KeeWee is a keyphrase extraction tool from abstract-like text paragraphs. It is based on an unsupervised learning approach and supports two different implementations: one using k-means clustering and another one using spectral clustering. Two baseline methods are provided as well: a frequency-based and a random method.
The performance of the system is comparable to existing state-of-the-art approaches to keyphrase extraction and accounts for a F-measure of about 37%.
Back to the top