Micropublications: a semantic model for claims, evidence, arguments and annotations in biomedical communications

Abstract

Background: Scientific publications are documentary representations of defeasible arguments, supported by data and repeatable methods. They are the essential mediating artifacts in the ecosystem of scientific communications. The institutional "goal" of science is publishing results. The linear document publication format, dating from 1665, has survived transition to the Web. Intractable publication volumes; the difficulty of verifying evidence; and observed problems in evidence and citation chains suggest a need for a web-friendly and machine-tractable model of scientific publications. This model should support: digital summarization, evidence examination, challenge, verification and remix, and incremental adoption. Such a model must be capable of expressing a broad spectrum of representational complexity, ranging from minimal to maximal forms.

Results: The micropublications semantic model of scientific argument and evidence provides these features. Micropublications support natural language statements; data; methods and materials specifications; discussion and commentary; challenge and disagreement; as well as allowing many kinds of statement formalization. The minimal form of a micropublication is a statement with its attribution. The maximal form is a statement with its complete supporting argument, consisting of all relevant evidence, interpretations, discussion and challenges brought forward in support of or opposition to it. Micropublications may be formalized and serialized in multiple ways, including in RDF. They may be added to publications as stand-off metadata. An OWL 2 vocabulary for micropublications is available at http://purl.org/mp. A discussion of this vocabulary along with RDF examples from the case studies, appears as OWL Vocabulary and RDF Examples in Additional file 1.

Conclusion: Micropublications, because they model evidence and allow qualified, nuanced assertions, can play essential roles in the scientific communications ecosystem in places where simpler, formalized and purely statement-based models, such as the nanopublications model, will not be sufficient. At the same time they will add significant value to, and are intentionally compatible with, statement-based formalizations. We suggest that micropublications, generated by useful software tools supporting such activities as writing, editing, reviewing, and discussion, will be of great value in improving the quality and tractability of biomedical communications.

Keywords: Annotation; Argumentation; Data citation; Digital abstract; Methods citation; Nanopublications; Research reproducibility; Scientific discourse; Scientific evidence.

Figure 1

Representation of statements and evidence in a nanopublication format. Nanopublications include a named graph called “Support”, but this is actually a set of Qualifiers used for filtering, rather than the supporting evidence and citations. <CHEBI: 68481>, <UNIPROT: P42345>, etc., are abbreviations for the respective fully-qualified URIs.

Figure 2

Activity lifecycle of biomedical communications linked to use cases, activity inputs and activity outputs. The main information content generated, enhanced and re-used in the system is shown. Bolded inputs and outputs represent micropublication-specific content.

Figure 3

Minimal form of a micropublication: formalizing a statement and its attribution.

Figure 4

A Micropublication supported by a Statement referenced to the domain literature; empirical Data; and a reusable Method.

Figure 5

Major classes and relationships in the model. A Claim is the main Statement argued by a Micropublication. A Statement is a truth-bearing Sentence, which may be variously qualifiedBy some Qualifier. A Sentence is a well-formed sequence of symbols, intended to convey meaning; and is not necessarily either complete or truth-bearing. A Micropublication hasElements consisting of those Representations it asserts or quotes. A Representation supports or challenges other Representations. The supporting Representations which are elementOf the Micropublication will be in its SupportGraph; challenging elements, will be in its ChallengeGraph. Dashed-line boundaries indicate graphs instantiated by query.

Figure 6

Representation of statements and evidence in a micropublication format, based on text from Spilman et al.[27]. Micropublications show the graph of support for a Statement, including the author’s Data and Methods, constituted by a set of supports relations in the SupportGraph. The Claim is qualifiedBy SemanticQualifiers < CHEBI:68481 > and < INO:0000736>, which are abbreviations for the respective fully-qualified URIs.

