Adapt emerging tech win highlights academic roots and industry experience
Adapt, the Research Ireland Centre for AI-driven digital content technology, in collaboration with Huawei, was awarded the emerging technology innovation of the year prize at this year’s Tech Excellence Awards. We speak with AI linguist at Adapt Dr Filip Klubička about why he thinks the project resonated with the judges and what the use cases are.
What do you think made your project with Huawei stand out?
I think the project shone a light on our expertise in academia and industry – this was a collaborative, intersectoral research project after all – and we have a long history of collaborating with industry partners. This experience helped us to develop a solution that satisfied the business needs of Huawei Ireland while using the research expertise and talent we have in-house. In the project, our team stayed up to date with the literature, they weren’t afraid to adapt when necessary, and they tapped into the latest developments in the field to create cutting-edge solutions.
Can you describe the project and its goals?
Our goal was to build a state of the art solution for the natural language to SQL (NL2SQL) translation task, which aimed to understand the meaning of natural language sentences and map them to meaningful executable queries, i.e. SQL. In a way, it can also be seen as a source code generation task.
In a typical NL2SQL scenario, given a relational database and a natural language question (NLQ), the goal is to find an SQL statement that, when executed on the database, will answer the NLQ. The challenge was twofold: understanding the meaning and intention of the NLQ and translating that meaning into an SQL query. While this is still an open problem in the research field, advancements in AI have accelerated the field of NL2SQL in the last decade.
We have used these advancements in our solution by leveraging the power of a large language model, combined with pipeline optimisation processes, allowing us to build a capable NL2SQL system.
What applications do you see for this system?
This system can be integrated into relational database management systems and become part of a natural language interface. This will make these databases much more accessible for end users.
Whether the user is a customer or employee, it can help improve the user experience by allowing them to use natural language questions to query the database, instead of writing SQL queries which they may not be proficient in. This can ultimately save a lot of time.
What does the future hold for the system?
As far as we’ve heard, the system is already being used in production by Huawei, and the hope is that it will continue to be used and iterated on for improvements. Beyond that, we are hoping to organise another intersectoral project with Huawei to continue the partnership and build systems based on cutting-edge GenAI technology.
What does winning a Tech Excellence award mean to the Adapt team?
It’s a wonderful accolade and a real feather in our cap, as it firmly positions the Adapt centre as a national leader in technological innovation. It shows that we operate with strong, high quality talent and are able to work successfully and effectively with industry partners.
Subscribers 0
Fans 0
Followers 0
Followers