June 10, 2021


NLP is 71% correct

Unveiling the Future of Declarative Process Discovery: A Collaborative Breakthrough

The collaborative efforts of BSc. students at Department of Computer Science, Copenhagen University in Denmark and team members Jean-Marie Niyodusenga, Rasmus Strømsted, and Morten Marquard have led to a groundbreaking breakthrough in the field of declarative process discovery. The work introduces an extensible Natural Language Processing architecture, powered by BERT language models to enable declarative process discovery using natural language texts.


Revolutionizing Declarative Process Discovery

Declarative process discovery plays a pivotal role in enhancing efficiency, compliance, and automation within organizations. Traditionally, this task has been labor-intensive and time-consuming, involving manual analysis of business documents and guidelines. However, the team’s research endeavors have led to the development of an innovative solution that harnesses the power of natural language processing and state-of-the-art BERT language models.


The Extensible Natural Language Processing Architecture

The research work centers around an extensible nlproc (natural language processing) architecture, designed specifically for declarative process discovery. By utilizing BERT language models, they were able to extract valuable insights from natural language texts, such as laws, legal documents, and business guidelines. This advanced architecture empowers organizations to automate the process discovery phase, saving significant time and resources.


State-of-the-Art Benchmarks

The initial benchmarks have demonstrated exceptional performance, positioning the solution as state-of-the-art in the field. Over the test datasets, the system achieved an impressive precision/accuracy rate of 71%.  This achievement is a testament to the efficacy and robustness of this approach. It should be noted that these benchmarks represent the initial results, and it will be continued to refine the models through ongoing training with additional data, including laws and other legal documents.


Future Implications 

The extensible nlproc architecture developed holds immense potential for various industries, ranging from finance and healthcare to manufacturing and government sectors. By automating the process discovery phase, organizations can streamline their operations, ensure compliance with regulations, and achieve unprecedented levels of efficiency.

Moving forward, the team will continue to train the models with more data, including a comprehensive collection of laws and legal documents. This will further enhance the accuracy and reliability of the system, making it an invaluable tool for organisations seeking to optimise their processes.



Get ready for a game-changing revolution in organizational efficiency! These state-of-the-art models are set to transform the way businesses automate and streamline processes. Embracing unprecedented benchmarks and constant improvements, we’re excited to shape a future where efficiency and compliance seamlessly intertwine. Get ready to witness the fusion of innovation and effectiveness!