Talk about reasoning in biological discourse for Dagstuhl Symposium on Debating Technology, December 14, 2015

1. (Rhetoric, bias, and reasoning in Scientific)
Argumentation in Biology Papers
Anita de Waard
VP Research Data Collaborations, Elsevier
Jericho, Vermont, USA

2. Caveat; Thank You
Physicist, publisher; goal since start of web (’93):
How can we improve scholarly communication,
using information technologies?
– Avoid cognitive load on scientists: reduce time to
process “Soul-crrrushingly boring” articles
– Offer better overview of progress in a (new) area of
science
– Allow for more accurate citation practices so
Chinese whispers are reduced
NWO (2006-2010) U Utrecht: “Semantic articles”
Since 2008: power of NLP (thank you, Ed!) > here.

3. • An exchange of diverging or opposite
views, typically a heated or angry one
• A discussion involving differing points
of view; debate
• An address or composition intended to
convince or persuade
• A reason or set of reasons given with
the aim of persuading others that an
action or idea is right or wrong
• A process of reasoning; series of
reasons
1.A communica-
tion: Either a
one-way
presentation
or a two-way
discussion
2.A cognition:
The reasoning
(process or the
result of that
process) that
leads to a
conclusion
Argumentation in Biology is…
© Ed Hovy

4. • What is an opinion?
How can they conflict?
• What is the basic
argumentation unit?
(Is it a claim?) Can we
represent it simply
enough to be workable?
• How do the units
compose, to form an
argument? (By what
reasoning, etc.?)
“We Need to Study:”
© Ed Hovy
• Claim, defined by an
author; can be supported
or not supported*
• Claims, backed by
(experimental) evidence
and a shared worldview!
• Document overall;
Citation networks
(*) Elizabeth White’s Thesis
Scientific text:

5. Three Levels of
Experimental Argumentation:
I. Macrostructure: Stories, that Persuade
with Data
II. Microstructure: Cycles of Scientific
Investigation
III. Diachronic structure: Claim-Evidence
Networks

6. Three Levels of
Experimental Argumentation:
I. Macrostructure: Stories, that
Persuade with Data
II. Microstructure: Cycles of Scientific
Investigation
III. Diachronic structure: Claim-Evidence
Networks

7. Macrostructure:
From General to Specific to General
Discussion:
• Statement of principal findings
• Strengths and weaknesses of the study
• Relation to other studies
• Unanswered questions and future research
Introduction: “Create a Research Space”
• Establish a research territory
• Establish a niche
• Occupy the niche
Methods and Results:
“Cycles of Scientific Investigation”
(see part II)

8. Story Grammar The Story of Goldilocks and
the Three Bears
Setting Time Once upon a time
Character a little girl named Goldilocks
Location She went for a walk in the forest.
Pretty soon, she came upon a
house.
Theme Goal She knocked and, when no one
answered,
Attempt she walked right in.
Episode Name At the table in the kitchen, there
were three bowls of porridge.
Subgoal Goldilocks was hungry.
Attempt She tasted the porridge from the
first bowl.
Outcome This porridge is too hot! she
exclaimed.
Attempt So, she tasted the porridge from
the second bowl.
Outcome This porridge is too cold, she
said
Attempt So, she tasted the last bowl of
porridge.
Paper
Grammar
The AXH Domain of Ataxin-1 Mediates Neurodegeneration
through Its Interaction with Gfi-1/Senseless Proteins
Background The mechanisms mediating SCA1 pathogenesis are still not fully
understood, but some general principles have emerged.
Objects of
study
the Drosophila Atx-1 homolog (dAtx-1) which lacks a polyQ tract,
Experimental
setup
studied and compared in vivo effects and interactions to those of
the human protein
Research
goal
Gain insight into how Atx-1's function contributes to SCA1
pathogenesis. How these interactions might contribute to the
disease process and how they might cause toxicity in only a
subset of neurons in SCA1 is not fully understood.
Hypothesis Atx-1 may play a role in the regulation of gene expression
Name dAtX-1 and hAtx-1 Induce Similar Phenotypes When
Overexpressed in Files
Subgoal test the function of the AXH domain
Method overexpressed dAtx-1 in flies using the GAL4/UAS system
(Brand and Perrimon, 1993) and compared its effects to those
of hAtx-1.
Results Overexpression of dAtx-1 by Rhodopsin1(Rh1)-GAL4, which
drives expression in the differentiated R1-R6 photoreceptor cells
(Mollereau et al., 2000 and O'Tousa et al., 1985), results in
neurodegeneration in the eye, as does overexpression of hAtx-
1[82Q]. Although at 2 days after eclosion, overexpression of
either Atx-1 does not show obvious morphological changes in
the photoreceptor cells
Data (data not shown),
Results both genotypes show many large holes and loss of cell integrity
Experimental papers are stories…

