Probabilistic models for semantic visualization and its applications
Abstract
Visualization of high-dimensional data, such as text documents, is useful to map out the similarities among various data points. In the high-dimensional space, documents are commonly represented as bags of words, with dimensionality equal to the vocabulary size. Classical document visualization directly reduces this into visualizable two or three dimensions. Recent approaches consider an intermediate representation in topic space, between word space and visualization space, which preserves the semantics by topic modeling. These approaches consider the problem of semantic visualization which attempts to jointly model visualization and topics. With semantic visualization, documents with similar topics will be displayed nearby. This dissertation focuses on building probabilistic models for semantic visualization by modeling other aspects of documents (i.e., document relationships and document representations) in addition to their texts. The objective is to improve the quality of similarity-based document visualization while maintaining topic quality. In addition, we find applications of semantic visualization to various problems. For document collection visualization, we develop a system for navigating a text corpus interactively and topically via browsing and searching. Another application is single document visualization for visual comparison of documents using word clouds.