V - AI&DS UNIT I SNS

UNIT I INTRODUCTION
Introduction to Semantic Web: Limitations of current Web - Development of Semantic Web - Emergence of the
Social Web - Social Network analysis: Development of Social Network Analysis - Key concepts and measures
in network analysis - Historical overview of privacy and security, Major paradigms for understanding privacy
and security

1.0 Introduction to Semantic Web: [Explain about Sematic web. Compare the sematic web with
traditional approach.] (08)(Or) What is the need of semantic web? Explain with Example. Nov-
2019.

The Semantic Web is the application of advanced knowledge technologies to the Web and distributed
systems in general.
The Semantic Web provides a common framework that allows data to be shared and reused
across application, enterprise, and community boundaries. It is a collaborative effort led by W3C with
participation from a large number of researchers and industrial partners.

The Semantic Web is a Web of data. There is a lot of data we all use every day,
and it's not part of the Web. For example, I can see my bank statements on the web, and
my photographs, and I can see my appointments in a calendar. But can I see my photos in a
calendar to see what I was doing when I took them? Can I see bank statement lines in a
calendar? Why not? Because we don't have a web of data. Because data is controlled by
applications, and each application keeps it to itself.

The vision of the Semantic Web is to extend principles of the Web from documents
to data. Data should be accessed using the general Web architecture using, e.g., URI-s; data
should be related to one another just as documents (or portions of documents) are already.
This also means creation of a common framework that allows data to be shared and reused
across application, enterprise, and community boundaries, to be processed automatically
by tools as well as manually, including revealing possible new relationships among pieces
of data.
1.1 Limitations of the current Web

1.1 What’s wrong with the Web?

The questions below are specific for the sake of example, but they represent very general
categories of search tasks.

1. Who is Saran N?
To answer such a question using the Web one would go to the search engine and enter the most logical
keyword: Saran N. The results returned by Bing are shown in Figure 1.1 and Google by Figure 1.2
If this question and answer would be parts of a conversation, the dialogue would sound like this:

Q: Who is Saran N?

A: I don’t know but there are over a 155000 documents with the word “Saran N” on
them and I found them all really fast (0.95s)
V - AI&DS UNIT I SNS

Figure 1.3. Search results for the keyword paris using Google Image Search.

1.2 Diagnosis: A lack of knowledge

The questions above are arbitrary in their specificity but they illustrate a general problem in
accessing the vast amounts of information on the Web. Namely, in all five cases we deal with a
knowledge gap: what the computer understands and able to work with is much more limited than the
knowledge of the user.

2 .0 The semantic solution. : [Discuss the need of semantic solution and it’s different approaches.] (16)

The idea of the Semantic Web is to apply advanced knowledge technologies in order to fill the knowledge
gap between human and machine.

➢ This means providing knowledge in forms that computers can readily process and reason with.
➢ This knowledge can either be information in the content of the Web pages but difficult to extract
or additional background knowledge that can help to answer queries in some way.
➢ In the following we describe the improvement one could expect in case of our four queries based
on examples of existing tools and applications that have been implemented for specific domains
or organizational settings.
✓ The Friend-of-a-Friend (FOAF) project provides a widely accepted vocabulary for such
descriptions.
✓ FOAF profiles includes attributes such as the name, address, interests of the user can be linked to
the web page or even encoded in the text of the page.
✓ As we will see several profiles may also exist on the Web describing the same person.
✓ As all profiles are readable and comparable by machines, all knowledge about a person can be
combined automatically.
V - AI&DS UNIT I
SNS

For example, Frank van Harmelen has such a profile attached to his homepage on the Web. This allows a
search engine to determine that the page in question is about a person with specific attributes. (Thus pages
about persons and villages would not be confused.) Assuming that all other van Harmelens on the Web would
provide similar information, the confusion among them could also be easily avoided. In particular, the
search engine could alert us to the ambiguity of our question and ask for some extra information about the
person we are looking for. The discussion with the search engine would be very different:

Q: Who is Frank van Harmelen?

