Madhusudanan, N and Gurumoorthy, Balan and Chakrabarti, Amaresh (2019) From natural language text to rules: knowledge acquisition from formal documents for aircraft assembly. In: JOURNAL OF ENGINEERING DESIGN, 30 (10-12). pp. 417-444.
Full text not available from this repository. (Request a copy)Abstract
Knowledge acquisition is a well-acknowledged bottleneck in the building of knowledge-based systems. Documents are a useful source of knowledge from experts. This paper targets the reuse of knowledge from the assembly phase of a product in the design and planning phases. Issues, their causes and the parameters involved are necessary to be acquired for reusing the knowledge so acquired. This paper discusses a method for knowledge acquisition, as a pipeline of existing tools in natural language understanding and processing. The acquired knowledge is expected to help in the decision making for a smart manufacturing system. The process of knowledge acquisition involves recognising the presence of issues and their causes using a combination of sentiment analysis and text patterns. The causes are then dissected to identify the constraints and constituent parameters. These pieces of knowledge are then reconstructed to form rules in a knowledge base. This paper demonstrates progress towards realising the method, by developing the cause dissection and rule-writing components, and validation of the issue-cause acquisition component with human subjects. A discussion is then presented on the potential integration and validation of the overall knowledge acquisition pipeline with a smart manufacturing system.
Item Type: | Journal Article |
---|---|
Publication: | JOURNAL OF ENGINEERING DESIGN |
Publisher: | TAYLOR & FRANCIS LTD |
Additional Information: | Copyright of this article belongs to TAYLOR & FRANCIS LTD |
Department/Centre: | Division of Mechanical Sciences > Centre for Product Design & Manufacturing |
Date Deposited: | 23 Dec 2019 09:14 |
Last Modified: | 23 Dec 2019 09:14 |
URI: | http://eprints.iisc.ac.in/id/eprint/63324 |
Actions (login required)
View Item |