Examples of links to topic pages from articles on Science Direct – see, for example, “fullerenes”.

Should you add functionality just because you can? Here is an example of why perhaps you should not. Todd Carpenter has written an excellent post on Scholarly Kitchen about Elsevier automatically adding links to published academic articles. It seems that many articles on Science Direct include links, which were not added by the author, and which take the reader to Elsevier’s topic pages. Not only are the links automatically added, but the topic pages themselves appear to be automatically generated (more about that later).

Todd Carpenter’s post is about the ethical implications of adding links to a published paper that is the copyright of an author. Is adding a link a change to the content? I would have thought it was, but I might forgive the publisher if they were adding value to the article. As readers, we frequently encounter terms and phrases we don’t understand, so links are potentially useful.

When the links are not helpful, it becomes an annoyance. Wikipedia started adopting the practice some years ago of adding links to other Wikipedia entries whenever a term in mentioned. I noticed this originally for city and country names – in the entry for France, for example, any mention of Paris became a hyperlink. The practice seems to have become slightly more subtle (sometimes not every instance of a term is converted to a link), but not much better. For example, the Wikipedia entry for French philosopher Voltaire contains no fewer than eight links in the first sentence, while the entry for Paris, France, has ten links in the first two sentences:

François-Marie Arouet 21 November 1694 – 30 May 1778, was a French Enlightenment writer, historian, and philosopher famous for his wit, his criticism of Christianity—especially the Roman Catholic Church—as well as his advocacy of freedom of speechfreedom of religion, and separation of church and state.

Paris is the capital and most populous city of France, with an estimated population of 2,175,601 residents as of 2018, in an area of more than 105 square kilometres (41 square miles). Since the 17th century, Paris has been one of Europe’s major centres of financediplomacycommercefashiongastronomyscience, and arts

Wikipedia content

Do such links help, or are they simply annoyances? Do we not know what “wit” is, or the “Roman Catholic Church”, or “the arts”? Why does Wikipedia do this? Largely because they can. The editors at Wikipedia, proud of their wealth of articles on all kinds of subjects, are anxious to point to all the careful editorial work they have done. So the reference to “most populous city” is actually a link to a table of the largest communes in France. Do I need to know this? No, but Wikipedia certainly wants to tell me. Somebody put a lot of work into compiling this table, and you might not notice.

Better would be to have a rule that terms used in their standard dictionary meaning (“capital”, “population”, “Roman Catholic Church”) do not require a link. Whether a machine could distinguish between links that are to “standard” content and links that contain something of genuine relevance to the article discussed is probably not very likely. Some cross references will have to be curated by hand.

What are “topic pages”? They were introduced by Elsevier in 2017: “a free layer of content which provides a quick snapshot of definitions, terms and excerpts on scientific topics, built on Elsevier’s highly trusted book content.” They are a brave attempt to create reference-type articles entirely automatically. Certainly, no human would create such poor pages. There are over 356,000 of them, but that doesn’t excuse their frequently poor quality.

The idea is to provide a brief definition for those unfamiliar with the topic. If any publisher has brief definitions, Elsevier should have – but the resulting topic pages are often less than satisfactory. For example, I encountered the term “electrodermal activity”, and clicked on it to the relevant topic page. The highlighted text at the top of the page appears to be a definition – but sometimes the selected text is nothing of the kind:

If you don’t know what an aardvark is, the following topic page definition isn’t going to help you:

It doesn’t help that there are three separate topic pages for “aardvark”. They all seem to be about the animal, but buried in one of the pages is a sense that looks rather different – although you would have to be a detective to find this other meaning.

A commercial publisher such as Elsevier is creating topic pages to provide more engagement with the user. So the issue is not just the ethics; if the automatically generated topic pages are good enough to be helpful, the exercise is a success. But the pages are, as far as I have sampled them, not very good. Will it help a researcher to be redirected to a hotchpotch of extracts loosely around the term “aardvark”? Like the Wikipedia hyperlinks, Elsevier’s topic pages look like the first attempt at an idea that has never been sorted out for publication – despite being launched four years ago.

Are there any occasions where automated links are a good idea? Perhaps one of the best examples is the Kindle e-book reader software. Simply click on any word to see a dictionary definition of that word. It’s not intrusive, but it exploits the capability of running an app to provide functionality only dreamed of in the age of print.

Kindle definitions – simply highlight a word to see a definition