![](bga.png)
3. It does not belong to the requesting insider domain and does not contain
knowledge units that are relevant to his/her domain, but contains knowledge
units that are not sensitive.
To clarify things suppose that there exists an object O that belongs to domain A.
Object O contains knowledge units K
1
, K
2
, K
3
, and K
4
with relevance values as: 0.6,
0.4, 0.8, and 0.2 respectively. These knowledge units are relevant to different
domains, namely; K
1
and K
2
are relevant to domain B while K
3
and K
4
are relevant to
domain C. Domain A threshold value equals to 0.7, while domain B threshold value
equals to 0.65, and domain C threshold value equals to 0.75. Suppose that two
insiders S
1
ϵ dom(B) and S
2
ϵ dom(C) try to access object O. Since object O belongs
to neither domain B nor domain C, then their requests might be denied unless it
contains knowledge units relevant to their domains or they satisfy the sensitivity
check. The following cases can be distinguished:
• Insider S
1
belongs to domain B. Both K
1
and K
2
are relevant to domain B.
Neither K
3
nor K
4
are relevant to domain B (they are relevant to domain C)
• Insider S
2
belongs to domain C. Both K
3
and K
4
are relevant to domain C.
Neither K
1
nor K
2
are relevant to domain C (they are relevant to domain B).
Since knowledge units K
1
and K
2
are relevant to domain B (which insider S
1
belongs
to) then both K
1
and K
2
are accessible by S
1
. However, since K
3
and K
4
are not
relevant to his/her domain then they might/might not be accessible. After applying the
sensitivity check: relevance(K
3
) = 0.8 which is greater than domain C threshold value
= 0.75. That is, K
3
is considered sensitive and hence, K
3
will not be accessible. Also,
relevance(K
4
) = 0.2 which is less than 0.75. Hence, K
4
is accessible because it does
not contain enough sensitive information related to domain C that must not be
revealed to insider S
1
. Thus, our model compares relevance of a knowledge unit with
the threshold value of the domain it is relevant to. The reason is to make sure that this
knowledge unit does not contain enough sensitive information that should be kept
hidden from insiders of other domains. That is, the knowledge an insider can gain
from this knowledge unit is less than the sensitivity threshold of that domain. The
same procedure is applied to requests of S
2
. It is found that knowledge units K
3
and
K
4
are accessible by S
2
because they are relevant to his/her domain. However, for K
1
and K
2
the sensitivity check is applied. Relevance(K
1
) = 0.6 which is less than domain
B threshold value. Also relevance(K
2
) is less than domain B threshold value.
Therefore, both K
1
and K
2
are accessible by S
2
besides K
3
and K
4
. That is, S
2
can
access the entire object O.
From the above example, it can be concluded that insider S
2
can access the whole
object O. However, insider S
1
gets partial access to the object O. In fact, he will get
access to knowledge units K
1
, K
2
, and K
4
. Since K
3
has sensitive information that he
should not access, a filtration process will be initiated which filters the object O out
by removing knowledge unit K
3
from that object. The remaining content of object O
is then presented to insider S
1
.
Based on the above example, there are three cases to consider: objects that
contain knowledge units that are only relevant to the domain of access of the
requesting individual, objects with knowledge units some of which are relevant to the
domain of access and some others are not relevant, and objects with knowledge units
all of which are not relevant to the domain of access of the requesting individual.
108