KETOS
Use Neural Networks to Find the Documents You Need
Get More out of Your Documents
CCRi’s Ketos uses deep learning neural networks to analyze the semantic content of your documents and to generate keywords and similarity rankings that make it easy for your users to find the documents they need. Ketos especially offers benefits to Solr-based systems by adding new features unavailable in Solr as well as replacing some Solr features with superior alternatives.
Use Your Content's Semantics
Ketos analyzes documents with the same word embedding technology used at Facebook, storing word semantics as vectors that make it easier to find similarities and connections. Precision and recall with this technique shows marked improvement over the BM25 algorithm currently used by Solr as well as the Term Frequency-Inverse Document Frequency (TF-IDF) algorithm used by older versions of Solr and other applications since the 1970s.
Combine Keyword Metadata, Entity Recognition, and Other Data Sources to Improve Search Even More
Going beyond simple identification of each document's key terms, Ketos also performs entity recognition to find people, places, and organizations named in both your documents and queries. The distributed representation of document semantics enables fusion of this data with other data sources—for example, the browsing histories and interest profiles of others in a user's workgroup—to aid in finding relevant documents, or even to find connections between the people, places, and organizations mentioned in a document collection.
Bring These Benefits to Your System
Ketos offers a RESTful API to integrate with a variety of systems. If your system already uses the Apache Solr search platform, Ketos can be integrated at the JVM layer to replace Solr's BM25 or TF-IDF implementation and give you even better performance than the standard API. (Learn more about the technical details behind Ketos in our blog post Ketos: neural networks for document retrieval.)
Contact us at info@ccri.com to learn more about how Ketos can help your organization get more value out of their documents.