Adaptive Evolutionary Navigation in Knowledge Representation Systems
Taras Filatov graduated from Kharkiv National University of Radioelectronics (KNURE), Ukraine and passed his Master of Philosophy at the Wessex Institute of Technology (WIT). Taras continued his research at WIT and has now successfully passed his PhD viva with the thesis ‘Adaptive Evolutionary Navigation in Knowledge Representation Systems’. His external examiner was Dr Gary Wills from the University of Southampton and the internal was Dr Andrés Peratta.
The main objective of this work was to develop architecture and carry out the implementation of a fully automated knowledge representation system, capable of rendering textual documents from various unstructured sources, online or offline, into a single web knowledge base where navigation was automatically provided by the system. This navigation was carried out using the principles of natural browsing.
The system architecture was developed in two main parts: data preparation and interactive knowledge representation. Novel approaches were introduced in both the data compression and representation parts to overcome difficulties that had prevented the construction of comprehensive full cycle systems in the past.
The data preparation employed modern information retrieval techniques, such as search engine elements, and represented the documents corpus as vector space. However, a distinguishing part was the application of the dimensionality reduction technique to reduce the vector space while keeping the relevance calculation features by reducing the dimensionality on initial vector space to intrinsic dimensionality value.
The interactive knowledge representation implemented the innovative concept of a ‘live’ navigational panel powered with evolutionary algorithms. This was achieved through the system of rules divided into two algorithms serving both dormant and active cycles of user-system interaction.
As a result of his research, both examiners recommended the award of Doctor of Philosophy.