talk at iswc 2012 workshop on semantic technologies applied to biomedical informatics and...

36
Formalising Uncertainty: An Ontology of Reasoning, Certainty and A9ribu<on (ORCA) Anita de Waard Disrup<ve Technologies Director Elsevier Labs, Jericho, VT, USA Jodi Schneider PhD Researcher DERI, Galway, Ireland

Upload: anita-de-waard

Post on 10-May-2015

875 views

Category:

Documents


1 download

DESCRIPTION

Formalising Uncertainty: An Ontology of Reasoning, Certainty and Attribution (ORCA), November 12, 2012, Boston, MA

TRANSCRIPT

Page 1: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Formalising  Uncertainty:    An  Ontology  of  Reasoning,  

Certainty  and  A9ribu<on  (ORCA)  

Anita  de  Waard  Disrup<ve  Technologies  Director  Elsevier  Labs,  Jericho,  VT,  USA    

Jodi  Schneider  PhD  Researcher  

DERI,  Galway,  Ireland    

Page 2: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Outline  •  Background:    

– Metadiscourse,  epistemic  modality,  and  knowledge  a9ribu<on,  oh  my!  

–  Some  related  work:  genre  studies,  linguis<cs,  NLP  •  Our  model:  

– What  it  models  –  The  ontology  – How  can  we  find  this  in  text?  

•  Possible  applica<ons:    –  Possible  uses  – Next  steps  

Page 3: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Background  

Page 4: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Scien<sts  make  uncertain  claims  Uncertainty  

These  miRNAs  neutralize  p53-­‐mediated  CDK  inhibi;on,  possibly  through  direct  inhibi;on  of  the  expression  of  the  tumor-­‐suppressor  LATS2.    

Page 5: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

But  uncertainty  gets  lost  while  ci<ng  Uncertainty  

These  miRNAs  neutralize  p53-­‐mediated  CDK  inhibi;on,  possibly  through  direct  inhibi;on  of  the  expression  of  the  tumor-­‐suppressor  LATS2.    

Certainty  

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)  

Page 6: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Uncertainty  in  ac<on:  

•  Voorhoeve  et  al.,  2006:  “These  miRNAs  neutralize  p53-­‐  mediated  CDK  inhibi<on,  possibly  through  direct  inhibi<on  of  the  expression  of  the  tumor  suppressor  LATS2.”  

•  Kloosterman  and  Plasterk,  2006:  “In  a  gene<c  screen,  miR-­‐372  and  miR-­‐373  were  found  to  allow  prolifera<on  of  primary  human  cells  that  express  oncogenic  RAS  and  ac<ve  p53,  possibly  by  inhibi<ng  the  tumor  suppressor  LATS2  (Voorhoeve  et  al.,  2006).”  

•  Yabuta  et  al.,  2007:    “[On  the  other  hand,]  two  miRNAs,  miRNA-­‐372  and-­‐373,  func<on  as  poten6al  novel  oncogenes  in  tes<cular  germ  cell  tumors  by  inhibi<on  of  LATS2  expression,  which  suggests  that  Lats2  is  an  important  tumor  suppressor  (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  ..  fic<on  into  fact  just  by  adding  or  subtrac<ng  references”,  Bruno  Latour  [1]

Page 7: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Uncertainty  =  Hedging:  •  Why  do  authors  hedge?  

–  Make  a  claim  ‘pending  […]  acceptance  in  the  community’  [2]  –  ‘Create  A  Research  Space’  –  hedging  allows  authors  to  insert  themselves  into  

the  discourse  in  a  community  [3]  –  ‘the  strongest  claim  a  careful  researcher  can  make’  [4]  

•  Hedging  cues,  specula<ve  language,  modality/nega<on:  –  Light  et  al  [5]:  finding  specula<ve  language  –  Wilbur  et  al  [6]:  focus,  polarity,  certainty,  evidence,  and  direc<onality  –  Thompson  et  al  [7]:  level  of  specula<on,  type/source  of  the  evidence  and  

level  of  certainty      

•  Sen<ment  detec<on  (e.g.  Kim  and  Hovy  [8]  a.m.o.):    –  Holder  of  the  opinion,  strength,  polarity  as  ‘mathema<cal  func<on’  ac<ng  on  

main  proposi<onal  content    –  Wide  applica<ons  in  product  reviews;  but  not  (yet)  in  science!  

