Trains of Thought: Generating Information Maps

Опубликовано 12 августа 2016, 0:07

As data becomes increasingly prevalent, people can be easily overwhelmed by the flood of available information. The challenge of making sense of large amounts of data spans entire sectors -- from scientists trying to stay on top of the evolving literature to news readers who struggle to follow the twists and turns of news stories. Search engines are effective in retrieving nuggets of knowledge, but fitting those nuggets together into a single coherent picture remains difficult. Complex stories spaghetti into branches, side stories, and intertwining narratives; in order to explore these stories, one needs a map to navigate unfamiliar territory. I will describe the construction and evaluation of structured summaries of information, named metro maps. Metro maps explicitly show the relations among retrieved pieces in a way that captures story development and the interconnections among ideas and perspectives. I will formalize characteristics of good metro maps ! and formulate their construction as an optimization problem. Then, I will provide efficient methods with theoretical guarantees for generating maps. Finally, I will explore personalization methods, enabling users to custom-tailor the maps in accordance with their interests. Pilot user studies on real-world datasets (news stories and scientific publications) validate the value and promise of the methodology for producing maps that help users to acquire knowledge efficiently.