Limits...
BioCreative V track 4: a shared task for the extraction of causal network information using the Biological Expression Language.

Rinaldi F, Ellendorff TR, Madan S, Clematide S, van der Lek A, Mevissen T, Fluck J - Database (Oxford) (2016)

Bottom Line: Given the complexity of the task, we designed an evaluation methodology which gives credit to partially correct statements.We identified various levels of information expressed by BEL statements, such as entities, functions, relations, and introduced an evaluation framework which rewards systems capable of delivering useful BEL fragments at each of these levels.The aim of this evaluation method is to help identify the characteristics of the systems which, if combined, would be most useful for achieving the overall goal of automatically constructing causal biological networks from text.

View Article: PubMed Central - PubMed

Affiliation: Institute of Computational Linguistics, University of Zurich, Zurich, Switzerland fabio.rinaldi@uzh.ch juliane.fluck@scai.fraunhofer.de.

No MeSH data available.


An example output of the sentence-based evaluation. The screenshot contains the detected true positive (green), false positive (red) and false negatives (yellow) entries for the term and relationship level.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4940434&req=5

baw067-F5: An example output of the sentence-based evaluation. The screenshot contains the detected true positive (green), false positive (red) and false negatives (yellow) entries for the term and relationship level.

Mentions: The output of the evaluation page shows results per evidence text and an overall performance statistics. The overall performance statistics contains values for true positives, false positives, false negatives and the evaluation metrics recall, precision and F-score for all different structural levels. The statistics includes the performance statistics for each evidence text. In addition, further information is provided, such as the evidence text itself, the gold standard BEL statement derived from the chosen evaluation set and the predicted BEL statements taken from the user's input. Furthermore, true positive, false positive and false negative entries for the various structural levels are displayed, as can be seen in Figure 5. The overall performance statistics shows the combination of the results of all evidence texts.Figure 5


BioCreative V track 4: a shared task for the extraction of causal network information using the Biological Expression Language.

Rinaldi F, Ellendorff TR, Madan S, Clematide S, van der Lek A, Mevissen T, Fluck J - Database (Oxford) (2016)

An example output of the sentence-based evaluation. The screenshot contains the detected true positive (green), false positive (red) and false negatives (yellow) entries for the term and relationship level.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4940434&req=5

baw067-F5: An example output of the sentence-based evaluation. The screenshot contains the detected true positive (green), false positive (red) and false negatives (yellow) entries for the term and relationship level.
Mentions: The output of the evaluation page shows results per evidence text and an overall performance statistics. The overall performance statistics contains values for true positives, false positives, false negatives and the evaluation metrics recall, precision and F-score for all different structural levels. The statistics includes the performance statistics for each evidence text. In addition, further information is provided, such as the evidence text itself, the gold standard BEL statement derived from the chosen evaluation set and the predicted BEL statements taken from the user's input. Furthermore, true positive, false positive and false negative entries for the various structural levels are displayed, as can be seen in Figure 5. The overall performance statistics shows the combination of the results of all evidence texts.Figure 5

Bottom Line: Given the complexity of the task, we designed an evaluation methodology which gives credit to partially correct statements.We identified various levels of information expressed by BEL statements, such as entities, functions, relations, and introduced an evaluation framework which rewards systems capable of delivering useful BEL fragments at each of these levels.The aim of this evaluation method is to help identify the characteristics of the systems which, if combined, would be most useful for achieving the overall goal of automatically constructing causal biological networks from text.

View Article: PubMed Central - PubMed

Affiliation: Institute of Computational Linguistics, University of Zurich, Zurich, Switzerland fabio.rinaldi@uzh.ch juliane.fluck@scai.fraunhofer.de.

No MeSH data available.