A: Your question is ambiguous: there is a great deal of information about a Frank van Harmelen
who is a professor at the Vrije Universiteit. However, there are other persons named Harmelen and
also a village in the munici- pality of Woerden. Which one did you mean?

✓ Similarly, the solution in the second case is to attach metadata to the images in question.
✓ For example, the online photo sharing site Flickr allows to annotate images using geographic
coordinates. After uploading some photos users can add keywords to describe their images (e.g.
“Paris, Eiffel-tower”) and drag and drop the images on a geographic map to indicate the location
where the photo was taken.
✓ Searching of photos of Paris becomes a breeze: we can look up Paris on the map and see what other
photos have been put there by other users.
✓ Although in this case the system is not even aware that Paris is a city, a minimal additional
information about photos (the geo-coordinates) enables a kind of visualization that makes the
searching task much easier.
✓ And if over time the system notes that most images with the keyword “Paris” fall in a specific
geographic area on the map, it can even conclude that Paris is a place on the map.

Figure 1.4. Searching for the keyword Paris using the geographic search of Flickr.
V - AI&DS UNIT I SNS

2.1 Development of the Semantic Web. [Discuss the different development approaches of
Semantic web. (12)]

2.1.1 Research, development and standardization

✓ The vision of extending the current human-focused Web with machine processable descriptions
of web content has been first formulated in 1996 by Tim Berners-Lee, the original inventor of
the Web [BLFD99].
✓ The Semantic Web has been actively promoted since by the World Wide Web Consortium (also
led by Berners-Lee), the organization that is chiefly responsible for setting technical standards on
the Web.
✓ As a result of this initial impetus and the expected benefits of a more intelligent Web, the
Semantic Web has quickly attracted significant interest from funding agencies on both sides of
the Atlantic, reshaping much of the AI research agenda in a relatively short period of time.
✓ In particular, the field of Knowledge Representation and Reasoning took center stage, but
outcomes from other fields of AI have also been put into to use to support the move towards the
Semantic Web: for example, Natural Language Processing and Information Retrieval have been
applied to acquiring knowledge from the World Wide Web.
✓ The core technology of the Semantic Web, logic-based languages for knowledge representation and
reasoning have been developed in the research field of Artificial Intelligence.
✓ As the potential for connecting information sources on a Web-scale emerged, the languages that have
been used in the past to describe the content of the knowledge bases of stand-alone expert systems have
been adapted to the open, distributed environment of the Web.
✓ Since the exchange of knowledge in standard languages is crucial for the interoperability of tools and
services on the Semantic Web, these languages have been standardized by the W3C as a layered set of
languages.

2.1.2 Technology adoption

✓ The Semantic Web was originally conceptualized as an extension of the current Web, i.e. as the
application of metadata for describing Web content.
✓ That the Semantic Web is formulated as a vision points to the problem of boot- strapping the
Semantic Web.
✓ Part of the problem is that as a technology for developers, users of the Web never experience the
Semantic Web directly, which makes it difficult to convey Semantic Web technology to
stakeholders.
✓ Further, most of the times the gains for developers are achieved over the long term, i.e. when
data and services need to reused and re-purposed.
✓ More significantly, as many other modern technologies, the Semantic Web suffers from what the
economist Kevin Kelly calls the fax-effect.
✓ So is it with the Semantic Web: at the beginning the price of technological investment is very high.
One has to adapt the new technology which requires an investment in learning.
✓ What makes the case of the Semantic Web more difficult, however, is an additional cost factor.
✓ While the research effort behind the Semantic Web is immense and growing dynamically,
Semantic Web technology has yet to see mainstream use on the Web and in the enterprise.
✓ To follow the popularity of Semantic Web related concepts and Semantic Web standards on the
Web, we have executed a set of temporal queries using the search engine Altavista.
V - AI&DS UNIT I SNS

✓ The queries contained single terms plus a disambiguation term where it was necessary.
✓ We have divided all counts with the number of documents with the word web to account for the
general growth of the Web.
2.2 The emergence of the social web (Or) Discuss about the limitation of current web and emergence of
social web?

