Asset 1
Digital Reasoning Wants To Help You Teach Machines Natural Language Understanding

Launches Beta program for Synthesys Studio at Big Data Conference in New York City

NEW YORK CITY, October 1, 2015 – Digital Reasoning, a leader in cognitive computing, today announced a beta program for Synthesys Studio: a tool that will enable knowledge workers to quickly and intuitively train new language models. Launched at Strata + Hadoop World 2015, Synthesys Studio will be explored in a session entitled Cognitive Computing: From Theory to Ubiquity, delivered today at 4:35pm ET by Tim Estes, CEO and founder of Digital Reasoning, at the Javits Center Room 1 E6 /1 E7. Digital Reasoning can also be found at booth #665 in the Exhibit Hall.

Before the creation of Synthesys Studio, training of domain-specific language models was usually time consuming and required skills in computer programming. This led to a bottleneck when developing meaningful applications. Synthesys Studio removes this logjam in two ways. First, it provides an intuitive visual interface that allow users, from novice to expert, to quickly and easily train and test language models. Second, instead of needing a connection to the powerful Synthesys platform, Synthesys Studio runs on a personal computer, which gives users the freedom to build models with their own data, at no cost and without support from sophisticated IT resources.

“The Synthesys Studio beta program is an open invitation to experiment with parts of our cognitive computing platform and discover its power to help you achieve your goals,” said Estes. “Digital Reasoning is driving the progress of cognitive computing from what was an intriguing yet esoteric concept towards a future where computers that understand and learn are a ubiquitous reality.”

Using Synthesys Studio begins by taking a sample of the data to be analyzed. The user tags the language in order to train Synthesys to recognize significant features, such as domain-specific words and terms. This process, which previously took a week or more, can be completed within a few hours. The next stage is to start building analytical models; a range of interfaces aligns the task with the technical skills of the user. Simple versioning facilitates easy creation and testing of model variations, quickly determining which approach delivers the best analytical accuracy. Synthesys Studio even helps to identify which data is most effective in delivering accurate findings and recommends which additional relevant data samples should be added to further improve the model. Once a satisfactory model has been built, it can be transferred to the Synthesys server, ready to be used on full scale data sets.

“Synthesys Studio fills a need that we’ve been hearing from our customers in making NLP [Natural Language Processing] more accessible,” said Greg Lamp, founder and CTO of Yhat – a provider of tools and systems that help enterprises turn data insights into data-driven products. “We’re excited to partner with Digital Reasoning and participate in the Synthesys Studio beta program, and are confident it will move the market beyond legacy tagging applications and empower users to interactively teach a system to understand human language.”

The beta program will initially allow users to train Synthesys Studio to identify entities (people, places, things) in their data. For example, a lab technician in a pharmaceutical company seeking to understand what is being said about a particular drug could train Synthesys Studio to recognize relevant mentions by providing examples. Once trained, the software can analyze the unstructured text contained in a set of documents and extract accurate findings, knowing, for example, that “Penicillin,” “the antibiotic” and “the drug” resolve to the same entity. Future updates to the beta software will give users the ability to try out sentence-level classification.

According to Steve Culp, Senior Managing Director, Accenture Finance and Risk Services, “Accenture was an early supporter and advocate of Digital Reasoning through our New York FinTech Innovation Lab and we are excited to be working closely with Digital Reasoning as we leverage their technology to help our clients mitigate critical risks such as Human Conduct Risk and Insider Threats. We are looking forward to our continued relationship with Digital Reasoning and the release of their latest Synthesys offering, Synthesys Studio, which will help us accelerate the development of models and improve safeguards for our clients and enable improved insights around their customers and employees.”

Users can sign up now at beta.digitalreasoning.com.