Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


SNA Under The Microscope

No description

Alex Ross

on 21 April 2014

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of SNA Under The Microscope

Come Closer to

Presented By:
Mariam Fatima,
Ishaq Khan &
Mohammed Rashid
Social Network Analysis
Social Network
Data Mining
Web Mining
Community Mining
Feature Extraction
Relation Extraction

Types Of Social Networks
Homogenous Networks
Heterogeneous Networks

Supervised Learning
Unsupervised Learning
Semi-Supervised Learning

Basic Metrics
Metrics For SN over the Web
Social Network
Data Mining
Web Mining
Community Mining
Feature Extraction
Relation Extraction
Homogenous Social Networks
Heterogeneous Social Networks
Types Of Social Networks
Collection of people and relations
Uses web interface to allow people to connect with each other
Represented By a Graph
Nodes - Individuals
Edges - Relationships
Popular SN - Facebook, LinkedIn, Twitter
Social Network
Methodical analysis of SN
Applies analytical techniques to relationships between individuals
Focused on uncovering patters related to interaction
Concepts regarding
Actors and Actions
Relational Ties
Network Models
Social Network Analysis
Actors - smallest unit of analysis in SN, who is an individual
Actions - the task performed by an actor
Relational Ties - linkage between actors for flow of resources
Network Model - conceptualizes the structure of the network
SNA Concepts
It is the non-trivial process of identifying valid, novel, potentially useful, and ultimately understandable patterns in data.
Types include pattern recognition, classification and so forth.
Information is obtained by identifying a common attribute.
Data Mining
Popular Social Networks
Data Mining applied on web documents to uncover general patterns.
It is the discovery and analysis of useful information from the internet.
It is generally of two types:
Web Content Mining - mining performed on web documents
Web Usage Mining - mining performed on web usage data
Web Mining
Content of actual web pages.
Intrapage Structure - includes HTML or XML code
Interpage Structure - refers to actual linkage between web pages.
User profiles include demographic and registration information about the user.
Web Data
It is the process of identifying communities which are grouped together based on some common properties.
Allows us to find clusters and identify how the people are connected to each other, i.e densely or sparsely.
Community Mining
The technique needed to transform data with multiple variables into data with lesser number of variables.
Feature Extraction
Regression in data mining is a process that predicts a value.
It is used to map a data item into a real valued prediction variable.
It is the process of finding links or relationships between two entities so that communities can get identified.
We measure the importance of each relationship so as to know in which community they fall.
Relation Extraction follows the following process:
Target relation is given to RE system as input
Hand-Crafted or learned extraction patterns are applied to obtain output.
Relation Extraction
These networks consist of only a single type of network.
When data mining is applied, show only direct relationships.
The results produced are independent of user requirements.
Homogenous Social Networks
These type of social networks consist of multiple networks.
Also known as multi-relational social networks.
There exist various kinds of relationships.
Each relation can constitute its own relation network.
When data mining is applied, the output is dependent of the user requirements.
The output is a result of query based analysis.
Heterogeneous Social Networks
Under The Microscope
Friend Suggestion in social community websites.
Targeting market trends and generating context based advertisements.
Predicting social trends.
Implementing social marketplaces to increase sales by improving customer relations.
Applications Of
Heterogeneous Social Networks
(cc) image by anemoneprojectors on Flickr
Supervised Learning
Unsupervised Learning
Semi-Supervised Learning
Nearest Neighbor Classifier
Supervised Learning
It says that similar samples should be classified into the same class.
Similarity can be measured by using metrics based on the data.
When a classifier is built, the training samples are indexed by some indexing technique like kd-tree.
Nearest Neighbor Classifier Algorithm
Easily implemented.
Accuracy is higher as parameters can be fine tuned.
Internet Marketing in the form of contextual advertising.
Hierarchical Clustering
Density Based Clustering
k-Means Algorithm
Unsupervised Learning
Constructs a hierarchy of clusters in form of a dendrogram.
Each node represents a cluster.
Parent-Child relationship enables to explore clustering granularity.
Algorithm performed in:
Bottom-Up Manner - starts with objects and then clustering occurs
Top-Down Manner - Reverse of above
Hierarchical Clustering
In semantic analysis.
Closeness of nodes to objects determines how well the object is liked.
Objects are grouped by recognizing their density.
Those areas with a higher density are qualified clusters.
Those with lower density are treated as outliers.
Density Based Clustering
In online social marketing networks like Overstock Auctions.
It is a top-down algorithm.
Classifies the objects into 'k' groups with regard to attributes.
Grouping done by minimizing the sum of square of distance between the object and corresponding cluster centroid.
The distance is calculated using Euclidean Distance metric.
The centroid location is revised with regard to updated information.
k-Means Algorithm
Determining how a certain object is used by the various nodes.
Semi-Supervised Learning
A classifier is trained on labeled data.
Then the classifier is used to classify unlabeled data.
The confident unlabeled data with its corresponding label class in inserted into labeled data.
Both labeled and unlabeled sets of data are updated and procedure is repeated.
Self - Training
Based on experience, a search engine can generate result to query
based on similarity with previous query in a community.
Self -Training
Graph Based Model
Learning is performed in a graph.
Each node represents a piece of data in the labeled and unlabeled datasets.
The graph is minimum cut and the nodes in each component should have the same label .
Graph Based Methods
Predicting relationships in communities that may occur due to common ties
Basic Metrics Of Social Network
Metrics for Social Network on the Web
Basic Metrics
Of Social Network
It refers to the number of vertexes present in a network.
Gives a rough indication of the social power of a vertex in a network.
Density - denotes the ratio of number of all existing edges to the total possible number of edges in the network.
Degree - actual number of edges contained in the network.
Density and Degree
Both represent magnitude measures that reflect the relationships of one vertex to the others in the network.
Betweenness and Closeness
Measure of how likely one vertex belongs to a community in a network
Measures the ratio of correct predictions to the total number of cases evaluated
Fraction of the documents retrieved that are relevant to the user.
Fraction of the documents that are relevant to the query that are successfully retrieved.
It is the weighted average of precision and recall.
Metrics for
Social Network on the Web
Three types:
Degree Centrality
Closeness Centrality
Betweenness Centrality
Focuses on inlinks.
Actor's prestige is affected by the ranks of the involved actors.
Refers to Hyperlink-Induced Topic Search.
Assigns two scores for each web page:
Authority - value of content of page
Hub Value - value of links to other pages
Basic understanding of Social Networks
Understanding the importance of relationships between the nodes in the network
Formation of Communities using various Algorithms
Application of Social Network Analysis
Superiority of Heterogeneous Social Networks over traditional networks
Possibility of mining data from virtual world and utilizing it as information in the respective physical environment
Full transcript