While most documents in eDiscovery data sets today are readily searchable, a significant and growing portion still requires separate workflows and intensive manual review.
In a recent session at Relativity Fest London I spoke about the growing problem of non-searchable data, and how CDS Vision AI helps legal teams quickly and easily unlock these rich evidentiary sources. Here are some highlights.
Non-Searchable Data: The Blind Spot in Your Documents
With all the incredible advancements in eDiscovery technology over the years, it’s easy to overlook documents with missing or poor-quality extracted text that may be relevant to a matter but are essentially invisible to search, analytics and many GenAI tools.
In a recent analysis of over a dozen CDS RelativityOne instances comprising more than 1.5 billion documents, image-only files accounted for 10% of the total population with HEIC files from mobile collections driving a 162% year-over-year increase. Another 10% were composed of PDF files, also a hidden risk especially when considering poor-quality scans that may contain handwritten annotations.
Not properly addressing these document types can result in significantly higher review costs, incomplete productions, or worse, the inadvertent production of potentially privileged or sensitive information.
Addressing Non-Searchable Data with AI
Understanding how to solve the problem of non-searchable files starts with understanding where traditional solutions like OCR fall short.
OCR is the technology that converts images into machine-readable text. OCR engines scan pixels, matching shapes against character templates. They require ideal document conditions with standard typewritten fonts and lack the ability to fill in gaps caused by damaged pixels from poor quality scans. So handwritten text, poor-quality scans, or anything else that doesn’t conform to those templates like image only documents fail.
CDS Vision offers next-generation image analysis and OCR technology, leveraging AI-enhanced capabilities to solve non-searchable data challenges. AI Image Analysis generates text-searchable image captions for image only documents, and Vision AI OCR leverages deep learning models trained on millions of documents to accurately interpret and transcribe poor quality scans and handwritten notes into good quality searchable text.
CDS Vision is also Relativity ECA-approved and fully integrated with RelativityOne, unlocking previously unsearchable data sets within the secure Microsoft Azure R1 Cloud environment.
Evidence isn’t confined to clean, preset templates. From photos to notes, eDiscovery files typically include large volumes of documents that can’t be read by traditional OCR software, adding countless hours of manual review time.
CDS Vision sees what those legacy technologies can’t, unlocking previously invisible files so they can readily be searched and applied as evidence. To learn more about how CDS Vision works and get started transforming your non-searchable data, please reach out to .
