Forthcoming beyond 5G (B5G) network systems are expected to introduce a new era in the mobile networks, supporting new and challenging use cases and applications. This new era will provide ubiquitous connection of up to billions of devices that are estimated to generate vast amounts of data, paving the way towards a data-driven network system design. In such network design, Artificial Intelligence (AI) will act as a key enabler with regard to network system automation and optimizations. One of the biggest challenges for AI in such complex and dynamic network environments, is the interoperability in terms of seamless exchange and analysis of information between two or more systems. Technological interoperability requires that the format in which data is generated or stored is standardized and that interconnected systems are aware of the nature of transmitted information and are able to analyze it. This paper proposes a novel, comprehensive Information Model (IM), named as SIM+, that extends all across the network, describing different network entities, attributes, capabilities and operations in an implementation-agnostic and standardized way towards assisting AI-driven optimizations. Seven key abstraction layers have been identified in an attempt to hide infrastructure complexities. The adaptability of the proposed SIM+ is showcased through its application in two state-of-the-art AI-driven solutions.