Abstract:Every organization has organizational networks for exchange of ideas and information. It is believed that organizational network analysis (ONA) can help the business be more effective. While considerable research efforts have been made for visualizing and analyzing relationships in organizational networks, it lacks a holistic way to model the complex social structures and rich semantic information of these networks. Indeed, employee behaviors can occur across different communication platforms, such as email and instant messaging systems, which naturally lead to the multiplex structure of organizational social networks. Meanwhile, it is also a challenge to model the impact of semantic information, such as employee attributes and organization charts, and the collaboration relationships of employees. To this end, in this paper, we propose a Multiplex Attentive Network Embedding (MANE) approach for modeling organizational social networks in a holistic way. Specifically, we first develop a multiple attributed random walk approach to jointly model multiple networks, with the integration of external work information. Then, we preserve the network structure by maximizing the probability of predicting the central node based on the surrounding context nodes. In particular, we introduce an attention mechanism to assign a weight to each context node in the training process, according to its attributed relation and structural relation with the central node by utilizing the k-core algorithm and the shortest path algorithm. In this way, the embedding results can be kept consistent with their structural relationships. Furthermore, to solve some department-level tasks, we introduce an attentive relational transition method to learn the representation of departments in the organizational networks. Finally, we evaluate the performance of MANE with extensive experiments on real-world data for three important talent management tasks, namely employee performance prediction, employee turnover prediction and department performance prediction. We also conduct a link prediction task to validate the effectiveness of employee embedding. Experimental results clearly show the effectiveness and interpretability of MANE for organizational network analysis.

MINE: A Method of Multi-Interaction Heterogeneous Information Network Embedding

Community Mining From Multi-Relational Networks

GraphHINGE: Learning Interaction Models of Structured Neighborhood on Heterogeneous Information Network

Multi-Relation Aware Temporal Interaction Network Embedding

MAHE-IM: Multiple Aggregation of Heterogeneous Relation Embedding for Influence Maximization on Heterogeneous Information Network

Multiplex network infomax: Multiplex network embedding via information fusion

OLMNE+FT: Multiplex Network Embedding Based on Overlapping Links

Self-supervised Nodes-Hyperedges Embedding for Heterogeneous Information Network Learning

CMG2Vec: A composite meta-graph based heterogeneous information network embedding approach

Mining Hidden Community in Heterogeneous Social Networks

Multimodal interaction aware embedding for location-based social networks

Clustering Heterogeneous Information Network by Joint Graph Embedding and Nonnegative Matrix Factorization

Joint Embedding of Meta-Path and Meta-Graph for Heterogeneous Information Networks

Multimodal Deep Network Embedding with Integrated Structure and Attribute Information.

Learning Node Embeddings in Interaction Graphs

Multi-View Collaborative Network Embedding

IGE+: A Framework for Learning Node Embeddings in Interaction Graphs.

Multi-Stage Network Embedding for Exploring Heterogeneous Edges

MANE: Organizational Network Embedding with Multiplex Attentive Neural Networks

Mining E-Commercial Data: A Text-Rich Heterogeneous Network Embedding Approach

InteractNet: Social Interaction Recognition for Semantic-rich Videos