Review Comment:
Review of Forest Logging: A Trace-Based Analysis of
Large Rule-Based Computations (paper by T Swift)
The revision has addressed all my concerns, hence I recommend
acceptance. Still, the paper was not carefully enough proofreaded and
minor corrections (see below) are needed.
DETAILS
p1-
Much of the literature on knowledge representation
and reasoning (KRR) has concerned
--> better?
Much of the literature on knowledge representation
and reasoning (KRR) is concerned with
For description
logics an example
->
For description
logics, an example
p2
overview: what about section 5?
As SLG and its exten-
sions have been presented in the literature our review
is largely an informal overview;
-->
As SLG and its exten-
sions have been presented in the literature, our review
is largely an informal overview;
p4 def 2.1
and V
is the underlying set of E.
--> better?
and V
is the underlying set of nodes of E.
p5
Fig 2: last literla of last clause
not p(Y) -->
not p(Z)
The
final DELAYING operation is applied to the literal not
p(a) in node 5 is delayed, so that the new selected lit-
eral for its child, node 13, is not p(b).
-->
The
final DELAYING operation is applied to the literal not
p(a) in node 5; in its child, node 15, not p(b) is the new selected
literal.
p6
the conditional answer of node 12,
-->
the conditional answer of node 14,
p9
def 3.1: S and S1 should be identical,
i.e., "(cmp(S , Sscc , c'' )"
def 3.2
(c) If n is an answer or failure node whose near-
est ancestor
the "nearest ancestor" is nothing else than the parent?
d) Otherwise, H(n) is closest ancestor
-->
d) Otherwise, H(n) is the closest ancestor
2. If there is an edge between nodes n1 and n2 in T
there is an edge between H(n1 ) and H(n2 ).
?You mean?
2. If there is an edge between nodes n1 and n2 in T
there is either an edge between H(n1 ) and H(n2 ) or H(n1 )=H(n2 )
p10
by condition 1(c) f A is in H(T ),
-->
by condition 1(c) if A is in H(T ),
an SLG tree for the can be constructed
-->
an SLG tree for the goal can be constructed
char-
acterize the informational maintained in a forest log,
H(T ) there are practical motivations for constructing
H(T ).
-->
char-
acterize the information maintained in a forest log,
H(T ) there are as well practical motivations for constructing
H(T ).
p12
footnote 13:
This statement is true in local evaluation but another common
scheduling strategy called batched evaluation.
-->?
This statement is not only true in local evaluation but also for another common
scheduling strategy called batched evaluation.
The overview also provides the distributions of
tabled subgoals across SCCs of the formed by the
SDGs of the various forests in the evaluation.
-->
The overview also provides the distributions of
tabled subgoals across SCCs formed by the
SDGs of the various forests in the evaluation.
p13
two-valued a ell-founded
-->
two-valued well-founded
p14
which omits facts
produced by the ANSWER RESOLUTION operation the
new answer event.
--> ???
which omits facts
produced by the ANSWER RESOLUTION operation immediately after a
new answer event.
p15
check_variant/2 begins
--> better avoid a lower case to start a sentence
The check_variant/2 predicate begins
p16
The maximal cost of traversing each logged
fact this can be treated as a constant function
-->
The maximal cost of traversing each logged
fact can be treated as a constant function
can reduce the
the logging overhead,
-->
can reduce the
logging overhead,
p17
and under certain conditions has the
information available
-->
and, under certain conditions, has the
information available
ref 7
clp(fd)
-->?
CLP(FD)
p20
a descendent of N such and any intermediate
-->
a descendent of N and any intermediate
p21
as Node’s children..
^^ 2 dots at end of sentence
| 
