Sinequa adds Neural Search capabilities to Cloud Platform

Enterprise search vendor Sinequa has announced the addition of optional neural search capabilities to its Search Cloud Platform, using four deep learning language models. These models are pre-trained and ready to use in combination with Sinequa’s advanced Natural Language Processing (NLP) and semantic search for the best relevance and question-answering capability, optimized to run efficiently even at scale.

“Quality and breadth of information retrieval and search have long been recognized as primary drivers of productivity, but relevance is key to enabling business insights and more informed decision-making,” said Alexandre Bilger, President and CEO of Sinequa.

“With Sinequa’s Neural Search capabilities, we’ve added best-in-class neural search to our existing best-in-class statistical search.”

Neural search models have been used in internet searches by Google and Bing since 2019, but computing requirements rendered them too costly and slow for most enterprises, especially at production scale. Sinequa optimized the models and collaborated with the Microsoft Azure and NVIDIA AI/ML teams to deliver a high performance, cost-efficient infrastructure to support intensive Neural Search workloads.

Neural Search is optimized for Microsoft Azure and the latest NVIDIA A10 or A100 Tensor Core GPUs to efficiently process large amounts of unstructured data as well as user queries.

Sinequa’s Neural Search improves relevance and is often able to directly answer natural language questions. It does this with deep neural nets that go beyond word-based search to better leverage meaning and context. Sinequa’s Search Cloud platform combines neural search with its extensive NLP and statistical search. This unified approach provides more accurate and comprehensive search results across a broader range of content and use cases.

Sinequa’s four deep learning models are trained for specific tasks and work in concert for the best possible relevance for any enterprise scenario. All four models are fully pre-trained, configured and optimized for enterprise content.

This eliminates the laborious and costly process of tagging large training sets, training custom models, and updating them over time. With Sinequa’s Neural Search, organizations can now deploy intelligent search-based solutions with cutting-edge deep learning technology quickly and easily.