Automated Research Assistant
Research assistants are already used in various sectors, particularly the legal and finance industries. Typically they are largely manual systems with graphical user interfaces allowing human researchers to enter relationships between typically, people, organisations and places into graph data-bases.
Clearly a system that performed a lot of the reading and graph formation automatically could save an awful lot of the legwork. The SemantEx system is being developed for this purpose and a sample of the results that can be obtained is shown in the plot below. The system has been asked if there are any links between Oleg Deripaska and George Osborne. A number (50+) web pages (mostly Wikipedia pages - it is a good place to start) have been read and the following link has been unearthed.
Note that a lot of relevant information has been structured about this link. Oleg Deripaska and Nathan Rothschild were both involved in the hosting and the system has recorded them as a 'collection', and Peter Mandelson was also hosted. The place and circumstance of the meeting is also recorded.
This information is not derived from the structured element of the Wikipedia pages but from the unstructured text. It can if necessary be combined with structured or semi-structured information from open source or privileged sources. Persons of significant control information from Companies house might be a good example.
