People ask me: why has this never been done before?
Only recently started to become feasible the construction of semantic models based on large-scale data (what we all call Big Data), and this due to the cost reduction of hardware in the form of accessible GPUs.
Other parameters: the achieved maturity of these models and techniques that use deep learning for natural language processing and what we at IntelliSem are best at: distributional semantics.
To learn more follow us on Twitter: or write to us: Dr André Freitas.