2
0
mirror of https://github.com/boostorg/graph.git synced 2026-02-26 16:52:12 +00:00
Files
graph/quickbook/reference/degree_centrality.qbk
Andrew Sutton 7777457d50 Importing quickbook docs from SOC 2007
[SVN r51250]
2009-02-14 13:53:55 +00:00

440 lines
16 KiB
Plaintext

[/
/ Copyright (c) 2007 Andrew Sutton
/
/ Distributed under the Boost Software License, Version 1.0. (See accompanying
/ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
/]
[section Degree Centrality]
[template ex_degree_centrality[] [link
boost_graph.reference.algorithms.measures.degree_centrality.examples.degree_centrality
Degree Centrality Example]]
[template ex_influence_prestige[] [link
boost_graph.reference.algorithms.measures.degree_centrality.examples.influence_and_prestige
Influence and Prestige Example]]
[template degree_centrality_measures[] [link
boost_graph.reference.algorithms.measures.degree_centrality.measures
Measures Section]]
[heading Overview]
The /degree centrality/ measure is used in social network analysis to help
determine which vertices (or more specifically actors) are the most central
or important to a network. Degree centrality of a vertex is computed over the number
of connections (adjacent vertices) of that vertex. This definition varies for
undirected and directed graphs. For undirected graphs, the degree centrality of a
vertex is equivalent to its degree. For directed graphs it is equivalent to either
its in- or out-degree.
Consider the social network represented by an undirected graph in Figure 1.
[figure
images/reference/social_network.png
*Figure 1.* A network of friends.
]
This graph depicts a set of friends and their relationships (friendship, dating, etc.)
The degree centrality of each person in this graph is their number of friends.
Here, Frank has the highest degree centrality with five friends, whereas Laurie,
Jill, Bill, Anne, and Howard are the least central with two friends apiece. See
the [ex_degree_centrality] for details on computing the degree centrality of
undirected graphs.
For directed graphs, there are related two interpretations of degree centrality.
When degree centrality is measured over the out-degree of a vertex, it is called
/influence/. When the in-degree is measured, the measure is called /prestige/.
See the [degree_centrality_measures] for information types and functions related
to the computation of influence and prestige.
Consider the information network represented by the directed graph in Figure 2.
[figure
images/reference/info_network.png
*Figure 2.* A network of informational websites.
]
This graph represents links between information websites, presumably regarding a
specific topic. The most influential (out-degree centrality) is blogger with
three outgoing links while wikipedia is the least influential. The most prestigous
(in-degree centrality) is wikipedia with five incoming references. The least
prestigous websites are myspace, blogger, and blogspot each of which have no
incoming links. See the [ex_influence_prestige] for details on computing the
influence and prestige of directed graphs.
[section [^degree_centrality()]]
#include <boost/graph/degree_centrality.hpp>
template <typename Graph, typename Vertex>
typename graph_traits<Graph>::degree_size_type
degree_centrality(const Graph& g, Vertex v);
template <typename Graph, typename Vertex, typename Measure>
typename Measure::degree_type
degree_centrality(const Graph& g, Vertex v, Measure measure);
The `degree_centrality()` function computes this measure for a single vertex,
returning the value to the caller. Optionally, a degree measure can be given,
specializing the computation of degree centrality. See the [degree_centrality_measures]
section for more details on the influence and prestige measures.
[heading Parameters]
[table
[[Type] [Parameter] [Description]]
[
[required, in] [`const Graph& graph`]
[
The graph object for which the degree centralities will be computed.
*Requirements:* The `Graph` type must be a model of the [IncidenceGraph]
concept. If the `Measure` parameter is given as `prestige_measure`,
then the graph must be a model of the [BidirectionalGraph] concept.
]
]
[
[required, in] [`Vertex v`]
[
The vertex descriptor for which the degree centrality is computed.
*Requirements:* The `Vertex` type must be the same as the `vertex_descriptor`
of the `Graph` parameter.
]
]
[
[optional, in] [`Measure measure`]
[
The `measure` parameter allows the caller to override the default
computation of degree centrality for a vertex.
*Requirements:* The `Measure` type must be a model of the
[DegreeMeasure] concept. The `degree_type` of `Measure` must
be the same as the `value_type` of the `CentralityMap` type.
]
]
]
[heading Return Value]
The `degree_centrality()` function returns the centrality of a vertex.
If no `measure` parameter is given, the return type is the same as the
`degree_size_type` of the `Graph`. Otherwise, it is the `degree_type` of
the `Measure`.
[heading Complexity]
The `degree_centrality()` returns in /O(M)/. If the measure is not given,
or the measure is given as `influence_measure` or `prestige_measure`, this function
returns in constant time.
