Given the complex nature of the public sector with several distributed governmental institutions and manifold semantic differences of interpretation, the achievement of interoperability and integration is a key challenge for a comprehensive and effective Electronic Government (e-Government). Promising technologies that could be used to tackle this issue are: the powerful concept of ontology and the advanced Artificial Intelligence (AI) systems. Ontologies contribute to a common understanding of heterogeneous resources, while AI techniques make process integration dynamic and automated. However, up until now the use of AI along with ontologies has been fairly limited in e-Government. There is still, then, untapped potential in this field which worth to be exploited. In this paper, we present a dynamic approach for semantically integrating e-Government Web services based on AI techniques. The overall objective of our approach is to improve the citizen centric e-Government vision by providing a conceptual framework for automatically discovering, composing and optimizing e-Government services. Within the proposed approach, special emphasis is put on personalization aspects and evaluation criteria for e-Government platform.