What is transliteration?

Transliteration consists in representing the characters of a given script by the characters of another, while keeping the operation reversible. The use of diacritics or digraphs solves the problem of different number of characters between the alphabets of the two writing systems.
The main goal of this conversion operation is to enable the automatic and unambiguous recreation of the original (which is also known as retroconversion). In a word, the transliteration of a transliterated text should return the original text. This is why transliteration standards are used.
These standards define the transliteration rules: which characters are replaced by which others, and in which case. There are many, but the most used for romanization are ISO-9 (for cyrillic), ALA-LC (to represent bibliographic names in the English-world libraries), and BGN/PCGN (for geographical names and personal names).
Romanization (or Latinization) is the transliteration of a non-Latin script into a Latin script.

Transliteration vs. translation

The transliteration process means that a word written in a character set like the Latin alphabet is transposed in another, say the Inuktitut syllabics. In other words, there is no translation involved. If the source word means nothing in the given language, its transliterated form will also mean nothing, even though it will look like a word in that language as it will be written in its alphabet or syllabic system.

What is transliteration used for?

Transliteration is particularly used by libraries or for the processing of textual data.
When a user performs a search or indexes content, the transliteration process can find the information written in a different alphabet and returns it into the user’s script.
Transliteration also enables the use of a keyboard in a given script to type in a text in another one. For example, it is possible with this technique to use a qwerty keyboard to type text in Cyrillic.