---
title: "Variable-Depth Natural Language Understanding"
authors:
  - "Daniel Kayser"
  - "Daniel Coulon"
doi: ""
url: "https://ijcai.org/proceedings/1981/15"
pdf: "https://ijcai.org/proceedings/1981/0015.pdf"
venue: "Proceedings of the Seventh International Joint Conference on Artificial Intelligence"
year: 1981
pages: "64-65"
track: "Natural Language"
keywords:
---

# Variable-Depth Natural Language Understanding

Daniel Kayser, Daniel Coulon

*Proceedings of the Seventh International Joint Conference on Artificial Intelligence*, pages 64-65.

## Abstract

S tandard A I repr e s e ntations of knowl e dge oper ate a t fixed depth (i.e. th e ob je c ts manipulat ed are des c ribed by an amount of information whi c h r emains constant for every task). Contrary to thi s approach, Variable Depth Pro ce ssing (VDP) us es a progressive des c ription of objects, tries different strategies according to the quality of the re s ult it needs, a nd continually controls this quality by means of an evaluation of th e approximati ons it makes. -- --Contextual Production Rules are shown t o be effect iv e way to implement some features of VDP. We are currently developing a VDP questionanswering sys tem which works on texts c on ce rnin g a no n-technical subje ct , namel y an e x ce rpt of a genera l public - oriented encyclopaedia.