C# Server side HTML filter(repost)

We have made a CMS system for client which includes a WYSIWYG editor. Our client often pastes rich textual content from Microsoft Word into this editor, and the result is bad looking content on the front-end side of the CMS system. Although the? WYSIWYG? editor contains a button the can filter out this bad looking Microsoft Word HTML code, our client often forgets to use this button, and now wants the filtering to be applied automatically. We want you to create a C# server side HTML filter (in a separate .NET assembly), that can filter out all of MS Word's unneeded HTML. Moreover, we need to be able to control the filitering process, by specifing from code which tags, attributes and style properties are to be allowed.