Next steps for BioStor: citation matching

Thinking about next steps for my BioStor project, one thing I keep coming back to is the problem of how to dramatically scale up the task of finding taxonomic literature online. While I personal find it oddly therapeutic to spend a little time copying and pasting citations into BioStor's OpenURL resolver and trying to find these references in BHL, we need something a little more powerful.

One approach is to harvest as many bibliographies as possible, and extract citations. These citations can come from online bibliographies, as well as lists of literature cited extracted from published papers. By default, these would be treated as strings. If we can parse them to extract metadata (such as title, journal, author, year), that's great, but this is often unreliable. We'd then cluster strings into sets that we similar. If any one of these strings was associated with an identifier (such as a DOI), or if one of the strings in the cluster had been successfully parsed into it's component metadata so we could find it using an OpenURL resolver, then we've identified the reference the strings correspond to. Of course, we can seed the clusters with "known" citation strings. For citations for which we have DOIs/handles/PMIDs/BHL/BioStor URIs, we generate some standard citation strings and add these to the set of strings to be clustered.

We could then provide a simple tool for users to find a reference online: paste in a citation string, the tool would find the cluster of strings the user's string most closely resembles, then return the identifier (if any) for that cluster (and, of course, we could make this a web service to automate processing entire bibliographies at a time).

I've been collecting some references on citation matching (bookmarked on Connotea using the tag "matching") related to this problem. One I'd like to highlight is "Efficient clustering of high-dimensional data sets with application to reference matching" (doi:10.1145/347090.347123, PDF here). The idea is that a large set of citation strings (or, indeed, any strings) can first be quickly clustered into subsets ("canopies"), within which we search more thoroughly:

When I get the chance I need to explore some clustering methods in more detail. One that appeals is the MCL algorithm, which I came across a while ago by reading PG Tips: developments at Postgenomic (where it is used to cluster blog posts about the same article). Much to do...

Linnaeus meets the Internet: PLoS + Botany = #fail

To much fanfare (e.g., Nature News, "Linnaeus meets the Internet" doi:10.1038/news.2010.221), on May 5th PLoS ONE published Sandy Knapp's "Four New Vining Species of Solanum (Dulcamaroid Clade) from Montane Habitats in Tropical America" doi:10.1371/journal.pone.0010502. To quote the Nature News piece:

The paper represents the culmination of a campaign to institute the electronic publication of scientific names, a case Knapp and others have made in journals including Nature[doi:10.1038/446261a]. Allowing electronic publication should make accessing information easier for scientists worldwide — especially those in developing countries who may not have access to fully stocked libraries. This, in turn, will aid conservation efforts, Knapp says.

Given the profile of this paper, "...the first time new plant names have been published in a purely electronic journal and still complied with ICBN rules", you'd think the participants would ensure the electronic aspects of the publication worked. Sadly, this is not the case.

The four names in question have apparently been deposited in IPNI with the following LSID's:

• Solanum aspersum: urn:lsid:ipni.org:names:77103633-1


• Solanum luculentum: urn:lsid:ipni.org:names:77103634-1


• Solanum sanchez-vegae: urn:lsid:ipni.org:names:77103635-1


• Solanum sousae: urn:lsid:ipni.org:names:77103636-1

Today is May 6th. None of these names are returned by a search of IPNI, for example http://www.ipni.org/ipni/simplePlantNameSearch.do?find_wholeName= returns this:

Resolving the LSID returns this:

<?xml version="1.0" encoding="UTF-8"?>
<rdf:RDF xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:dc="http://purl.org/dc/elements/1.1/" 
xmlns:dcterms="http://purl.org/dc/terms/"
xmlns:tn="http://rs.tdwg.org/ontology/voc/TaxonName#"
xmlns:tm="http://rs.tdwg.org/ontology/voc/Team#"    
xmlns:tcom="http://rs.tdwg.org/ontology/voc/Common#"    
xmlns:p="http://rs.tdwg.org/ontology/voc/Person#"    
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:owl="http://www.w3.org/2002/07/owl#">
<tn:TaxonName rdf:about="urn:lsid:ipni.org:names:77103633-1">	
<tcom:versionedAs rdf:resource="urn:lsid:ipni.org:names:77103633-1:1.2"/>
<tcom:Deleted>Yes</tcom:Deleted>
</tn:TaxonName>  
</rdf:RDF>

Hmmm, so apparently this record has been "deleted"?

