Title: Finding a team of experts in social networks
Venue: ACM SIGKDD
Year: 2009
Main Problem
The author describes the problem formally as Given the set of n individuals X = {1, . . . , n}, a graph G(X,E), and task T, find X0 belongs to X, so that C (X0, T) = T, and the communication cost Cc (X0) is minimized.
Input
A set of individuals, a connected graph G(X, E) and a Task T.
Output
Minimized communication cost Cc (X0) for an individual X0 belonging to X.
Motivation
The author starts with a large group of individuals to make a perfect team given some constraint. This problem is prevalent in many different fields of work where organizations struggle to find the best suited subset of individuals for a particular task. In order to solve this problem the author comes up with TEAM FORMATION Problem and proposes algorithms to solve this problem.
Previous works and their drawbacks
Recent such as by [A. Baykasoglu, et al], using simulated annealing, branch-and-cut by [A. Zzkarian and A. Kusiak], or genetic algorithms used by[ H. Wi et al]works ignore the social structure or social bonds among the individuals.
Approach by Fitzpatrick and Askin, Chen and Lin takes into account the necessity of effective collaboration among individuals in a team but they also ignore the existing graph structure among individuals.
Gaston et al. takes into account the network structure between individuals in a workforce but fails to take into account the computational problem of finding a team of experts in a given network.
Proposed Method
The authors propose two algorithms for the Diameter-Tf and Mst-Tf problems.
Algorithm for Diameter-Tf Problem(RarestFirst Algorithm).
Algorithm for Mst-Tf Problem(CoverSteiner and EnhancedSteiner)
Experimentation
Dataset: - DBLP Snapshot taken on April 12, 2006.
Entries of papers published in the areas of Database (DB), Data mining (DM), Artificial intelligence (AI) and Theory (T) conferences were selected.
Metrics Used: - Communication Cost of the proposed algorithm, Cardinality of the teams generated from the proposed algorithms.
The authors do not explain the rationale behind selecting the communication cost. However, they do provide two alternative approaches in calculating the cost via the graphs. This does not explain the cost that a team must have or how they are calculating the minimum cost required by the team to function optimally.
the authors do not provide any substantial quantitative evidence that the teams chosen by their algorithms are most optimal.
Title: Finding a team of experts in social networks Venue: ACM SIGKDD Year: 2009
Main Problem The author describes the problem formally as Given the set of n individuals X = {1, . . . , n}, a graph G(X,E), and task T, find X0 belongs to X, so that C (X0, T) = T, and the communication cost Cc (X0) is minimized.
Input A set of individuals, a connected graph G(X, E) and a Task T.
Output Minimized communication cost Cc (X0) for an individual X0 belonging to X.
Motivation The author starts with a large group of individuals to make a perfect team given some constraint. This problem is prevalent in many different fields of work where organizations struggle to find the best suited subset of individuals for a particular task. In order to solve this problem the author comes up with TEAM FORMATION Problem and proposes algorithms to solve this problem.
Previous works and their drawbacks
Proposed Method The authors propose two algorithms for the Diameter-Tf and Mst-Tf problems.
Experimentation
Code Link Unavailable
Dataset http://arnetminer.org/citation - 2006 Snapshot
Gaps of the Work