The graph: influence and authority between people and content

Nodes and edges

A social graph is a graph that represents the relations between people and things.^⁠[2] In ContentSignals the nodes are people and pieces of content, and the edges are the signals between them: a read, a share, a citation, a rating. The more signals pass along an edge, the stronger it is.

The 2012 filing the model comes from treats this activity as worth something in itself: "The content and commerce we consume and share represents social currency."^⁠[1] The graph is where that worth becomes visible.

Hubs and authorities

The filing classifies a graph by hubs and authorities, much as search engines classify the web.^⁠[1] The idea comes from link analysis: a good hub points to many other pages, and a good authority is pointed to by many different hubs.^⁠[3]

In content terms, a curator who shares widely and well is a hub. A piece that many good curators share and cite is an authority. Each makes the other count for more: a share from a strong hub is worth more than a share from a stranger, and a hub that keeps pointing to authorities becomes a stronger hub.

Influence

Influence is how far and how fast signals travel out from a node. A creator whose work is passed on, and passed on again, has influence even before anyone cites them. Network analysis measures a node's position with centrality, which assigns numbers or rankings to the nodes of a graph and is used, among other things, to find the most influential people in a social network.^⁠[4]

How close a community is

The filing also uses the clustering coefficient to judge "the intensity of the graph holders network".^⁠[1] The clustering coefficient measures how much the nodes of a graph tend to cluster together.^⁠[5] In a close community the people around a creator also know and share with each other, so their signals reinforce each other. The filing weighs exactly this: "the authority and influence within their community".^⁠[1]

Ranking and indexing

The same graph serves search. Because it records which pieces are authorities and which people are hubs, it can rank content by the standing that real signals give it, rather than by how often a page repeats a word. Swomi's search, GSERP, on swomi.ai, is built to read signals in this way.

People's data and owners' preferences

Measuring a piece of content in the graph works from counts and totals: how often it is shared, cited and passed on is what shows it to be an authority in its field. Where signals feed patterns beyond a member's own use, they are combined above a minimum group size. A member's own data is handled as Swomi's Privacy Policy sets out. Each piece's owner preferences, from the robots.txt line on the signals page, stay attached to it in the graph.