I hate these questions, I always thought relevance and validity meant the same thing, where research or experiment being carried out is relevant to the intended study, i.e. the method actually investigates the aim.
As for reliability of sites and books, if they are written by a person of authority in that area i.e. a scientist or a well respected research institute such as the CSIRO, reliability is high as opposed to information found on wiki for example.
accuracy can be found in the correct technique in procedure - but thats in terms of first hand investigations. for information from sites i guess accuracy can be assessed in terms of how thorough the information is? e.g. a survey of 1000 people will have a more accurate indication than a survey of 10 people
hope that helped :S