英文摘要 |
The local broadcast ICRT (International Community Radio Taipei) in Taipei has their news scripts sent to their listeners in ALL CAPS, which makes the articles more difficult to read. Therefore, we think it may facilitate the readers if we transform the text into normal cases that we are familiar with. In this prototype system, we established a practical method of restoration of case information, using different techniques from NLP and statistics. The system can apply many different kinds of approach, however, in this prototype, we focus our analysis and test data on broadcast transcription. |