Graphics Reference
In-Depth Information
type of link in the data—some graph software will be able to analyze and
explicitly represent these different types of links, such as Cytoscape.
Potential Hierarchy Data Sources
Hierarchies are a special type of graph. Though hierarchies may be
embedded into data in a number of the techniques discussed previously,
they may also be embedded in data with other techniques.
Links within a Node (for Example, Employee Data)
In one approach, a node may be described as containing a link to the next
higher node in the hierarchy (that is, the “parent” for that node).
For example, a human resources database has each record indicating an
employee, and the manager of each employee indicated in one of the fields.
The successive chains of managers can be used to create a graph:
Person, Age, Income, Manager
Ann, 22, 20000, Ben
Ben, 33, 30000, Zoe
Tim, 44, 40000, Zoe
...
Flattened Hierarchies (for Example, Pivot Tables)
Flattened hierarchies are sometimes found in public data sets and from
spreadsheets, such as exported pivot tables. In flattened hierarchies, each
successive column represents the next level of the hierarchy. Each row
represents a node. Links must be extracted by identifying each unique pair
across successive pair of columns.
In the following example, the links are as follows:
• Technology
⇒
Software, Software
⇒
Application_Software
• Technology
⇒
Hardware, Hardware
⇒
Computer_Hardware
• Financials
⇒
Insurance and Insurance
⇒
Life_Insurance
Note that the root level is sometimes not shown as a distinct column or row,
andsothenodeandlinktothetoplevelmustbecreated.Followingfromthe
same example, the node would be Portfolio and the links would be Portfolio
⇒
Technology and Portfolio
⇒
Financials:
