A collaborative research approach is helping Fidelity use natural language processing and AI to improve the customer experience and build regtech solutions.
Natural language processing (NLP) is an AI sub-field that holds big promise for financial services companies. By being able to detect, understand and respond to the natural language that people use in spoken or written form, NLP applications can help companies improve or even rethink how they support customers.
One company that has embraced the opportunities that NLP presents is Fidelity. The financial services giant has embarked on multiple NLP-enabled projects in recent years. For example, it has built a virtual assistant and chatbot infrastructure that can help quickly answer customers’ questions. While also exploring on the regulatory side, how NLP can be used to help avoid potential violations of content standards rules in marketing materials prior to communication with the public or submission to regulators.
Like many industrial sectors, financial services have a language all of their own. For that reason, companies like Fidelity cannot simply use off-the-shelf NLP products for their applications. Instead, companies must build bespoke solutions that required novel machine-learning NLP approaches and specialized expertise to understand the unique terminology and regulatory requirements of the financial services domain.
To accelerate the development of its in-house NLP expertise, Fidelity tapped into the Irish research ecosystem to leverage years of world-class research that is readily available within Ireland’s academic institutions.
Why Ireland?
Fidelity has been innovating and operating in Ireland since the 90s. In that time, it has grown to over 1,000 technologists and has successfully leveraged Ireland’s unique research centre model for industry-and-academia collaboration, which is designed to maximize collaboration and ease the process of connecting industry with the right academic expertise.
Science Foundation Ireland, which is Ireland’s national STEM funding agency, supports 16 academic research centers in the country. Each center has a specific advanced technology focus and connects academics across universities in Ireland that share a particular research focus. This creates a critical mass of expertise under one virtual roof, from which the industry can find relevant researchers that match their business needs.
For its chatbot and virtual-assistant NLP project, Fidelity connected with the Insight SFI Research Centre for Data Analytics, one of the largest Data Analytics centers in Europe. Through the Center, Fidelity worked closely with the University of Galway, which has a focus on NLP industry applications.
The research teams involved in these projects value the ability to collaborate with industry not only locally, but also with AI and Data Science experts in the U.S. The academics also value opportunities to work with real data and confirm their research findings on real-world industry problems. Meanwhile, Fidelity gains access to state-of-the-art research that can improve how it serve and supports customers.
Doing this work in Ireland also makes good business sense to multi-national companies. Research projects in Ireland can be co-funded by the Irish Government through agencies like SFI and IDA Ireland. Additionally, research work that companies undertake in-house in Ireland can qualify for more government supports, like a 25% R&D tax credit.
Collaboration in action
Throughout Fidelity’s collaboration with Irish academia, university researchers have been effectively embedded in Fidelity, working with the company’s Data Scientists and Solution Architects. In addition to reducing barriers to collaboration, this approach delivers other benefits. For example, “it can help big organizations from being bound by an enterprise mindset”, as pointed out by Fintan O’Malley, Director of Research, Development and Innovation at Fidelity Investments Ireland.
“When you bring researchers and academics into the conversation, they help open up conversations to a lot more possibilities,” O’Malley said. “You may start a project with one specific objective and realize that there is much more possible here by drawing from the ‘state of the art’ in academic research. It’s not absolutely prescribed from beginning to end, and that’s the beauty of it.”
In complex domain spaces such as compliance, Fidelity leveraged expertise across multiple SFI research centers in areas such as NLP and Image Processing. “The research ecosystem worked seamlessly – giving the Fidelity product-development team access to a multi-disciplinary expertise with minimal effort” said O’Malley, identifying this as another benefit of this unique collaborative research process.
“SFI encourages multi-disciplinary collaboration in order to provide not only domain-specific research but offer industry cross-sectoral research,” he said. “For instance, in an NLP project where you may draw from social media data, you may also benefit from input from the social sciences schools. This holistic research offering can be a game changer.”
Results so far
Key outputs of Fidelity’s academic engagements have included publications and proof of concepts. The work has also led to the follow-on development by Fidelity of patents and new products, and the release of an open-source software tool for knowledge graph extraction called Saffron, which can be used to build out chatbot solutions.
