in these paragraphs, we have a clean knowledge base or
ontology. But what if our ontology has to be queried, merged
or linked with another one? Answer to this question is
ontology alignment (a.k.a. ontology matching) and it has to
be done in agreement to Big Data requirements (a recent and
relevant review of schema alignment with structured data in
Big Data era is presented in [9]). A deep and recent review of
ontology matching is presented in [90]. Aspects of ontology
matching which present an interest for us are mentioned
there in terms of challenge. Some of those aspects like the
use of external resources have a direct impact on ontology
matching in the context of Big Data. It is the case of (i) matcher
selection, combination and tuning and (ii) user involvement.
Challenge (i) is relevant to us because matcher uses different
techniques and to combine/tune them can improve results.
Moreover, the improvements of these techniques can focus on
specific aspects (volume, uncertainty) of ontologies. But these
combinations can have a negative impact on processing time.
The same remark can be done in the second “challenge” since
the user can resolve matching errors but it is difficult to rely
on users in large ontologies alignment.
In addition, Shvaiko and Euzenat [90] mention the lack of
evaluation of scalability as a challenge. Likewise, all these
remarks could be made ours, after we have presented main
aspects of Big Data semantic management. Surely, all the
techniques and tools aforementioned can be improved by
various parameters or heuristics, but in Big Data era, a
significant place must be made to optimization. Tools must
handle exabytes of data, streaming data, fast changing ones,
very informal data, etc.