Text data sets can be represented using models that do not preserve text structure, or using models that preserve text structure. Our hypothesis is that depending on the data set nature, there can be advantages using a model that preserves text structure over one that does not, and vice versa. The key is to determine the best way of representing a particular data set, based on the data set itself. In this work, we proposde B“orjae to investigate this problem by combining text distortion and algorithmic clustering based on string compression. Specifically, a distortion technique previously developed by the authors is applied to destroy text structure progressively. Following this, a clustering algorithm based on string compression is used to analyze the effects of the distortion on the information contained in the texts. Several experiments are carried out on text data sets and artificially-generated data sets. The results show that in strongly structural data sets the clustering results worsen as text structure is progressively destroyed. Besides, they show that using a compressor which enables the choice of the size of the left-context symbols helps to determine the nature of the data sets. Finally, the results are contrasted with a method based on multidimensional projections and analogous conclusions are obtained.

Compression distances have been widely used in knowledge discovery and data mining. They are parameter-free, widely applicable, and very effective in several domains. However, little has been done to interpret their results or to explain their behavior. In this paper, we take a step toward understanding compression distances by performing an experimental evaluation of the impact of several kinds of information distortion on compression-based text clustering. We show how progressively removing words in such a way that the complexity of a document is slowly reduced helps the compression-based text clustering and improves its accuracy. In fact, we show how the nondistorted text clustering can be improved by means of annealing text distortion. The experimental results shown in this paper are consistent using different data sets, and different compression algorithms belonging to the most important compression families: Lempel-Ziv, Statistical and Block-Sorting.

Multimedia Information Retrieval is currently a hot research topic due the popularity of the World Wide Web and the huge amount of multimedia data available. There exists an increasing interest to design and develop new methods and techniques to represent and classify this kind of information. Among the different sources of multimedia information currently available, we have decided to work with music audio files. Three different music representations (binary code, wave information, and SAX) have been used to study how the selection of a particular representation could affect a clustering process based on a set of similarity clusters. Two different algorithms (a hierarchical clustering method based on the quartet tree method and a genetic algorithm) have been applied to automatically perform the clustering. A compression distance, the Normalized Compression Distance (NCD), has been used to generate the similarities among the music files. This distance is parameter-free and widely applicable so we can use it directly with different formats and representations. The paper shows some experimental results using these representations and compares the behavior of both clustering methods.