Traditionally DOIs (Digital Object Identifiers) have been associated with published papers in the digital era, but papers are not the only research objects that physicists may want to search, use, and cite. We talked with Jim Simone of Fermilab about his efforts to get DOIs assigned to MILC collaboration datasets and to get records of them uploaded to INSPIRE.

How is Jim involved with the MILC collaboration?

Jim is a member of FERMILAB-LATTICE collaboration, which works closely with MILC on scientific projects involving matrix elements and flavor physics. MILC generates data sets consisting of lattice gauge configuration files, which the collaboration has made openly available for others to use, as is increasingly becoming required for federally funded research in the U.S.

What is the MILC collaboration’s connection to the International Lattice Data Grid (ILDG)?

Jim was an early organizer of the ILDG, which is intended as a data grid to enable collaborations to share gauge configurations. The ILDG metadata catalog had its limitations; it only held limited kinds of metadata, sometimes making it difficult for people to find what they were looking for. People involved with the project have been trying to fill in the gaps, including the biggest problem: connecting scientific papers produced by the data to the datasets.

Rather than reinventing the wheel, ILDG is considering to use INSPIRE as a catalog to connect papers with datasets, making the data usable and findable by all physicists, including HEP and nuclear phenomenologists, as ILDG is currently only used by lattice scientists. In INSPIRE the datasets and associated papers can be searched starting with the papers in order to see what configurations were used to get the results, though in the upcoming version of INSPIRE, the Data collection will be made public and searching will also be possible starting with the individual datasets and from there finding what papers were produced from these configurations.

Lattice1INSPIRE record of MILC dataset that has been cited. http://dx.doi.org/10.15484/milc.asqtad.en05b/1178157

Lattice2References in INSPIRE record of a paper that cited MILC datasets.

Why and how did Jim go about getting DOIs assigned to the datasets? What challenges did he face?

Jim believes DOIs, as public, persistent identifiers, are a natural mechanism to identify the datasets, which are public, first class data objects, and permanent. With DOIs, the configurations will be better integrated into the ILDG and INSPIRE.

In the case of published papers, DOIs are assigned by publishers, but this route would not work for datasets. While INSPIRE is equipped to directly issue DOIs, MILC’s direct connection to the U.S. Department of Energy (DOE) made it practical for DOIs to be issued by DOE Office of Scientific and Technical Information (OSTI). In either case, DOIs are registered with the central agency DataCite.

ILDG has started a discussion on how other groups can get DOIs for their datasets. Outside the DOE, CERN also issues DOIs, and regional ILDG groups can help members get DOIs and serve as gatekeepers to keep the metadata clean and clear. DataCite can also help researchers find registration organizations.

For Jim it was a learning experience working with OSTI and interacting with their web services. As one of his main focuses was findability, Jim wanted to include lots of searchable metadata in the dataset records so to help physicists find the particular configurations they wanted. This amount of metadata was more than OSTI was used to receiving when minting DOIs, but they were able to work with Jim’s requests and he considered them a great help through the entire process

Beyond getting the DOIs assigned, another challenge was figuring out how citations should be marked up in papers, both written and digitally. With the goals of making the datasets findable and identifiable, Jim and the ILDG wanted people to be able to see the DOI in a print version of a reference list as well as click it in a digital version. In order to make the process as transparent as possible for people citing the datasets, Jim worked with us to include instructions in the metadata of the INSPIRE records and OSTI records.

Lattice3

For researchers unsure of how to cite datasets that do not include specific citation guidelines in their metadata, DataCite and CrossRef have developed a DOI citation formatter that can take a DOI registered by either of these services and format its citation in a variety of styles.

When going through the publication process with a paper that used MILC configurations, Jim found the referees and copy editors weren’t familiar with how the citations should appear. Most objects with a DOI are published papers that can be cited in written format using a journal reference, volume, page range, etc., so the DOI is often left out of the text of a reference list. However, following this standard would not make the datasets adequately identifiable to the human eye.

The community known as FORCE 11 (Future of Research Communication and e-Scholarship) has developed eight principles of data citation practices with equal emphasis on human readability and machine-actionability. As these recommendations become more widely endorsed in research communities and researchers become accustomed to citing datasets in their papers, the issue of human identifiable data citations will most likely be resolved.

What advice does Jim have for others looking to make their datasets more findable and citable?

Jim has two pieces of advice: get DOIs and mark up the metadata in a way that’s sensible for the community who will use the datasets. DataCite makes this simple by being explicit about its mandatory metadata requirements, while also allowing for additional recommended and optional metadata.

