Towards combining web classification and web information extraction: a case study