Synthexis meets with hundreds of companies every year. This page chronicles our thoughts on some of these companies and their place in the search, text analytics and cognitive computing markets.
Nara Logics. This neural net start-up bases its technology on neuroscientist and co-founder Dr. Nathan Wilson’s research on the behavior of neurons in the brain. Their Synaptic Intelligence Network differs from traditional neural networks in that it is massively recurrent and is inspired by how the brain remembers entities and their relationships to each other. Relationship weights are set through biological learning rules, rather than traditional statistical supervised learning approaches, in a bottom up rather than top down process. The implications for this technology, still in its early stages of productization, are intriguing. The first use is in creating a highly accurate recommendation engine that suggests personal, contextual options, based on previous outcomes. Nara Logics calls this “assistive decision making.” The architecture is highly scalable, and can accommodate tens of millions of entities, people, attributes and relationships in the same graph. It is also designed for exploration rather than demanding precise queries–the kind of interaction that people do naturally as they gather possibilities before coming to a decision.
Sinequa is a well-established technology company that has always been ahead of the market’s understanding of why search is more than a search box. Their platform applies text analytics and search to both structured and unstructured data, and acts as a foundation technology that gathers, organizes, integrates, analyzes and explores. Although Sinequa is pigeon-holed as a search company, it was one of the earliest to demonstrate the importance of unified access to all types of information and the analysis of the most diverse data sources, the precursor to cognitive computing. From the beginning, they pragmatically went about developing the connectors, analytics, API’s and administrative tools so that their customers could connect the dots between structured and unstructured sources. Their customers typically have a range of information needs that they want to satisfy with a single technology platform and multiple front ends, often implemented as Search Based Applications to integrate with core business processes. Their products are probabilistic, interactive, adaptive and contextual. The company is moving into the US market.
VirtualWorks, founded in 2010, joins search luminaries from FAST with Citrix product developers. Their emphasis is on productized search that is downloadable, scalable and extensible. The product, ViaWorks, is available as an Azure cloud service, or as an on-premise enterprise search product. In either case, it is designed to work out of the box. It consists of a platform of connectors, OCR, database, search, a rules engine, and analytics that are well integrated. However, the product is both search- and database-agnostic, so that it is possible to swap out, for instance, the SOLR index for Elastic Search or another search product. On Jan. 6th 2016, they launched a partnership with Hitachi Solutions that delivers a new encrypted search version, solving a difficult problem in information access: accessibility vs. security. Ordinarily, it is not possible to search encrypted text without first decrypting the query, the documents and the index. This is a pretty big security hole, especially since in today’s mobile world, it’s likely that devices that need to access information will be outside the firewall. Hitachi’s Searchable Encryption technology, embedded in ViaWorks, is based on NIST’s AES encryption standard. This ViaWorks option allows similarity matching without decryption. Decryption only happens when the information arrives at an authorized device. The applicability in healthcare, finance, and government is obvious.