Olga Pustylnikov

Geography of social ontologies: Testing a variant of the Sapir-Whorf Hypothesis in the context of Wikipedia

In this article, we test a variant of the Sapir-Whorf Hypothesis in the area of complex network theory. This is done by analyzing social ontologies as a new resource for automatic language classification. Our method is to solely explore structural features of social ontologies in order to predict family resemblances of languages used by the corresponding communities to build these ontologies. This approach is based on a reformulation of the Sapir-Whorf Hypothesis in terms of distributed cognition. Starting from a corpus of 160 Wikipedia-based social ontologies, we test our variant of the Sapir-Whorf Hypothesis by several experiments, and find out that we outperform the corresponding baselines. All in all, the article develops an approach to classify linguistic networks of tens of thousands of vertices by exploring a small range of mathematically well-established topological indices.

Evolution of Romance Language in Written Communication: Network Analysis of Late Latin and Early Romance Corpora

In this paper, the authors induce linguistic networks as a prerequisite for detecting language change by means of the Patrologia Latina, a corpus of Latin texts from the 4th to the 13th century.

Classification of Documents Based on the Structure of Their DOM Trees

In this paper, we discuss kernels that can be applied for the classification of XML documents based on their DOM trees. DOM trees are ordered trees in which every node might be labeled by a vector of attributes including its XML tag and the textual content. We describe five new kernels suitable for such structures: a kernel based on predefined structural features, a tree kernel derived from the well-known parse tree kernel, the set tree kernel that allows permutations of children, the string tree kernel being an extension of the so-called partial tree kernel, and the soft tree kernel as a more efficient alternative. We evaluate the kernels experimentally on a corpus containing the DOM trees of newspaper articles and on the well-known SUSANNE corpus.

Structure-sensitive learning of text types

In this paper, we discuss the structure based classification of documents based on their logical document structure, i.e., their DOM trees.We describe a method using predefined structural features and also four tree kernels suitable for such structures. We evaluate the methods experimentally on a corpus containing the DOM trees of newspaper articles, and on the well-known SUSANNE corpus. We will demonstrate that, for the two corpora, many text types can be learned based on structural features only.

Structural Differentiae of Text Types – A Quantitative Model

The categorization of natural language texts is a well established research field in computational and quantitative linguistics (Joachims 2002). In the majority of cases, the vector space model is used in terms of a bag of words approach. That is, lexical features are extracted from input texts in order to train some categorization model and, thus, to attribute, for example, authorship or topic categories. Parallel to these approaches there has been some effort in performing text categorization not in terms of lexical, but of structural features of document structure. More specifically, quantitative text characteristics have been computed in order to derive a sort of structural text signature which nevertheless allows reliable text categorizations (Kelih & Grzybek 2005; Pieper 1975). This “bag of features” approach regains attention when it comes to categorizing websites and other document types whose structure is far away from the simplicity of tree-like structures. Here we present a novel approach to structural classifiers which systematically computes structural signatures of documents. In summary, we present a text categorization algorithm which in the absence of any lexical features nevertheless performs a remarkably good classification even if the classes are thematically defined.