9. Aristotle Quintilian Scientific Paper
prooimion
Introduction
/ exordium
The introduction of a speech, where one announces the subject
and purpose of the discourse, and where one usually employs the
persuasive appeal to ethos in order to establish credibility with
the audience.
Introduction:
positioning
prothesis
Statement of
Facts/narrati
o
The speaker here provides a narrative account of what has
happened and generally explains the nature of the case.
Introduction:
research question
Summary/
propostitio
The propositio provides a brief summary of what one is about to
speak on, or concisely puts forth the charges or accusation.
Introduction:
Summary of contents
pistis
Proof/
confirmatio
The main body of the speech where one offers logical arguments
as proof. The appeal to logos is emphasized here.
Results
Refutation/
refutatio
As the name connotes, this section of a speech was devoted to
answering the counterarguments of one's opponent.
Related Work
epilogos peroratio
Following the refutatio and concluding the classical oration, the
peroratio conventionally employed appeals through pathos, and
often included a summing up.
Discussion: summary,
implications.
… that persuade…
• Format has co-evolved: predator-prey relationship with reviewers
• Goal of the paper is to be published; it uses author/journal as a host

10. … with data:
10

11. Three Levels of
Experimental Argumentation:
I. Macrostructure: Stories, that Persuade
with Data
II. Microstructure: Cycles of Scientific
Investigation
III. Diachronic structure: Claim-Evidence
Networks

12. Cycles of Scientific Investigation (CoSI):
© Gully Burns, 2011

13. CoSI in action in a paragraph:
© Gully Burns, 2011
3. We used for this experiment BJ/ET cells containing
p14ARFkd because, following RASV12 treatment, in
those cells p53 is still activated but more clearly
stabilized than in parental BJ/ET cells (Voorhoeve and
Agami, 2003), resulting in a sensitized system for slight
alterations in p53 in response to RASV12.
1. Importantly, our results so far indicate that the
expression of miR-372&3 did not reduce the activity
of RASV12, as these cells were still growing faster
than normal cells and were tumorigenic, for which
RAS activity is indispensable (Hahn et al, 1999 and
Kolfschoten et al, 2005).
2. To shed more light on this aspect, we
examined the effect of miR-372&3 expression on
p53 activation in response to oncogenic
stimulation.
4. Figure 4A shows that following RASV12
stimulation, p53 was stabilized and activated, and its
target gene, p21cip1, was induced in all cases,
indicating an intact p53 pathway in these cells.

14. Who Why What
Teufel: Argumentative
zones
Liakata: CoreSC
Identify main components of
scientific investigation for
machine learning
Sentence
Ananiadou:
MetaKnowledge/BioE
vents
Enhance information extraction
for biomedical texts to enable
metadiscourse annotation
Events: can be several
per sentence, or one in
more sentences
de Waard:
Discourse Segment
Types
Identify mechanisms of
conveying (epistemic)
knowledge in scientific
discourse
Clause
A few models of
scholarly discourse annotation:

15. In defense of the clause
as the unit of thought:
1. Importantly, our results so far indicate that the expression of miR-
372&3 did not reduce the activity of RASV12, as these cells were
still growing faster than normal cells and were tumorigenic, for
which RAS activity is indispensable (Hahn et al, 1999 and
Kolfschoten et al, 2005).
2. To shed more light on this aspect, we examined the effect of miR-
372&3 expression on p53 activation in response to oncogenic
stimulation.
3. We used for this experiment BJ/ET cells containing p14ARFkd
because, following RASV12 treatment, in those cells p53 is still
activated but more clearly stabilized than in parental BJ/ET cells
(Voorhoeve and Agami, 2003), resulting in a sensitized system for
slight alterations in p53 in response to RASV12.
4. Figure 4A shows that following RASV12 stimulation, p53 was
stabilized and activated, and its target gene, p21cip1, was induced
in all cases, indicating an intact p53 pathway in these cells.
• More than one ‘thought unit’ per sentence.
• Verb tense changes within sentence (several times).
• Attribution, actions/states, and prepositions all contained within a sentence.

