BabelNet and Friends: A manifesto for multilingual semantic processing

Abstract. Semantic processing is one of the most compelling and ambitious objectives in today’s Natural Language Processing. Being able to process and understand text at the machine level can potentially enable powerful applications like semantically-aware statistical machine translation and semantic information retrieval, thereby having the potential to change the lives of everyday computer users. In this paper I present a manifesto for the multilingual semantic processing of text. I illustrate the research vision that is pursued in my research group at the Sapienza University of Rome and describe the most recent results obtained. In the last part of the paper I outline a likely future for multilingual semantic processing focusing on the current directions and successes and highlighting on the major obstacles that make this task so hard