This repository is my system for SemEval-2014 Task 8:Semantic DependencyParsing (Oepen et al., 2014), as the second course project for the Empirical Methods for Natural Language Processing course in Peking University, spring 2020, directed by Professor Yansong Feng and Weiwei Sun. My submission is an ensemble system made up of arc-factored classification models. The system is implemented in nearly pure Python without any packages about machine learning or natural language processing. It reaches 78.32% precision and 0.6888 F1 score on the test set provided by Oepen et al. (2014),and the prediction results for the newly released test set(esl.output and cesl.output) has been uploaded with the report and system code.
For more details, please see the report.