Ion Androutsopoulos, February 2014 NLP Group, Department of Informatics, Athens University of...

Ion Androutsopoulos, February 2014 www.bioasq.org

NLP Group, Department of Informatics, Athens University of Economics and Business

Ion Androutsopoulos

BioASQ

Intelligent Information Management Targeted Competition Framework ICT-2011.4.4(d)

Ion Androutsopoulos, February 2014

A challenge on large-scale biomedical semantic indexing and question answering

www.bioasq.org


Biomedical articles per year

2/29

• In 2011: approx. 20 000 new PubMed articles per week!


Questions of biomedical experts

• Yes/no question: “Are there any DNMT3 proteins present in plants?”– Exact answer: “Yes”– Ideal answer: “Yes. The plant DOMAINS

REARRANGED METHYLTRANSFERASE2 (DRM2) is a homolog of the mammalian de novo methyltransferase DNMT3. DRM2 contains a novel arrangement of the motifs required for DNA methyltransferase catalytic activity.”

3/29


Questions of biomedical experts (II)

• Factoid question: “What is the methyl donor of DNA (cytosine-5)-methyltransferases?”– Exact answer: “S-adenosyl-L-methionine”– Ideal answer: “S-adenosyl-L-methionine (AdoMet, SAM)

is the methyl donor of DNA (cytosine-5)-methyltransferases. DNA (cytosine-5)-methyltransferases catalyze the transfer of a methyl group from S-adenosyl-L-methionine to the C-5 position of cytosine residues in DNA.”

4/29


Questions of biomedical experts (III)• List question: “Which species may be used for the

biotechnological production of itaconic acid?”– Exact answer: { “Aspergillus terreus”, “Aspergillus

niger”, “Ustilago maydis” }– Ideal answer: “In 1955, the production of itaconic acid

was firstly described for Ustilago maydis. Some Aspergillus species, like A. itaconicus and A. terreus, show the ability to synthesize this organic acid and A. terreus can secrete significant amounts to the media. Itaconic acid is mainly supplied by biotechnological processes with the fungus Aspergillus terreus. Cloning of the cadA gene into the citric acid producing fungus A. niger showed that it is possible to produce itaconic acid also in a different host organism.”

5/29


Questions of biomedical experts (IV)• Summary question: “How do histone

methyltransferases cause histone modification?”– Exact answer: none– Ideal answer: “Histone methyltransferases (HMTs) are

responsible for the site-specific addition of covalent modifications on the histone tails, which serve as markers for the recruitment of chromatin organization complexes. There are two major types of HMTs: histone-lysine N-Methyltransferases and histone-arginine N-methyltransferases. The former methylate specific lysine (K) residues such as 4, 9, 27, 36, and 79 on histone H3 and residue 20 on histone H4. The latter methylate arginine (R) residues such as 2, 8, 17, and 26 on histone H3 and residue 3 on histone H4. Depending on what residue is modified and the degree of methylation (mono-, di- and tri-methylation), lysine methylation of histones is linked to either transcriptionally active or silent chromatin.”

6/29

Ion Androutsopoulos, February 2014 www.bioasq.org 7/6


Finding relevant snippets

8/29


Not only texts: ontologies, linked data, …

9/29


Information from structured dataList question: “Which forms of cancer is the Tpl2 gene associated with?”• Related RDF triple:

– Subject: http://www4.wiwiss.fu-berlin.de/diseasome/resource/diseases/3003 (lung cancer)

– Predicate: http://www4.wiwiss.fu-berlin.de/diseasome/resource/diseasome/associatedGene

– Object: http://www4.wiwiss.fu-berlin.de/diseasome/resource/genes/TPL2"

• Related concepts:– http://www.disease-ontology.org/api/metadata/DOID:162

(cancer) – http://www.uniprot.org/uniprot/M3K8_RAT (TPL2 synonym)

11/29


What is BioASQ?

13/29

A competition funded by the European Union (FP7).

Task A: Hierarchical text classification• Organizers distribute new unclassified PubMed articles.• Participants assign MeSH terms to the articles.• Evaluation based on annotations of PubMed curators.

