“Meanings are relativized to scenes”
FrameNet 101
FrameNet is the computational application of the theory of Frame Semantics. In FrameNet, knowledge about the semantics of lexical items is modeled against the background semantic frames which they evoke. A frame is a sort of scene, a set of concepts related to each other in such a way that the presence of one of them makes all the other concepts readily available. The frames are also connected to each other by several types of formally defined relations, so they can be thought of as a network of nodes linked by different types of arcs.
The FrameNet project started in 1997, at the International Computer Science Institute in Berkeley, California, under the leadership of Prof. Charles Fillmore. During the past two decades, FrameNets have been built for many other languages, including Japanese, German, Spanish, Swedish, Brazilian Portuguese, Chinese, French, Hebrew, Korean and Latvian. FrameNet is applied in several areas of Natural Language Understanding. Global FrameNet is an effort to bring together all existing - and yet to be created - FrameNets in a common multilingual setting, focused on the development of collaborative research, shared tasks, and applications.