The first integer is the number of samples n. Following, we have n pairs that
describe the message ID and topic. The topic information is not used for
any specific propose. After that, the similarities between each object are
described. We assume similarity 1.0 between an object and itself.

Number of samples
ID1 topic
ID2 topic
...
...
...
ID1 ID2 similarity
ID1 ID3 similarity
ID1 ID4 similarity
...
ID2 ID3 similarity
...
