Unlocking Inter-Divisional Synergy: A Hybrid Social Network Approach to Corporate Communication

TL;DR

This research tackles the "silo effect" in corporate environments by combining a word-of-mouth Information Propagation Model (IPM) with structured Discussion Forums (DF). By deploying this system in a real-world company, the authors demonstrated that unifying spontaneous sharing with curated discussion can bridge the gap between isolated departments and turn "stranger" colleagues into collaborators.

Background & Motivation: The Limits of E-mail

While 97% of Japanese companies utilize digital communication, nearly 60% still report insufficient inter-divisional interaction. The core issue is that traditional tools—like e-mail or basic BBS—require a pre-existing reason to reach out. If you don't know someone in the Engineering department, you are unlikely to start a thread with them.

The authors identify a secondary hurdle: Negative Users. Most Social Networking Services (SNS) like LinkedIn rely on active, self-starting users. Highly reserved employees often remain on the periphery, never expanding their professional network despite having valuable expertise.

Methodology: The Two-Pronged Strategy

The system's brilliance lies in its dual-layer architecture, designed to catch different types of users in its net.

1. The Information Propagation Model (IPM)

Think of this as "Digital Word-of-Mouth."

• Trust-Based: You only receive information from people you explicitly register as "neighbors."
• Traceability: Unlike a random rumor, you can see the exact path a message took to reach you.
• Logic: It allows information to hop across the social graph organically, reaching distant divisions through mutual acquaintances.

2. Discussion Forums (DF)

Where IPM is organic, DF is intentional.

• Strategic Allocation: Administrators assign users to forums based on age, post, and department—not social ties.
• Thematic Focus: Using prerequisite questionnaires, the system identifies "hot topics" (e.g., "how to improve office efficiency") to give people a reason to talk.

Figure: The system architecture combining the Web-based server and the IPM/DF logic.

Experimental Insights: Does it Actually Work?

The authors conducted a two-month field study at NTT DATA SANYO SYSTEM CORPORATION. They mapped the resulting social network (Figure 4), revealing a non-centralized, healthy distribution of connections.

Figure: The actual social network constructed during the experiment, showing 83 participants.

Key Findings:

• Breaking the Silo: Users reported a noticeable impact on communication between far-flung branch offices and different divisions—areas typically ignored by daily business routines.
• Different Strokes for Different Folks:
  • IPM was most effective for "Social Butterflies" (users with many neighbors) to discover new contacts.
  • DF was a lifeline for "Isolated Users" (those with few neighbors), providing them a structured environment to build their first few connections.
• Synergy: The study proved that running Discussion Forums actually increased the frequency of spontaneous messages sent via the IPM (Accuracy p < 0.05).

Critical Analysis & Conclusion

Takeaway

The paper proves that horizontal communication doesn't happen by accident; it requires a mix of pull (IPM) and push (DF) mechanisms. The "Traceability" of the IPM provides the safety and trust necessary in a corporate culture, while DF provides the necessary nudge for introverts.

Limitations & Future Work

The system currently relies on a human administrator to curate forum themes and assign users. In a modern context, this is a clear bottleneck. One could imagine an LLM-based agent modernizing this by:

1. Automatically identifying trending topics within the IPM.
2. Using semantic analysis to pair users with complementary expertise but no social ties.

Despite being an earlier study, the fundamental insight remains relevant: to break down corporate walls, we must design for the social graph, not just the org chart.

References

• Takeuchi, S., et al. "In-house communication support system utilizing social network."
• Takeuchi, S., et al. (2003). "Human-Network-based Filtering." SAINT-2003.