Acquisition and Validation of Knowledge from Data

Reference: Walker, M. G. & Wiederhold, G. Acquisition and Validation of Knowledge from Data. 1990.

Abstract: Data provide a basis for most of our knowledge. Although most knowledge is obtained through personal experience and education, its direct extraction from databases also has an important role. The process of extracting knowledge from data may take several forms: 1. Automated abstraction and summarization; for example, extracting a concise description of a patient's history from a lengthy medical chart 2. Discovery of new knowledge about relationships; for example, discovering drug side effects by retrospective examination of medical databases 3. Discovery of new abstractions; for example, determining the need for a new factor to explain inconsistencies 4. Quantitative knowledge acquisition; for example, deriving likelihood ratios or other statistical parameters for rules in knowledge bases by statistical analysis of a database 5. Knowledge validation; for example, monitoring the database to ensure that knowledge-base rules entered in the past remain adequate, or testing the accuracy of new rules proposed by domain experts. In each case, we begin with a database of case examples, and apply a combination of statistical analysis and domain knowledge to extract the knowledge implicitly present in the data. In this chapter, we describe systems that we have implemented to perform these tasks.

Jump to... [KSL] [SMI] [Reports by Author] [Reports by KSL Number] [Reports by Year]
Send mail to: ksl-info@ksl.stanford.edu to send a message to the maintainer of the KSL Reports.