Skip Headers
Oracle® Communications Data Model Adapters and Analytics User's Guide
Release 11.3.2

E37699-03
Go to Documentation Home
Home
Go to Book List
Book List
Go to Table of Contents
Contents
Go to Index
Index
Go to Feedback page
Contact Us

Go to previous page
Previous
Go to next page
Next
PDF · Mobi · ePub

7 Social Network Analytics

Oracle Communications Data Model provides the Social Network Analytics (SNA) Add-On adapter that is optimized for social network analysis across large volumes of call record data (CDR) associated with an Oracle Communications Data Model data warehouse.

Overview of Oracle Communications Data Model Social Network Analytics

Oracle Communications Data Model Social Network Analytics supports the following:

About Social Network Analytics Features

  • Identification of social network communities.

  • Network metrics characterizing the social graph.

  • Network distribution information.

  • Predictive scores for churn and influence at a node level, and potential revenue/value at risk.

  • User interface targeted at business users and flexible ad-hoc reporting.

You can use Oracle Communications Data Model Social Network Analytics and the associated reports and tools to assist you with:

  • Acquiring Customers

    • Member-get-member campaigns

    • Profile offnet prospects using community membership information

  • Retaining Customers

    • Enhance churn prediction capabilities

    • Focus retention marketing on high social network revenue segments

    • Augment existing customer segmentation capabilities

    • Identify rotational churners through social signatures

  • Growing Your Customer Base

    • Increase conversion rates for cross-sell and up-sell campaigns

    • More focused product offer development

Social Network Analytics Opt-Out Features

  • Opt-Out field added at subscriber level to help address privacy concerns.

  • All members still included in aggregate analysis (that is, when determining communities, connections, and so on).

  • Individual scores not computed, and subscribers excluded from reports with individual subscriber details.

SNA Concepts and Terms

Table 7-1 SNA Concepts and Terms

SNA Concept or Term Description

Churn Pressure

Measure by number of immediate neighbors for a node that have churned recently (during the period used to construct the graph)

Connectedness

The average number of connections of a node's immediate neighbors

Connections

The number of direct connections a node has with other nodes (aka, first-order centrality

Degree Influence

Number of connections a node has (aka, degree centrality

Eigen Influence

Correlated with degree influence, but taking into account the pattern of connectivity around the node; For example, A may have more connections than B, but if B is located in highly connected region of the graph while A is located in isolated region, A would have higher degree influence but lower Eigen influence

Network Value

individual node value, also factoring in the value of a node's immediate neighbors

Network Value at Risk

Individual node's network value multiplied by that node's churn probability

Product Viral Influence

The number of times an immediate neighbor acquires a product within a pre-determined window of time after the original node had purchased that product. Returns an integer score for each product considered in the analysis

Subscriber Viral Influence

Computed as the number of times an immediate neighbor of a node acquires a product (any product) within a pre-determined window of time after the original node had purchased the product.

Value

Monetary measurement of a node's value or importance. User definable; typically will use ARPU or a subscribers lifetime value

Value at Risk

the value of a node multiplied by its probability of leaving the network (churning)

Viral Influence

Measure of a node's ability to influence others in a network, and the degree the adoption of a product displayed viral behavior characteristics


Social Network Analytics Process Flow

Overview of Social Network Analytics Process Flow

Figure 7-1 shows the Oracle Communications Data Model Social Network Analytics process flow and major components. The process flow includes steps as follows:

  1. Call detail records, subscriber demographics, purchase history, and so on are loaded into Oracle Communications Data Model foundation tables.

  2. Oracle Communications Data Model intra-ETL populates Social Network Analytics input tables

  3. OBIEE dashboard used to enter Social Network Analytics input parameters

  4. Social Network Analytics application back-end process schedules jobs to run through mining models; populates results to Social Network Analytics output tables.

  5. OBIEE dashboards used to display Social Network Analytics results.

Figure 7-1 Social Network Analytics Process Flow

Description of Figure 7-1 follows
Description of "Figure 7-1 Social Network Analytics Process Flow"

About SNA Roles

In social network analysis, a role describes the position of a node in relationship to its neighbors and the network as a whole. For example, common metrics used for determining node roles include a node's centrality, popularity, and authority.

SNA Subscriber Roles Leaders and Role Strength

The Oracle Communications Data Model SNA add-on provides community-based node roles based on Scripps, Tan and Esfahanian (2007)Foot 1 . Thus, SNA defines the community-based role of a node according to the number of communities they connect to and the number of connections they have to direct neighbors.

As shown in Figure 7-2, SNA defines four node roles by dividing a community-degree chart into four quadrants. Each quadrant is assigned a different role from the available roles listed in Table 7-2. The vertical axis represents relative degree, a measure of how many immediate neighbors a node has. This measure is scaled between 0 and 1. The horizontal axis is a community score representing how connected a node is to other communities. This metric is also scaled between 0 and 1. The four quadrants are defined by spitting the graph vertically and horizontally at 0.5.

SNA Role Strength

In addition to assigning a role to a node SNA provides a measure of role strength. The higher this measure the more the node displays the characteristics of the role. With respect to the graph in Figure 7-2, the closer a node is to one of the corners the stronger the role strength for the role assigned to it. For example, a node that is placed close to the upper left hand corner based on its value for relative degree and community score will be assigned a role of Big Fish with role strength close to 1. A node classified as Big Fish but with both relative degree and community score close to 0.5 will have role strength close to 0.

SNA Leaders

A node is considered to be a leader in a community if it is the most connected node in the community and it is also substantively more connected than the second most connected node in the community.

Figure 7-2 SNA Subscriber Node Roles

Description of Figure 7-2 follows
Description of "Figure 7-2 SNA Subscriber Node Roles"

Table 7-2 SNA Role Descriptions

Role Description

Ambassador

These are nodes with a high degree score, high number of connections, and a high community score (connections with nodes in other communities). In this way they act as ambassadors, between many different communities. They also help propagate information rapidly over the network.

Big Fish

These are nodes that are very important only within a community. They are very connected, high degree score, to nodes in their own community but have very few connections with nodes in other communities (low community score). The term comes from the cliché "big fish in a small pond."

Bridge

These are nodes with a low degree of connections, small by number of connections, but a high community score (a relatively high number of connections to other communities). As such, they behave as bridges between small numbers of communities.

Loner

These are nodes with a low degree of connections and most of them are only in their own communities.




Footnote Legend

Footnote 1: Scripps, Tan and Esfahanian (2007). Exploration of Link Structure and Community-Based Node Roles in Network Analysis. ICDM 2007. Seventh IEEE International Conference on Data Mining.