Dynamic Display of Text using Lexical Chains

Abstract

With the advent of digitization of text documents and digital communication technologies, an enormous quantity of information can be stored and transferred. However, digesting this huge amount of information is becoming an increasingly difficult task due to the time needed to read them. Furthermore, it is observed that people who searched for information does not necessary read all the information retrieved and this is especially true during the initial search for information. During the initial phase, people are keen to find out how relevant is the information retrieved. This research describes SmartView, a system that displays only a portion of a text file which is of interests to the reader. SmartView opens a text file, extracts and displays a list of useful words contained in the file. It then compiles and displays its initial view of the file. A view of the file is compiled based upon sentences that contain the words of interests and their anaphors and their lexical chains. Initially, the word that appears most frequently in the text (ignoring simple words such as prepositions, articles, etc.) is used. User can click on the list of useful words displayed to change the view and in doing so, the user sees the file according to his/her interests. A view of the file is displayed in a (linear) tree form so that only the key sentence in each paragraph is displayed. Nonetheless, user can click on the node of a tree to read more. User can also decide how much of a view to see. Options include 25%, 50%, 75% and 100%. In addition to a list of sentences being displayed, a view file also contains a graphic display showing the approximate whereabouts in the text of each of the sentences being displayed in the view.