If the success of Google proves anything about how to structure information, it's that it doesn't matter how you structure information if your search algorithm is good enough. The difficulty is to create a search system that will catch all the results relevant to the search request without overwhelming the user with irrelevant matches. You also need to present the results to users in such a way that they can tell which results are relevant and which are spurious.
Within Rails, the goal is to search the ActiveRecord objects based on text input that might match one or more fields of the record. For the main text box, you want an unstructured match against multiple fields. You can then provide an advanced search form to specify particular fields to use for the match. This section describes two useful methods for handling a search. One method uses SQL queries. It's simple, and requires nothing beyond the existing ActiveRecord find mechanism. However, it doesn't easily distinguish the importance of the match, and it's hard to add support for Boolean logic in the search.
The second, more complex method uses the Ferret text-indexing tool to support a full-text search. Ferret scores each match based on importance, plus it allows a rich language for logic within the query. However, it requires a Ruby gem, a Rails plugin, and some additional files to support the indexes.
Both methods may have performance issues in high-traffic production sites.