Diagnostic Evaluation of Machine Translation Systems Using Automatically Constructed Linguistic Check-Points

摘要

We present a diagnostic evaluation platform which provides multi-factored evaluation based on automatically constructed check-points. A check-point is a linguistically motivated unit (e.g. an ambiguous word, a noun phrase, a verb~obj collocation, a prepositional phrase etc.), which are pre-defined in a linguistic taxonomy. We present a method that automatically extracts check-points from parallel sentences. By means of checkpoints, our method can monitor a MT system in translating important linguistic phenomena to provide diagnostic evaluation. The effectiveness of our approach for diagnostic evaluation is verified through experiments on various types of MT systems.