✓ the Web was a read-only medium for a majority of users.

✓ The web of the 1990s was much like the combination of a phone book and the yellow pages (a mix
of individual postings and corporate catalogs) and despite the connecting power of hyperlinks it
instilled little sense of community among its users.
✓ This passive attitude toward the Web was broken by a series of changes in usage patterns and
technology that are now referred to as Web 2.0.
✓ The resulting increase in our capacity to obtain information and social support online can be
quantified.
✓ The first wave of socialization on the Web was due to the appearance of blogs, wikis and other
forms of web-based communication and collaboration.
✓ Blogs and wikis attracted mass popularity from around 2003, for adding content to the Web:
editing blogs and wikis did not require any knowledge of HTML any more.
✓ Blogs and wikis allowed individuals and groups to claim their personal space on the Web and fill
it with content at relative ease.
✓ The first online social networks (also referred to as social networking services) entered the field
at the same time as blogging and wikis started to take off.
✓ In 2003, the first-mover Friendster attracted over five million registered users in the span of a
few months, which was followed by Google and Microsoft starting or announcing similar services.
✓ Although these sites feature much of the same content that appear on personal web pages, they
provide a central point of access and bring structure in the process of personal information
sharing and online socialization.
✓ Following registration, these sites allow users to post a profile with basic information, to invite
others to register and to link to the profiles of their friends.
✓ The system also makes it possible to visualize and browse the resulting network in order to
discover friends in common, friends thought to be lost or potential new friendships based on
shared interests.
✓ These vastly popular systems allow users to maintain large networks of personal and business
contacts and started relationship among them.
✓ Explicit user profiles make it possible for these systems to introduce rating mechanism whereby
either the users or their contributions are ranked according to usefulness or trustworthiness.
✓ Ratings are explicit forms of social capital that regulate exchanges in online communities in
much the same way that reputation moderates exchanges in the real world.
✓ For example, in the technology news platform users give thumbs up or thumbs down to news
items provided by other users. Further, it provides a way for users to see what other users are
’digging’ real-time, catering to the creation of a similar sense of simultaneous social presence as
in the case of instant messaging.
✓ The design and implementation of Web applications have also evolved in order to make the user
experience of interacting with the Web as smooth as possible.
V - AI&DS UNIT I SNS

2.2..1 Web 2.0 + Semantic Web = Web 3.0?

✓ Web 2.0 is often contrasted to the Semantic Web, which is a more conscious and carefully
orchestrated effort on the side of the W3C to trigger a new stage of developments using semantic
technologies.
✓ In practice the ideas of Web 2.0 and the Semantic Web are not exclusive alternatives: while Web
2.0 mostly effects how users interact with the Web, while the Semantic Web opens new
technological opportunities for web developers in combining data and services from different
sources.
✓ In the following, we point out some of the opportunities that arise by the combination of ideas
from these two developments.
1. Firstly, we note that the basic lesson of Web 2.0 is that users are willing to provide
content as well as metadata. This may take the form articles and facts organized in tables and
categories in Wikipedia, photos organized in sets and according to tags in Flickr or structured
information embedded into homepages and blog postings using microformats.
2. Secondly, due to the extensive collaborations online many applications have access
to significantly more metadata about the users. Information about the choices, preferences,
tastes and social networks of users means that the new breed of applications are able to build
on a much richer user profiles. Clearly, semantic technology can help in matching users with
similar interests as well as matching users with available content. Semantics, however, is not
enough: as Golbeck shows, there are components of trust that are beyond of what can be
inferred based on profile similarity alone [GH06]. Therefore social-semantic systems that can
provide recommendations based on both the social network of users and their personal
profiles are likely to outperform traditional recommender systems as well as purely
network-based trust mechanisms.
3. Lastly, in terms of technology what the Semantic Web can offer to the Web 2.0
community is a standard infrastructure for the building creative combinations of data and
services. Standard formats for exchanging data and schema information, support for data
integration, along with standard query languages and protocols for querying remote data
sources provide a platform for the easy development of mashups.

