you'll note that Zotero imports fine via the DOIs where they're present. For anything else, we'll need ORCID to implement some structured metadata into the page (unAPI, ideally, COinS at a minimum), we're not going to parse the citation data from the page.
It is impossible to parse the information on an ORCID record page because publication information is mostly in narrative form. Citations differ in format from one ORCID user record to another and even within a single user's publication list.
Right now, the ORCID folks have expressed less than little interest in any sort of structure within their database, much less interest in facilitatiing export of the publication data in each user's record. Hand-entered publication data has almost no structure. Even the material that gets brought in from CrossRef seems to be imported into a free form flat citation format. There is no standard format for auther names -- when I signed up for an ORCID ID, there was a single field to enter my name and absolutely no guidance as to best practices for entry (last first middle etc.) Publications are entered using a single "citation" form with some of the useful metadata fields (journal name, pub year) but journal volume, issue and pagination are entered into a free text field. This single "citation" free-text field is where co-authors are entered. Again there is no guidance for a standard entry format.
Through SafetyLit, I was one of the ORCID "launch partners". I pleaded with the staff to delay the launch so that there could be room for structured metadata in the database. I even offered to donate some time of the team that develops and maintains my database to help with getting structured data into the db tables. My pleas and offers were rejected. My high hopes for the great potential of this project have wilted somewhat. I still hope to convince the decision-makers that structured metadata is necessary to begin to meet their goals.
I have volunteered to participate in an ORCID workgroup. I hope to lead the participants to an understanding of the importance metadata structure. From the feedback I received, it is clear that the decision makers belive that requiring a structured system for data entry and storage makes the site less user friendly. When I questioned the data entry process for entering the various names by which an author has published (a single field for all names and no guidance concerning name order or punctuation) their reply was that naming practices differ across the world. They were unaware that there are detailed cataloging guidelines for recording names that are internationally accepted. With the active participation of so many publishers and databases I would have thought that metadata standards would have been a given. Instead, the system was designed from the beginning by savvy database folks but who had no background in indexing or cataloging. So far, I have failed in my attempts to help them understand the need for standards. I remain convinced that if the right person at ORCID is reached the work model will change. I believe that the ORCID staff are dedicated and hard-working but that in their worldview metadata structure standards are not relevant to the project. That there are universally accepted standards for bibliographic information didn't seem to occur to he ORCID folks I corresponded with.
I'm now part of an ORCID working group, for works metadata, and these issues are being actively addressed. I'm happy to reeve communication via the details in my profile, as I only call by here occasionally.
From our perspective this is pretty simple:
COinS would be good, unAPI with some better metadata like MODS or bibtex would be even better, and if you implement any type of metadata - even if it's just google highwire tags - we can easily write a translator, just let us know.
The biggest problem right now is that all authors' names are entered and stored in a single field with no guidance as to how first name / last name should be included and multiple authors simply separated by ";" (but not necessarily). However, if the author name information is imported the name metadata can be in almost any format and order (full names, names with initials and periods, names with initials and no periods. Journal article volume, issue, and page metadata is all one field with no guidance to inform the person entering the data for the preferred pattern and punctuation. I too have been working with ORCID but the comments of @pigsonthewing are the first hint that someone agrees that structured data is important.
Upgrade Storage
Right now, the ORCID folks have expressed less than little interest in any sort of structure within their database, much less interest in facilitatiing export of the publication data in each user's record. Hand-entered publication data has almost no structure. Even the material that gets brought in from CrossRef seems to be imported into a free form flat citation format. There is no standard format for auther names -- when I signed up for an ORCID ID, there was a single field to enter my name and absolutely no guidance as to best practices for entry (last first middle etc.) Publications are entered using a single "citation" form with some of the useful metadata fields (journal name, pub year) but journal volume, issue and pagination are entered into a free text field. This single "citation" free-text field is where co-authors are entered. Again there is no guidance for a standard entry format.
Through SafetyLit, I was one of the ORCID "launch partners". I pleaded with the staff to delay the launch so that there could be room for structured metadata in the database. I even offered to donate some time of the team that develops and maintains my database to help with getting structured data into the db tables. My pleas and offers were rejected. My high hopes for the great potential of this project have wilted somewhat. I still hope to convince the decision-makers that structured metadata is necessary to begin to meet their goals.
COinS would be good, unAPI with some better metadata like MODS or bibtex would be even better, and if you implement any type of metadata - even if it's just google highwire tags - we can easily write a translator, just let us know.