16. In defense of the clause
as the unit of thought:
1. Importantly, our results so far indicate that the expression of miR-
372&3 did not reduce the activity of RASV12, as these cells were
still growing faster than normal cells and were tumorigenic, for
which RAS activity is indispensable (Hahn et al, 1999 and
Kolfschoten et al, 2005).
2. To shed more light on this aspect, we examined the effect of miR-
372&3 expression on p53 activation in response to oncogenic
stimulation.
3. We used for this experiment BJ/ET cells containing p14ARFkd
because, following RASV12 treatment, in those cells p53 is still
activated but more clearly stabilized than in parental BJ/ET cells
(Voorhoeve and Agami, 2003), resulting in a sensitized system for
slight alterations in p53 in response to RASV12.
4. Figure 4A shows that following RASV12 stimulation, p53 was
stabilized and activated, and its target gene, p21cip1, was induced
in all cases, indicating an intact p53 pathway in these cells.
Head: premise, motivation,
attribution (matrix clause)
Middle: main
biological statement
End: interpretation, elaboration,
attribution (reference)

17. In defense of the clause
as the unit of thought:
1. Importantly, our results so far indicate that the expression of miR-
372&3 did not reduce the activity of RASV12, as these cells were
still growing faster than normal cells and were tumorigenic, for
which RAS activity is indispensable (Hahn et al, 1999 and
Kolfschoten et al, 2005).
2. To shed more light on this aspect, we examined the effect of miR-
372&3 expression on p53 activation in response to oncogenic
stimulation.
3. We used for this experiment BJ/ET cells containing p14ARFkd
because, following RASV12 treatment, in those cells p53 is still
activated but more clearly stabilized than in parental BJ/ET cells
(Voorhoeve and Agami, 2003), resulting in a sensitized system for
slight alterations in p53 in response to RASV12.
4. Figure 4A shows that following RASV12 stimulation, p53 was
stabilized and activated, and its target gene, p21cip1, was induced
in all cases, indicating an intact p53 pathway in these cells.
Regulatory
clause
Fact Goal Method Result Implication

18. Concepts, models, ‘facts’: Present tense
Experiment: Past tense
Transitions: non-finite tense
(1) Both seminomas
and the EC component
of nonseminomas
share features with ES
cells.
(2) b. the detection of
miR-371-3 merely
reflects its expression
pattern in ES cells,
Fact Problem
(3) c. miR-371-3
expression is a
selective event
during
tumorigenesis.
Implication
(2) a. To exclude that
Goal
(3) b. suggesting that
Regulatory-Implication
(2) c. we tested by RPA
miR-302a-d, another ES
cells-specific miRNA cluster
(Suh et al, 2004).
(3) a. In many of the miR-371-3
expressing seminomas and
nonseminomas, miR-302a-d
was undetectable (Figs S7 and
S8),
Method Result
Clause, realm and tense:

19. Facts in the
eternal present
Endogenous small RNAs (miRNAs) regulate
gene expression by mechanisms conserved
across metazoans.
Events in the
simple past
Vehicle-treated animals spent equivalent
time investigating a juvenile in the first and
second sessions in experiments conducted in
the NAC and the striatum: T1 values were
122 ± 6 s and 114 ± 5 s.
Events with
embedded
facts
We also generated BJ/ET cells expressing the
RASV12-ERTAM chimera gene, which is only
active when tamoxifen is added (De Vita et al,
2005).
Attribution in
the present
perfect
miRNAs have emerged as important
regulators of development and control
processes such as cell fate determination and
cell death (Abrahante et al., 2003, Brennecke
et al., 2003, Chang et al., 2004, Chen et al.,
2004, Johnston and Hobert, 2003, Lee et al.,
1993]
Implications
are hedged,
and in the
present tense
These results indicate that although miR-
372&3 confer complete protection to
oncogene-induced senescence in a manner
similar to p53 inactivation, the cellular
response to DNA damage remains intact
Tense in science and mythology
I sing of golden-throned Hera whom Rhea bare.
Queen of the immortals is she, surpassing all in
beauty: she is the sister and the wife of loud-
thundering Zeus, --the glorious one whom all the
blessed throughout high Olympus reverence and
honor.
Now the wooers turned to the dance and to
gladsome song, and made them merry, and waited
till evening should come; and as they made merry
dark evening came upon them.
And she took her mighty spear, tipped with sharp
bronze, heavy and huge and strong, wherewith she
vanquishes the ranks of men-of warriors, with
whom she is wroth, she, the daughter of the
mighty sire.
In this book I have had old stories written down, as
I have heard them told by intelligent people,
concerning chiefs who have held dominion in the
northern countries, and who spoke the Danish
tongue; and also concerning some of their family
branches, according to what has been told me.
Now it is said that ever since then whenever the
camel sees a place where ashes have been
scattered, he wants to get revenge with his enemy
the rat and stomps and rolls in the ashes hoping to
get the rat

