Visualization of text document corpus

Automated text processing is commonly used when dealing with text data written in a natural language. However, when processing the data using computers, we should be aware of the fact that many words having different form share a common or similar meaning. For a computer this can be difficult to handle without some additional information--background knowledge. Latent Semantic Indexing (LSI) is a technique for extracting this background knowledge from text documents. It employs a technique from linear algebra called Singular Value Decomposition (SVD) and the bag-of-words representation of text documents for extracting words with similar meanings. This can also be viewed as the extraction of hidden semantic concepts from text documents. Visualization of a document corpus is a very useful tool for finding the main topics that the documents from this corpus talk about. Different methods were proposed for visualizing a large document collection using different underlying methods. For instance, visualization of large document collection based on document clustering [3], or visualization of news collection based on visualizing relationships between named entities extracted from the text [4]. Another example used in our work is visualization of European research space [5].