Syntactic simplification for machine translation
José Camacho-Collados · ORCA Online Research @Cardiff (Cardiff University) · 2013
This paper presents a new rule-based Syntactic Simplification system for the Spanish language. The system basically consists of splitting complex sentences into several simple ones preserving its original meaning. It is envisioned to be a preprocessing tool for other Natural Language Processing applications such as Text Summarization, Information Extraction, Parsing and Machine Translation. According to the evaluation, the application of the syntactic rules proposed clearly improves Machine Translation from Spanish to far target languages such as Korean and Chinese. A parallel corpus of Spanish original and simplified texts has been manually built and validated by native Spanish speakers for this research. In addition, an experiment was carried out among two groups of Spanish learners comparing their reading comprehension of original and simplified texts. Their answers show a null correlation between simplification and text comprehension. The automatic Syntactic Simplification system has also been intrinsically evaluated obtaining promising results. Key-words