Page 8: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Our  Model  

Page 9: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Our  model  for  epistemic  evalua<ons:  

For  a  Proposi<on  P,  an  epistemically  marked  clause  E  is  an  evalua<on  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     Model  suggested  by  Eduard  Hovy,    

Informa;on  Sciences  Ins;tute  University  South  Califormia  

Page 10: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Adding  Epistemic  Evalua<on  

Together,  Lats2  and  ASPP1  shunt  p53  to  proapopto<c  promoters  and  promote  the  death  of  polyploid  cells  [1].  (…)    

Value  =  3  Source  =  N  Basis  =  0    

Further  biochemical  characteriza<on  of  hMOBs  showed  that    only  hMOB1A  and  hMOB1B  interact  with  both  LATS1  and  LATS2  in  vitro  and  in  vivo  [39].  (…)    

Value  =  3  Source  =  N  Basis  =  Data      

Our  findings  reveal  that  miR-­‐373  would  be  a  poten<al  oncogene  and  it  par<cipates  in  the  carcinogenesis  of  human  esophageal  cancer  by  suppressing  LATS2  expression.        

Value  =  1  Source  =  Author  Basis  =  Data      

Furthermore,  we  demonstrated  that  the  direct  inhibi<on  of  LATS2  protein  was  mediated  by  miR-­‐373  and  manipulated  the  expression  of  miR-­‐373  to  affect  esophageal  cancer  cells  growth.      

Value  =  2  (3?)  Source  =  Author  Basis  =  Data      

Page 11: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Finding  hedges  in  text  [9]:  •  Modal  auxiliary  verbs  (e.g.  can,  could,  might)    •  Qualifying  adverbs  and  adjec<ves  (e.g.  interes;ngly,  possibly,  likely,  poten;al,  somewhat,  slightly,  powerful,  unknown,  undefined)  

•  References,  either  external  (e.g.  ‘[Voorhoeve  et  al.,  2006]’)  or  internal  (e.g.  ‘See  fig.  2a’).    

•  Repor<ng/epistemic  verbs  (e.g.  suggest,  imply,  indicate,  show)    –  either  within  the  clause:  ‘These  results  suggest  that...’    –  or  in  a  subordinate  clause  governed  by  repor<ng-­‐verb  matrix  clause  ‘{These  results  suggest  that}  indeed,  this  represents  the  true  endogenous  ac;vity.’  

Page 12: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Manual  iden<fica<on:  Value   Modal  

Aux    Repor6ng  Verb  

Ruled  by  RV  

Adverbs/Adjec6ves  

References  

None   Total    

Total  value  =  3   1  (0.5%)   81  (40%)   24  (12%)   7  (4%)   41  (20%)   47  (24%)  201(100%)  

Total  Value  =  2   29  (51%)   23  (40%)   1  (2%)   4(7%)   57(100%)  

Total  Value  =  1   9(27%)   11(33%)   11(33%)   1(3%)   1(3%)   33(100%)  

Total  Value  =  0   9  (64%)   3  (21%)   1(7%)   1(7%)   14(100%)  

Total  No  Modality   16(37%)   3(7%)   0   3(7%)   22(50%)   44(100%)  

Overall  Total   10  (2%)   146(23%)   64(10%)   10(2%)   50(8%)   69(11%)  640(100%)  

Page 13: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Most  prevalent  clause  type:    “These  results  suggest  that...”  

