A Shape-Based Visual Interface for Text Retrieval

n recent ye a rs, visualization has ra p i d ly evo lved from a specialized re s e a rch nich e i n to a viable tool for analysis. In the emerging discipline of information visualization, for example, researchers use visual analysis for ex p l o ring nonspatial info rm at i o n. 1 As visualization te ch n i ques mature and powe r-ful, affordable, personal computers become common, e ffo rts in applying visual te ch n i ques ex tend to bro a d e r problem domains. The explosion in the number, size, and availability of info rm a t i o n s o u rces has sparked inte re st in data mining, which allows inte ra c t i ve ex p l o ration and discove ry of hidden k n owl e d ge and relationships buri e d w i thin this data glut. Vi s u a l i z a t i o n augments this analysis process. Coupling visualization with info rm a t i o n mining and analysis te ch n i qu e s fo rms a unique and powe rful new p a radigm for ex p l o ration and disc ove ry—visual analys i s. Visual data mining re qu i res a t i g h t ly coupled visual inte rface with u n d e rlying info rmation ret ri ev a l and analysis engines. To be useful, these te ch n i ques must augment human ex p l o ration and discove ry b eyond ex i sting methods. Since d a ta mining aims to provide humans with support i n g tools to think and ex p l o re, human perc e ptual issues remain an imp o rtant component of effe c t i ve visual inte r-faces for such syste m s. In this article, we describe a shape-based visual inte r-face for info rmation ret ri eval and inte ra c t i ve ex p l oration that exploits shape recognition. Our ex p l o ra to ry s ystem uses pro c e d u ra l ly ge n e ra ted shapes coupled w i th an underlying tex t-ret ri eval engine. A visual inte r-face based on …