April 2, 2020
ROSS Intelligence is an AI-first company. According to ROSS co-founder Jimoh Ovbiagele, that means “We live and die based on our AI capabilities. We didn’t pivot into AI. Our value proposition relies on it.”
That value proposition involves using AI to help lawyers quickly find answers to complex legal research questions in a body of over 14 million cases. Underpinning the platform are algorithms that belong to one of the most exciting and quickly-evolving AI subdomains – natural language processing (NLP) — the focus of which is the accurate extraction of meaning from language as its naturally spoken or written.
ROSS’ platform brings intelligent automation to the painstaking research processes that lawyers employ to determine how courts have interpreted statutes. Lawyers typically conduct legal research using Boolean searches, a type of search that allows multiple ideas to be included in one query and retrieves legal documents accordingly. An example search for a case related to a traffic accident may be: “autonomous vehicle AND injury AND finger.” Depending on the specificity of the case, queries can become long and complex, as they allow a string of key terms to be included, other terms to be excluded, and any synonyms the user can imagine to be listed as well.
The problem with this complexity is that a lawyer manually entering keywords may not include all terms and word relationships required to get a complete picture of case law. A user may overlook words in the long-tail of synonyms leaving potentially crucial documents undiscovered. What if a judge’s ruling substitutes “self-driving car” for autonomous vehicle? Or “digit” for finger?
ROSS’s platform replaces the need for a user to cover every possibility by using machine learning and NLP models to learn how to read and answer questions about legal documents. The result: faster, more accurate, and more complete results.
In 2017, ROSS became a founding sponsor of the Vector Institute. This sponsorship has allowed them to work with top foundational AI experts, including many premier NLP researchers. Ovbiagele says, “We’re interested in applied research. That’s the end of the spectrum we’re focused on rather than fundamental research. By having access to Vector faculty and being able to leverage them, we’re able to focus on the areas where we can create the most value for our organization.”
ROSS has leveraged Vector’s researchers and programs to get feedback on product development and to experiment with advanced NLP models in a consortium comprised of Vector industry sponsors. Through Vector’s Face-to-Face program, sponsors present highly-defined AI-related business opportunities or challenges to a Vector researcher with relevant expertise. Through this program, ROSS met with Dr. Frank Rudzicz, an award-winning NLP researcher, to get guidance on the most-advanced NLP developments related to linking specific parts of legal documents together and uncovering relationships among them.
Regarding the program, Ovbiagele says, “It’s like technically we have a direction we want to go, but we’re not sure how to get there and we don’t know what pitfalls we’re going to encounter. Then we go to the Face-to-Face program and we have people who have different maps of the research space, and not only do they have maps, they have commentary about the different pathways to where we want to go. That’s extremely helpful because we are not map makers, but they are.”
In another program, ROSS works in a consortium of 25 industry sponsor participants and 15 Vector researchers on NLP projects that directly relate to their core value proposition. Over 14 working sessions, knowledge transfer between Vector researchers and industry participants occurs as sponsors train an advanced natural language processing model (BERT — bidirectional encoder representations from transformers) in a cost effective manner, and practice reproducing and fine tuning a model trained in one domain for use in others, including health, law and, finance.
Ovbiagele says, “A lot of what Vector is working on is pretty advanced, so it allows us to quickly build our expertise and knowledge with these new techniques so when we are trying to commercialize them, we’re not starting from zero.” The program also saves the company time down the line. Ovbiagele continues: “We encounter problems in the course of those projects in the consortium, and then when we’re working on analogous problems inside our company, we’ve seen it before and we know the solution.”
ROSS’ Vector sponsorship also allows the company to participate in an important part of Vector’s mission: to build and sustain AI-based innovation, growth and productivity in Canada. For ROSS, this means doing good while doing well. Regarding the decision to sponsor Vector, Ovbiagele explains, “To stop the brain drain, to create IP, to make Canadian companies competitive in global AI landscape, to help bridge the divide between research and academia and the businesses that are commercializing it — that’s something we’re very passionate about.”
 ROSS Intelligence is a founding Bronze-level sponsor of the Vector Institute, which grants the company access to Vector’s education, training, and networking opportunities. Bronze sponsors support the Vector community by contributing their time and expertise as experienced entrepreneurs.