Applying Deductive Techniques to the Creation of Realistic Historical 3D Spatiotemporal Visualisations from Natural Language Narratives

Abstract

This paper builds on previous research into the development of the TMap3D system which allows for the creation of historical 3D spatiotemporal visualisations from natural language narratives [1]. This paper addresses the use of deductive 3D modelling techniques to generate the realism of the object behaviours, the environmental conditions and their interactions within the visualisation. The authors define deductive 3D modelling techniques as follows: (i) Given an object and its specification the Vist3D system (formally TMap3D) can deduce the object's visualised behaviour in response to a given "real world" instruction from the natural language narrative. (ii) The Vist3D system can deduce visualised environmental conditions from the natural language narrative. (iii) The Vist3D system can deduce the interactions between the objects (e.g. ship) and the environmental conditions (e.g. rough sea) to generate, for example, a ship rolling in rough seas.

Visualizing Natural Language Descriptions

“…Vist3D [54,59] is a domain-specific system for creating 3D animation of historical naval battles from narratives provided by the users. The system uses a manually populated ship specification database and a temporal database.…”