Information Technology Reference
In-Depth Information
Prolog implementation in section 4). During semantic inference, the left-side-
parameter will be instantiated into a set of constant entities. The meaning of other
meta rules in table 2 is easily read based on the introduction of predicates in section
3.1.
There are two kinds of queries need to be answered by the semantics of REAL05.
The predicates in queries are
isMember
and
allowAccess
respectively. The query
containing
isMember
will be raised when an entity tries to activate a role, and the
query containing
allowAccess
will be raised when an entity tries to access resources.
Given a REAL05 rule set P, the computational complexity of P denotes the time
needed to answer a given query Q based on P, i.e. P
⊢
Q or P
⊬
Q. REAL05 is a subset
of Datalog
C
, its computational complexity lies on Datalog
C
. The computational com-
plexity of Datalog
C
is closely connected with constraint domain it contains. The con-
straint domains that can be evaluated with safe Datalog in polynomial data complex-
ity (PTIME) include: (1) equality constraints, order and inequality constraints over
dense linear order domains [15], (2) linearly decomposable domain [13]. It's clear
that REAL05 only contains these two constraint domains, and each rule in REAL05
has finite variables. Therefore we get the following result:
Proposition 1.
Given a set of REAL05 rules P, its computational complexity is
PTIME on size(P), where size(P) = |P|*V, |P| is the number of the rules in P, V is the
upper-bound of the sum of the variables in a rule.
3.3 Samples
A comprehensive example is introduced in this section to illustrate how REAL05 can
be used to express more controllable delegation policies, both for DoA and DoC. The
sample shares the same entity names and delegation paths in fig. 2.
Suppose S is an online digital library and wants to give 20% discount to the stu-
dents of its cooperative universities. But these students must have papers indexed by
S and can only download the discounted papers through the proxy servers of his/her
certifying university. M
1
is a cooperative university of S. S only trusts M
1
and its
branch campus to certify a student of M
1
. M
n
is one of the branch campuses of M
1
. P
1
is a proxy server of M
1
. U is a student of M
1
and studying at M
n
. One of U's papers
has been indexed by S. The above policies can be specified in REAL05 as follows.
(1)
S.canHold
(
dR
(S, discount) ,
pm
(download, 20%)).
(2)
S.canActivate
(M
1
,
dR
(S, co-university)).
(3)
S.canActivate
(?x,
dR
(S, contributor)) ←
S.
author-of-indexed-papers(?x
).
The rules (1, 2, 3) are basic authorization policies defined by S. The predicate “au-
thor-of-indexed-papers” is a user-defined-predicate, which will query database to
answer whether S has indexed papers of an entity. The policies for
n
~
u
in fig. 2 are
specified with following rules.
nop
: The rules (4, 5) specify constrained DoA policies from S to M
1
. The rules
(6, 7) specify constrained DoA policies from M
1
to M
n
. The rules (8, 9, 10, 11) define
the authorization policies of M
1
and M
n
. The rule (10) shows that M
n
makes authori-
zation based on the authority originated from S.
