Chapter 4, Metalinguistic Abstraction

Exercise 4.60


Duplication can be explained by looking at the rule:

1
2
3
4
(rule (lives-near ?person-1 ?person-2)
      (and (address ?person-1 (?town . ?rest-1))
           (address ?person-2 (?town . ?rest-2))
           (not (same ?person-1 ?person-2))))

We lookup for ?person-1 and ?person-2 such that they have same ?town. This can happen even when ?person1 and ?person2 are same so we removed that possibility with (not (same ?person1 ?person2)).

But we still have not removed the possibility that we can have two combinations (x y) and (y x), where x and y are two person sharing the same address and:

  • ?person1 is instantiated to x and ?person2 instantiated to y.
  • ?person1 is instantiated to y and ?person2 instantiated to x.

How to avoid this?

Similar to the check of (not (same ..)), we can instead use something like (less-than ?person1 ?person2) such that `less-than** is defined for symbols.

Not doing the implementation as it is not asked and it’s not interesting too :)



Note that if we implement the above change than, lives-near won’t work for cases like (lives-near ?x (Fect Cy D)) because Hacker Alyssa P is greater than Fect Cy D!

This can be fixed by implementing a new lives-near that calls the above change version say lives-near-ordered like this:

1
2
(rule (lives-near ?x ?y)
 (or (lives-near-ordered ?x ?y) (lives-near-ordered ?y ?x)))

Well these are speculations because unlike last section(non-deterministic computing), I have not read the implementation details of query evaluator before using it!



So far, this approach of first understanding the use and then the implementation is also having its own advantages. For example: I changed a rule in one of the previous exercise and without restarting/cleanup the code/database the rule seems not to work.

This made me to speculate that evaluator does some preprocessing and save those results in database. These results don’t get removed even when the rule is modified.

Else, I might had different answers if i had some idea of the implementation! This gives me slightly different perspective.