Toward integrated multimedia understanding for intuitive human-system interaction

摘要

The mental image directed semantic theory (MIDST) proposes an omnisensory mental image model, and its description language L_md is intended to facilitate intuitive human-system interaction such that which happens between nonexpert people and geographic information systems (GISs). This paper presents a systematic method for formulating and computing natural concepts (i.e., mental images) of physical reality in L_md and its application to spatial language understanding in view of cross-media operations in text and pictures.