We have a new player in the semantic search arena; "semantic search" refers to the idea that a better way to find the most relevant material is to not just look at word frequency, location, links to a page and so on but to understand the sense of what is being asked and the sense of what is on web pages, and then to match the most applicable pages. For example, if I typed in a query for stock, a semantic search engine would see that this was in the context of livestock, and would find other pages that have to do with animals on farms rather than about shares of ownership in a company, flowers, or racing cars.

PowerSet uses a technology that attempts to mine meaning out of web pages -- specifically, Wikipedia articles. This isn't a general-purpose search engine, but it is a tremendous tool if you are looking for information on a subject that spans a number of Wikipedia articles, or that has a number of meanings. Take "Barack Obama" as an example. There are thousands of articles that mention Obama -- the Wikipedia article specifically about him, as well as articles about politicians who have endorsed him, his early career, the roles of race and religion in the US presidential campaign, and so on.

Rather than going through all of those entries, including the extensive article specifically on Obama, head over to PowerSet and you will get a single search result that pulls together what it judges to be the most relevant information about Barack Obama. There is a nice summary of who he is, the key facts about his career, and a relevance-ranked list of Wikipedia pages that mention him.

What I find particularly interesting is the "Factz" section of the page, which attempts to pull together significant statements about Barack Obama in the Wikipedia articles. So, for example, the Factz for Barack Obama include information that he introduced the following pieces of legislation: