Free text has been and continues to be for humans the traditional and natural mean of representing and sharing knowledge. However, the knowledge encoded in free text hinders its accessibility and usage, since the retrieval of information from a large corpus is a tedious and time-consuming task for humans and a hard and prone to error task for machines.

Besides the exponential growth of knowledge bases (KB) and the initiatives to connect them (e.g. Linked Data), most of our knowledge is still locked in free text. The task of identifying the most appropriate KB entry for describing a given entity mentioned in text is usually referred as Named Entity Disambiguation (NED), but is also named as entity disambiguation, resolution, mapping, matching, linking or even grounding.

Linked text will enable us to more effectively navigate, retrieve information, find evidence, updates or even discern true from fake information. Effectively linking text to KBs will also enhance the computer’s ability to infer new knowledge. However, all these benefits require in-depth NED solutions that are still not in place.