When citing anything, provide arXiv (or doi) link

Question

It's more about culture of site than anything else, but I think we should encourage people to give links to paper they cite. Either arXiv, or by doi. The first one (i.e. arXiv abs page link) should be recommended, as:

• it is open access,
• it contains doi as well.

This post has been migrated from (A51.SE)

Comments

Would linking to the abstract from a journal be acceptable in place of a link to the arXiv? Some times a pre-print is not available, especially for older papers. I know it is not necessarily open access, but most journals that I deal with provide doi's.

This post has been migrated from (A51.SE)

---

@rcollyer: My opinion is such: if both arXivID and DOI are available, then there should be link to arXiv. If there is only one - there is no choice anymore.

This post has been migrated from (A51.SE)

---

I'm all for encouraging, but strongly against insisting. I don't think we really want to do anything to increase the barrier for posting answers or interesting questions.

This post has been migrated from (A51.SE)

Answer 1

Is this really a problem? If I see a paper mentioned, SE provides reputation incentives to edit the post to provide links. This way a new user doesn't have to think about it, and will naturally notice the practice after some time participating.

I prefer DOI and arXiv of course. DOI because it's robust against change and arXiv because it's open. Perhaps some standard way of including both should be used when possible. I suggest including both because the DOI seen in arXiv abstracts is supplied by the authors. Some arXiv documents may be published, yet still lack a DOI on the abstract page.

Perhaps something along the lines of:

Andersson et al. [arXiv][DOI]

which is kinda fiddly or

Andersson et al. (arXiv)(DOI)

If we're serious about formalising this, perhaps it would be worth chatting with the SE people to see if they can add a tool. I noticed that there has been a discussion on backtracks to the arXiv, and if we ask SE to make a tool then perhaps we should consider integrating the two. Bibliography management sites like CiteULike use a scraper which is rather complex, but I reckon if we just ask for the abstract page or the arXiv paper ID then a little javascript magic can do the rest of the work, and even pluck a DOI if it is available.

This post has been migrated from (A51.SE)

Comments

I think we should not be very serious about formalizing it. First, it rises the initial barrier. Second, it is not a journal and (I guess) most people won't care the form of the link, as long as the link is there.

This post has been migrated from (A51.SE)

---

@PiotrMigdal: I mean having a markup for it. Otherwise my first point stands, and citations can be put in by editors.

This post has been migrated from (A51.SE)

Answer 2

Since this is a theoretical physics resource, it might be useful to consider links to Spires/Inspire records. For eg: here. This forms the standard database for publications in high energy physics and will provide links to copies of the article on arxiv and a publication copy, if any. We can insist on Inspire, since it will replace Spires soon, and has a better interface.

This post has been migrated from (A51.SE)

Comments

To be honest, I don't see why its beneficial to use as a link _anything_ but arXivID/DOI.

This post has been migrated from (A51.SE)

---

The Inspire record for the document gives the arXiv link, doi, link to actual publication, citations for the document, references cited by the document, all in one place. The database is also searchable by author, title, etc. It is the de facto standard in the high energy physics community. It contains all the information about an article, more than arXiv. So it's nothing less than what you suggested, but in some cases, might be more helpful.

This post has been migrated from (A51.SE)

---

OK, but are you sure that it is robust enough? (E.g. in 10-20 years is it going to work in the same form.) And what is the percentage of all papers it covers (DOI covers virtually all papers published in respectable journals)?

This post has been migrated from (A51.SE)

---

@PiotrMigdal: Your point is well taken. Spires has served the HEP community very well for almost 40 years, so I think that they have a very good track record in future-proofing the database. Having said that, I wonder whether links to Spires will automatically redirect to INSPIRE once the transition takes place.

This post has been migrated from (A51.SE)

---

@Jose: Inspire is supposed to formally replace Spires sometime in October. So we might as well have people using Inspire right away. And Inspire is more useful than Spires due to the way they present all the data in their new interface.

This post has been migrated from (A51.SE)

---

@Siva: Yes, but my question was about the many links to Spires already in existence in other places. For instance, the arXiv links to Spires right now. I'm pretty sure that the transition has been well thought out and that the issue of dead links has been addressed, but I have not read anything about it.

This post has been migrated from (A51.SE)

---

@Piotr Migdal: "OK, but are you sure that it is robust enough? (E.g. in 10-20 years is it going to work in the same form.) " Are you sure that arXiv is going to work in the same form in 10-20 years?

This post has been migrated from (A51.SE)

---

@MarcinKotowski: I'm not (anyway, the life has proven it is hard to predict the lifespan of a certain web service). However, I can bet beer than in 15 years arXiv links (or IDs) will be backward compatible. When it comes to DOI, it _seems_ to be as stable as ISBN (so give DOI at least 30ys).

This post has been migrated from (A51.SE)

---

@siva: To be clear - I don't have anything _against_ using INSPIRE link. I just believe that arXiv/DOI is a safer solution. In any case, you can provide both, e.g. _arXiv:9999.9999 (or INSPIRE)_ (if you believe that INSPIRE links are more useful and worth promoting it may be a good solution).

This post has been migrated from (A51.SE)

---

The plans are to expand INSPIRE and make it more comprehensive and useful. There seems to be at least a 25 year vision plan to make it a full text service like arXiv, etc. Check [this](http://poynder.blogspot.com/2008/09/open-access-interviews-annette-holtkamp.html). They might probably even make documents searchable. Personally, I imagine this to be a huge opportunity for open science and it would be great if we can encourage it's use. The initiative has the support of the HEP community.

This post has been migrated from (A51.SE)

---

@Piotr: Either way. In essence, we should encourage people to link articles which they refer to (preferably open access).

This post has been migrated from (A51.SE)