Knowledge discovery from historical data: an algorithmic approach

The authors report the results of a study that involved the discovery of knowledge and the creation of knowledge bases from large, archived, textual databases. The knowledge base, which was based on a semantic network structure, was created automatically using two statistical algorithms. Each of the knowledge bases has more than 8900 terms (concepts) and 80000 relationships (links) in the area of East-bloc computing. With the help of four East-bloc computing experts, the authors evaluated the two knowledge bases in an experiment based on recall and recognition tests. Both knowledge bases were robust in capturing the concepts in the domain. The discovered knowledge was brought back into the textual database to create a tightly integrated and intelligent system. The authors believe their algorithmic approach to knowledge discovery can be applied to large-scale textual databases, in which the information is voluminous and the semantics are embedded in the texts. Current research efforts include the creation of a meta knowledge base, and design of semantic network and neural network based inferencing algorithms