. The well-known Quinlan's paper [Qui89] discusses quite a few routines for processing of unknown attribute values in the TDIDT family and analyzes seven of them. This paper introduces five routines for processing of unknown attribute values that have been designed for the CN4 learning algorithm, a large extensio...

This paper concerns methods for inferring decision trees from examples for classification problems. The reader who is unfamiliar with this problem may wish to consult J. R. Quinlan’s paper (1986), or the excellent monograph by Breiman et al. (1984), although this paper will be self-contained. This work is inspired by Rissanen’s work on the Minimum description length principle (or MDLP for short...

STRUCT is a system that learns structural decision trees from positive and negative examples. The algorithm uses a modification of Pagallo and Haussler's FRINGE algorithm to construct new features in a first-order representation. Experiments compare the effects of different hypothesis evaluation strategies, domain representation, and feature construction. STRUCT is also compared with Quinlan's ...

The decision tree output of Quinlan's ID3 algorithm is one of its major weaknesses. Not only can it be incomprehensible and difficult to manipulate, but its use in expert systems frequently demands irrelevant information to be supplied. This report argues that the problem lies in the induction algorithm itself and can only be remedied by radically altering the underlying strategy. It describes ...

We discuss a particular approach to automatic abstracting, where an abstract is created by extracting hnportant sentences from a text. A primary purpose of the paper is to demonstrate that the reliability of human supplied annotations on corpora has crucial effects on how well an automatic abstracting system performs. The corpus is developed through human judgements on possible s~,mmary sentenc...

Ensemble classification techniques such as bagging, (Breiman, 1996a), boosting (Freund & Schapire, 1997) and arcing algorithms (Breiman, 1997) have received much attention in recent literature. Such techniques have been shown to lead to reduced classification error on unseen cases. Even when the ensemble is trained well beyond zero training set error, the ensemble continues to exhibit improved ...

We examine the relationship between gender of the barrister and appeal outcomes on the High Court of Australia. We find evidence of asymmetries in that gender matters if a female barrister presents oral argument for the appellant opposed to a respondent, for which one or more male barristers present oral argument, but the reverse is not true. Specifically, we find that an appellant represented ...

A necessary function of ILP systems is to test whether rulesets intensionally cover examples. In general, when R contains recursive clauses, the above test may not terminate. In order that ILP techniques can serve end users directly in applications such as knowledge discovery in databases, general methods must be provided to ensure that the learned rulesets are executable and, in particular, th...

Employability development is a strategic priority for universities across advanced western economies. Despite this, there is no systematic study of employability development approaches internationally. In this study, we considered how universities portray employability on the public pages of their websites. We undertook website content analysis of 107 research-intensive universities in Australi...

Label semantics is a random set framework for modelling with words. In previous work, several machine learning algorithms based on this semantics have been proposed and studied. In this paper, we introduce a new linguistic rule induction algorithm based on Quinlan’s FOIL. According to this algorithm, a set of linguistic rules are generated for classification problems. The new model is empirical...