Robust Requirements Analysis in Complex Systems through Machine Learning - Trustworthy Eternal Systems via Evolving, Software Data and Knowledge - page 50

the analysts in (1) accessing the extracted information as well as (2) in providing
feedback to the system in form of acceptance or rejection of some of its decisions.
4 Semantic Technologies in a Real Application Scenario
New Generation Naval Combat Systems are very complex systems based on a sw
component able to manage all the Combat System Equipment (CSE) in different
mission scenarios: the Combat Management System (CMS) as in [15]. The main
objective of the CMS is to enable the Command Team to manage the ships CSE
to conduct the missions in the scenarios. The CMS is mainly composed of a real
time component (C2S) which provides the Combat System with facilities for
the management of short term activities (Conduct of action); and a Command
Support System (CSS), which provides the Combat System with facilities for the
management of medium and long term activities in the conduct of operational
tasks. From the functional point of view the C2S is decomposed into application
segments that allow the system to perform the following functions:
1. global tactical picture compilation,
2. warfare missions conduction in different domains (Air, Surface, Submarine,
Land) at platform and force level,
3. Tactical Data Link exchange data functions.
Each functional requirement of CMS is allocated to Computer Software Con-
figuration Items (CSCI). CSCIs communicate exchanging data over the ship
network through a common application layer. The communication principles are
different according to the relationships among the components that communi-
cate each other. Independently of the communication model the strategy is that
each sw component shares data with the other system components to enable
them, i.e. allow them to carry out their own functionalities. A key aspect for
managing the overall CMS complexity is the design and description of CSCI
interactions in terms of data each component has to publish for the benefit of
the users. A system like the CMS has a large number of users, a large number of
connections to CSE and heavy requirement on the processing applications that
must be executed in real-time. Further constraints are given by the demanding
performances, security requirements and by the incorporation of Off-The-Shelf
software. It is clear how this class of systems needs clear requirement descrip-
tion and management throughout its entire system life cycle. The introduction
of Semantic Technologies such as IE (as described in previous sections) in the
Requirement Analysis process of naval CMS is bringing significant benefits in
different phases of project life cycle. In particular, the application of Machine
Learning techniques in the initial phase of the project has allowed filling the gap
between the contractual technical specifications and system design description.
Thanks to the machine learning method, the tool illustrated in this paper auto-
mates the process of recognition of several inferences about system components
directly from the texts that characterize them, i.e. the requirements.