A User-centered Approach To Nlp-based Visualization Design

As our work and personal lives increasingly move online,humans use online texts more and more for communication,knowledge transfer,learning,and recording of information.However,this “natural language” data is difficult to aggregate and parse because it is unstructured,in comparison to structured formats such as tables or forms,which is much more difficult for machines to process.The field of artificial intelligence has supplied some solutions to the problem of processing large amount of text.In particular,the subfield of natural language processing(NLP)focuses on the problem of finding ways for machines to parse human language in a meaningful way.Visualization has been proposed as a method to reveal more of the inner workings of advanced data processing algorithms to users,as has been explored in the field of visual analytics.However,current research in this area does not integrate user-centered design into the process of developing such interfaces.Furthermore,the evaluation of the interfaces has been limited.This thesis considers the design of an NLP-based visualization user interface element for online media sites,through a user-centered approach.The methodology integrates a user-centered design process with NLP data processing and data visualization,components which are all necessary for the development of the proposed feature.In the design case study,the methodology is applied to the design of the search interface for a particular news website.User research is conducted using the interview,questionnaire,and contextual inquiry methods.Then,through an iterative process of prototyping and evaluation,a visualization is developed to enhance the user experience of information seeking for news articles.The evaluation of the final design shows that NLP-based visualization can benefit users in information seeking activities: research results show that task efficiency and the user experience of novelty and stimulation were improved by the novel interface.