✓ The Semantic Web first appeared ten years ago as the ambitious vision of a future Web where
all content is described in a way that computers can easily access, including the information that
is currently only described in human language, captured by images, audio and video, or locked
up in the many databases behind complex websites.

Although the Web has been already naturally evolving toward a separation of content and
representation, the vision of the Semantic Web is rather revolutionary: not only major technological
changes are required to realize this vision, but also a social adoption process had to be bootstrapped.
In terms of technology, the past years have seen the formation of a substantial Semantic Web research
community that delivers the innovative solutions required to realize the Semantic Web. The World Wide
Web Consortium (W3C) as the main body of standardization in the area of web technologies has also
proceeded to issue official recommendations about the key languages and protocols that would
guarantee interoperability on the Semantic Web.
V - AI&DS UNIT I SNS

3.0 Social Network analysis. [Define Social network analysis and explain the different approaches
involved in it. / What are the key concepts and measures in network analysis? (Or) Discuss
key concepts and measures in network analysis.

What is network analysis?

✓ Social Network Analysis (SNS) is the study of social relations among a set of actors.
✓ The key difference between network analysis and other approaches to social science is the focus
on relationships between actors rather than the attributes of individual actors.
✓ Network analysis takes a global view on social structures based on the belief that types and
patterns of relationships emerge from individual connectivity and that the presence (or absence)
of such types and patterns have substantial effects on the network and its constituents.

SNS is thus a different approach to social phenomena and therefore requires a new set of concepts
and new methods for data collection and analysis.
✓ Network analysis provides a vocabulary for describing social structures, provides formal
models that capture the common properties of all (social) networks and a set of methods
applicable to the analysis of networks in general.
✓ The concepts and methods of network analysis are grounded in a formal description of
networks as graphs.
✓ Methods of analysis primarily originate from graph theory as these are applied to the graph
representation of social network data. (Network analysis also applies statistical and
probabilistic methods and to a lesser extent algebraic techniques.)

Social Network Analysis is a branch of sociology that is formalized to a great extent based on the
mathematical notions of graph theory.

3.1. Development of Social Network Analysis

✓ The field of Social Network Analysis today is the result of the convergence of several streams of
applied research in sociology, social psychology and anthropology.
✓ Many of the concepts of network analysis have been developed independently by various
researchers often through empirical studies of various social settings.
✓ The vocabulary, models and methods of network analysis also expand continuously through
applications that require to handle ever more complex data sets.
✓ An example of this process are the advances in dealing with longitudinal data.
✓ New probabilistic models are capable of modelling the evolution of social networks and
answering questions regarding the dynamics of communities.

3.2 Key concepts and measures in network analysis

✓ Social Network Analysis has developed a set of concepts and methods specific to the analysis of
social networks.
✓ In the following, we introduce the most basic notions of network analysis and the methods from the
macro level to the micro level.

3.2.1 The global structure of networks

A (social) network can be represented as a graph G = (V, E) where V denotes the finite set of
vertices and E denotes a finite set of edges such that E ⊆ V × V . Recall that each graph can be
associated with its characteristic matrix
V - AI&DS UNIT I SNS

Some network analysis methods are easier to understand when we conceptualize graphs as matrices (see
Figure 3.1).

Figure 3.1. Most network analysis methods work on an abstract, graph based representation of real
world networks.

