A computer-implemented method for selecting a desired Roman or non-Roman-alphabet character or objects from a set of non-Roman characters or objects may include steps of providing an association database that includes, for each non-Roman-alphabet character of the set, a Roman alphabet or other phonetic transliteration associated with each said non-Roman-alphabet character and a plurality of entries that are associated with each said non-Roman-alphabet character; receiving a phonetic transliteration of the desired non-Roman-alphabet character or data object and at least one associated entry that is associated with the desired non-Roman-alphabet character or other similar symbolic input; accessing the association database and identifying as candidate characters those characters of the set that are associated with the received phonetic transliteration and with the at least one received associated entry; if a number of candidate characters is greater than one, receiving additional associated entries and repeating the accessing and identifying step until a number of candidate characters is narrowed down to a single candidate character, and providing the single candidate character as the desired non-Roman-alphabet character. Also, derived from the principles described above, this invention includes a variety of methods for improving the efficiency of search engines through use of associations and other means of providing context for the item(s) being searched.