Giuseppe Castellucci is a Machine Learning, Natural Language Processing engineer and an Artificial Intelligence enthusiast. Since 2016 he's a NLP/ML Engineer in Almawave R&D team, where he's contributing to the companyâs products evolution in managing and leveraging on unstructured data to solve business oriented natural language problems with Machine Learning and Deep Learning. He holds a PhD from the University of Rome, Tor Vergata, where he investigated how to leverage on Machine Learning and Deep Learning both to gain new knowledge and to acquire effective classifiers for opinion mining over social media data.
Speeches di Giuseppe Castellucci
Machine Learning and Knowledge Representation: a synergic approach for Business Oriented Natural Language Processing
Business oriented natural language applications require historically high precision, especially when supporting critical decisions. For this reason, they have been mainly developed with Knowledge Management (KM) and reasoning approaches, as they enable to model precisely the rules of languages and their usage with very little ambiguity at the expense of coverage. In last years, data driven approaches tackled the coverage problem in Natural Language Processing (NLP) by 1) enabling to learn language generalization patterns and 2) tailoring such patterns on different domains. However, data-driven approaches, like Machine Learning (ML) and Deep Learning (DL), works at the expense of precision and control. In Almawave we've developed a complex pipeline of orchestrable NLP modules made of KM, reasoning and ML/DL. This pipeline enables to treat each business case either with KM, ML/DL or both, resulting in a flexible and reusable architecture. Once information is discovered and extracted in texts with these modules, it is modeled and used to enable complex business processes. Here, we present the Almawave architecture and how data-driven approaches help classical KM systems in defining complex decision processes in different application domains.Lingua speech: Italian