| Summary: | 碩士 === 國立成功大學 === 工程科學系碩博士班 === 100 === As more and more information including news articles, personal stories and so on gathered on the Internet, how a user perceives and memorizes every details of what he or she reads is of significant challenge. In view of this information overload problem, recent advances in the research field of information retrieval have resolved most difficulties when a user is able to provide appropriate keywords of his/her search target. Nevertheless, some important events which are etched deeply in one's memory may not be clearly defined as a few keywords or even easily recalled. Thus, we propose in this work to provide some visual suggestions to help users associate and recall their own stories from the memory while reading current news articles. In addition to the usage of fragment text, we propose to extract similarities from news photos. Consequently, resulting suggestions are effective for a user to associate the current news event with previous ones. As such implicit relationships are gradually found, the topic map is then constructed to help a user organize relevant concepts he or she has ever known or learned.
|
|---|
