SOLVIZ: A Document Visualization System Based On Topic Modelling

Abstract

Abstract
The digital age has led to a voracious appetite for data as the world is gradually becoming a global village. Technology giants, businesses, schools and organization harness data in order to run their daily activities and beat competitors. Data has now become the crude oil of this global village. However, data could be structured, semistructured or unstructured. Understanding contents and features in the stack of data and getting hidden information stack could be very tedious and time consuming. Fortunately, a lot of information visualization techniques have been implemented to bring out meaning from the various forms of data. Meaningful information can be derived
from data with the applications of these techniques. However, the application of some of these techniques pose some level of difficulties for the users. The presentation provided by some of these visualization techniques has rendered little help while leaving the user confused on getting around the information needed. This paper presents an information visualization abstraction for document visualization using the solar system. The abstraction led to the development of a document visualization tool tagged Solviz that provides an effective and powerful discovery of information in documents using topic modelling for extracting thematic elements from documents. It enables
a user to explore hidden information in documents using a three dimensional space. Users can drill down into the document from topics to keywords to paragraphs in the document. The implementation also provides a data format which would allow anyone make use of this system efficiently.