Adverb/Connec<ve   thus,  therefore,  together,  recently,  in  summary    

Determiner/Pronoun     it,  this,  these,  we/our  

Adjec<ve   previous,  future,  beeer  

Noun  phrase   data,  report,  study,  result(s);  method  or  reference  

Modal   form  of    ‘to  be’,  may,  remain  

Adjec<ve   ogen,  recently,  generally  

Verb   show,  obtain,  consider,  view,  reveal,  suggest,  hypothesize,  indicate,  believe  

Preposi<on     that,  to  

Page 14: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Repor<ng  verbs  vs.  epistemic  value:  Value  =  0  (unknown)  

establish,  (remain  to  be)  elucidated,    be  (clear/useful),  (remain  to  be)  examined/determined,  describe,  make  difficult  to  infer,  report  

Value  =  1  (hypothe<cal)  

be  important,  consider,  expect,  hypothesize  (5x),  give  insight,  raise  possibility  that,  suspect,  think  

Value  =  2  (probable)  

appear,  believe,  implicate  (2x),  imply,  indicate  (12x),  play  a  role,  represent,  suggest  (18x),  validate  (2x),    

Value  =  3  (presumed  true)  

be  able/apparent/important  /posi<ve/visible,  compare  (2x),  confirm  (2x),  define,    demonstrate  (15x),  detect  (5x),  discover,  display  (3x),  eliminate,  find  (3x),  iden<fy  (4x),  know,  need,  note  (2x),  observe  (2x),  obtain  (success/results-­‐  3x),  prove  to  be,  refer,  report(2x),    reveal  (3x),  see(2x),  show(24x),    study,  view  

Page 15: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Finding  Claimed  Knowledge  Updates  [10]:  Defini<on:    1)  A  CKU  expresses  a  proposi<on  about  biological  en<<es    2)  A  CKU  is  a  new  proposi<on  3)  The  authors  present  the  CKU  as  factual:  =>  Strength  =  Certainty  4)  A  CKU  is  derived  from  experimental  work  described  in  the  ar<cle:  =>  Basis  =  Data  5)  The  ownership  is  a9ributed  to  the  author(s)  of  the  ar<cle.    =>  Source  =  Author,  Explicit  3),  4)  and  5)  are  either  explicitly  expressed  or  structurally  conveyed:  Here  we  used  mass  spectrometry  to  iden:fy  HuD  as  a  novel  SMN-­‐interac;ng  partner  

Our  analysis  of  known  HuD-­‐associated  mRNAs  iden:fied  cpg15  mRNA  as  a  highly  abundant  mRNA  in  HuD  Ips  

     

Page 16: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Automa<c  hedge  detec<on  with  The  Xerox  Incremental  Parser:  

Concept-­‐matching:  

Match  concept  pa9erns  with  rules  

Assign  features  to  keywords,  dependencies  and  sentences  

 

 

General  linguis<c  analysis  of  running  texts:  

Extract  syntac<c  dependencies  between  words  

Chunking  

Part-­‐of-­‐speech  disambigua<on  

Segment  the  sentences  into  words  

Segment  the  text  into  sentences  

Page 17: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Title Abstract Intro. Results Figures Discussion Citation

Interaction of survival of motor neuron (SMN) and HuD proteins [with m RNA cpg15rescues motor neuron axonal deficits]

Here we used mass spectrometry to identify HuD as a novel neuronal SMN-interacting partner.

Here we identify HuD as a novel interacting partner of SMN,

Together with our co-IP data, these results indicate that SMN associates with HuD in motor neurons.

SMN interacts with HuD.

Our MS and co-IP data demonstrate a strong interaction between SMN and HuD in spinal motor neuron axons.

Furthermore, these findings are consistent with recent studies demonstrating that the interaction of HuD with the spinal muscular atrophy (SMA) protein SMN …

Result:  CKUs  appear  throughout  the  paper    bio-event

interaction

entity 1 entity 2 location

HuD SMN motor neurons

event name

Page 18: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

The  Xerox  Incremental  Parser:  Concept-­‐matching:  

Match  concept  pa9erns  with  rules  

Assign  features  to  keywords,  dependencies  and  sentences  

 

 

General  linguis<c  analysis  of  running  texts:  

Extract  syntac<c  dependencies  between  words  

Chunking  

Part-­‐of-­‐speech  disambigua<on  

Segment  the  sentences  into  words  

Segment  the  text  into  sentences  

Page 19: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

The  formal  model  

©  Jodi  Schneider,    with  thanks  to  Siggi  Handschuh  

Page 20: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

orca  [11]    vocab.deri.ie/orca    

Page 21: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Example  Usage  

     <claim>  orca:hasBasis  orca:Data  .  

