Information mining platforms
Corinna Cortes, Daryl Pregibon · ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD) · 1999
Experiencehas shown that the data extraction, parsing, cleaning and analysis steps of a KDD problem account for a much larger expenditure of resources (time and money) than the statistical modeling or machine learning part.Couple this statement with the need for fast turn-around time in commercial applications, and the obvious conclusion is that it is impractical to start from scratch for each new KDD application.To deal with this situation, we propose the use of an information mining platform that amortizes several of the critical pre-and post-processing steps needed to apply KDD.Thus, new KDD applications can leverage the platform for efficiency and robustness.
Cited by 9 works. doi:10.1145/312129.312270 Open-access PDF
Authors
References
- An Intrusion-Detection Model
- Organization and maintenance of large ordered indexes
- Adaptive Fraud Detection
- Organization and maintenance of large ordered indices
- The Alternating Decision Tree Learning Algorithm
- Approximate Bayes factors and accounting for model uncertainty in generalised linear models
- Catching the boat with Strudel
- A simple, fast, and effective rule learner
- Proceedings of the Fourth International Conference on Knowledge Discovery and Data Mining
- Daytona and the fourth-generation language Cymbal