20. Facts
Hypothesis
Problem
Method
Results
Implication
Goal These results suggest that....Transitions
Experiment
Concepts
Three-Realm Model Mapped to CoSI:

21. Big Mechanism: Detecting CoSI Models:
Train CRF and Structured SVM models with:
● observed variables: clauses
● latent variables: discourse types
● chains: passages
Segment Type SVM F1
(Clause)
SSVM F1
(Passage)
CRF F1
(Passage)
Result 0.7332 0.7544 0.7929
Method 0.6984 0.7111 0.7188
Fact 0.3373 0.3810 0.4941
Implication 0.5741 0.6364 0.6275
Hypothesis 0.4528 0.5517 0.4364
Goal 0.6429 0.5806 0.6780
Problem 0.0000 0.0000 0.0000
Overall 0.5987
0.6572 0.6845
● Training data:
○ 179 passages
○ 1828 clauses
● Test data:
○ 44 passages
○ 449 clauses
● Features
○ POS tags
○ Identity of verb and
adverb
○ Presence of figure
pointers and citations
○ Manually created lexicons
MLP with no feature engineering for
classifying clauses gets an F-score of
0.50
Pradeep Dasigi, Ed Hovy, CMU

22. Issues with the CoSI Cycle:
What conclusions
can you draw from
this experiment?
How can you test
these hypotheses?
Need models of/relating
Domain Reasoning to
Experimental Design!
(cf Stede, coherence)

23. Three Levels of
Experimental Argumentation:
I. Macrostructure: Stories, that Persuade
with Data
II. Microstructure: Cycles of Scientific
Investigation
III. Diachronic structure: Claim-Evidence
Networks

24. Some (Early) Experiments With
Argumentation Modeling in Science:

25. Hedging And Citation
• Voorhoeve et al., 2006: “These miRNAs neutralize p53- mediated CDK inhibition,
possibly through direct inhibition of the expression of the tumor suppressor
LATS2.”
• Kloosterman and Plasterk, 2006: “In a genetic screen, miR-372 and miR-373
were found to allow proliferation of primary human cells that express
oncogenic RAS and active p53, possibly by inhibiting the tumor suppressor
LATS2 (Voorhoeve et al., 2006).”
• Okada et al., 2011: “Two oncogenic miRNAs, miR-372 and miR-373, directly
inhibit the expression of Lats2, thereby allowing tumorigenic growth in the
presence of p53 (Voorhoeve et al., 2006).”
“[Y]ou can transform .. fiction into fact just by adding or
subtracting references”, Bruno Latour, Science in Action

26. Hedging and Segment Types:

27. Voorhoeve et al. (116) employed a novel
strategy by combining an miRNA vector library
and corresponding bar code array…
miR-372 and miR-373 were consequently found
to permit proliferation and tumorigenesis of
these primary cells carrying both oncogenic RAS
and wild-type p53,
probably through direct inhibition of the
expression of the tumor-suppressor LATS2 and
subsequent neutralization of the p53 pathway.
to identify miRNAs that when overexpressed
could substitute for p53 loss and allow continued
proliferation in the context of Ras activation
TAC Corpus:
Curated Collection of 500 Citing > 50 Cited Papers
Voorhoeve et al. (2006), A Genetic
Screen …
In mammals, a near-perfect complementarity between
miRNAs and protein coding genes almost never exists, making
it difficult to directly pinpoint relevant downstream targets of
a miRNA. Several algorithms were developed that predict
miRNA targets, most notably TargetScanS, PicTar, and
miRanda (John et al., 2004, Lewis et al., 2005 and Robins et al.,
2005).
These programs predict dozens to hundreds of target genes
per miRNA, making it difficult to directly infer the cellular
pathways affected by a given miRNA. Furthermore, the
biological effect of the downregulation depends greatly on the
cellular context, which exemplifies the need to deduce miRNA
functions by in vivo genetic screens in well-defined model
systems.
The cancerous process can be modeled by in vitro neoplastic
transformation assays in primary human cells (Hahn et al.,
1999). Using this system, sets of genetic elements required for
transformation were identified. For example, the joint
expression of the telomerase reverse transcriptase subunit
(hTERT),
oncogenic H-RASV12, and SV40-small t antigen combined with
the suppression of p53 and p16INK4A were sufficient to
render primary human fibroblasts tumorigenic (Voorhoeve
and Agami, 2003).
Goal
Method
Result
Con
clusion
Citing PapersReference Paper

28. Hedging:
• Hedging: ‘the strongest claim a careful researcher
can make’:
– Authors need to downplay claims to be acceptable in
the literature
– ‘Create A Research Space’ – hedging allows authors to
‘insert themselves into the discourse in a community’
• Hedging cues, speculative language,
modality/negation:
– Light et al: finding speculative language
– Wilbur/Shatkay: focus, polarity, certainty, evidence, and
directionality
– Thompson/Ananiadou: level of speculation, type/source of the
evidence and level of certainty

29. A Model of Hedging (=Epistemic Markup):
For a Proposition P, an epistemically marked clause
E is an evaluation of P, where EV, B, S(P), with:
– V = Value:
3 = Assumed true, 2 = Probable, 1 = Possible, 0 = Unknown,
(- 1= possibly untrue, - 2 = probably untrue, -3 = assumed untrue)
– B = Basis:
Reasoning
Data
– S = Source:
A = speaker is author A, explicit
IA = speaker author, A, implicit
N = other author N, explicit
NN = other author NN, implicit
(Ed’s model!)
= “Modal Qualifier”

30. Adding metadiscourse to triples:
Biological statement with BEL/ epistemic
markup
BEL representation: Epistemic
evaluation
These miRNAs neutralize p53-mediated CDK
inhibition, possibly through direct inhibition
of the expression of the tumor-suppressor
LATS2.
r(MIR:miR-372) -
|(tscript(p(HUGO:Trp53)) -|
kin(p(PFH:”CDK Family”)))
Increased abundance of miR-
372 decreases abundance of
LATS2
r(MIR:miR-372) -|
r(HUGO:LATS2)
Value =
Possible
Source =
Unknown
Basis =
Unknown
Biological statement with
Medscan/epistemic markup
MedScan Analysis: Epistemic
evaluation
Furthermore, we present evidence that the
secretion of nesfatin-1 into the culture
media was dramatically increased during the
differentiation of 3T3-L1 preadipocytes into
adipocytes (P < 0.001) and after treatments
with TNF-alpha, IL-6, insulin, and
dexamethasone (P < 0.01).
IL-6  NUCB2 (nesfatin-1)
Relation: MolTransport
Effect: Positive
CellType: Adipocytes
Cell Line: 3T3-L1
Value =
Probable
Source =
Author
Basis = Data

31. Three Levels of
Experimental Argumentation:
I. Macrostructure: Stories, that Persuade
with Data
II. Microstructure: Cycles of Scientific
Investigation
III. Diachronic structure: Claim-Evidence
Networks

32. Putting It All Together:
Introduction
Methods and
Results
Conclusion
Context
Interpretation
Experiment
Experiment
Experiment
Context
Interpretation
Experiment
Experiment
Experiment
Introduction
Methods and
Results
Conclusion
Context
Interpretation
Experiment
Experiment
Experiment
Context
Interpretation
Experiment
Experiment
Experiment

33. 3333
Ontologies: HCLS Ontology of Rhetorical Blocks
http://www.w3.org/TR/hcls-orb/
orb:Introduction:
The section describing why was the study in the publication undertaken, what was
the tested hypothesis or what was the purpose of the research. It lays down the
rationale behind the existence of the publication.
orb:Methods:
The section describing when, where, and how was the study done. It includes the
materials used as part of the study and who was included in the study groups
(patients, etc.).
orb:Results:
The part synthesizing the results of the study presented in the paper.
orb:Discussion:
The section analyzing whether the tested hypothesis was confirmed. It also
interprets the results to understand their consequences and importance. And
finally, it shows how the approach and results fit to what other researchers in the
field have discovered, including possible perspectives of future research.

34. Ontology for Reasoning,
Certainty and Attribution
vocab.deri.ie/orca

35. Three Levels of
Experimental Argumentation:
I. Macrostructure: Stories, that Persuade with Data
II. Microstructure: Cycles of Scientific Investigation
III. Diachronic structure: Claim-Evidence Networks
Identifying contradictions, controversies and
counterarguments??
– Need to understand the knowledge structure these
arguments are based on: which experiments are valid?
What does ‘as a control’ mean? What other explanations
would explain these results?
– Need to find ‘rebuttal by omission’, recognize schools of
thought, subtle pragmatics, ‘family trees’
– How does the background of the person you are talking to
influence the best form of argumentation?

