Starmind

Zurich, February 4, 2010. Today, Starmind is officially announcing the new Starmind Release 2.1. For the first time, details on the internal architecture are published:

The novelty of the Starmind portal 2.1 is based on a unique combination of efficient database algorithms and intelligent neural network architecture, which have been developed at the Department of Informatics at the University of Zurich, Switzerland.

Intelligent artificial neural network to identify know-how hotspots

Marc Vontobel, Starmind Head IT: “Starmind users are virtually connected to subject areas (tags) on the basis of the questions they have solved. As a result, the generated connections between users and subject areas are strengthened or weakened according to the success rate of a question solver. For instance, a user who has repeatedly and successfully solved a question concerning ‘Web 2.0 technology’ has a stronger connection to the subject ‘Web 2.0’ than a user whose solution related to the subject area has been rated negatively by the question poser. In addition, the mechanism of forgetting occurs with every learning process, analogous to the natural process. This process ensures that users who are more active in a subject area are favoured over less active users. When a new question is posed, a very limited number of qualified question solvers are informed, which leads to an efficient solution of the question with minimal communication effort.”

Global competition for outstanding solutions

The larger the Community, the higher the level of selection and therefore, the higher the qualification of the located individual is. The intelligent neural network enables an efficient mapping of experts worldwide, which allows forming groups of experts. The network memorises the created structures (“Know-How Hotspots”) while the allocation of know-how and questions is continuously optimised. As a part of the Starmind Release 2.1, the neural network forms another basis for the systematic expansion of the Starmind Community.