The definition of the concept of 'publication' is not entirely obvious when it is applied to a document that is only published electronically, by placing it on one or more servers that distribute them over the Internet. Can the electronic file be considered as "the publication"? What happens if the file stored on-line is changed from time to time? What happens when it is moved to another computer? and so on.
A committee within ICSU (the International Council of Science) was appointed to study this question, and the report of the committee was published in the journal Learned Publishing; it can be accessed here.
I was a member of this committee, partly due to the fact that I had previously published some reports on the same topic, and I am generally in agreement with the committee's report. However, I would have wished that we could have gone a bit further in our recommendations. In my opinion, the only logically coherent way of defining "electronic publication" is to first define the concept of an "electronic publisher" as an organization that performs certain well-specified functions in certain well-specified ways. After this one can rely on the concept of electronic publisher for defining the activity of electronic publication as that what an electronic publisher does, and also for defining the document object that is an electronic publication, as a document that has been duely published by an electronic publisher.