Preface
As humans, we build relationships with one another. A fruitful line of research
in the last 90 years has been to consider the structure that emerges from the aggregate
of these relationships for a group of individuals, their social network. These networks
are social because they are the result of human individual and social properties and
so are quite different in structure from other kinds of networks (computer networks,
electrical distribution networks, and so on).
The payoffs from analyzing social networks come because, although each relationship is formed based on the purely local choice of the individuals concerned,
the resulting structure has properties that are not obvious from the individual relationships. These properties do not come from nowhere; rather they are consequences
of deep human properties of the way that we, as humans, relate to one another. Thus
social network analysis reveals much about the way that humans in groups arrange
ourselves, mostly unconsciously. Understanding the macro-structure of a social network also provides a context to revisit each individual relationship, and understand
it in a new way.
Much of the technical work on analyzing social networks has modelled the
relationship between a pair of individuals as being of only a single kind. For example, Facebook considers all relationship as “friends” while LinkedIn considers
all relationships as “colleagues”. Even in a tribal society, the archetypal social network setting, relationships are of multiple kinds: hierarchical control relationships,
relatives, friends, hunting partners, and so on. In today’s world, we also have relationships with others of multiple kinds; all of those present in tribal societies, but also
team members at work, telephone acquaintances, fellow enthusiasts about sports or
hobbies, as well as variations of online connections. It is reasonable to consider an
individual as participating in a collection of non-intersecting social networks, one at
work, one in social life, one in family gatherings, but these networks often overlap
— the same person can be both a friend and a colleague. To gather these separate
social networks into a larger one requires a way to represent all of these different
kinds of connections within a single framework.
We have already hinted that we need a way to model relationships of qualitatively different kinds (friends, family, colleagues). Relationships also naturally have
an associated intensity; A’s relationship to B may be much stronger or more important than A’s relationship to C, and we need to be able to model this. Relationships
can also be asymmetric: the intensity of A’s relationship to B may be different from
xi
As humans, we build relationships with one another. A fruitful line of research
in the last 90 years has been to consider the structure that emerges from the aggregate
of these relationships for a group of individuals, their social network. These networks
are social because they are the result of human individual and social properties and
so are quite different in structure from other kinds of networks (computer networks,
electrical distribution networks, and so on).
The payoffs from analyzing social networks come because, although each relationship is formed based on the purely local choice of the individuals concerned,
the resulting structure has properties that are not obvious from the individual relationships. These properties do not come from nowhere; rather they are consequences
of deep human properties of the way that we, as humans, relate to one another. Thus
social network analysis reveals much about the way that humans in groups arrange
ourselves, mostly unconsciously. Understanding the macro-structure of a social network also provides a context to revisit each individual relationship, and understand
it in a new way.
Much of the technical work on analyzing social networks has modelled the
relationship between a pair of individuals as being of only a single kind. For example, Facebook considers all relationship as “friends” while LinkedIn considers
all relationships as “colleagues”. Even in a tribal society, the archetypal social network setting, relationships are of multiple kinds: hierarchical control relationships,
relatives, friends, hunting partners, and so on. In today’s world, we also have relationships with others of multiple kinds; all of those present in tribal societies, but also
team members at work, telephone acquaintances, fellow enthusiasts about sports or
hobbies, as well as variations of online connections. It is reasonable to consider an
individual as participating in a collection of non-intersecting social networks, one at
work, one in social life, one in family gatherings, but these networks often overlap
— the same person can be both a friend and a colleague. To gather these separate
social networks into a larger one requires a way to represent all of these different
kinds of connections within a single framework.
We have already hinted that we need a way to model relationships of qualitatively different kinds (friends, family, colleagues). Relationships also naturally have
an associated intensity; A’s relationship to B may be much stronger or more important than A’s relationship to C, and we need to be able to model this. Relationships
can also be asymmetric: the intensity of A’s relationship to B may be different from
xi