✓ Milgram (American Psychologist) went out to test the common observation that no matter where
we live, the world around us seems to be small: we routinely encounter persons not known to us
who turn out to be the friends of our friends.
✓ Milgram thus not only wanted to test whether we are in fact all connected but he was also
interested in what is the average distance between any two individuals in the social network of
the American society.
In order to find out he devised an experiment in which he gave letters to a few hundred randomly
selected persons from Boston in Massachusetts and Omaha in the state of Nebraska.
✓ The participants were asked to send the letter to a single target person, namely a stockbroker in
Sharon, Massachusetts.
✓ They were not allowed to send it directly to the target person, however. Instead every participant
was asked to forward the letter to a person he or she knew on a first name basis.
✓ That person would then also need to follow the same instructions and send the letter to someone
who was more likely to be acquainted with the stock broker.
✓ In the end, the letter would reach someone who knew the target person in question and would
V - AI&DS UNIT I SNS

hand the letter to him. In other words, Milgram devised a chain-mail similar to the ones that now
aimlessly circle the planet. However, these letters had a target and the chains stopped when they
reached their final destination.

✓ Milgram calculated the average of the length of the chains and concluded that the experiment
showed that on average Americans are no more than six steps apart from each other.
✓ While this is also the source of the expression six degrees of separation the actual number is
rather dubious: not only was Milgram’s sample too small, but even only 20% of the those letters
have made it to their destination.
✓ Thus the number could be actually larger: those letters that did not make it would probably have
resulted in longer paths. But the number could be also smaller as it is not guaranteed that the
letters have travelled the shortest possible path from their source to the tar- get. Still, Milgram’s
experiment had a tremendous impact on social network research and sociology as a whole as it
showed that the number is orders of magnitude smaller than the size of the network.

• Formally, what Milgram estimated is the size of the average shortest path of the network,
which is also called characteristic path length.
• An open (simple) path in a graph is a sequence of vertices vi0 , vi2 , . . . , vin such that ∀j =
0 . . . n −1(vij , vij+1 ) ∈ E and ∀j, k = vij j= vik , in other words every vertex is connected to
the next vertex and no vertex is repeated on the path.
• The shortest path between two vertices vs and vt is a path that begins at the vertex vs and ends
in the vertex vt and contains the least possible number of vertices.
• The shortest path between two vertices is also called a geodesic.
• The longest geodesic in the graph is called the diameter of the graph: this is the maximum
number of steps that is required between any two nodes. The average shortest path is the average
of the length of the geodesics between all pairs of vertices in the graph. (This value is not possible
to calculate if the graph is not (strongly) connected, i.e. in case there exists a pair of vertices with
no path between them.)

The two dimensional lattice model shown in Figure 3.2, for example, does not have the small world
2
property: for a network of size n the characteristic path length is 3 ∗ √𝑛 which is still too large a
number to fit the empirical finding.

Figure 3.2. The 2D lattice model of networks (left). By connecting the nodes on the opposite borders of the
lattice we get a toroidal lattice (right).
V - AI&DS UNIT I SNS

Another simple model could be the tree graph shown in Figure 3.3 However, a tree is unrealistic
because it shows no clustering: we all know from practice that our friends are likely to know each other
as well because we tend to socialize in groups.
• Clustering for a single vertex can be measured by the actual number of the edges between the
neighbors of a vertex divided by the possible number of edges between the neighbors.
• When taken the average over all vertices we get to the measure known as clustering coefficient.
The clustering coefficient of a tree is zero, which is easy to see if we consider that there are no
triangles of edges (triads) in the graph. In a tree, it would never be the case that our friends are
friends with each other.
• The lattice and the tree also have the rather unappealing characteristic that every node has the
same number of connections.
• The random graph model can be generated by taking a set of vertices with no edges connecting
them. Subsequently, edges are added by picking pairs of nodes with equal probability. This way
we create a graph where each pair of vertices will be connected with an equal probability. (This
probability is a parameter of the process.)

Figure 3.3. A tree is a connected graph where there are no loops and paths leading from a vertex to
itself.

