Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


My work

A machine learning approach to negation and speculation detection in biomedical and review texts

Noa Cruz Diaz

on 18 January 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of My work

And Finally A Machine Learning Approach to Negation and Speculation Detection in Biomedical and Review Texts Why, how... Noa P. Cruz Díaz Negation in biomedical texts

Sentiment analysis and opinion mining SFU review corpus site http://www.sfu.ca/~mtaboada/research/SFU_Review_Corpus.html 400 documents (50 of each type)

Annotation in a similar way as in BioScope corpus What does it consist of? Movies, books and consumer product reviews from 'epinions.com'

In total, 17,263 sentences and 303,289 words Oslo '12 Negation Speculation Agreement analysis 10% annotated by another expert High value of Kappa [>0.81 (almost perfect)] 18% (cc) image by nuonsolarteam on Flickr
University of Huelva (Spain) 1 2 6 7 3 4 5 8 First annotator annotated 20%
adapting BioScope guidelines
annotated 100% Second annotator annotated 10% selected randomly gold standard inter-annotator agreement
guidelines were corrected Annotation process Negated sentences
Speculative sentences 22% Journal of the American Society for Information Science and Technology
Cruz et al., 2012 Research Group for Language Technology BioScope corpus
"A review corpus annotated for negation, speculation and their scope" LREC Conference. Konstantinova et al., 2012 Results Agreement Disagreement More details in paper Corpus corrected
after discussion Automatic negation and speculation detection systems Where to use it? Biomedical domain ... small (cc) photo by medhead on Flickr BioScope corpus "A machine learning approach to negation and speculation detection in clinical documents" System Architecture Two consecutive tasks Cue detection

Scope detection Patient may have continued vesicoureteral reflux. Patient may have continued vesicoureteral reflux. Clinical documents (BioScope corpus) Classification Algorithms Naïve Bayes
C4.5 SVM Kernels:
Linear, polynomial, RBF and sigmoid Text collection 1954 documents
Negation sentences 13.55%
Speculation sentences 13.39% Attributes Scope detection Token Context Lemma
Other tags
Class Lemma
Other tags Cue detection Cue Token Tokens between

Others Lemma
Neg/Esp Lemma

Class Distance (#tokens)
Chain of POS Results Negation Precision Recall F-score F-score Recall Precision Speculation Cue detection Scope detection Gold-Standard cues Predicted cues Negation Speculation Speculation Negation Conclusion There is much to do Biomedical domain Sentiment analysis Improve results in papers and abstracts Develop systems to automatic negation detection 2 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Beginning negation cue Outside Outside Outside Outside Outside Inside scope cue Outside
Full transcript