A filter bubble is when a website algorithm guesses what information a user wants based on his/her location, online behaviour, personalised searches and search history. The result is a 'filter bubble' which can isolate users from information not a part of their search backgrounds.
In Eli Pariser's 2011 TED Talk 'Beware of Filer Bubbles', Pariser notes the dangers of personalised searches. Algorhythms used by Facebook and Google as well as others may limit and change searches to suit our supposed preferences.
According to Pariser, Google uses 57 'signals' including the computer you're working on, the browser you're using, your location, etc. to set up personalised searches that filter information. While this can help a user deal with lots of unneeded information, it also means that an algorithm is determining search results and it may block out more results than a user may consider necessary.
As a result of personal preferences, it is important to realise that two people searching for the same information may retrieve very different results.