Nested conceptual graphs for information fusion traceability
InSyTo is a toolbox of algorithms for information fusion and query relying on the conceptual graphs formalism and subgraph isomorphism search. InSyTo was used in order to develop many applications in different domains.
Although the framework was used in several application domain and well received by end-users, they highlighted an urgent need for traceability within the information fusion process. We propose here an improvement of the toolbox, in order to embed traceability feature inside the fusion algorithm.
The underlying conceptual graph representation of the information was extended from basic conceptual graph to Nested Typed Graphs. A lineage nested graph is added to each concept of the initial information graph, that contains it's processing history through the several processing steps. The lineage graph contains the information concerning the initial sources of each elementary information item (concept), as well as the fusion operations that were applied on them. The main advantage of this new development is the capacity of having a trustworthy framework aware of the current observed situation, as well as the interpretations that were used to build this situation from elementary observations coming from different sources.
In this paper, after presenting the context of our work, we recall of the InSyTo toolbox approach and functionalities. We then define the new information representation and operations that we proposed for a matter of traceability handling.