In the world of data-driven solutions, progress and innovation are critical. At snapADDY, we are constantly looking for ways to improve our products and solutions accordingly. Because we know you count on our products to make your work more efficient and productive.
Today we want to share our latest optimisations with you: Title recognition quality has increased by 11%, first names by 7% and last names by 4% – with an already high overall recognition rate in the 90% range.
This progress is mainly due to targeted technological adjustments. Our Data Team replaced our previous model with a quantised, multilingual DistilBERT model. This new model shows improved generalisation capabilities, which means it can better handle unknown input data and still deliver accurate results. This change has made our system more efficient and flexible overall, as well as making the quality of our predictions more precise.
The advances have a direct impact on aspects of our products, including duplicate checking, which enables duplicate data entries to be identified and cleansed more efficiently, and our LinkedIn integration, which optimizes the collection of contact data from LinkedIn.
In an effort to continually optimize our models, we take inspiration from our own best practices. Take our business card scanner, the snapADDY CardScanner, for example: it outperforms our competition in terms of recognition quality, as shown by the CardScanner study. This high benchmark motivates us to get the best out of our latest models and to continuously optimize them.
Of course, new models and techniques are always emerging in the field of Machine Learning and Data Science. We see this rapid development as a challenge, but also as an opportunity to test innovations and ultimately integrate them into our products. Our team is already working on adapting our products to the latest developments. After all, the goal is to ensure that our models are always up to date so that you can benefit from the most advanced and efficient solutions.