Naive Graph Miner

An application that mines data from a graph naively.

A traditional miner would look at the data in each node. A NaiveGraphMiner uses only the linked relationships between nodes to mine data. A NaiveGraphMiner would be ideally suited for automated mining of a wiki. It could even be used to find high-level pattern patterns or potentially as a PatternSolver.

NaiveGraphNode? Information

• Complexity: # of links to other nodes
• Importance: # of links from other nodes
• Cyclicity: # of levels to back-trace to the node

NaiveGraphSet? Information

• Cohesiveness: how linked the set of nodes is to each other (Boolean searches)
• Independence: compared to global (DataIslands?, WalledGardens)
• Directedness: cyclicity of a path the set contains

NaiveGraphGlobal? Information

---

Prototype Search form:

Solve from node to [ ]

Weights : [ ]

Gather groups : A[ ] B[ ] C[ ]

then:

1. Display weighted relevance of gathered nodes. Output is "relevant reading material" or solutions - if gathered nodes are solutions (say with a *Pattern RegularExpression)
2. Gathered groups result in a set of sets
3. A Discerner makes a final processing of the gathered groups

---

See WikiMines