I'm working on a dictionary tool that provides, as an auto-assist, transcription of Chinese characters into corresponding (Modern Standard Mandarin) pinyin+tones on a naive, character-per-character basis without lexical parsing (the tool is intended for indexing of new terms/phrases, lexical parsing is overkill).
In order to make the auto-assist feature as useful as possible, I would like the assisted-transcription operation to choose the most frequent reading [1] of any individual character as the preferred default in an effort to reduce the amount of time users will spend correcting the inevitable [2] mistakes.
Python's pinyin package, based on Mandarin.dat, seems to provide reasonably good performance in that regard; but I'm curious whether there's a standard (issued by a language authority in PRC or Taiwan) list of such frequencies available, and/or whether any list exists that includes the other possible readings, in decreasing order of frequency, by way of providing a helpful dropdown list of alternates in cases where the preferred reading doesn't happen to be correct.
[1] i.e., assuming a large and representative set of textual data, if a character has multiple possible readings, the pronunciation most often assigned the character based on more advanced lexical analysis of the function of the character within a larger semantic block. For example, if within the set there are three observed instances of 为[wei4]什么 and one observed instance of 作为[wei2], choose "wei4" as the default transcription option for 为.
[2] I'm aware of characters such as 调 for which alternate readings (in this case diao4/tiao2) are equally likely, so in these cases such "assistance" will likely never have an accuracy rate better than 50%.
