Hey BERT! Meet the Databases: Explorations of Bidirectional Encoder Representation from Transformers Model Use in Database Search Algorithms

Abstract

Bidirectional Encoder Representations from Transformers (BERT) is an open-source algorithm that allows for easier searching in search engines as it looks for natural language combinations in Google. The same algorithm could be applied to free database resources and commercial databases to make searching easier for users. This article examines the possibilities of BERT in resources other than Google. Many of our users come to database searching using Google type searches only to find controlled vocabulary is the way in which many databases are structured. If BERT is incorporated into popular databases along with controlled vocabulary, searching for information and finding it would be much easier for novice and advanced users, though careful consideration should be given to the search bias the algorithm may induce.