[endsect]
[section [^influence()]]
template <typename Graph, typename Vertex>
typename graph_traits<Graph>::degree_size_type
influence(const Graph& g, Vertex v);
The `influence()` function the influence of the vertex to the caller. The influence
of a vertex is the same as its out-degree.
[heading Parameters]
[table
[[Type] [Parameter] [Description]]
[
[required, in] [`const Graph& graph`]
[
The graph object for which the degree centralities will be computed.
*Requirements:* The `Graph` type must be a model of the [IncidenceGraph]
concept. If the `Measure` parameter is given as `prestige_measure`,
then the graph must be a model of the [BidirectionalGraph] concept.
]
]
[
[required, in] [`Vertex v`]
[
The vertex descriptor for which the degree centrality is computed.
*Requirements:* The `Vertex` type must be the same as the `vertex_descriptor`
of the `Graph` parameter.
]
]
]
[heading Return Value]
The `influence()` function returns the influence of the vertex, which is the same
as its out-degree.
[heading Complexity]
The `influence()` function returns in constant time.
[endsect]
[section [^prestige()]]
template <typename Graph, typename Vertex>
typename graph_traits<Graph>::degree_size_type
prestige(const Graph& g, Vertex v);
The `prestige()` function the prestige of the vertex to the caller. The prestige
of a vertex is the same as its in-degree.
[heading Parameters]
[table
[[Type] [Parameter] [Description]]
[
[required, in] [`const Graph& graph`]
[
The graph object for which the degree centralities will be computed.
*Requirements:* The `Graph` type must be a model of the [IncidenceGraph]
concept. If the `Measure` parameter is given as `prestige_measure`,
then the graph must be a model of the [BidirectionalGraph] concept.
]
]
[
[required, in] [`Vertex v`]
[
The vertex descriptor for which the degree centrality is computed.
*Requirements:* The `Vertex` type must be the same as the `vertex_descriptor`
of the `Graph` parameter.
]
]
]
[heading Return Value]
The `prestige()` function returns the influence of the vertex, which is the same
as its out-degree.
[heading Complexity]
The `prestige()` returns in constant time.
[endsect]
[section [^all_degree_centralities()]]
#include <boost/graph/degree_centrality.hpp>
template <typename Graph, typename CentralityMap>
void all_degree_centralities(const Graph& g, CentralityMap cent);
template <typename Graph, typename CentralityMap, typename Measure>
void all_degree_centralities(const Graph& g, CentralityMap cent, Measure measure);
The `all_degree_centralities()` computes the degree centrality for each vertex in the
graph, assigning the values to an output property map. Optionally, a degree measure
can be given, specializing the computation of degree centrality. See the
[degree_centrality_measures] section for more details on the influence and
prestige measures.
[heading Parameters]
[table
[[Type] [Parameter] [Description]]
[
[required, in] [`const Graph& graph`]
[
The graph object for which the degree centralities will be computed.
*Requirements:* The `Graph` type must model the [VertexListGraph]
and [IncidenceGraph] concepts. If the `Measure` parameter is given
as `prestige_measure`, then the graph must be a model of the
[BidirectionalGraph] concept.
]
]
[
[required, out] [`CentralityMap cent`]
[
The property map that contains the degree centralities of all vertices
in a graph after calling the `degree_centrality()` function.
*Requirements:* The `CentralityMap` type must be a model of the
[WritablePropertyMap] concept. The `key_type` of the map must
be the same as the `vertex_descriptor` of the `Graph` type. If the
`Measure` parameter is given in the call, then the `value_type`
of `CentralityMap` must be the same as `Measure::degree_type`.
Otherwise, the `value_type` must be the same as the `degree_size_type`
of the graph.
]
]
[
[optional, in] [`Measure measure`]
[
The `measure` parameter allows the caller to override the default
computation of degree centrality for a vertex.
*Requirements:* The `Measure` type must be a model of the
[DegreeMeasure] concept. The `degree_type` of `Measure` must
be the same as the `value_type` of the `CentralityMap` type.
]
]
]
[heading Complexity]
The `all_degree_centralities()` function returns in /O(n*O(M))/ where /n/ is the
number of vertices in the graph and /M/ is a measure of a vertex. If the measure
is not specified or is `influence_measure` or `prestige_measure`, this function
returns in linear time.
[endsect]
[section [^all_influence_values()]]
#include <boost/graph/degree_centrality.hpp>
template <typename Graph, typename CentralityMap>
void all_influence_values(const Graph& g, CentralityMap cent);
The `all_influence_values()` function computes the influence of each vertex in the
graph, assigning the values to an output property map.
