Auto-Relevancy Baseline: A Hybrid System Without Human Feedback

摘要

On obtaining a Request for Production and automatically emulating a typical eDiscovery workflow, a simple application of the classical Bayes algorithm upon the pseudo-hybridization of Semantic and Latent Semantic Indexing systems should smooth out historically high yet noisy Recall of some LSI models and their derivatives and produce a tighter linear distribution when assessing relevant documents unsupervised.