One aspect of RePEc that has grown in importance over the last years is its citations analysis, provided by the CitEc project, in particular due to their use in rankings. Citations extractions is a complex process. First, one needs to be able to access texts and find where references are (see details), then one needs to be able to interpret those references and match them with some work already listed in RePEc (see details). At this time, 5,400,000 references could be extracted from 240,000 works, with 2,300,000 matched to an item listed in RePEc. While these numbers may sound impressive, it still means that only about a third of online texts could be parsed successfully. To improve on this we rely on the RePEc archive maintainers to help us do a better job. Here is some advice in this regard that they should heed, as any linked reference allows links back and forth between the citing and cited works, thus increasing visibility.
- Check out how successful CitEc is in extracting references from your series and journals. Maintainers receive every months statistics about coverage that they can monitor. In addition, they can look up on CitEc the reasons why some items were not processed. For the series with the best coverage, see here.
- Make sure links in the metadata go directly to a pdf file, and not to an intermediate abstract page. CitEc does not go further than the link that is provided to it. If you really want the abstract page present in the metadata, provide it as a second link.
- Make sure that CitEc is actually allowed to get to the pdf. If the pdfs are gated, consider allowing CitEc to access with its IP, which will be provided upon request.
- The above are not possible, or if for some other reason references cannot be parsed, one can also transfer references to CitEc by using the X-File-Ref construct in the metadata, as described here.
- For larger archives, an alternative way of transferring references can be arranged.
- Also, CitEc sometimes grabs too many references. This happens for working papers when a list of other papers in the series is appended. This is also a waste of paper. We strongly recommend not to have such lists and, where they are present, to alert CitEc so that these errors can be remedied.
Any request should be send to José Manuel Barrueco, who is in charge of the CitEc project.