I'm afraid it's very unlikely that you'll be able to find an existing dataset of abbreviations for every language, simply because there are so many languages (more than 6,000) and variations within each language. For example, even in English 'Avenue' can be abbreviated to 'Av', 'Ave' or 'Avn' at least. That's before you get to the issue of different alphabets – do you want to abbreviate 'улица' (the Russian word for Street) to 'ул' as it would be in Cyrillic or 'Ul' as it would be transliterated into the Latin alphabet.
A useful starting point would probably be to think carefully about exactly what languages you need for your purpose and whether the benefit of including a particular language in your abbreviations dataset out-weighs the time required to find or generate the abbreviations for that language. You might want to look at a list of languages by total number of speakers or to gather data about the most-common languages that you will be applying your abbreviations dataset to. This different might matter if, for example, Hindi is a rare language in your context even though it's the third-most common language in the world.
Once you've done that you can concentrate your search on particular languages that are important in your context. For English, ESRI has an abbreviation dictionary for commonly used street prefixes that might be helpful.