On August 7th, 2023, Google announced research for a new ranking framework called Term Weighting BERT (TW-BERT) that will help improve search results. The other benefit of this announcement is that it will be easy to deploy in existing ranking systems.
Google has not confirmed that it is currently using the research yet. However, this research is a new breakthrough that improves ranking processes across the SEO board, including query expansion.
What is TW-BERT?
TW-BERT is research that has been written to possibly be a future ranking framework that assigns scores (called weights) to words within a search query. This helps the ranking framework more accurately determine the relevant documents for that search query.
It also is helpful in areas such as query expansion. Query expansion is a process that adds more words to a search query to help make a user’s search query more defined and help them find the right source they’re looking for.
TW-BERT Bridges Two Paradigms
The research ultimately brings two search methods to work better together. One method is statistically based, and the other is deep learning models. Bridging these two paradigms will help to determine which terms are the most relevant or non-relevant search terms in the query.
TW-BERT Will Be Easy to Add
One of the advantages of TW-BERT is that it can be inserted right into the current information retrieval rank processing if it is ever to be added as a ranking framework. Essentially like a drop-in component. This is a benefit that it doesn’t require specialized software or updates to the hardware to add to a ranking algorithm.