Keyword-aware semantic search solution by Pinecone provides advanced hybrid search capabilities
Machine learning search infrastructure company, Pinecone Systems Inc., is unveiling its keyword-aware semantic search solution to bring to smaller organizations the sort of highly relevant search results that tech giants with extensive investments in data science and engineering work were previously only able to achieve.
Pinecone addresses a common issue in the intersection between “vector search,” semantic search, and keyword search. While vector search elicits search results based on semantic, or similar meanings, it neglects to incorporate keywords, leaving out results such as names or industry-specific terms. Most enterprises are forced to pick between semantic search and keyword search or running both systems in tandem.
Ultimately, this trend offered incomprehensive search results; in the case of running two systems simultaneously in an effort to combat incomplete searches, the costs and complexities significantly increased.
According to the company, this clear obstacle in obtaining holistic search results prompted Pinecone to innovate keyword-aware semantic search, a hybrid search technology that generates a merged list of relevant results.
“Our research shows that keyword-aware semantic search is superior to either semantic search or keyword search separately,” said Edo Liberty, founder and CEO of Pinecone. “This release finally allows businesses to provide their users with the most relevant possible results no matter how specific the query or how unique the topic.”
Pinecone’s keyword-aware semantic search solution searches through two data types, “dense vectors” and “sparse vectors,” to create a ranked list of the results. “Dense vectors,” generated by ML models to represent meaning, and “sparse vectors,” generated by traditional keyword-ranking models such as BM25, locate relevant results and automatically fuse them together in the final result list.
“Most companies do not have the resources to provide their users with the types of advanced search solutions that tech giants do, and this release from Pinecone changes all of that,” said Liberty.
The Pinecone hybrid search feature is currently available in beta and can be requested for access via Pinecone’s website.
For more information about Pinecone’s keyword-aware semantic search solution, please visit https://www.pinecone.io/.