There are several algorithms available that can generate summaries of traditional documents. But I believe Wikis are a bit different. Wikis are themselves a collection of all the information relevant to it. We could leverage this to create more coherent and lucid summaries of articles by "crawling" through the links to get more relevant information. The infoboxes and categories should also give us a fair idea about the context and important parts of the article. Here is some research done on it.