11791-04 / project-team04 Goto Github PK

Evaluate retrieved answers with the gold standard using the given MAP and GMAP metrics.

Reader annotated question from resources/BioASQ-SampleData1B.json, and generate JCas.

Rank the retrieved information to produce final answer according question type.

Make sure we can process question.json - unannotated input

Retrieve concepts using the helper class and add them to CAS

Retrieve documents using the helper class and add them to CAS

https://piazza.com/class/hyvsubeilei6dd?cid=345

Read the annotations (answers) and store it in UIMA JCas index

Create a consumer that reads the candidate answer and also the gold standard.

Retrieve triples using the helper class and add them to CAS

evaluate using exact_match for M3

Modify Metic.java to evaluate snippets

The API shows that document 24939007 has full text, but in fact it doesn't...

To fix that, in the isFullTextAvailable() a secondary check may be added.

Write final answer to a JSON file.

When using the GoPubMedService class, the findUniprotEntitiesPaged() function causes an illegal state exception and the whole thing crashes...

Stack Trace:

uhh... seems to have gone away. but let's leave this here for now.

Triples retrieved from the PubMed API seems incorrect. New way required to evaluate triples.