I think in one sense the answer is it always depends who or what you are asking for your answers. The early Web wrestled with this, early on it was going to be

I think in one sense the answer is it always depends who or what you are asking for your answers. The early Web wrestled with this, early on it was going to be directories and meta keywords. But that quickly broke down (information isn't hierarchical, meta keywords can be gamed). Google rose up because they use a sort of reputation system based index. In between that, there was a company called RealNames, that tried to replace domains and search with their authoritative naming of things, but that is obviously too centralized. But back to Google, they now promote using schema.org descriptions of pages, over page text, as do other major search engines. This has tremendous implications for precise content definition (a page that is "not about fish" won't show up in a search result for fish). Google layers it with their reputation system, but these schemas are an important, open feature available to anyone to more accurately map the web. Schema.org is based on Linked Data, its principle being each piece of data can be precisely "followed." Each schema definition is crafted by participation from industry and interest groups to generally reflect its domain. This open world model is much more suitable to the Web, compared to the closed world of a particular database (but, some companies, like Amazon and Facebook, don't adhere to it since apparently they would rather their worlds have control; witness Facebook's open graph degeneration to something that is purely self-serving).