Page 22: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Basis  

Page 23: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Source  

Page 24: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

ConfidenceLevel  

Page 25: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

How  to  represent  the  hierarchy?  

lack  of  knowledge  <  hypothe;cal  knowledge    <  dubita;ve  knowledge  <  doxas;c  knowledge    

•  skos:broaderThan  –  not  appropriate  •  skos  Collec<ons  add  an  unwanted  layer  of  complexity.  

•  Our  approach:  transi<ve  proper<es  “lessCertain”  and  “moreCertain”  

Page 26: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Transi<ve  proper<es  used  for  ConfidenceLevel  

Page 27: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

ConfidenceLevel  &  its  Rela<onships  

Page 28: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Possible  Applica<ons  

Page 29: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Add  knowledge  value/basis/source    to  a  bio-­‐event  

 Biological  statement    with  epistemic  markup   Epistemic  evalua6on  

Our  findings  reveal  that  miR-­‐373  would  be  a  poten<al  oncogene  and  it  par<cipates  in  the  carcinogenesis  of  human  esophageal  cancer  by  suppressing  LATS2  expression.      

Value  =  Probable  Source  =  Author  Basis  =  Data      

Further  biochemical  characteriza<on  of  hMOBs  showed  that  only  hMOB1A  and  hMOB1B  interact  with  both  LATS1  and  LATS2  in  vitro  and  in  vivo  [39].  

Value  =  Presumed  true  Source  =  Reference  Basis  =  Data    

Moreover,  the  mechanisms  by  which  tumor  suppressor  genes  are  inhibited  may  vary  between  tumors.  

Value  =  Possible  Source  =  Unknown  Basis  =  Unknown  

Page 30: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

E.g.  to  augment  Medscan  [13]  Biological  statement  with  Medscan/epistemic  markup  

MedScan  Analysis:   Epistemic  evalua6on  

Furthermore,  we  present  evidence  that  the  secre;on  of  nesfa:n-­‐1  into  the  culture  media  was  drama<cally  increased  during  the  differen<a<on  of  3T3-­‐L1  preadipocytes  into  adipocytes  (P  <  0.001)  and  a{er  treatments  with  TNF-­‐alpha,  IL-­‐6,  insulin,  and  dexamethasone  (P  <  0.01).  

IL-­‐6  è  NUCB2  (nesfa;n-­‐1)  Rela<on:  MolTransport  Effect:  Posi<ve  CellType:  Adipocytes  Cell  Line:  3T3-­‐L1    

Value  =  Probable  Source  =  Author  Basis  =  Data      

Page 31: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Or  Biological  Exchange  Language  [14]:    Biological  statement  with  BEL/  epistemic  markup  

BEL  representa6on:   Epistemic  evalua6on  

These  miRNAs  neutralize  p53-­‐mediated  CDK  inhibi;on,  possibly  through  direct  inhibi;on  of  the  expression  of  the  tumor-­‐suppressor  LATS2.    

Increased  abundance  of  miR-­‐372  decreases:  Increased  ac;vity  of  TP53  decreases  ac;vity  of  CDK  protein  family  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    

Page 32: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Using  ORCA  for  Nanopublica<ons  [15]:  

•  Use  to  indicate  Strength,  Basis,  Source  of  Asser<ons:    

Knowledge  Strength,  Basis,  Source   Methods   Authors,  DOIs  

Page 33: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Next  steps:    

•  Con<nuing  experiments  with  automated  detec<on  

•  Can  be  used  in  Claim-­‐Evidence  network  projects,  e.g.  Data2Seman<cs  or  DIKB  

•  Could  replace  more  complicated  models  of  argumenta<on  

•  Ontology  is  available  for  all  to  use!    

