Starmind enables users to focus on their core expertise. The application therefore autonomously learns as much as possible about corporate content, users and informal know-how networks with little initialization or customization. This results in a corporate brain with the ability to solve virtually any posed business-related problem.


Predictive Analytics

Starmind is an expertise prediction engine that tells you with up to 98% accuracy who is the right person to solve your current problem within your company.

Neural Know-how Network

The basic source for the prediction model is Starmind’s neural know-how network: a representation of how topics are related to each other within a company. Based on Hebbian Learning rules, topics that often occur together grow together and build clusters. On the other hand, the more two topics occur separately, the less they are connected. Just like brain cells.

Observing Interactions

To find out areas of interest and expertise of its users, Starmind observes how people are interacting with the neural know-how network. Posing a question for example is a good indication for the author being interested in the topics involved. Solving a question on the other hand is a signal for expertise. Getting recommended as an expert by someone else, receiving good ratings, commenting, searching or just even reading a solution - everything leaves traces in the system and builds up a personal know-how profile.

Retrieval Induced Forgetting

Inspired by latest brain research, Starmind developed a retrieval induced forgetting algorithm. That’s why both, the neural know-how network as well as the personal know-how profile keep updated automatically. That’s why we are also called the real-time know-how network.

Multidimensional Routing

Just to know “who knows what” is not enough to have a high resolution rate. Starmind introduced several other decision layers to the routing process to optimize both: Time to solution and the overall resolution rate.

Availability Patterns

An important dimension when it comes to routing a question to the potential best expert is the availability of the targeted person: It might not be the best idea to send a request at 8 p.m. from New York to Berlin where everyone is asleep. That’s why Starmind also takes availability patterns into account for effective routing.

Load Balancing

Another important dimension for having great response rates is to balance the load on experts fairly. Sometimes it’s better to send a request to the second or third best person in the list to not overload a single person. Therefore, Starmind developed a personalized load balancing mechanism.

Fast Initialization

In order to accelerate learning at the beginning of a new know-how network, Starmind developed several algorithms and processes to pre-populate a new environment with existing content and helping users to make implicit know-how highways explicit by recommending colleagues as solvers for a certain question.


Multi-language Support

The Starmind user interface is available in English, German and French. The algorithm identifies the language of each question and considers users’ language skills when searching for experts.