Crossing the Cognitive Chasm in Language Technologies
October 10th 2017 13:35 - 13:55
Sectors like Banking and Insurance are facing increasing challenges derived from the need to manage ever-growing amounts of data, multiple languages, channels and modalities. In this context, language technologies and particularly Expert System’s Cogito play a key role, enabling more natural human-machine interaction and increasing productivity by automatically processing large volumes of unstructured information. Successfully deployed in many of the world’s largest banking and insurance organizations, Cogito understands text in a similar way to how human experts read, based on a structured knowledge graph with millions of explicitly represented concepts and relations. In parallel, statistical Machine Learning has proven to be a powerful, cheap and fast way to address similar goals, based on learning mathematical functions that predict how data are correlated. However, ML is limited in several ways including the lack of a true understanding of real-world semantics and the capability to reason in ways understandable by humans. In this talk, we show how we are bringing together the best of both worlds at Expert System, supporting multilingual text understanding that is both deep and broad as well as flexible and tractable.