Named Entity Recognition in Broadcast News Using Similar Written Texts

We propose a new approach to improving named entity recognition (NER) in broadcast news speech data. The approach proceeds in two key steps: (1) we detect block alignments between highly similar blocks of the speech data and corresponding written news data that are easily obtainable from the Web, (2) we employ term expansion techniques commonly used in information retrieval to recover named entities that were initially missed by the speech transcriber. We show that our method is able to find the named entities missing in the transcribed speech data, but also to correct incorrectly assigned named entity tags. Consequently, our novel approach improves state-of-the-art results of NER from speech data both in terms of recall and precision.