Yafang Wang

About

Mrs. Yafang Wang currently performs her Master Thesis (Specialization: Security and Trust) within a cooperation project between UL and the University of Shadong, China.

She is currently a PhD student at the Max Planck Institute in Saarbruecken: http://www.mpi-inf.mpg.de/~ywang/ (2009)

Project

• MINT - Associative Mindmaps for Trust Modeling

Abstract

Assume that Person 1 talks to Person 2 (and vice versa) in a way that Person 1 understands Person 2 (and vice versa). Person 1 may then forget incoming content signals but may also extract some aspects out of it, summarizes it and keep these aspects in mind. This is a sub jective process, since the extraction and management of aspects depend on personal things like interests, knowledge level, etc. Furthermore, the process occurs in real-time since the conversation with Person 2 is lost once words and sentences are pronounced. Both trust - depending on the associative mind-map of the other, respectively. Assume further on that both Person 1 and Person 2 store aspects of the other’s textual contribution in mind. Then Person 1 knows something about Person 2 and has a certain estimation of what Person 2 thinks; on the other hand, Person 2 knows something about Person 1 and has a certain estimation of what Person 1 thinks. Based on this, each of them may decide of either to stop or to continue the conversation. Moreover, both may decide on to what level they trust each other.

The Master Thesis will be concerned with the part of trust modeling while taking the ANIMA system as the underlying representation engine. ANIMA stands for the creation of an artificial mind-map, where a textual data stream is read and represented in an associative dynamic network:

• Each incoming stream is decomposed to its items, for example a text stream may be decomposed to words.
• The collection of items form a mini-network, consisting of item cells that are fully connected, sharing a connection weight between the items.
• Once te mini-network is established, it is sent to the mind-map: if an item of the mini-network exists in the mind-map, then it is merged. The activation status of the item is increased.
• If two adjacent items of the mind-map is activated by the mini-network, the connection weight between them is increased (Hebbian Learning).
• If two adjacent items remain unactivated, their connection weight is decreased until the association is forgotten.

The ANIMA system is a real-time engine that always represents a current snapshot of existing associations - depending on the incoming data. It might be large or small, consist of strong or weak associations, and produce temporal constructs (skeletons).

The aim of MINT is to provide a prototype that demonstrates the use of associative mind-maps for the modeling of trust.

"Yafang Wang" is mentioned on: Members