Optimal Text Classification
-
Non-Metric Optimization
The foundation for Notiora's optimized classification, Non-Metric Optimization (NMO) is a unique approach to refining and perfecting standard classification models (such as Naive Bayes, and Mixture of Experts). NMO transforms the base classification model to generate optimized accuracy of classification - accuracy not obtainable with the base model alone. NMO technology is also applicable to many process and queuing problems where rank ordering is critical.
-
Optimized Bayesian Text Classification
Naive Bayesian text classification is a widely used (and publicly available) technology for text classification. Although Naive Bayes has seen success on numerous problems, the NB algorithm has significant limitations. Notiora's Optimized Bayesian technology overcomes these limitations by transforming the core Bayes classifier in to an optimized engine, with demonstrably superior classification accuracy.
-
Hybrid Classification and Expert Evidence Combination
In many classification problems, highly specialized domain knowledge plays a critical role in achieving classification success. "In-house" solutions often contain implementations of human expert knowledge embedded within the code. For example, spam filters may detect certain message headers, return addresses, etc, and flag these as spam. However, as the amount of codified knowledge increases, reconciling this evidence becomes intractable. Notiora technology leverages your existing investment, creating optimized classifiers that combine the available evidence from both human-engineered and traditional statistical classifiers.
