This paper presents an automated natural language question generator for Tagalog informational texts. It takes a target text as an input and generates set of what, where and who question and answer pairs extracted from the source text. The resulting questions are also ranked using a proposed acceptability criterion. The implementation consists of three modules: a part of speech tagger, an anaphora resolver, and a question ranker. Questions generated from the experiment were rated by three evaluators. Of the generated questions, 33% were rated as acceptable by the evaluators, 96% of which were in need of modifications.
展开▼