36. Some work that would benefit science:
1. Macrostructure: Parse (or at least define) main
Rhetorical Blocks in papers to identify
macrostructures
2. Microstructure: Find and read ‘cycles of scientific
investigation’ inside Results sections
3. Diachronic structure: Identify ‘contextual citations’:
experimental context accompany a claim
4. We still need good, working models and methods
for argumentation!
How to start?
- Some ontologies and corpora are available for 1-3
- Maybe this workshop can be a start for 4?
Thank you! a.dewaard@elsevier.com

37. References (1/2):
• Swales and Feak, 2000 (CARS Model, English in Today’s Research World: a writing guide
• Docherty, M (BMJ 1999) The case for structuring the discussion of scientific papers (summary of
discussion roles):s http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1115625/
• Latour, B. and Woolgar, S., Laboratory Life: the Social Construction of Scientific Facts, 1979, Sage
Publications
• Latour, B. (1987). Science in Action, How to Follow Scientists and Engineers through Society,
(Cambridge, Ma.: Harvard University Press, 1987)
• Teufel. S. (2005). Argumentative Zoning for improved citation indexing. In ``Computing Attitude and
Affect in Text: Theory and Applications'' James G. Shanahan, Yan Qu, Janyce Wiebe (Eds.) Springer,
Dordrecht, The Netherlands, 2005. Pp 159-170.
• Liakata, M. (2010) Zones of conceptualisation in scientific papers: a window to negative and
speculative statements. In Proceedings of the Workshop on Negation and Speculation in Natural
Language Processing (Uppsala, Sweden, July 10 - 10, 2010). R. Morante and C. Sporleder, Eds. ACL
Workshops. ACL
• Waard, A. de, From Proteins to Fairytales: Directions in Semantic Publishing. IEEE Intel Systems 25(2):
83-88 (2010)
• Buckingham Shum, SJ.; Uren, V; Li, Gangmin; Sereno, B and Mancini, Clara (2007). Modelling
naturalistic argumentation in research literatures: representation and interaction design issues.
http://oro.open.ac.uk/6463/1/
• Visualizing Argumentation: Software Tools for Collaborative and Educational edited by Paul A.
Kirschner, S J. Buckingham Shum, Springer Verlag (2003)
• Groza, T., Handschuh, S., Decker, S. (2011) Capturing Rhetoric and Argumentation Aspects within
Scientific Publications. In Journal of Data Semantics XV, 2011 (to appear)

38. References (2/2)
• Harmsze, F.A.P., van der Tol, M..C. and Kircz, J.G. 1999. A modular structure for electronic scientific
articles. Conf. Informatiewetenschap 1999. CWI, Amsterdam, 1999. In: P. de Bra and L. Hardman
(eds). TUE Report 99-20, 2-9.
• Ciccarese, P., Wu, E., Wong,G. Ocana, M. Kinoshita, J., Ruttenberg, A. Clark, T. (2008) The SWAN
biomedical discourse ontology, JBiomed Inf. 2008 Oct;41(5):739-51.
• Light M, Qiu XY, Srinivasan P. (2004). The language of bioscience: facts, speculations, and
statements in between. BioLINK 2004: Linking Biological Literature, Ontologies and Databases
2004:17-24.
• Wilbur WJ, Rzhetsky A, Shatkay H (2006). New directions in biomedical text annotations:
definitions, guidelines and corpus construction. BMC Bioinformatics 2006, 7:356.
• Thompson P., Venturi G., McNaught J, Montemagni S, Ananiadou S. (2008). Categorising modality in
biomedical texts. Proc. LREC 2008 Wkshp Building and Evaluating Resources for Biomedical Text
Mining 2008.
• de Waard, A and Pander Maat, H.P.M. 2012. Epistemic modality and knowledge attribution in
scientific discourse: A taxonomy of types and overview of features. In Proceedings of the Workshop
on Detecting Structure in Scholarly Discourse, ACL 2010 47–55. Association for Computational
Linguistics
• Ciccarese, Groza, Clark, de Waard, W3C Interest Group 20/10/2011 Ontology of Rhetorical Blocks:
http://www.w3.org/TR/hcls-orb/
• de Waard, A. and Schneider, J. (2012) An Ontology of Reasoning, Certainty and Attribution (ORCA),
ISWC 2012, http://ceur-ws.org/Vol-930/p2.pdf