It means Site1.com/page1.html links to SiteB.com/download/file1.pdf which redirects to file1.pdf on mysite.com. It's a little odd, coming from irrelevant sites, so it could be web spam, but even then I wouldn't worry about it. My backlink reports in GWT/SC are full of spammy sites, but I have no control over who links to me so I just ignore them.

No redirection to my site. That's the strange thing. I am just guessing Google identify pdf on SiteB as duplicate to my site. But there is nothing to verify this.

Oh, I see. It's possible Google recognizes the file on mysite.com as the original (canonical) version, and the others as duplicates, and then attributes the link value to the original document rather than (or on top of) the file on SiteB.com. Interesting! I can find other reports of this, but no word from Google. I do wonder if this is related to the PDF spam I'd been seeing a lot of earlier this year. And also if this could potentially be similar for other web content (zip, exe, doc, xls, images, perhaps even web text). Attribution is a bit of an issue online; it's certainly not always given, and finding the source for any given piece of (redistributed) content may be very important to Google.

Not aware about PDF spam but if Google does this on a pdf file and without rel="canonical" tag, then it has to be a smart thing from Google. I was worried because the links are from low quality sites (Site1, Site2,..) to the duplicate content site (SiteB) and showing as my backlinks. So, indirectly I am getting low quality backlinks. Sounds a bizarre logic though!