There are two types of incoming interaction:
A story can be:
- free-text post
- custom post with verbs and nouns
An engagement can be a:
Facebook users can:
- like a story
- create a story
- comment on a story
User can also like a page but PYLON does not offer filtering on Facebook page likes; this is something that is available using DataSift’s Managed Sources.
The main namespace families in this target are:
Topics are individual items in the Facebook social graph. A topic can be referred to with a FBID (Facebook id). Please refer to Namespace Families for further information.
Hint: to find the FBID for a page or topic you can use the Facebook Graph API Explorer. Take a look at our guide on Discovering Topics to learn more.
Topic information is deduced by Facebook's natural language processing algorithms. Facebook automatically infers topics that a story is likely to be referring to. For example, if a story mentions 'Countach', 'Sesto Elemento', and 'Aventador', it is reasonable to infer that the author is discussing Lamborghini supercars so Facebook makes information from that manufacturer's Facebook page available for filtering and data delivery.
The JSON elements that Facebook offers are shown on this Lamborghini graph page: https://graph.facebook.com/Lamborghini
You can filter on any of these that appear in the fb.topics.* target namespace.
For stories we populate fields in these namespaces:
- fb.topics.* (if Facebook's NLP engine infers any topics)
For engagements (comments, page likes, and reshares) we populate fields in these namespaces:
|Filtering for stories||Filtering for engagements|
|fb.\*|| Use this filter:
fb.media_type == "reshare"
|fb.parent.\*||No data.|| Use this filter:
fb.type == "reshare"