A task-based evaluation of the ecm database. Effect on parsing performance

One project aim was to establish a theory-neutral lexical representation of mwes so that the resulting lexical database could be used in various nlp systems in an effective way. We set up a small scale task-based evaluation to find out to what extent the wide-coverage Alpino parser can benefit from using the ecm lexical database. In the remainder, we first explain some needed notions. Further, we deal with the automatic conversion evaluation and next describe the methodology used to evaluate parsing performance. The results and error analysis follow. We finalize this report by summarizing the implications and our conclusions.