Acquiring and Filtering Knowledge: Discovery and Case-based Reasoning

E. Ajala and K. Ahmad (UK)

Keywords

: carbon nanotubes, corpus linguistics, knowledge management, casebased reasoning

Abstract

A method for semi-automatically acquiring knowledge directly from natural language texts has been developed and an initial study reported here. The hypothesis used in the development of the method is that a certain class of texts specifically learned articles reporting work carried out in a laboratory, is written according to a format developed by a specialist community in the first instance and, perhaps, according to a format developed by experimental scientists across disciplines. A specialist community has its discernible lexical preferences; specialist terms used to write sentences and phrases according to a local grammar. The lexical preferences assist in automatically populating a case-frame with objects as slot values. The local grammar helps in identifying meaning relationships between the objects and facilitates the slot `names' in the frame. The method has been tested on tracking developments in the emergent subject of nano-technology where the vocabulary has been drawn from different subject domains and where the experimental report follows consensually adapted formats in chemistry and experimental physics. The results of the analysis have been used to populate a case-based reasoning system.

Important Links:



Go Back