Page 34: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Thank  you!  •  Funding:    

–  Elsevier  Labs  –  NWO  Casimir  programme  

•  Collaborators:    –  Henk  Pander  Maat,  UU  –  Agnes  Sandor,  XRCE  –  Siegfried  Handshuh,  DERI  –  Rinke  Hoekstra  &  co,  VU  –  Richard  Boyce  &  co,  UPi9  – Maria  Liakata,  EBI  –  Sophia  Ananiadou  &  co,  NaCTeM  

 

•  Discussion  partners:    –  Phil  Bourne,  UCSD  –  Ed  Hovy,    –  Gully  Burns,  ISI  –  Joanne  Luciano,  RPI  –  Tim  Clark  et  al.,  Harvard  

Page 35: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

Ques<ons?      

Anita  de  Waard  [email protected]  

h9p://elsatglabs.com/labs/anita/      

Jodi  Schneider  [email protected]    

h9p://jodischneider.com/jodi.html      

Page 36: Talk at ISWC 2012 Workshop on Semantic Technologies Applied to Biomedical Informatics and Individualized Medicine (SATBI+SWIM 2012)

References  [1]  Latour,  B.  and  Woolgar,  S.,  Laboratory  Life:  the  Social  Construc<on  of  Scien<fic  Facts,  1979,  Sage    [2]  Myers,  G.  (1992).  ‘In  this  paper  we  report’:  Speech  acts  and  scien<fic  facts,  Jnl  of  Pragmatlcs  17  (1992)  295-­‐313  

[3]  Swales,  J.  (1990).  Genre  Analysis,  English  in  Acad.  and  Res.Se}ngs,  Cambridge  University  Press,  1990.    [4]  Salager-­‐Meyer,  F.  (1994),  Hedges  and  Textual  Communica<ve  Func<on  in  Medical  English  Wri9en  Discourse,  English  for  Specific  Purposes,  Vol.  13,  No.  2,  pp.  149-­‐170,  1994.    [5]  Light  M,  Qiu  XY,  Srinivasan  P.  (2004).  The  language  of  bioscience:  facts,  specula<ons,  and  statements  in  between.  BioLINK  2004:  Linking  Biological  Literature,  Ontologies  and  Databases  2004:17-­‐24.  [6]  Wilbur  WJ,  Rzhetsky  A,  Shatkay  H  (2006).  New  direc<ons  in  biomedical  text  annota<ons:  defini<ons,  guidelines  and  corpus  construc<on.  BMC  Bioinforma<cs  2006,  7:356.  [7]  Thompson  P.,  Venturi  G.  et  al.  (2008).  Categorising  modality  in  biomedical  texts.  Proc.  LREC  2008  Wkshp  Building  and  Evalua<ng  Resources  for  Biomedical  Text  Mining  2008.  [8]  Kim,  S-­‐M.  Hovy,  E.H.  (2004).  Determining  the  Sen<ment  of  Opinions,COLING  conference,  Geneva,  2004.    [9]    de  Waard,  A.  and  Pander  Maat,  H.  (2012).  Epistemic  Modality  and  Knowledge  A9ribu<on  in  Scien<fic  Discourse:  A  Taxonomy  of  Types  and  Overview  of  Features.  Workshop  on  Detec<ng  Structure  in  Scholarly  Discourse,  ACL  2012.    [10]  Sándor,  À.  and  de  Waard,  A.,  (2012).  Iden<fying  Claimed  Knowledge  Updates  in  Biomedical  Research  Ar<cles,  Workshop  on  Detec<ng  Structure  in  Scholarly  Discourse,  ACL  2012.    [11]  de  Waard,  A.  and  Schneider,  J.  (2012)  Formalising  Uncertainty:  An  Ontology  of  Reasoning,  Certainty  and  A9ribu<on  (ORCA),  SATBI+SWIM,  ISWC  2012.    [12]  Medscan  [13]  Biological  Expression  Language  –  h9p://www.openbel.org    [14]  Groth  et  al  (2010)  'The  anatomy  of  a  nanopublica<on'  Informa<on  Services  &  Use  30:51-­‐6