The Way Forward - Discrimination and Privacy in the Information Society - page 334

previously unknown and sometimes completely unexpected patterns and relations.
Specifying purposes before the start of any data mining exercise is basically
impossible, because it is not yet known what will be discovered.
Note that it may be argued that the purpose in such cases could be specified as
something like: “we plan to use your data for data mining purposes”, but
obviously it is impossible for data subjects to know what the outcome of such an
exercise will be and may raise difficulties for them assessing the positive and
negative effects of disclosing information. Consent in such cases is unlikely to be
informed . Furthermore, the knowledge discovered with data mining is likely to
be used for further decision-making. For data subjects it is even more difficult to
overview how data mining results will be used and may affect them at a stage
where these data mining results are not yet known.
19.1.4 Focus on Transparency and Accountability
In the previous subsections, we have shown that a priori limiting measures
(particularly access controls, anonymity and purpose specification) are failing
solutions for privacy and discrimination issues. In this subsection, instead, we
argue that a focus on (a posteriori) accountability and transparency may be more
useful. 17 Instead of limiting access to data, which (as shown above) is increasingly
hard to enforce in a world of automated and interlinked databases and information
networks, we must stress the question as to how data can and may be used.
Whereas a more traditional approach focuses on the concepts of access controls,
anonymity, 'need to know' and 'select before you collect', our approach focuses
on other legal concepts, such as those used in tort law: accountability, liability,
redress, etc. Another option is that of considering data mining as (legal or illegal)
search. 18 The use of such concepts is familiar to legal experts and, as such, may
help them understand what data mining and profiling are about.
From a technological perspective, transparency and accountability can be
improved in several of the technological measures suggested in this topic. For
instance, the architecture of data mining technologies can be adjusted ('solutions
in code') 19 to create a value-sensitive design, that incorporates legal, ethical and
social aspects in the early stages of development of these technologies. This is
exactly what privacy preserving data mining techniques strive to achieve. 20 These
may aim at protecting identity disclosure or attribute disclosure, but also
at prevention or protection of the inferred data mining results. Similarly,
discrimination-free data mining techniques have been developed, by integrating
legal and ethical concerns and interests in data mining algorithms. 21 Such
17 Weitzner, D.J., Abelson, H. et al. (2006).
18 See Chapter 18.
19 See Chapters 11-14. For more information, see also:
http://wwwis.win.tue.nl/~tcalders/dadm/doku.php
20 See Chapter 11.
21 See Chapters 12-14 and Calders, T., and Verwer, S. (2010).