Figure 7

Example 1: The argument “Rapamycin is an inhibitor of the mTOR pathway” represented as a micropublication, with semantic qualifiers. This argument is taken from Spilman et al. 2010 [67]. C1 is the Claim; A_C1 is the Attribution of the Claim; Ref5 is the Claim’s supporting reference; SG1 is the SupportGraph. At the time the micropublication was extracted, its claim was assigned the Qualifiers Q1 and Q2. Note that the Claim Attribution for C1, represents the attribution of the article in which the text of C1 appears, not the article cited in support of C1.

Figure 8

Example 2: Scientific Evidence consisting of Data and Methods supporting C3, a Claim from [67]. D1 is a composite image of graphs produced from the primary data. The Methods M1 (a procedure) and M2 (a transgenic mouse strain), both support D1.

Figure 9

Example 3: Micropublication for the principal Claim of Spilman et al. 2010 [67]. Claim C3 from the abstract is supported by Statements S1 and S2, with support in the literature (Ref5, Ref9 and Ref10); and S3, which interprets the Data in D1. D1 in turn is supported by Methods M1 and M2. All elements of the Support Graph have Attribution A_C3; for simplicity we show only the relationship supports(A_C3, C3) in this figure.

Figure 10

Example 4: Transition from document-level backing references, to claim-level backing references, to construct a citation network.

Figure 11

Example 4: Connected support relations of three arguments give a Claim network across three publications. Micropublication MP6 asserts the new information that C1.1 is support for S1; and C2.1 is the support for S2. The dotted lines connecting elements of MP3, MP4 and MP5 are given by the support relations in MP6.

Figure 12

Example 5: Similarity Group with representative (Holotype) Claim. The Holotype is from one of three additional publications, outside the Claim Lineage C1→ S1.

Figure 13

Example 5: Group of homologs as a micropublication, with a user-defined holotype (representative) Claim. C4.H is the Claim. The other members of the homology group are members of its SupportGraph, SG7. The Attribution of MP7, A_MP7, shows that the author and curator are the same person.

Figure 14

Example 6: BEL representation of a Claim from Spilman et al. 2010 [67] with support at document level only. Attribution of support is based on resolving the PubMed ID from a database.

Figure 15

Example 6: BEL representation of a Claim from Spilman et al. 2010 [67] with support resolved to the Claim level. S4 is the specific Claim translated into the BEL statement.

Figure 16

Example 7: Annotation using micropublication relations. C10 is a Claim made by the annotator, “A. Neuroscientist”. It is supported by Statement C1 in Spilman et al. 2010, which it annotates.

Figure 17

Example 8: Claim C11, taken from[[64]], Bryan et al.’s review of transgenic mouse models of AD, challenges Statement MP3:S3 in Micropublication MP3. The Claim notes an issue with PDAPP mice tending to confound their performance in the Morris Water Maze (MWM). R48, R49 and R50 are document-level citations.

Figure 18

Example 8: Annotation of inconsistency between Spilman et al. [27] and Bryan et al. [107] as an independent micropublication. Challenge relationships between MP3:S3 and MP11:C11 are included in the SupportGraph SG12, because what MP12:C12 argues is precisely that these two Statements are in conflict.

Figure 19

Example 9: Contextualizing a claim within a full-text document using the Open Annotation Ontology.

Figure 20

Citation distortion: Graph of a claim network from Greenberg 2009 [3] showing citation distortion. The claim presented in this lineage states that amyloid beta deposition in IBM muscle fibers precedes other pathological changes – from which can be inferred that it is the causative factor. The foundational publications, which would be expected to contain supporting data, do not. Needham and Massaglia’s article from Lancet Neurology provides no support at all for Claim C11 at all, treating it as a pure fact. In the Claim network, we assert C11 as a Holotype, or representative Statement, for C12. In fact, that assertion could be made for all Claims in the network.

Figure 21

Software implementation: Creating a micropublication in Domeo version 2. The micropublication asserts a Claim, which was simply selected by highlighting a few lines of full text from the original publication [27]. Support is provided by the cited Reference, which was retrieved and instantiated as an annotatable object when Domeo loaded the citing publication. Additional support is provided by the images of data shown, which was also loaded and instantiated on load of the containing publication. Original raw datasets, if cited and stored in a stable repository, may also provide support.