Discover v2.0 builds on the success of our first Discover solution to expand the power and applicability of the platform by adding several key new features.
In certain circumstances, especially when dealing with very short documents, our research showed that by using embedding based on analysing full sentences (“sBERT”) had the potential to improve result accuracy and relevancy. Discover v2.0 now adds the option of using these alternative embeddings to further allow us to ensure users obtain the best possible results.
Discover v2.0 also adds the ability to perform Smart Data Extraction (SDE) from unstructured information in your data sets. This means that for a complex, diverse, and disparate set of unstructured documents key information that is buried within the document (eg author, customer names, dates etc) can be extracted without the need for laborious human intervention. In Beta trials of this functionality with real-world customers we achieved accuracy rates greater than 85%, way exceeding that of manual processes and of course at a fraction of the time and cost. And since this is an AI application that learns from its usage, this accuracy rate will of course increase over time.
Discover v2.0 marks a significant step forward for iKVA’s solutions and technology, enhancing the value we can provide users and adding to our solution portfolio.
Jon Horden, iKVA CEO.
If you’d like to know more, please contact us, or alternatively you can sign up for our free, upcoming webinars:
‘A Voyage Into Discovery‘ hosted by Prof Richard Mortier
‘Vector Search: The Information Revolution’ hosted by Ian Firth and Jon Horden