[heading Parameters]
[table
[[Type] [Parameter] [Description]]
[
[required, in] [`const Graph& graph`]
[
The graph object for which the degree centralities will be computed.
*Requirements:* The `Graph` type must model the [VertexListGraph]
and [IncidenceGraph] concepts. If the `Measure` parameter is given
as `prestige_measure`, then the graph must be a model of the
[BidirectionalGraph] concept.
]
]
[
[required, out] [`CentralityMap cent`]
[
The property map that contains the degree centralities of all vertices
in a graph after calling the `degree_centrality()` function.
*Requirements:* The `CentralityMap` type must be a model of the
[WritablePropertyMap] concept. The `key_type` of the map must
be the same as the `vertex_descriptor` of the `Graph` type. If the
`Measure` parameter is given in the call, then the `value_type`
of `CentralityMap` must be the same as `Measure::degree_type`.
Otherwise, the `value_type` must be the same as the `degree_size_type`
of the graph.
]
]
]
[heading Complexity]
The `all_influence_values()` function returns linear time with respect to the
number of vertices in the graph.
[endsect]
[section [^all_prestige_values()]]
#include <boost/graph/degree_centrality.hpp>
template <typename Graph, typename CentralityMap>
void all_prestige_values(const Graph& g, CentralityMap cent);
The `all_prestige_values()` function computes the prestige of each vertex in the
graph, assigning the values to an output property map.
[heading Parameters]
[table
[[Type] [Parameter] [Description]]
[
[required, in] [`const Graph& graph`]
[
The graph object for which the degree centralities will be computed.
*Requirements:* The `Graph` type must model the [VertexListGraph]
and [IncidenceGraph] concepts. If the `Measure` parameter is given
as `prestige_measure`, then the graph must be a model of the
[BidirectionalGraph] concept.
]
]
[
[required, out] [`CentralityMap cent`]
[
The property map that contains the degree centralities of all vertices
in a graph after calling the `degree_centrality()` function.
*Requirements:* The `CentralityMap` type must be a model of the
[WritablePropertyMap] concept. The `key_type` of the map must
be the same as the `vertex_descriptor` of the `Graph` type. If the
`Measure` parameter is given in the call, then the `value_type`
of `CentralityMap` must be the same as `Measure::degree_type`.
Otherwise, the `value_type` must be the same as the `degree_size_type`
of the graph.
]
]
]
[heading Complexity]
The `all_prestige_values()` function returns linear time with respect to the
number of vertices in the graph.
[endsect]
[section Measures]
template <typename Graph> struct influence_measure;
template <typename Graph> struct prestige_measure;
template <typename Graph>
influence_measure<Graph> measure_influence(const Graph&);
template <typename Graph>
prestige_measure<Graph> measure_prestige(const Graph&);
These functions in this module can be specialized throught the use of different
/measure/ functors. These funtors are called to compute the degree centraltiy of a
vertex within the computational framework. This module provides the two default
measures for /influence/ and /prestige/ and functions to help instantiate them.
Custom measures can also be used.
The `influence_measure` returns the out-degree of a vertex. Using this measure
requires the `Graph` to model the [IncidenceGraph] concept.
The `prestige_measure` returns the in-degree of a vertex. Using this measure
requires the `Graph` to model the [BidirectionalGraph] concept.
The `measure_influence()` and `measure_prestige()` functions can be used to
create the corresponding measure objects.
[endsect]
[section Examples]
[heading Degree Centrality]
This example reads a graph from standard input (in a simple edge-list format),
computes the degree-centrality of each vertex, and prints them to standard
output. This example includes of the following files:
* [^examples/degreee_centrality.hpp]
* [^examples/helper.hpp]
[degree_centrality_example]
If given the `social_network.graph` file as input, the program will produce the
following output:
[pre
Scott: 4
Jill: 2
Mary: 2
Bill: 2
Josh: 2
Frank: 5
Laurie: 1
Anne: 2
Howard: 2
]
[heading Influence and Prestige]
This example is nearly identical to the previous in that it computes the degree
centrality over vertices in a graph. However, this example constructs a directed
graph and computes both influence and prestige. This example consist of the following
files:
* [^examples/influence_prestige.hpp]
* [^examples/helper.hpp]
[influence_prestige_example]
If given the `info_network.graph` file as input, the program will produce
the following output where the second column is influence and the third is prestige:
[pre
myspace 1 0
digg 2 2
blogger 3 0
slahsdot 0 1
wikipedia 0 5
slashdot 2 2
blogspot 2 0
bbc 1 1
]
[endsect]
[endsect]