Re:List of Computer Science courses with video lectures
by lnternet on Tuesday November 22, 2016 @10:03AM
(#13014561)
Previous discussion: https://news.ycombinator.com/item?id=13000158
[ Parent
| on HN ]
Re:List of Computer Science courses with video lectures
by turingbook on Tuesday November 22, 2016 @10:40AM
(#13014796)
Sorry for the replication. It seems that HN should have a better duplicates removal algorithm.
Re:List of Computer Science courses with video lectures
by pkd on Tuesday November 22, 2016 @11:36AM
(#13015245)
You posted the link to the README, which is different from the link to the repo itself. Since these two aren't really duplicates, I don't think it's a fault of HN's algorithm.
Re:List of Computer Science courses with video lectures
by xapata on Tuesday November 22, 2016 @12:32PM
(#13015688)
Since the root page of a repo renders the README, the two URLs are not very different. A sophisticated duplication detection algorithm could recognize that.
Re:List of Computer Science courses with video lectures
by wuliwong on Tuesday November 22, 2016 @03:25PM
(#13017340)
The algorithm would have to examine the HTML returned by the URLs to determine they are duplicate pages. There is no way to determine they are duplicate pages by just analyzing the URLs. It is not impossible to have two different posts to HN that are legitimately different content that have the same domain and even partially matching paths. In order to work in a dependable way, the algorithm would have to examine the HTML returned by the URLs to determine they are duplicate pages. In this case, that wouldn't even work because they aren't actually duplicate pages. In order to mark these two URLs as duplicate I think you might need to use some machine learning and then it would only yield some level of confidence that these were indeed duplicate pages as a result.
Re:List of Computer Science courses with video lectures
by xapata on Tuesday November 22, 2016 @11:57PM
(#13020166)
Or they could just special-case GitHub repos and their readme files. But what is duplicate detection if not machine learning? One can't be sure the same URL is the same document if submitted at different times.