The paper also states that:

The IPNI LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID contained in this publication to the prefix http://ipni.org/.

This sentence mirrors similar ones in other PLoS ONE papers saying we can resolve ZooBank LSIDs by appending the LSID to http://zoobank.org (e.g., see doi:10.1371/journal.pone.0001787).

Thing is, URLs such as http://ipni.org/urn:lsid:ipni.org:names:77103633-1 return a 404 from Kew (any IPNI LSID I've tried does this).

Update As per Alan Paton's comment below, the http://ipni.org prefix now works.

So, to recap:

1. The names aren't in IPNI


2. The LSIDs state the record has been deleted


3. The LSID's can't be resolved by the means stated in the paper

Now, I don't know what happened (perhaps IPNI wanted to hold off until the paper actually appeared before releasing the names), but the paper is out, the buzz in Nature is out, and IPNI doesn't have the resolver in place, yet alone the names.

Given the milestone this paper represents, and the fuss over the publication of the name Darwinius, you'd expect the bioinformatics side of it to be, you know, actually working. In these circumstances, how on Earth do we make the case that the LSID and name databasing side of taxonomic publication is useful?

Mendeley Open API and the Biodiversity Heritage Library

Mendeley have called for proposals to use their forthcoming API. The API will publicly available soon, but in a clever move Mendeley will provide early access to developers with cool ideas.

Image by Mendeley.com

Given that the major limitation of the Biodiversity Heritage Library (from my perspective) is the lack of article-level metadata, and Mendeley has potentially lots of such data, I wonder whether this is something that could be explored. My BioStor project takes article metadata and finds articles in BHL, so an attractive work flow would be:

1. People upload bibliographies to Mendeley (e.g., bibliographies for particular taxa, journals, etc.)


2. BioStor uses Mendeley's API to find articles liklely to be in BHL, then locates the actual article in Mendeley.


3. The user could then grab a PDF of the article from BioStor that contains XMP metadata (which Mendeley, and other tools, can read)

Users would gain a tool to manage their bibliographies (assuming that they prefer Mendeley to other tools, or are happy to sync with Mendeley), they would be contributing to a database of taxonomic (and biological literature in general, BHL's content is pretty diverse), and also gain easy access to PDFs for BHL content (this last feature depends on whether Mendeley can associate a PDF with an existing bibliographic record automatically). In the same way, a tool such as BioStor (and, by implication, BHL) could gain usage statistics (i.e., who is reading these articles?).

Our communities efforts at assembling bibliographies haven't amounted to much yet. The tools we use tend to be poor. I find CiteBank to be underwhelming, and Drupal's bibliographic modules (used by CiteBank and ScratchPads) lack key features. We also seem reluctant to contribute to aggregated bibliographies. Perhaps encouraging people to use a nicer tool, and at the same time providing additional benefits (e.g., XMP PDFs) might help move things forward.

Anyway, food for thought. Perhaps other tools might make more sense, such as using the API to upload metadata and PDFs direct from BioStor to Mendeley, and making the collection public. But, if I were Mendeley, what I'd be looking for are tools that enhance the Mendeley experience. There's some obvious scope for visualising the output and social networks of authors, such as the sparklines and coauthor graphs I've been playing with in BioStor (for example, for W E Duellman):

Before this blog post starts to veer irretrievably off course, I'd be interested in thoughts of anyone interested in matters BHL. There's nothing like a deadline (Friday, May 14th) to concentrate the mind...