It very much depends on the lab that created the data.
If it's from ChIP-seq data, a WIG file could have the number of reads (possibly with each read extended 3' to be the size of an average fragment in the library). It may or may not be scaled to library size or scaled to other tracks in the same experiment. If it's ChIP-chip, then the numbers could be log2 ratio of IP to input.
Since the numbers you're showing are integers, my guess is that these values are raw read counts that are neither scaled to library size nor normalized by control. But to find the actual answer, you need to look closely at the methods of how the track was created. If that information isn't available, you'll need to contact the authors.