The idea of a data ecosystem is an extension of the WikiPhilosophy across the whole of the Internet to a theoretical ideal to include all the web and its interactions. It is composed of data and the objects, which encode, programmatically, the functionality for interaction.
Among those who work with social data, two main axes have emerged: "most recent" and "most popular".
As many community-driven sites have found out, these two cannot be combined after scaling beyond approximately a few dozen items without ThreadMess.
One way to solve this problem is to from a CrowdSourced with both PerItemVoting and UserRanking. Most sites implement the former, but not the latter. Another way is with an AttentionManagementSystem.
Since there isn't a way to combine a 2-dimensional VotingModel on a single page, our vast, interactive knowledge repository would require a ThreeDimensionalVisualizationModel. Throw away the tired desktop metaphor for interacting with your data. When AllDataRelatesToOtherData, we can build a better, KillerUserInterface. Under the post-Internet model of OperatingSystem's, data rules over applications. This upturns the traditional monolithic OperatingSystem stack.
FairCredit: The term "data ecosystem" was copied from an IEEE member who coined the term, as far as I can tell. <find citation>.
Google reports: About 91,900,000 results (0.34 seconds) for a search on data ecosystem and there seem to be a number of commercial organisations using the phrase. It looks like a crowded field.
See PangaiaProject, UnifiedDataModel, GlassBeadGame, FlikiBase, ExtendingTheWikiParadigm