Web-based extraction of technical features of products
S. Schmidt
and H. Stoyan
Abstract
We present a novel symbolic approach to extract domain-specific technical features of products from large German corpora. Our prototypical implementation extracts terms like “Auflösung” (resolution), “Speicherplatz” (storage capacity), etc. The proposed methods depend on manually added lists of technical measures of the target domain (in our case measures like “Megapixel” or “MB”).We applied the extraction in the domain of digital cameras on the internet using Google.
Full Text: PDF