[Taxacom] Automated monitoring of the taxonomic literature

[Roderic Page]

Hi Carlos,

Thanks for the detailed response. This gives me a sense of the expectations for such a service.

In a sense I’ve gone about this the opposite way to what you suggest. The classification in BioRSS is being assembled as the literature collection grows, and it follows GBIF’s classification. That is, the treemap diagram on the left summarises what taxa are in the database so far (based on taxonomic names found in the title and/or abstract). As new papers are added this diagram will expand to include more taxa.

Regarding integration with a virtual research environment, this is a good point. At the moment I have focussed on RSS as the API, that is, each combination of taxon and geography has a corresponding RSS feed, the idea being that one could simply grab the URL to a feed of interest and poll that for publications.

The desire to have access to the data in a bibliographic format such as BibTex or RSS has also been mentioned by Tony Rees. Initially I was avoiding this as the RSS feeds I use typically lack bibliographic details, but there are ways around this. The data could be stored in BioRSS, or elsewhere such as Wikidata (which is rapidly becoming a repository of a lot of taxonomic literature).

Any automated service will have limitations, and the taxonomic literature is particularly challenging given how many journals publish taxonomy. Zootaxa dominates animal taxonomy making it an obvious source, but there is a very long tail of other publications. I am looking at adding Google Scholar as a source, which brings its own challenges.

Regarding taxonomic databases as a source, this is something I am thinking about, especially databases that regularly get updated (ZooBank being an obvious one). But I guess the real proof of utility for BioRSS would be that people find references in BioRSS to add to those databases, not the other way around.

Thanks again for the feedback, I’ve clearly got a lot of work to do.

Regards,

Rod


Hi Rod,
I have a similar interest and thus a similar problem to that of Peter Uetz: Myriapoda & Onychophora are around 15K valid species (very rough number). Of course, the new works on reptiles are definitely more than those on multipedes.
When I see new tools, I always ask myself if they could be wrapped in a virtual research environment (VRE), and not just exist as a standalone application or web service. That's a good thing to keep in mind when developing something like BioRSS.
Finding both new and old names is important in the course of a research workflow. That is one of the reasons why I like that Scratchpads has both ReFindIt and BHL search integrated with a customizable classification (= taxonomic backbone). Because of that, if I were to create something like BioRSS, the first thing that I would do is to import a classification as complete as possible, at least down to the family level, but that would take BioRSS closer to being a virtual research environment, which takes me back to the idea of rather having the service as a module and adding it to an existing VRE with a taxonomic backbone. Then something like BioRSS could be an alternative or addition to integration with ReFindIt.
If I were to use a standalone BioRSS as a broker of other feeds, then I would very much like it to have "Myriapoda" in the classification. One issue that I see is deduplication, e.g., a broker should reduce and not add more reading burden on top of Google Scholar alerts, journal alerts, etc. If such a broker could deduplicate alerts, then that would already be a win, as users could subscribe to the broker and unsubscribe from journals + Google Scholar + other alerts.
Anyway, it seems that there is no replacement for a tight community committed to recording, as some publications (including new taxa) don't reach the internet and others do but are not findable by taxon. In the latter I include articles in technologically advanced journals which do taxon tagging for the "text" but not for the supplementary data. If authors (mainly in ecology and molecular biology) are trying to save on pricey page charges, taxa tend to be relegated to those supplementary files and thus not marked up. Then recorders have no other option than to take a closer look at unmatched but "suspicious" articles.
Next in automatically monitoring new literature for databasing purposes is the possibility to make a bulk download (BibTex, RIS) of the bibliographic references for closer inspection, or for direct, bulk upload into a database. It is inefficient to go back to the sources and download a BibTex or a RIS at the time. That also depends on how frequently one wants / can update the database. If there are two updates per year and few people, better to go to the publishers a couple of times, make a search, and export references in bulk. So, the preferences will depend on the tradeoff between "daily / weekly aggregated and downloadable references from a broker" versus "biannual / annual aggregated and downloadable references from a publisher or journal". If the broker can't provide bulk download, the users may tend to shift to other services.
Taxonomic databases as a source: In Myriatrix (a Scratchpad) I do try to add DOIs whenever available. I also try to record publication dates as exact as possible in a dedicated "Date Published:" field. Additionally, most bibliographic databases (all?) would try to record the publication year ("Year of Publication:" in Scratchpads). Those two fields should facilitate harvesting recent literature from Scratchpads sites, with the added benefit of the references being exportable in bulk in several interoperable formats. Combined with taxon search and matching in titles, abstracts, and keywords, that would make e.g., myriapod-reptile publications discoverable and downloadable for reuse in the Reptile Database.
Just my five cents.
Kind regards,
Carlos