Task B: IR, QA, summarization, …• Organizers distribute English biomedical questions.• Participants provide: relevant articles, snippets,

concepts, triples, “exact” answers, “ideal” answers. • Evaluation: both automatic (GMAP, MRR, ROUGE etc.)

and manual (by biomedical experts).


Two cycles

14/29

Evaluation infrastructure &

dry-run data

Start of the challenge

End of the challenge

BioASQ workshop

March 2013 June 2013 August 2013 September 2013

2013 Schedule

Start of Task 2A Start of Task 2B

End of the challenge

BioASQ workshop

February 2014 March 2014 May 2014 September 2014

2014 Schedule

► Both tasks run twice, in two cycles (two years).► 1st cycle completed, workshop collocated with CLEF-2013.► 2nd cycle starting! Part of CLEF QA track! (http://nlp.uned.es/clef-qa/)

► Participation can be partial (any task, subtask, response type). ► Prizes for each task/subtask.


More info

15/29

Questions (300+500) and gold articles, snippets, concepts, triples, “exact” and “ideal” answers prepared by biomedical experts from around Europe.► Using tools/infrastructure developed by BioASQ.

Data sources include both text and structured info.► PubMed abstracts, PubMed Central articles, MeSH.► Gene Ontology, UniProt, Jochem, Disease Ontology.

BioASQ datasets, infrastructure, evaluation services etc. available beyond the end of the project:► Plus social net to help extend data, set up new challenges.

Advisory board: both academia and industry.► NLM, NIST, CMU, IBM, MSR, NaCTeM etc.


Annotation tool

16/29


Annotation tool (II)

17/29


Annotation tool (III)

18/29


BioASQ social network

19/29


BioASQ social network (II)

20/29


How they all fit together

21/29


First year participants

22/29


First year participants (II)• Task 1A (46 systems, 11 teams)

– Mayo Clinic, USA– University of Alberta, CANADA– Aristotle University of Thessaloniki + Atypon, GREECE– University of Vigo, SPAIN– University of Colorado, USA– NCBI, NLM, USA– Université de Rouen, FRANCE– Fudan University, CHINA– UCSD, USA– Toyota Technological Institute, JAPAN– Imran, PAKISTAN

23/29


First year participants (III)• Task 1B, Phase A (4 systems, 2 teams)

– Mayo Clinic, USA– University of Alberta, CANADA

• Task 1B, Phase B (7 systems, 2 teams)– University of Alberta, CANADA– Toyota Technological Institute, JAPAN

• More participants needed in Task 1B, esp. from Europe!• Workshop (30 participants)• Invited speakers

– Lan Aronson, Lister Hill Center, U.S. National Library of Medicine, USA

– Jennifer Chu-Caroll, IBM T.J. Watson Research Center, USA

24/29


First year technology/results overview

• Task 1A– Mainly SVMs and learning-to-rank.– Mostly flat classification, ignoring class taxonomy.– Mediocre results by hierarchical methods.– One of the systems outperformed NLM’s system.

• Task 1B– Phase A (retrieve relevant documents, concepts, snippets,

triples): low performance (compared to baselines).– Phase B (formulate ‘exact’ and ‘ideal’ answers): poor performance

for ‘exact’ answers (except for yes/no questions); high performance for ‘ideal’ answers (paragraph-sized summaries), but starting with gold documents, snippets etc.

• Large scope for improvements, esp. in Task 1B.

25/29


“Exact” answer results (batch 2/3)

26/29


“Ideal” answer results (batch 2/3)

27/29


Project Consortium

1. National Centre for Scientific Research “Demokritos” -NSCR “D” (EL)

2. Transinsight GmbH – TI (D)

3. Universite Joseph Fourier- UJF (F)

4. University Leipzig - ULEI (D)

5. Universite Pierre et Marie Curie Paris 6 – UPMC (F)

6. Athens University of Economics and Business – Research Centre – AUEB-RC (EL)

29/29

Ion Androutsopoulos, February 2014 NLP Group, Department of Informatics, Athens University of...

Documents

Transcript of Ion Androutsopoulos, February 2014 NLP Group, Department of Informatics, Athens University of...