Serge Demeyer | Publications | E-mail Feedback
Last updated on Thursday, November 16, 2023
@inproceedings{LeyvaPernia2018FLORITECH,
author = {Diana Leyva Pernia and Serge Demeyer and
Olivier Schalm and Willemnien Anaf},
booktitle = {Proceedings {HERI-TECH 2018} (IOP Conference Series:
Materials Science and Engineering)},
pages = {012045},
title = {A data mining approach for indoor air assessment, an
alternative tool for cultural heritage conservation},
volume = {364 -- 1},
year = {2018},
abstract = {The exposure of cultural heritage to the environment
has a significant impact on its degradation process
and degradation rate. Consequently, managing the
indoor air quality is vital to minimize further
damage to historical artefacts and works of art.
Despite its potential impact, the traditional
assessment of the indoor air quality still represents
a challenge for most collection guardians. This
approach typically relays on the comparison of
measured environmental parameters and corresponding
acceptable values. However, determining the
acceptable values and relative importance of the
different environmental parameters turns out to be
quite complex since it depends on the material types
present in the collection and their preservation
state. Furthermore, the significant amount of data
generated during the measurements hampers the
application of traditional methods of analysis.
Considering all these, we propose the use of data
mining as an alternative method for the indoor air
quality assessment in cultural heritage studies. Data
mining can provide knowledge from vast volumes of
heterogeneous data, through high-speed processing,
detection, and analysis. Here we present its
application to identify dynamics and patterns
affecting the indoor air quality in a realistic case.
Using data from a measuring campaign held at a late
Gothic church in Belgium, we show that inappropriate
periods can be identified without using standards. In
addition, different types of periods can be
identified by studying the relation between multiple
parameters. For that we use the k-means clustering
method, interpreting the results with both visual and
statistical tools.},
annote = {internationalconference},
doi = {10.1088/1757-899X/364/1/012045},
}