Please enable JavaScript.
Coggle requires JavaScript to display documents.
Online Collaboration and Social Networks (Gopal et Al 2016 (Methodology…
Online Collaboration and Social Networks
Gopal et Al 2016
Definitions
Cascade Propagation: Any one incoming node has to be greater than a threshold
Malleable networks
Threshold Model: Sum of influence of all incoming nodes higher than threshold
Research Question
RQ: In addition to finding the right seeds, can making connections between nodes improve message propagation?
Methodology
Use of the Hop-Constrained Minimum Spanning Tree(HMST) model to select seed and new connections to make
Minimize Cost, with a constraint on hops
HMST is NP-hard. This paper looks for a computationally efficient heuristic.
Try three techniques: Prim (1957), FGV(2007, and Akgun (2011)
Then, compare against no-connection allowed
Data
Enterprise Network: Enron email network
Social Network: Facebook and Twitter Ego Networks - Networks of all people connected to one person
Results
CIT and Hops plot shows way better performance for malleable HMST
Dawande et Al 2012
Research Question
What are some fundamental set-based notions on which search in social networks is expected to be prevalent
How can we look for influence networks and centrally located networks?
Definitions
Degree Centrality: Number of nodes a node is connected to
Closeness Centrality: 1/(Sum of the shortest paths between node and all other nodes)
Betweenness Centrality: How often do shortest paths go through a node
Notion: Elite Group - a set of nodes where edges are coming in, but outgoing edges remain within the elite group.
Notion: Portal Problem: Highest level of group betweenness centrality
Social Network: A social structure where relationships between entities are represented as edges and nodes.
Findings
For Elite Group, convert to top-bottom tree. Leaf nodes are elite group. Or Convert to bottom-to-top directed tree. Interior nodes are elite
Find different rules to simplify the NP hard problem. eg: in connected cycle, atleast one is part of an elite group
Binary tree of height 4 - PP is the mid 4.
Elite problem can be solved polynomially, but its size constrained version is NP hard
PP is NP hard
Motivation
Participants of a network derive some utility from it
This utility derives from their ability to search. eg: Fb user wants to discuss topic of interest, developer may want to create a new project team
Social network search different from a directory search: Social search could also be looking for complex relationship between returned nodes
Useful to identify influence set. Eg: promoting an idea, product, message to other members of the network
Location may not equate to influence. eg: managers that are influential are not necessarily ones passing messages
Elite Group
Mongolian small town
School principal, vice principal,
Key players or opinion leaders - Everyone seeks advice from them, they seek advice from each other
Portal Group
Understanding of the paths that are used in message propagation so that there is no bottleneck in communication
Identifying a bad group's communication links, taking them out will slow them down.
Placing anti-virus solutions on network nodes - what are the best nodes?
Disease outbreak network - find points from where to best control the outbreak.
Finding a socially central group of monkeys
Yan and Tan 2014
Research Question
Does social support exchanged in an online healthcare community benefit patients' mental health?
Findings
Participation in the online community helps them to improve their health.
The form of social support exchanged has a differential impact on the patient’s well-being. Informational support is common, but emotional support has the greatest impact
Data
Every patient has a 0-3 star rating based on level of engagement with the system
Partial data is available about the mental state of each patient
Number of posts
Methodology
A directed social network is created based on posts on other's profile
Transition Matrices (Bad to Bad, Bad to good) Percentages are calculated. These show marked differences after posts
Faraj et Al 2011
Research Question
What makes knowledge collaboration in Online Communities different from knowledge collaboration traditional structures?
Argument
Solution
Changing management role
Quickly bring participants up to speed
Dynamic boundaries
Evolving technology affordances
Online Collaboration has fluid membership, causing
Temporary Convergence
Socially Ambiguous identities creates uncertainty
Time leads to fluctuations in collaborative process
Passion - changes makes this unpredictable
Kane et Al 2014
Research Question
How do online community contributors respond to the knowledge change-retain tension in order to facilitate coproduction?
Findings
The authors find three emergent interaction patterns of activities characterized by different foci of coproduction.
These patterns are heterogeneously distributed over the lifetime of the community
Contributors to the coproduction community tend to use particular patterns
Data
1 article in wikipedia, and its changes over a 10 year period
Methodology
Read all comments, categorize it manually, visually chart the events to find patterns of activity, ups and downs of patterns, and contributor characteristics
Chaotic Generating - multiple people create information
Content filtering - changes are made with consensus
Defensive Filtering - New and stupid ideas are shut out