Show me the data!
Header

With a first commit to github not so long ago (2015-04-13), getpapers is one of the newest tools in the ContentMine toolchain.

It’s also the most readily accessible and perhaps most immediately exciting – it does exactly what it says on the tin: it gets papers for you en masse without having to click around all those different publisher websites. A superb time-saver.

It kinda reminds me of mps-youtube: a handy CLI application for watching/listening to youtube.

Installation is super simple and usage is well documented at the source code repository on github, and of course it’s available under an OSI-approved open source MIT license.

An example usage querying Europe PubMedCentral

Currently you can search 3 different aggregators of academic papers: Europe PubMedCentral, arXiv, and IEEE. Copyright restrictions unfortunately mean that full text article download with getpapers is restricted to only freely accessible or open access papers. The development team plans to add more sources that provide API access in future, although it should be noted that many research aggregators simply don’t appear to have an API at the moment e.g. bioRxiv.

The speed of the overall process is very impressive. I ran the below search & download command and it executed it all in 32 seconds, including the download of 50 full text PDFs of the search-relevant articles!

You can choose to download different file formats of the search results: PDF, XML or even the supplementary data. Furthermore, getpapers integrates extremely well with the rest of the ContentMine toolchain, so it’s an ideal starting point for content mining.

getpapers is one of many tools in the ContentMine toolchain that I’ll be demonstrating to early career biologists at a FREE registration, one-day workshop at the University of Bath, Tuesday 28th July. If you’re interested in learning more about fully utilizing the research literature in scalable, reproducible ways, come along! We still have some places left. See the flyer below for more details or follow this link to the official workshop registration page: bit.ly/MiningWrkshp

mining-flyer

To prove my point about the way that supplementary data files bury useful data, making it utterly indiscoverable to most, I decided to do a little experiment (in relation to text mining for museum specimen identifiers, but also perhaps with some relevance to the NHM Conservation Hackathon):

I collected the links for all Biology Letters supplementary data files. I then filtered out the non-textual media such as audio, video and image files, then downloaded the remaining content.

A breakdown of file extensions encountered in this downloaded subset:

763 .doc files
543 .pdf files
109 .docx files
75 .xls files
53 .xlsx files
25 .csv files
19 .txt files
14 .zip files
2 .rtf files
2 .nex files
1 .xml file
1 “.xltx” file

I then converted some of these unfriendly formats into simpler, more easily searchable plain text formats:

 

Now everything is properly searchable and indexable!

In a matter of seconds I can find NHM specimen identifiers that might not otherwise be mentioned in the full text of the paper, without actually wasting any time manually reading any papers. Note, not all the ‘hits’ are true positives but most are, and those that aren’t e.g. “NHMQEVLEGYKKKYE” are easy to distinguish as NOT valid NHM specimen identifiers:

 

Perhaps this approach might be useful to the PREDICTS / LPI teams, looking for species occurrence data sets?

I don’t know why figshare doesn’t do deep indexing by default – it’d be really useful to search the morass of published supplementary data that out there!

Progress on specimen mining

June 14th, 2015 | Posted by rmounce in Content Mining - (0 Comments)

I’ve been on holiday to Japan recently, so work came to a halt on this for a while but I think I’ve largely ‘done’ PLOS ONE full text now (excluding supplementary materials).

My results are on github: https://github.com/rossmounce/NHM-specimens/tree/master/results – one prettier file without the exact provenance or in-sentence context of each putative specimen entity, and one more extensive file with provenance & context included which unfortunately github can’t render/preview.

 

Some summary stats:

I found 427 unique BMNH/NHMUK specimen mentions from a total of just 69 unique PLOS ONE papers. The latter strongly suggests to me that there are a lot of ‘hidden’ specimen identifiers hiding out in difficult-to-search supplementary materials files.

I found 497 specimen mentions if you include instances where the same BMNH/NHMUK specimen is mentioned in different PLOS ONE papers.

Finding putative specimen entities in PLOS ONE full text is relatively automatic and easy. The time-consuming manual part is accurately matching them up with official NHM collection specimens data.