Alpha Model
• The alpha model was successful in generating graphs with small path lengths and relative large
clustering coefficients.
• The beta-model is also generative and it starts with a one-dimensional toroidal lattice where
every node is connected not only to its neighbors but also to the neighbors of its neighbors.
• By choosing beta appropriately, the beta model allows to generate networks with small path
lengths and relatively large clustering coefficients.
• While the alpha and beta models could be considered perfect by these characteristics alone, they
too fail to recreate an important feature of networks in nature: the scale-free characteristic of
the degree distribution.
• To understand the scale-free phenomenon we have to look at the degree distribution of
networks. Such a diagram shows how many nodes in the network have a certain number of
neighbors (degrees).
• In other words, the higher the degree the least likely it is to occur and the vast majority of the
nodes have much fewer connections than the few hubs of the network. The exact correlation is a
power law,
i.e. p(d) = d−k where k > 0 is a parameter of the distribution.
V - AI&DS UNIT I SNS

As an example, the upper part of Figure 3.4 shows the degree distribution for the co-authorship
network of the Semantic Web research community. The chart is a histogram showing how many
people have a certain degree, i.e. a certain number of co-authors. The chart shows that most people
have only one or two co-authors while only very few researchers have more than twenty co-authors.
(One co-author is slightly less likely than two because the data comes from publications and the
average publication has about two authors.) The lower part of the Figure shows the degree distribution
of a random network of similar size.

Figure 3.4 The degree distribution of a real world scale-free network (upper part) and the degree
distribution of a random network of the same size (lower part).

3.2.2 The macro-structure of social networks

✓ Based on the findings about the global characteristics of social networks we now have a good
impression about what they might look like.
✓ In particular, the image that emerges is one of dense clusters or social groups sparsely connected
to each other by a few ties as shown in Figure 3.5.
✓ For example, this is the image that appears if we investigate the co-authorship networks of a
scientific community. Bounded by limitations of space and resources, scientists mostly co-operate
with colleagues from the same institute.
V - AI&DS UNIT I SNS

Figure 3.5. Most real world networks show a structure where densely connected subgroups are linked
together by relatively few bridges
✓ Network visualizations based on topographic or physical principles can be helpful in
understanding the group structure of social networks and pinpoint hubs that naturally tend to
gravitate toward the center of the visualization.
✓ Unfortunately, computer generated network visualizations rarely show the kind of image seen
in Figure 3.5. Displays based on multi-dimensional scaling, for example, attempt to optimize the
visualization in a way that distances on the paper correlate with the distances between the nodes
in the graph.
✓ In general, the more dense the graph and the fewer the dimensions of the visualization the more
likely the graph will degenerate into a meaningless “spaghetti bowl” tangle of nodes and edges.
✓ Fortunately, visualizations are also not the only kind of tools in the network analysts toolkit to
uncover subgroup structure. Various clustering algorithms exist for creating disjunct or
overlapping subsets of nodes based on different definitions of a what a subgroup is or how to find
one.

Several definitions are based on the observations that subgroups are densely connected and their members
are close to each other in the graph.
• For example, a clique in a graph is maximal complete subgraph of three or more nodes. As complete
sub- graphs are very rare, the definition of a clique is typically relaxed by allowing some missing
connections.
• For example, a k-plex is a maximal subgraph in which each node is adjacent to no fewer than gs
− k nodes in the subgraph, where gs is the number of nodes in the subgraph. The larger we set
the parameter k, the larger the k-plexes that we will find.
• Yet another way of defining cohesiveness is to compare the density of ties within the group to the
density of ties between members of the subgroup and the outside.
• The lambda-set analysis method we will use in our work is based on the definition of edge
connectivity. Denoted with the symbol λ(i, j), the edge connectivity of two vertices vi and vj is
the minimum number of lines that need to be removed from a graph in order to leave no path
between the two vertices.
• A lambda-set is then defined as a set of nodes where any pair of nodes from the set has a larger
edge connectivity than any pair of nodes where one node is from within the set and the other
node is from outside the set. Unlike the above mentioned k-plexes,lambda-setsalso have the nice
property that they are not overlapping.
V - AI&DS UNIT I SNS

• The betweenness of an edge is calculated by taking the set of all shortest paths in the graph and
looking at what fraction of them contains the given edge.
• An edge between clusters has a much higher betweenness than edges inside clusters because all
shortest paths between nodes in the different clusters have to go through the given edge.
• By progressively removing the edges with the highest betweenness the graph falls apart in
distinct clusters of nodes.

Clustering a graph into subgroups allows us to visualize the connectivity at a group level. In some cases
we already have an idea of what this macro-structure might look like. A typical pattern that often emerges
in social studies is that of a Core-Periphery (C/P) structure.
• A C/P structure is one where nodes can be divided in two distinct subgroups: nodes in the core are
densely connected with each other and the nodes on the periphery, while peripheral nodes are not
connected with each other, only nodes in the core (see Figure 3.6).
1 . matrix.
• The matrix form of a core periphery structure is a
. 0

• Algorithms for identifying C/P structures and other block models (structural patterns) work by
dividing the set of nodes in a way that the error the between the actual image and the “perfect”
image is minimal. The result of the optimization is a classification of the nodes as core or periphery
and a measure of the error of the solution.

Figure 3.6. A Core-Periphery structure that would be perfect without the edge between nodes p1 and p2.

• Affiliation networks contain information about the relationships between two sets of nodes: a set
of subjects and a set of affiliations. An affiliation network can be formally represented as a bipartite
graph, also known as a two-mode network.
• In general, an n-partite graph or n-mode network is a graph G = (V, E) where there exists a
partitioning 𝑣 = ⋃𝑛 𝑉I Such that ⋂ 𝑛 𝑉 =0 and (Vi × Vi) ∩ E = 0. In other words, the set of
𝑖=1 𝑖 =1 i

vertices is divided into n disjoint sets and there are no edges between vertices belonging to the
same set.

There are relative few methods of analysis that are specific to affiliation networks; when dealing with
affiliation networks they are typically transformed directly to a regular, one-mode network. This
transformation considers the overlaps between the affiliations as a measure of tie strength between the
V - AI&DS UNIT I SNS

actors (see Figure 3.7). For example, we can generate a network among scientists by looking at how many
interests they have in common. We would place an edge between two researchers if they have interests in
common and would weight the edge according to the number of shared interests. The analysis of such a
one-mode network would follow as usual.

Figure 3.7. A two-mode network of persons and affiliation (shown as a graph and a matrix above) can
be folded into a regular social network by considering the overlaps in affiliations (graph and matrix
below).

Interestingly, we can also use this technique to map the relationship between the affiliations
themselves. In this case, we create a network of the affiliations as the nodes and draw edges between
the nodes based on the number of individuals who share that affiliations. This technique is commonly
used, for example, to map interlocking directorates, overlaps in the board membership of companies. In
such a setting the analysis starts with data on persons and their positions on the boards of various
companies. Then a network of companies is drawn based on potential interactions through shared
members on their boards

3.2.3 Personal networks

• In many cultures the term networking gained a negative connotation as a reference to nepotism,
the unfair access to advantages through “friends of friends”.
• However, among others in the American society being a good “networker” has become to be seen
as an important personal skill that every professional should master.
V - AI&DS UNIT I SNS

Figure 3.8. The site theyrule.net visualizes the interconnectedness of the American political, business and
media elite by showing connections between executive boards of various institutions and their board
members. Connections between individuals can be inferred by the boards in which they co-participate.

