I believe the documentation regarding the usage of Open Graph types "article" and "website" is unclear and possibly contradictory, especially with regards to common web page content types that are not "transient" or represented by more specific types.
A quote from
https://developers.facebook.com/docs/opengraphprotocol/#types, which is titled "Open Graph Protocol":
"Use article for any URL that represents transient content - such as a news article, blog post, photo, video, etc. Do not use website for this purpose. website and blog are designed to represent an entire site, an og:type tag with types website or blog should usually only appear on the root of a domain."
This statement implicitly indicates that the "article" type should only be used for "transient" content. The definition of "transient" is "lasting only for a short time". This quote also implicitly states that "website and blog are designed to represent an entire site...(and) should usually only appear on the root of a domain."
Another quote from the same page:
"If your URL is a piece of content — such as a news article, photo, or similar — you should set og:type to article (see below)."
This second quote seems to concur with the first quote, but is slightly more vague: if a "piece of content" such as "news article, photo, or similar". What is "similar"? How does this statement correspond to the use of the adjective "transient" in the first quote? I'm not sure.
Furthermore, the following page, titled "How-To: Using Built-In Objects",
https://developers.facebook.com/docs/technical-guides/opengraph/built-in-objects/#article shows using og:type = article in conjunction with Publish Time, Expiration Time, and Author attributes, which support the idea that "article" should be used for pages that have transient and news-like content.
However, from the Open Graph protocol website (which is referenced on the first URL I quoted):
http://ogp.me/#types "Any non-marked up webpage should be treated as og:type website."
This quote seems to indicate a best practice that a "webpage," which I assume refers to an individual webpage that may exist within the context of a larger "website" with many pages, should be treated as og:type website. That seemingly contradicts the above statements in the FB Developers documentation.
Despite the Open Graph protocol website's statement, it seems that, if we want individual pages of a website to each have unique LIKE buttons then "website" is not the right og:type.
To make matters more complicated, I'm unable to find any documentation that specifically addresses and clarifies OG:type usage for a common real-world scenario in which the content of a single web page within a website contains content that is not an article (in the traditional sense), blog entry, or some similarly "dated" or "transient" content, but instead is long term content that is also not directly represented by any other og:type. Examples: an FAQ page, a page describing a company's services, a page of resources (links to other websites), a page that describes an organizations policies and procedures.
And to complicate matters even more, the following quote also comes from the first URL above:
"If your URL is a video, you should set og:type to video."
That statement is in direct contradiction to the first quote above, which states to use the "article" type for a video.
Also related, the "How-To: Using Built-in Objects" has an example at the bottom of the use of "website" as the type that supports the first quote above but confusingly includes an og:title of "Name of blog", when we're told elsewhere that a blog should have og:type=blog.
Finally, the list of available, pre-existing object types on the first and second URLs I quote from do not match. It is unclear to me from the documentation whether the "extra" types listed "Open Graph Protocol" page are or are not Built-In Types.
I apologize for the long report. I have been trying to determine conclusively for weeks the correct type to use for pages like the examples I listed above, but the available information does not make it clear to me.
Thanks, Fritz Green