Introducing arxiv-sanity

Key Takeaways

we all know and love archive it is a massive pre-print repository with a lot of wonderful papers unfortunately these papers are not very easy to skim search or sort new papers are also added every single day but unfortunately all you get is this huge featureless list it's easy to miss interesting papers that might be relevant to your research introducing archive sanity the function of this website is as you might imagine to preserve your sanity when dealing with archive as one of its features it offers the same view of recent new archive papers but all papers are now laid out in a nice way with a thumbnail preview of the paper for a quick skim and Abstract Below in green unlike archive there are a lot of nifty additional features included for example we can decide that this paper on human posst estimation is interesting and that we may want to see more papers like it for that we can come here to the right and click on Rank by tfid have similarity to this and archive sanity will rerank all papers in archive based on their similarity to this paper so in particular here we see all the papers in archive that are on human POS estimation this is based on Byram tfidf encoding of the full text of papers and works very well we can also search papers for example we can see what Jeff Hinton is up to by searching his name and here we see all of his papers recently posted on archive you'll notice on the top right that I am logged in with my user account this allows me to save any papers I like to my personal library for example if I like this human plus estimation paper I can click click on this diset to add this paper to my library if I go to my library here on this tab you can see it appear here on the top along with it there are many other papers I've already added as interesting to my work so these are all the papers currently in my library and there are many I can also decide that I'm actually not that Keen on human post estimation and remove this paper from my library by clicking on it again the library is not only useful as a way of keeping track of interesting paper archive sanity will take the papers in your library and use them to recommend related work internally all of these papers in your library are positives and all the papers outside of it are negative and archive San will train you a personalized svm based on Byram TF Ida features of papers we can then go to the recommended tab to see our results here we can see all the papers posted over the last week that archive sanity thinks would be most interesting to US based in our library we can also filter papers to different timer for example if we click on all papers posted over the last year or over all time we could find related work that we may have missed and we should really know about finally we have the top recent tab which shows the popular papers across all users and archive sanity for example these are the papers over the last week that many users have added to their libraries and these papers can of course similarly be filtered based on time ranges therefore even if you don't have an account yourself you can still see what papers other archive sanity users find interesting and that's it for a quick tour of archive sanity currently the project website is only hosting papers from machine learning parts of archive such as CS cvcl LG NE and stml this is mostly because these are the areas that I personally work in however the entire source code for archive sanity is on GitHub so anyone can easily create their own Fork of archive sanity for other parts of archive as well okay so I hope you enjoyed the tour and that you'll find the website useful byebye