can be done for firm entries. This methodology of pat-
tern recognition within networks could also be applied
in distinguishing different types of multiple job hold-
ers, allowing more accurate and detailed labour statis-
tics from administrative data. Thus the ability of se-
mantic web technology to analyse the network struc-
ture of data allows NSOs to streamline business prac-
tices.
4. Conclusion
NSOs face both a challenging and exciting future in
the big data and data integration spheres. They are ex-
pected to do more with less to meet the needs of in-
creasingly engaged and savvy users in a more com-
petitive data marketplace. These challenges can be met
by embracing technological advances to build the ca-
pacity to integrate big data and existing collections, to
enhance and expand upon traditional statistical prod-
ucts and streamline business practices. The ABS is
contributing to this research by developing a proto-
type GLIDE using semantic web technology, which
currently includes as a test case a prototype seman-
tic LEED. Semantic web technology supports integrat-
ing multiple datasets (including unstructured big data),
with flexible and adaptive structures, machine reason-
ing and inference, shared understanding of the data’s
meaning, reusable standards, easy multidimensional
data exploration and network analysis. By utilising this
type of technology NSOs can build capability to take
advantage of administrative and other big data sources.
The practical applications of data integration and
semantic web technology using the prototype GLIDE
have been described in two cases – longitudinal ex-
ploration and analysis of unit level data for firm
level performance, and network analysis and pattern
recognition for identifying true firm births and deaths.
These are just two examples of the potential in this
area. Other potential applications of these methods for
NSOs include:
–
Flexible and adaptive business registers
– Busi-
ness entities are both inherently complex – with
a variety of different legally-permissible organ-
isational structures – and essentially dynamic
– with changing characteristics, ownership, and
economic purposes. Semantic web approaches
provide the flexibility and conceptual rigour to
consistently capture the real world structure of
businesses and facilitate linking to a wide variety
of datasets, across the institutional level (enter-
prise group, legal entity) and production level (ac-
tivity unit, individual establishment). Developing
a prototype semantic statistical business register
is an area of current research for the ABS.
–
Enhanced automated classifications and coders
–
NSOs need to be able to transform free text fields
(e.g. occupations) into their classifications; but
current classification models can only link con-
cepts in a hierarchical structure and do not al-
low interconnections between related concepts
(e.g. chef and cook are similar in meaning, but
in different branches of the occupation classifica-
tion tree) – semantically linking similar concepts
