linkanalysis

Type Members

1. class Hits extends LinkAnalysis[HitsIterationState]

   

   Class for performing Hits algorithm.

   Class for performing Hits algorithm. Hits is a link analysis algorithm that returns two values characterizing each node. Each node receives both a hub value and an authority value. A node that is characterized by a large hub value is one that has many high quality outbound connections to other nodes, while a node that is characterized by a large authority value has inbound connections from high quality hub nodes.

2. case class HitsIterationState(hubs: Array[Double], authorities: Array[Double], iteration: Int, error: Double) extends IterationState with Product with Serializable

   

   Stores all values necessary to fully describe one Hits iteration

   Stores all values necessary to fully describe one Hits iteration

   hubs

   Array of values indexed by node id storing hubs values for each node

   authorities

   Array of values indexed by node id storing authorities values for each node

   error

   The T1 error for the current iteration vs the previous iteration.

3. case class HitsParams(maxIterations: Option[Int] = Some(100), tolerance: Double = 1.0e-8, normalize: Boolean = true) extends Params with Product with Serializable

   

   Stores all parameters for Hits algorithm

   Stores all parameters for Hits algorithm

   maxIterations

   The maximum number of times that the link analysis algorithm will run before termination

   tolerance

   The maximum error allowed.

   normalize

   Flag true to return normalized values

4. abstract class IterationState extends AnyRef

   

   The base class for all iterations through our iterative algorithms.

   The base class for all iterations through our iterative algorithms. These classes will simply hold all of the information needed to assess the number of iterations, the error, and the current set of values.

5. abstract class LinkAnalysis[T <: IterationState] extends AnyRef

   

   All link analysis algorithms should inherit from the LinkAnalysis base class.

   All link analysis algorithms should inherit from the LinkAnalysis base class.

   T

   LinkAnalysis must be generically typed by IterationState or one of its subclasses. An IterationState holds all of the pertinent information for a given algorithm.

6. class PageRank extends LinkAnalysis[PageRankIterationState]

   

   PageRank is a link analysis algorithm designed to measure the importance of nodes in a graph.

   PageRank is a link analysis algorithm designed to measure the importance of nodes in a graph. Popularized by Google.

   Unoptimized for now, and runs in a single thread.

7. case class PageRankIterationState(pageRank: Array[Double], error: Double, iteration: Int) extends IterationState with Product with Serializable

   

   Class containing information to fully describe a page rank iteration.

   Class containing information to fully describe a page rank iteration.

   pageRank

   The current set of pageRank values

   error

   The T1 error for the current iteration vs the previous iteration.

8. case class PageRankParams(dampingFactor: Double = 0.85, maxIterations: Option[Int] = Some(10), tolerance: Double = 1e-8) extends Params with Product with Serializable

   

   Parameters for PageRank

   Parameters for PageRank

   dampingFactor

   Probability of NOT randomly jumping to another node

   maxIterations

   The maximum number of times that the link analysis algorithm will run before termination

   tolerance

   The maximum error allowed.

9. abstract class Params extends AnyRef

   

   The base class for all parameters fed to our iterative algorithms.