PPLRE Evaluation Task
- The evalution task is designed around an iterative approach of automated and manual evaluation. Each iteration is designed to extract more and more OPL relations. However during the first few rounds the emphasis will be on Exploration for more optimal solutions to the task. It is during the latter rounds that the focus will shift to Exploitation of the best algorithm developed to date.
- Through each iteration there are two distinct subtasks:
- The PPLRE Automated Evaluation Task is developers of PPLRE Relation Extraction Algorithms use the PPLRE Automated Evaluation System to evaluate the Performance of their Relation Recognition Algorithms at recognizing OPL() relations. The result of this task is an Ensemble Model that is used to in part to grow the PPLRE Training Data.
- The PPLRE Manual Evaluation Task is where Subcellular Localization Domain Experts use the PPLRE Manual Evaluation System to evaluate predicted OPL() relations reported by an Ensemble of the PPLRE Relation Extraction Algorithms that have been optimized during the preceeding subtask of automated evaluation.
- See: PPLRE Evaluation.