Zotero Integration with the Aquifer Portal

Here are some miscellaneous notes on Integrating Zotero into the Aquifer portal:

Initially what we want is for Zotero to be able to easily slurp citations from the portal. Secondly we want the digital objects themeselves to be importable into Zotero. It appears there are several options for doing this.

1) Write a custom Zotero translator just for our portal site. This seems to be the most complex solution, plus the portal web site is undergoing constant changes right now, so this doesn't seem like the best solution. There may be some benefit to eventually doing this once the portal is solidified, but probably not.

2) The path of least resistance appears to be using COinS, so this was my initial solution. COins are OpenURL Context Objects inside of HTML span tags. For example:


<span class='Z3988' title='ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rfr_id=info%3Asid%2Fdlfaquifer.org%3Aportal&rft_id=info:oai/oai%3Aoai.dlib.indiana.edu%3Aarchives%2Fcushman%2FP10267&rft.title=St.%20Gaudens%27%20Lincoln%20in%20Lincoln%20Park%20Chicago&rft.identifier=http%3A%2F%2Fpurl.dlib.indiana.edu%2Fiudl%2Farchives%2Fcushman%2FP10267'></span>


The COinS contains the encoded metadata for the citation. Zotero is able to parse COinS from any web page and import them as citations into its database. A disadvantage of COinS is that OpenURL Context Objects do not support the full fidelity of metadata that might be desired, for example there is no way to embed the abstract.

3) Zotero also supports unAPI. This is a simple API for discovering and retrieving digital objects associated with a given web page. It is similar to COinS in that it embeds identifiers in special HTML tags (in this case abbr tags), but this API also requires a web service component which is able to retrieve the identified digital objects in different formats. This API probably holds the most promise for the long term, but requires some additional server side coding. We might be able to directly import MODS using unAPI, but I still need to investigate this.

4) A fourth option is embedding RDF metadata directly in the web pages. This has some potential in that it might support the full richness of metadata that we desire, but in my opinion is not quite as clean or simple as either the COinS or unAPI solutions.

So, right now we have implemented COinS, but the likely long term implementation will done using unAPI.