Search

Abstract: The paper describes an application of artificial neural networks (ANN) for natural language text reasoning. The task of knowledge discovery in text from a database, represented with a database file consisting of sentences with similar meanings but different lexico-grammatical patterns, was solved with ANNs which recognize the meaning of the text using training files with limited dictionary. The paper features recognition algorithms of text meaning from a selected source using 3-layer ANNs. Tests of the new method have also been described. [Copyright &y& Elsevier]

Abstract: In recent years a great effort has been made in order to create Intelligent Tutoring Systems that get close to human teaching. Some of the handicaps of the systems already created are the impossibility of sharing the courses between different Intelligent Tutoring Systems and the difficulty of creating them. Once the intelligent tutoring system is created, creating a new course is an expensive job that requires the intervention of many people that are expert in different areas. In this paper a generic and extensible authoring tool to create courses for different Intelligent Tutoring Systems is presented. This authoring tool allows the creation of courses for different types of intelligent tutoring systems. Once a course is created it can be exported to another intelligent tutoring system, reusing the domain model that the course represents. The prototype of the authoring tool has been tested with two simple Intelligent Tutoring Systems. [Copyright &y& Elsevier]

Abstract: The task of selecting relevant features is a hard problem in the field of unsupervised text clustering due to the absence of class labels that would guide the search. This paper proposes a new mixture model method for unsupervised text clustering, named multinomial mixture model with feature selection (M3FS). In M3FS, we introduce the concept of component-dependent “feature saliency” to the mixture model. We say a feature is relevant to a certain mixture component if the feature saliency value is higher than a predefined threshold. Thus the feature selection process is treated as a parameter estimation problem. The Expectation–Maximization (EM) algorithm is then used for estimating the model. The experiment results on commonly used text datasets show that the M3FS method has good clustering performance and feature selection capability. [Copyright &y& Elsevier]

Abstract: Finding previously unknown patterns in a time series has received much attention in recent years. Of the associated algorithms, the k-motif algorithm is one of the most effective and efficient. It is also widely used as a time series preprocessing routine for many other data mining tasks. However, the k-motif algorithm depends on the predefine of the parameter w, which is the length of the pattern. This paper introduces a novel k-motif-based algorithm that can solve the existing problem and, moreover, provide a way to generate the original patterns by summarizing the discovered motifs. [Copyright &y& Elsevier]

Abstract: Traditionally, interactive natural language systems assume a semantic model in which the entities referred to are in some abstract representation of a real or imagined world. In a system where graphical objects such as diagrams may be on the screen during the language interaction, there is possibility that the user may want to allude to these visual entities. Since graphical objects have their own properties (colour, shape, position on the screen, etc.) but may also represent items in a knowledge base which have other associated properties (price, geographical location, technical specifications, etc.), some systematic way is needed to enable such objects to be referred to in terms of either their screen properties or their associated attributes from the domain under discussion. In this paper, we present a formalisation for these arrangements, and show how our logical definitions can be used to generate constraints suitable for reference resolution within a natural language interpreter. [Copyright &y& Elsevier]

Abstract: Data pruning works with identifying noisy instances of a data set and removing them from the data set in order to improve the generalization of a learning algorithm. It has been well studied in supervised classification where the identification and removal of noisy instances are guided by available labels of instances. However, to the best knowledge of the authors’, very few work has been done on data pruning for unsupervised clustering. This paper deals with the problem of data pruning for unsupervised clustering under the condition that labels of instances are unknown beforehand. We claim that unsupervised data pruning can benefit for the clustering of the data with noise. We propose a feasible approach, termed as unsupervised Data Pruning using Ensembles of multiple Clusterers (DPEC), to identify noisy instances of a data set. DPEC checks all instances of a data set and identifies noisy instances by using ensembles of multiple clustering results provided by different clusterers on the same data set. We test the performance of DPEC on several real data sets with artificial noise. Experimental results demonstrate that DPEC is often able to improve the accuracy and robustness of the clustering algorithm. [Copyright &y& Elsevier]