Machine Learning for Emergent Middleware - Trustworthy Eternal Systems via Evolving, Software Data and Knowledge - page 16

Machine Learning for Emergent Middleware

Amel Bennaceur 1 ,Valerie Issarny 1 ,DanielSykes 1 ,
Falk Howar 2 ,MalteIsberner 2 , Bernhard Steffen 2 ,
Richard Johansson 3 , and Alessandro Moschitti 3

1 Inria, Paris-Rocquencourt, France
2 Technical University of Dortmund, Germany
3 University of Trento, Italy

Abstract. Highly dynamic and heterogeneous distributed systems
are challenging today's middleware technologies. Existing middleware
paradigms are unable to deliver on their most central promise, which
is offering interoperability. In this paper, we argue for the need to dy-
namically synthesise distributed system infrastructures according to the
current operating environment, thereby generating “Emergent Middle-
ware” to mediate interactions among heterogeneous networked systems
that interact in an ad hoc way. The paper outlines the overall archi-
tecture of Enablers underlying Emergent Middleware, and in particular
focuses on the key role of learning in supporting such a process, spanning
statistical learning to infer the semantics of networked system functions
and automata learning to extract the related behaviours of networked
systems.

Keywords: Machine learning, Natural language processing, Automata
learning, Interoperability, Automated Mediation.

1 Introduction

Interoperability is a fundamental property in distributed systems, referring to
the ability for two or more systems, potentially developed by different manufac-
turers, to work together. Interoperability has always been a challenging problem
in distributed systems, and one that has been tackled in the past through a com-
bination of middleware technologies and associated bridging solutions. However,
the scope and level of ambition of distributed systems continue to expand and
we now see a significant rise in complexity in the services and applications that
we seek to support.
Extreme distributed systems challenge the middleware paradigm that needs
to face on-the-fly connection of highly heterogeneous systems that have been
developed and deployed independently of each other. In previous work, we have
introduced the concept of Emergent Middleware to tackle the extreme levels of
heterogeneity and dynamism foreseen for tomorrow's distributed systems [13,4].
Emergent Middleware is an approach whereby the necessary middleware to
achieve interoperability is not a static entity but rather is generated dynamically
as required by the current context. This provides a very different perspective on