At INSPIRE we look forward to integrating more dataset DOIs into our records. Send your questions and comments about dataset DOIs in INSPIRE to feedback@inspirehep.net.

The 2013 Nobel Prize in physics was awarded to two particle physics theorists – François Englert and Peter W. Higgs today “for the theoretical discovery of a mechanism that contributes to our understanding of the origin of mass of subatomic particles, and which recently was confirmed through the discovery of the predicted fundamental particle, by the ATLAS and CMS experiments at CERN’s Large Hadron Collider.”

The theorists published their papers independently in 1964 – the first one by François Englert and Robert Brout, and, a month later, a pair of papers by Peter Higgs.

In the graph below you can see the number of citations that each of the papers received yearly since 1964 when they were published, peaking in 2012 with the latest Higgs boson search results.

Higgs-Nobel-plot-revised-final
Click to enlarge the picture

The theories were confirmed on 4th July 2012, by ATLAS and CMS, the two experiments at the LHC that were searching for the new particle. The two collaborations include more than 3000 people each.

The two papers published shortly after the first evidence presented by the experiments, accumulated enormous numbers of citations in just one year.

About a month ago the ATLAS experiment made the datasets behind the likelihood function associated to the Higgs boson property measurements available to the public in digital format. The datasets can be easily accessed on INSPIRE.

More about the Nobel prize in today’s CERN press release.

[A guest post from our power user and advisory board member Kyle Cranmer]

We are familiar with the critical role of INSPIRE in searching for papers, following references, tracking citations, and providing author profiles. Now INSPIRE is taking steps to extend this service to data, thus creating a rich new layer to the information system of high energy physics.

Historically, papers have been the primary means for scientific communication; however, it is common to augment papers with data. The Durham HepData project has hosted this type of data for several years, and since last year, HepData is integrated with INSPIRE. Some papers have several datasets associated to it, so each dataset is given a unique, persistent Digital Object Identifier (DOI). Not only do these DOIs ensure that you can find the data, but they also provide a clear way to cite the data.

Let’s look at an example. Last month, ATLAS took an important step forward and released a digital form of the likelihood function associated to the Higgs boson property measurements. You can find these likelihood functions by clicking on the Data tab of the ATLAS paper. There are three datasets associated to this paper, and each has its own DOI. For instance, the H→γγ likelihood’s DOI is 10.7484/INSPIREHEP.DATA.A78C.HK44. If you click, the DOI will “resolve” to the INSPIRE record for this specific dataset (not the paper). From this record you can:

  • download the data from the “Files” tab
  • check which papers are citing this dataset from the “Citation” tab
  • follow the link to the original paper
  • export a properly formatted citation to the dataset itself

In addition to data coming from HepData, INSPIRE now supports data hosted in other third-party data repositories such as Figshare or The Dataverse Network. To test this out, I put some data from a phenomenological study of the CMSSM onto Dataverse — yes, theorists create data too! In this case, Dataverse issued the persistent identifier to our data since they take on the responsibility to store it. I sent the persistent identifier to INSPIRE and now it shows up in the data tab of our original paper. INSPIRE can now track citations to this data, which is hosted remotely. Nice!

The last example comes from a not-so-high-energy experiment I was involved in called APEX Jefferson Lab, which is searching for evidence of a 5th fundamental force of nature together with similar experiments such as DarkLight, HPS, and MAMI. Unlike the enormous LHC experiments, APEX had 66 collaborators that contributed to the test-run for this small, special-purpose experiment. The results of the test run were published in 2011, and this week the raw mass distribution from those 770,509 events collected by APEX was released directly on INSPIRE.

These three examples illustrate the diversity of data in HEP ranging from low-level experimental data, to theoretical predictions, to the results of statistical analysis. They also demonstrate the richness of the data layer and the need for a robust information system. Looking to the future, we can imagine an extended author profile that includes details on datasets analogous to what we are already have with papers.

Datasets from HepData are now integrated into INSPIRE. HepData is a data repository hosted at Durham University which aggregates data extracted from publications in HEP and other datasets made available by the experiments. The data is presented in tables and available as data files for further reuse (plain format on INSPIRE; more formats are available on HepData). With the integration of HepData into INSPIRE, you can access data sets from HepData directly when browsing bibliographical records. If a record on INSPIRE has data from HepData attached, a corresponding tab is displayed next to the references and citation tabs. About 50,000 data files attached to about 7,000 records are available which are stored as regular records in INSPIRE. Datasets from 2012 e.g. can be found here.

What do you think about this new feature included in INSPIRE? We are looking forward to your feedback at feedback@inspirehep.net.