Email Contact Phone Company Visit Website

Neoris Spain Office

Avenida de Portugal 1




Juan J Morras
[email protected]
Back to all Neoris announcements

Seal Software Hosts Popular Meetup in Sweden to Explore Advancements in Machine Learning and Natural Language Processing

Event Receives Overwhelming Interest from Leaders in Advanced Technology 

Seal Software announced today that its first meetup session on advanced concepts in Machine Learning (ML) and Natural Language Processing (NLP) was a huge success, with additional sessions being scheduled to meet high attendance demand. The meetup, led by Seal’s ML team based in Gothenburg, Sweden, was designed to assemble some of the best minds in ML, describe and explore the newest techniques in the field, and help cultivate the next generation of advancements in this area. Interest in the meetup was also driven by Seal’s strong ties to Chalmers University in Gothenburg, a leading technology center in the advancement of data science.

The meetup was met with overwhelmingly high demand, with more than twice the number of registrants that could be accommodated. Because of this, Seal plans to schedule additional sessions in order to open the discussion to all community members interested in advanced technology. Attendee feedback from the first session was extremely positive.

“We’re thrilled to bring together some of the best minds in the field for discussion of the newest technologies in the area of data and sentiment extraction,” said Kevin Gidney, Chief Technology Officer and Co-Founder of Seal Software. “These ‘meetups’ between Seal’s dedicated Machine Learning team and data scientists in Gothenburg help us to further advance and explore the ever-changing field of data science, and ensure that Seal Software is leading the industry in the applications of NLP and ML for precise extraction of information from large quantities of unstructured content.”

Topics discussed by Seal’s ML team and other advanced technology leaders in the meetup included the paradigms proposed for the development of good hypotheses with less data, and an understanding of human learning and the process of formalizing it mathematically. The overall theme of the event was the prevention of “overfitting” by being economical about the amount of information extracted from a holdout set.

Additional discussion was based on Deep Learning algorithms, and how they have been applied to information extraction and sentiment analysis with varying degrees of complexity and success. Seal also shared its work in the variations on the Convolutional Neural Networks (CNN) architecture applied on top of word-vector features, and its experiences from applying CNNs to information extraction within the Seal platform.

Earlier in March, Seal announced the release of its V4.1, which featured significant enhancements to its Machine Learning framework, including adding new algorithms to dramatically improve the trainability of the platform, and to increase overall accuracy and speed. The development of the new ML framework was implemented by Seal’s ML team based in Gothenburg, which is comprised of data scientists and leaders in the field of ML technology.

More specifically, Seal’s V4.1 included the adding of ML models including Support Vector Machines, Word2Vec with CBOW, K-NN and Deep Learning within its Java framework. It also developed a proprietary ensemble algorithm to optimize the results from the layers of algorithms being used together to identify contract data. The result of this work dramatically improves how the system works to extract user-defined terms and provisions, and in particular, how it identifies how different combinations of words best fit a particular type of provision within a contract.