Time Limit: 2 Seconds
Memory Limit: 65536 KB
It's easy to tell if two words are identical - just check the letters. But
how do you tell if two words are almost identical? And how close is "almost"?
There are lots of techniques for approximate word matching. One is to determine
the best substring match, which is the number of common letters when the words
are compared letter-byletter.
The key to this approach is that the words can overlap in any way. For example,
consider the words CAPILLARY and MARSUPIAL. One way to compare them is to overlay
There is only one common letter (A). Better is the following overlay:
with two common letters (A and R), but the best is:
Which has three common letters (P, I and L).
The approximation measure appx(word1, word2) for two words is given by:
common letters * 2
length(word1) + length(word2)
Thus, for this example, appx(CAPILLARY, MARSUPIAL) = 6 / (9 + 9) = 1/3. Obviously,
for any word W appx(W, W) = 1, which is a nice property, while words with no
common letters have an appx value of 0.
The input for your program will be a series of words, two per line, until the
end-of-file flag of -1.
Using the above technique, you are to calculate appx() for the pair of words
on the line and print the result. For example:
The words will all be uppercase.
Print the value for appx() for each pair as a reduced fraction, like this:
appx(CAR,CART) = 6/7
appx(TURKEY,CHICKEN) = 4/13
appx(MONEY,POVERTY) = 1/3
appx(ROUGH,PESKY) = 0
appx(A,A) = 1
Fractions reducing to zero or one should have no denominator.
Source: Pacific Northwest 1999