email communications in a large scientific research ...public.lanl.gov/bsims/pdf/email...
TRANSCRIPT
![Page 1: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/1.jpg)
Email communications in a large scientific research organization
Benjamin Sims Nikolai Sinitsyn
Stephan Eidenbenz Los Alamos National Laboratory
![Page 2: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/2.jpg)
Abstract
We present findings from a large-scale study of email networks, interpreted as social networks, based on data from a large scientific research organization. We explore several methods for visualizing network structure at large and small scales, develop a model of how organizational hierarchy influences email communications, and examine temporal patterns in email traffic. Based on organization charts, we classify each organizational unit by function and organizational level. By visualizing and aggregating email traffic between organizational units using these categories, we derive several insights into how large subdivisions of the organization interact with each other and with outside organizations. In addition, we provide examples of how email data can be used to examine the internal structure of individual organizational units. We propose a power law model for predicting connectivity in organizations based on number of managers at each level and number of support staff associated with each manager, and compare it to actual email connectivity in the organization. Additional temporal analysis shows both diurnal and weekly cycles in email traffic within the organization.
![Page 3: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/3.jpg)
Overview
• Analyzed complete email traffic for Los Alamos National Laboratory over a one month period
• Used a variety of analytic tools to analyze data in relation to known organizational features
• Results demonstrate a variety of potentially useful approaches to understanding organizational communication patterns
• Useful for understanding organizational issues, as well as identification of potentially anomalous communications
![Page 4: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/4.jpg)
Email traffic between organizational units
![Page 5: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/5.jpg)
Email traffic between types of organizational units
![Page 6: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/6.jpg)
Email traffic with outside domains
![Page 7: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/7.jpg)
Internal email traffic, small organizational unit
![Page 8: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/8.jpg)
Internal email traffic, larger organizational unit
![Page 9: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/9.jpg)
Node connectivity distribution as a function of organizational hierarchy • Top manager has highest connectivity N • l lower-level managers report to each manager, each with
connectivity N/l • a administrators assist each manager, also with connectivity
N/l • Therefore, at level x, (al)x managers (or their proxies) send
emails to N/(lx) people under their supervision • Consider a plot of number of nodes n vs. outdegree w.
Excluding the variable x, we find:
log 𝑛 =log (𝑁)log (𝑙)
−log 𝑎𝑙log 𝑙
log (𝑤)
![Page 10: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/10.jpg)
Node connectivity distribution as a function of organizational hierarchy (cont.)
• Outdegree distribution is then, generally, a power law with exponent log (𝑎𝑎)
log (𝑎)> 1
• At some level x, the hierarchy terminates around the point where 𝑎𝑙 𝑥 = N/(𝑙𝑥), so number of managers does not exceed number of employees
• Therefore, power law is expected to hold only for nodes with large outdegree (e.g. >40)
![Page 11: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/11.jpg)
Out-degree distribution, with linear fit for >40
![Page 12: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/12.jpg)
Fitting the model to organizational data
Fitting model to our data, we find: • 𝑙 ≈ 4 (4x greater managers at each level than
level above) • 𝑎 ≈ 7 (each manager has 6 support staff) • Power law terminates at 𝑥 ≈ 3, indicating about
four levels of management above each employee • Based on organizational observation, l appears to
be too small, a somewhat large, and levels of management approximately correct
• Further work needed to refine this model
![Page 13: Email communications in a large scientific research ...public.lanl.gov/bsims/pdf/Email communications in a large scientific... · scientific research organization Benjamin Sims Nikolai](https://reader035.vdocument.in/reader035/viewer/2022081617/60259846990d2924ec381075/html5/thumbnails/13.jpg)
Email traffic and number of senders, one week