• These terms reflect one of the basic intuitions behind network analysis: that social structure plays a
key role in defining our opportunities and constraints with respect to access to key resources.
• In most settings network ties serve as important conduits for knowledge and other intangible
resources such as social support.
• More often than not establishing social links to those who control tangible resources in a community
provides competitive advantages to the individual, especially if those resources are scarce.
• In turn, a person with such access to key players will be increasingly sought after by other
individuals.
• Conversely, the absence of such relations means the lack of access to the intangible resources
“flowing” through social networks and to those tangible resources controlled by certain individuals
or groups.
• The structural dimension of social capital refers to patterns of relationships or positions that
provide benefits in terms of accessing large, important parts of the net- work.
• Common to structural measures of social capital is that they put a single node (the ego) in the center
and provide a measure for the node based on his connectivity to other nodes (the alters).
First method is a simple, but effective measure is the degree centrality of the node.
• Degree centrality equals the graph theoretic measure of degree, i.e. the number of (incoming,
outgoing or all) links of a node.
• This measure is based on the idea that an actor with a large number of links has wider and more
efficient access to the network, less reliant on single partners and because of his many ties often
participates in deals as a third-party or broker.
V - AI&DS UNIT I SNS

• Degree centrality does not take into account the wider context of the ego, and nodes with a high
degree may in fact be disconnected from large parts of the network. However, the degree measure
features prominently in the scale-free model, which makes it an important measure to investigate.

A second, more intuitive measure of centrality is closeness centrality,

• which is obtained by calculating the average (geodesic) distance of a node to all other nodes in
the network.
• Closeness centrality thus characterizes the reach of the ego to all other nodes of the network.
• In larger networks it makes sense to constrain the size of the neighborhood in which to measure
closeness centrality. It makes little sense, for example, to talk about the most central node on the
level of a society. The resulting measure is called local closeness centrality.

Two other measures of power and influence through networks are related to the similar advantages of
broker positions and weak ties.
• A broker gains advantage by standing in between disconnected communities. As we have seen
above ties spanning communities tend to be sparse compared to the dense networks of
cohesive subgroups and as a result there are typically few bridges across communities.
• Brokers controlling these bridges are said to be in an advantageous position especially because
of the value attributed to the information flowing across such ties.
• A structural hole occurs in the space that exists between closely clustered communities.
According to Burt, a broker gains advantage by bridging such holes. Therefore this measure
favors those nodes that connect a significant number of powerful, sparse linked actors with only
minimal investment in tie strength. Burt suggests that structural holes show information
benefits in three forms: access to large, disparate parts of the network (knowing where to turn
for information, who can use some information), timing (receiving information sooner than
others) and reputation through referrals.
• Density can be measured by computing the clustering coefficient measure introduced above on a
neighborhood of the individual.
• Further, there is evidence that these are extreme views: overly dense networks can lead to
overembeddedness. On the other hand, structural embeddedness is necessarily present as
through transitivity our friends are likely to develop ties of their own over time.

The aspect of tie strength is an example of the relational dimension of social capital, which concerns
the kind of personal relationships that people have developed with each other through a history of
interaction. The relationships of pairs of actors who occupy similar network positions in similar
network configurations may significantly differ based on their past interactions and thus their
possibilities for action might also differ.

• The Strength of Weak Ties , In the first study, Granovetter looked at two Boston communities and
their responses to the threat of urban development. He came to the conclusion that efficient
response was not so much a function of a dense, strong tie network but it rather depended on
occasional “weak” ties between individuals who only saw each other occasionally.
• In a second study he found the same effect when looking at how different individuals mobilize
their social networks when finding a job. While it has been known long before that most jobs are
found through personal connections, Granovetter showed that somewhat surprisingly close
friends play a minor role in this process compared to casual acquaintances.
V - AI&DS UNIT I SNS

The cognitive dimension of social capital refers to those resources providing shared representations,
interpretations and systems of meaning . In particular, cognitive ties are based on the existence of shared
languages, signs and narratives, which facilitate the exchange of knowledge.

In fact, these measures are typically used in a “cases-by- variables” analysis where these network
measures are correlated with some output variable using regression analysis. Besides network variables
other attributes of the individuals are also entered into the model as control variables (see Figure 3.9).

Figure 3.9. In a cases-by-variables analysis we fit a linear equation (below) to the model described by
the table containing the values of the variables for each case.