Text zoning for job advertisements with bidirectional LSTMs
Ann-Sophie Gnehm · Zurich Open Repository and Archive (University of Zurich) · 2018
We present an approach to text zoning for job advertisements with neural networks. Text zoning refers to segmenting texts into eight classes differing from each other regarding content. It aims at capturing text parts dedicated to particular subjects, e.g. the publishing company or qualifications wanted, and hence facilitates subsequent information extraction. We use BiLSTMs, a class of neural networks particularly suited for sequence labeling. Our best approach,vwith task-specific word embeddings and ensemble technique, reaches token-level accuracy of 89.8% and outperforms previous approaches with CRFs.