This is an automatically produced distributional thesaurus, which finds words that tend to occur in similar contexts as the target word. It is not a manually constructed thesaurus of synonyms. It was produced by Lexical Computing Ltd on the basis of the MLRS corpus.
The file contains lines with the lemma_freq and lemma in first two columns and the rest of the line is a list of up to 20 best matches pairs (lemma+score) that have score at least 0.2.