[Paper Review] Forming Compatible Teams in Signed Networks

Ioannis Kouvatis, Konstantinos Semertzidis+3 more|arXiv (Cornell University)|Jan 1, 2020

Complex Network Analysis Techniques13 references4 citations

TL;DR

This paper introduces a novel team formation problem in signed social networks, where compatibility between team members is defined using structural balance theory to account for both positive (friend) and negative (foe) relationships. The authors propose heuristic algorithms that prioritize compatibility and communication efficiency, demonstrating through experiments that compatibility-aware selection significantly improves team coherence, especially under strict compatibility definitions, with LCMD (least compatible distance) emerging as the most effective strategy for minimizing team cost while maintaining high success rates in forming compatible teams.

ABSTRACT

The problem of team formation in a social network asks for a set of individuals who not only have the required skills to perform a task but who can also communicate effectively with each other. Existing work assumes that all links in a social network are positive, that is, they indicate friendship or collaboration between individuals. However, it is often the case that the network is signed, that is, it contains both positive and negative links, corresponding to friend and foe relationships. Building on the concept of structural balance, we provide definitions of compatibility between pairs of users in a signed network, and algorithms for computing it. We then define the team formation problem in signed networks, where we ask for a compatible team of individuals that can perform a task with small communication cost. We show that the problem is NP-hard even when there are no communication cost constraints, and we provide heuristic algorithms for solving it. We present experimental results with real data to investigate the properties of the different compatibility definitions, and the effectiveness of our algorithms.

Motivation & Objective

• To address the gap in existing team formation research that assumes only positive social ties, by incorporating negative (foe) relationships in signed networks.
• To formalize user compatibility in signed networks using structural balance theory, ensuring that only compatible individuals are grouped into teams.
• To develop efficient heuristic algorithms that balance compatibility and communication cost in team formation.
• To evaluate the effectiveness of different compatibility definitions and algorithms on real-world signed network datasets.
• To demonstrate that compatibility-aware team formation leads to more cohesive and effective teams compared to standard approaches.

Proposed method

• Define a compatibility relation $Comp$ based on structural balance, where compatibility is inferred through paths connecting nodes, with even numbers of negative edges implying compatibility.
• Formalize the Team Formation in Signed Networks (TFSN) problem as an NP-hard optimization task minimizing communication cost while ensuring all team members are mutually compatible.
• Propose two heuristic algorithms: LCMD (Least Compatible Distance) and LCMC (Least Compatible Member), which iteratively select the most compatible or closest user to the current team.
• Use four real-world signed network datasets (Epinions, Slashdot, Flixster, Friendster) to evaluate the performance of the algorithms under varying task sizes and compatibility definitions.
• Implement a baseline Random selection strategy and compare results to assess the impact of compatibility-aware selection.
• Apply the TFSN framework to unsigned variants of signed networks (by ignoring or removing negative edges) to benchmark performance against standard team formation methods.

Experimental results

Research questions

• RQ1 How can compatibility between users be formally defined in a signed network using structural balance theory?
• RQ2 What is the impact of different compatibility definitions (e.g., DPE, DNE, SPO, SBP_H) on the success rate of team formation?
• RQ3 How do heuristic algorithms that prioritize compatibility and communication cost compare in forming effective teams in signed networks?
• RQ4 Can compatibility-aware team formation significantly improve team coherence and reduce communication cost compared to standard approaches?
• RQ5 How does the success rate of team formation vary with task size and the strictness of compatibility definitions?

Key findings

• The TFSN problem is NP-hard, even when communication cost is ignored, confirming the computational challenge of incorporating compatibility constraints.
• The LCMD algorithm outperforms LCMC and Random selection in minimizing team communication cost, achieving the lowest average diameter across all compatibility definitions.
• For $k=5$ skills, the success rate of forming compatible teams was highest under the NNE and SBP_H compatibility definitions, while stricter definitions like SPO and SPA reduced the number of solvable tasks.
• The percentage of tasks with compatible skill coverage (MAX) served as an upper bound, and LCMD achieved close to this bound, indicating high efficiency in compatible team discovery.
• When compared to standard unsigned team formation on transformed versions of signed networks, most returned teams were incompatible, highlighting the necessity of a dedicated framework for signed networks.
• As task size increased, the success rate dropped sharply for stricter compatibility relations (e.g., SPO, SPA), while remaining stable for NNE and SBP_H, indicating robustness of these definitions under scalability.