The fundamental purpose of Tripledex is to allow users to easily collect and meaningfully display data Triples. A "Triple" is a simple data model where two entities/nodes have a single relationship. The format is expressed as follows, entity ~ relationship ~ entity. The entities and relationships can be virtually anything. This generic and flexible design makes it a powerful means for exploring social data. These are are a few examples of triples:

In this collection of data, Jeff appears in the greatest number of triples. Another way of phrasing this is that the Jeff node has the highest connectivity. Connectivity of a node is an important statistic in social network analysis, and most network visualizations are designed in ways that place special emphahsis on nodes with the highest connectvity. These visualizations are useful if the user has question mostly related to connectivity, but they often leave litle room to meaningfully express more than that single characteristic. This effect becomes increasingly acute and exaggerated in large networks. Such designs suggest that connectivity is the most significant aspect of the data.

Tripledex always visualizes the total connectivity for each individual node, but it is designed primarily to emphasize the strength of the relationships between nodes.

Using the samples of triples above as a data set, consider this: what if Jonathan is the user and he wants to know how to get an invitation to the TED Conference? Even though Jeff attened the TED Conference, Nick is the node that Jonathan might want to contact. In most visualizaitons Jeff's higher connectivity would mean his node is portrayed in ways that suggest he is more relevant than Nick. Also keep in mind that it is not only that Nick attended the TED Conference that is significant. Jonathan and Nick also both live in New York and attended NYU, so even though Jeff also attended TED, Nick and Jonathan still have more relationships in common. The Jonathan and Nick nodes have a higher affinity for one another.

Affinity is how items are ordered in Tripledex. By default, nodes with the greatest affinity are placed at the top of the visual hierarchy.

The first clip shows the basic environment after a subject is entered into the search box. The first item to load is a list on the left of all the entities connected to that subject. A graph of the same data builds on the right with the subject in the middle and all the entities from the list in an ellipse around that subject. On each line connecting the subject and entities is at least one circle, one for each shared relationship. Mousing over the relationship circle (a.k.a. "bead") will reveal its specifics. The more beads, the higher the affinity. Those items with highest affinity are to the left.

[CLIP 1]

A small arrow appears beneath each of the entities. Clicking this arrow opens a panel with a list of all the items connected to that entity. All of these items are one degree away from the original subject on the graph. From inside that panel, users can select to either make that entity the new subject in the middle top of the graph, or they can select an item from the list in the panel and make it the new subject. In either case, a new graph is drawn (with new, relevant entities, of course).

[CLIP 2]

When searching on former O'Reilly Editor, Rael Dornfest, it's easy to see affinity ordering at work. The number of beads become almost abacus like.

[CLIP 3]

In some instances, the affinity values are much higher. In these cases, it is possible for the user to hide the list so they can expand the graph.

[CLIP 4]

Users can scroll the graph to see other entities that have the same relationship type to the subject node. Even just clicking inside the track that holds the scroll bar will reposition the nodes (sliding the graph). Also, results are returned in groups of 50. Subsequent results can be linked to at the bottom of the graph. These groupings of 50 can also be scrolled.

[CLIP 5]

Because instances of relationships often appear multiple times across a graph, it is helpful to be able to re-order the graph with those relationships prioritized (ordered at the beginning of the graph). In these cases, affinity becomes the secondary criteria for order.

[CLIP 6]

It is also helpful to gain some sense of how frequent a relationship occurs across the entire graph of 50 nodes. With this tool, the user can see and measure how often and where a relationship appears. On this panel shown in the clip below, instances of colored dots indicate a presence of that relationship across the entire set of 50 nodes. The gray dots mean there is no instance of that relationship between the subject and that node. This tool allows users to look at larger patterns.

This device also allows users to reposition the graph so they can navigate, in an informed way, based on relationships as well as by entity.

[CLIP 7]