I could only confidently link-up 314 of the 497 detected mentions, to their corresponding unique IDs / URLs in the NHM Open Data Portal Collection Specimens dataset. Approximately one third can’t be confidently be matched-up to a unique specimen in the online specimen collection dataset — I suspect this is mainly down to absence/incompleteness in the online collections data, although a small few are likely typo’s in PLOS ONE papers.

In my last post I was confident that the BM Archaeopteryx specimen would be the most frequently mentioned specimen but with more extensive data collection and analysis that appears now not to be true! NHMUK R3592 (a specimen of Erythrosuchus africanus) is mentioned in 5 different PLOS ONE papers. Pleasingly, Google Scholar also finds only five PLOS ONE papers mentioning this specimen – independent confirmation of my methodology.

One of the BM specimens of Erythrosuchus is more referred to in PLOS ONE than the BM Archaeopterx specimen

Now I have these two ‘atomic’ identifiers linked-up (NHM specimen collections occurrence ID + the Digital Object Identifier of the research article in which it appears), I can if desired, find out a whole wealth of information about these specimens and the papers they are mentioned in.

My next steps will be to extend this search to all of the PubMedCentral OA subset, not just PLOS ONE.

 

In this post I’ll go through an illustrated example of what I plan to do with my text mining project: linking-up biological specimens from the Natural History Museum, London (sometimes known as BMNH or NHMUK) to the published research literature with persistent identifiers.

I’ve run some simple grep searches of the PMC open access subset already, and PLOS ONE make up a significant portion of the ‘hits’, unsurprisingly.

Below is a visual representation of the BMNH specimen ‘hits’ I found in the full text of one PLOS ONE paper:

Grohé C, Morlo M, Chaimanee Y, Blondel C, Coster P, et al. (2012) New Apterodontinae (Hyaenodontida) from the Eocene Locality of Dur At-Talah (Libya): Systematic, Paleoecological and Phylogenetical Implications. PLoS ONE 7(11): e49054. doi: 10.1371/journal.pone.0049054

onepaper

I used the open source software Gephi, and the Yifan Hu layout to create the above graphical representation. The node marked in blue is the paper. Nodes marked in red are catalogue numbers I couldn’t find in the NHM Open Data Portal specimen collections dataset: 10 out of 34 not found.

Source data table below showing how uninformative the NHM persistent IDs are. I would have plotted them on the graph instead of the catalogue strings as that would be technically more correct (they are the unique IDs), but it would look horrible.

 

I’ve been failing to find a lot of well known entities in the online specimen collections dataset which makes me rather concerned about its completeness. High profile specimens such as Lesothosaurus “BMNH RUB 17” (as mentioned in this PLOS ONE paper, Table 1) can’t be found online via the portal under that catalogue number. I can however find RUB 16, RUB 52 and RUB 54 but these are probably different specimens. RUB 17 is mentioned in a great many papers by many different authors so it seems unlikely that they have all independently given the specimen an incorrect catalogue number – the problem is more likely to be in the completeness of the online dataset.

Another ‘missing’ example is “BMNH R4947” a specimen of Euoplocephalus tutus as referred to in Table 4 of this PLOS ONE paper by Arbour and Currie. There are two other records for that taxon, but not under R4947.

To end on a happier note, I can definitely answer one question conclusively:
What is the most ‘popular’ NHM specimen in PLOS ONE full text?

…it’s “BMNH 37001”, Archaeopteryx lithographica which is referred to in full text by four different papers (see below for details).

I have feeling many more NHM specimens are hiding out in separate supplementary materials files. Mining these will be hard unless figshare gets their act together and creates a full-text API for searching their collection – I believe it’s a metadata only API at the moment.

37001 in PLOS ONE papers

 

I’ve purposefully made very simple graphs so far. Once I get more data, I can start linking it up to create beautiful and complex graphs like the one below (of the taxa shared between 3000 microbial phylogenetic studies in IJSEM, unpublished), which I’m still trying to get my head around. The linked open data work continues…

Bacteria subutilis commonly used