Wikipedia provides a knowledge base for computing word relatedness in a more structured fashion than a search engine and with more coverage than WordNet. In this work we present experiments on using Wikipedia for computing semantic relatedness and compare it to WordNet on various benchmarking datasets. Existing relatedness measures perform better using Wikipedia than a baseline given by Google counts, and we show that Wikipedia outperforms WordNet when applied to the largest available dataset designed for that purpose. The best results on this dataset are obtained by integrating Google, WordNet and Wikipedia based measures. We also show that including Wikipedia improves the performance of an NLP application processing naturally occurring texts.

A central problem in machine learning is identifying a representative set of features from which to construct a classification model for a particular task. This thesis addresses the problem of feature selection for machine learning through a correlation based approach. The central hypothesis is that good feature sets contain features that are highly correlated with the class, yet uncorrelated with each other. A feature evaluation formula, based on ideas from test theory, provides an operational definition of this hypothesis. CFS (Correlation based Feature Selection) is an algorithm that couples this evaluation formula with an appropriate correlation measure and a heuristic search strategy. CFS was evaluated by experiments on artificial and natural datasets. Three machine learning algorithms were used: C4.5 (a decision tree learner), IB1 (an instance based learner), and naive Bayes. Experiments on artificial datasets showed that CFS quickly identifies and screens irrelevant, redundant, and noisy features, and identifies relevant features as long as their relevance does not strongly depend on other features. On natural domains, CFS typically eliminated well over half the features. In most cases, classification accuracy using the reduced feature set equaled or bettered accuracy using the complete feature set.

Current systems often create socially awkward interruptions or unduly demand attention because they have no way of knowing if a person is busy and should not be interrupted. Previous work has examined the feasibility of using sensors and statistical models to estimate human interruptibility in an office environment, but left open some questions about the robustness of such an approach. This paper examines several dimensions of robustness in sensor-based statistical models of human interruptibility. We show that real sensors can be constructed with sufficient accuracy to drive the predictive models. We also create statistical models for a much broader group of people than was studied in prior work. Finally, we examine the effects of training data quantity on the accuracy of these models and consider tradeoffs associated with different combinations of sensors. As a whole, our analyses demonstrate that sensor-based statistical models of human interruptibility can provide robust estimates for a variety of office workers in a range of circumstances, and can do so with accuracy as good as or better than people. Integrating these models into systems could support a variety of advances in human computer interaction and computer-mediated communication. Author Keywords Situationally appropriate interaction, managing human attention, sensor-based interfaces, context-aware computing, machine learning.

Essentially all data mining algorithms assume that the datagenerating process is independent of the data miner's activities. However, in many domains, including spam detection, intrusion detection, fraud detection, surveillance and counter-terrorism, this is far from the case: the data is actively manipulated by an adversary seeking to make the classifier produce false negatives. In these domains, the performance of a classifier can degrade rapidly after it is deployed, as the adversary learns to defeat it. Currently the only solution to this is repeated, manual, ad hoc reconstruction of the classifier. In this paper we develop a formal framework and algorithms for this problem. We view classification as a game between the classifier and the adversary, and produce a classifier that is optimal given the adversary's optimal strategy. Experiments in a spam detection domain show that this approach can greatly outperform a classifier learned in the standard way, and (within the parameters of the problem) automatically adapt the classifier to the adversary's evolving manipulations.

This paper introduces concepts and algorithms of feature selection, surveys existing feature selection algorithms for classification and clustering, groups and compares different algorithms with a categorizing framework based on search strategies, evaluation criteria, and data mining tasks, reveals unattempted combinations, and provides guidelines in selecting feature selection algorithms. With the categorizing framework, we continue our efforts toward building an integrated system for intelligent feature selection. A unifying platform is proposed as an intermediate step. An illustrative example is presented to show how existing feature selection algorithms can be integrated into a meta algorithm that can take advantage of individual algorithms. An added advantage of doing so is to help a user employ a suitable algorithm without knowing details of each algorithm. Some real-world applications are included to demonstrate the use of feature selection in data mining. We conclude this work by identifying trends and challenges of feature selection research and development.

In this paper, we identify two issues involved in developing an automated feature subset selection algorithm for unlabeled data: the need for finding the number of clusters in conjunction with feature selection, and the need for normalizing the bias of feature selection criteria with respect to dimension. We explore the feature selection problem and these issues through FSSEM (Feature Subset Selection using Expectation-Maximization (EM) clustering) and through two different performance criteria for evaluating candidate feature subsets: scatter separability and maximum likelihood. We present proofs on the dimensionality biases of these feature criteria, and present a cross-projection normalization scheme that can be applied to any criterion to ameliorate these biases. Our experiments show the need for feature selection, the need for addressing these two issues, and the effectiveness of our proposed solutions.

This paper describes experiments in Machine Learning for text classification using a new representation of text based on WordNet hypemyms. Six binary classification tasks of varying difficulty are defined, and the Ripper system is used to produce discrimination rules for each task using the new hypernym density representation. Rules are also produced with the commonly used bag-of-words representation, incorporating no knowledge from WordNet. Experiments show

Machine learning algorithms automatically extract knowledge from machine readable information. Unfortunately, their success is usually dependant on the quality of the data that they operate on. If the data is inadequate, or contains extraneous and irrelevant information, machine learning algorithms may produce less accurate and less understandable results, or may fail to discover anything of use at all. Feature subset selectors are algorithms that attempt to identify and remove as much irrelevant and redundant information as possible prior to learning. Feature subset selection can result in enhanced performance, a reduced hypothesis search space, and, in some cases, reduced storage requirement. This paper describes a new feature selection algorithm that uses a correlation based heuristic to determine the "goodness" of feature subsets, and evaluates its effectiveness with three common machine learning algorithms. Experiments using a number of standard machine learning data sets are pres...

We propose in this paper a very fast feature selection technique based on conditional mutual information.

Data engineering is generally considered to be a central issue in the development of data mining applications. The success of many learning schemes, in their attempts to construct models of data, hinges on the reliable identification of a small set of highly predictive attributes. The inclusion of irrelevant, redundant and noisy attributes in the model building process phase can result in poor predictive performance and increased computation. Attribute selection generally involves a combination of search and attribute utility estimation plus evaluation with respect to specific learning schemes. This leads to a large number of possible permutations and has led to a situation where very few benchmark studies have been conducted. This paper presents a benchmark comparison of several attribute selection methods. All the methods produce an attribute ranking, a useful devise for isolating the individual merit of an attribute. Attribute selection is achieved by cross-validating the rankings with respect to a learning scheme to find the best attributes. Results are reported for a selection of standard data sets and two learning schemes C4.5 and naive Bayes. 1