Demonstration CSCW 2017, February 25–March 1, 2017, Portland, OR, USA
Professional Network Analytics
Platform for Enterprise Collaboration
Gopakumar Gopalakrishnan, Krupa Benhur, ACM Classification Keywords
Abhishek Kaushik, Anjaneyulu Passala J.4 [Social AND Behavioral Sciences]: Psychology; K.4.3
Infosys Ltd, Bangalore, India.
[Organizational Impacts]: Computer-supported
{Gopakumar; krupa_bg; [Link]; Anjaneyulu_
collaborative work.
Pasala} @[Link]
Introduction
In large multinational organizations, work is increasingly
Abstract
organized around projects and members of those projects
A fundamental challenge to effective coordination of work
are often required to collaborate across geographical and
across various boundaries in large and geographically
temporal boundaries to get their work done. In such
distributed organizations is the ability of employees to
settings, locating and securing timely support of other
identify who knows what, who knows whom, and who is
relevant project team members is important for effective
doing what. In this demo, we introduce a professional
individual and project team performance. Leveraging
network analytics platform that analyzes large and diverse
existing contacts, forming new beneficial connections, and
set of operational records in an enterprise and constructs
traversing geographically dispersed organizational
professional networks between employees. Further, we
networks thus become the foci of collaboration challenges
illustrate certain collaboration support services built
faced by project teams in large distributed organizations.
around these networks in a contextually relevant manner.
Finding other employees with relevant skills in certain
Author Keywords areas is a very common problem faced by employees and
Professional social networks; Enterprise network analysis; their ability to connect with experts on time can improve
Project Collaboration; Big data engineering. project performance considerably. Balog et al., [1] have
Permission to make digital or hard copies of part or all of this work for proposed a document-centric analytic method using
personal or classroom use is granted without fee provided that copies
Bayesian probabilistic approach for finding people with
are not made or distributed for profit or commercial advantage and that
copies bear this notice and the full citation on the first page. Copyrights relevant skills. Other studies include vector space [3],
for third-party components of this work must be honored. For all other document retrieval [6] and social networks based [4]
uses, contact the owner/author(s). Copyright is held by the
author/owner(s). models.
CSCW'17 Companion, February 25–March 1, 2017, Portland, OR, USA. Likewise, social connection recommendations that help
ACM ISBN 978-1-4503-4676-4/17/02.
people form new beneficial connections has emerged as
[Link]
an important research topic. Yarosh al., [7] have
5
Demonstration CSCW 2017, February 25–March 1, 2017, Portland, OR, USA
examined the multiplexity of social connections among authored documents, attended meetings together etc. Our
users in enterprise social networks, which include both platform constructs networks between employees from
professional and personal closeness. Other friend different types of data stored in multiple databases within
recommendation works are done either on small-sized the enterprise that represent different types of
enterprise social networks, like Beehive [2], or survey relationships between them. A graphical view of an
based dataset [5]. employee’s professional network is presented as Fig.1.
We extend these findings and introduce a professional A number of challenges emerge while constructing
network analytics platform developed to support professional networks from different enterprise data
Figure 1:
collaboration between employees in a large multinational sources. A key source of challenges is the nature of data,
Professional Network of an
enterprise with over 200,000 employees working in more which is of different type, stored in different data bases,
Employee
than 50 countries. In large organizations like this, dynamic, and evolve with continuous usage by
enormous amount of internal operational data regarding employees. To develop an analytics platform using this
project teams such as project management documents, heterogeneous, dynamic, and time stamped data for
human resource records, work emails, common meetings, multiple users, we have addressed the following
instant messages etc. are collected routinely for various challenges.
purposes and stored in different enterprise databases.
i Constructing Professional Networks from
This data explicitly or implicitly represent social networks
Disparate Data
among project team members but they are rarely used in
The first challenge is how to construct networks between
a holistic way.
employees from different types of data that represent
In the following sections, we discuss our professional different types of relationships and stored in multiple
network platform development approach and how we databases within the enterprise. Different types of data
have addressed various challenges. This is followed by an include prior work related information such as alumni of
illustrative set of collaboration support services. We various educational institutions, prior work experience in
conclude by suggesting potential research opportunities same organizations, etc. and work related data include
that our platform opens up for advancing the information such as project co-assignments, common
Talent Network All Search
understanding of micro foundations of cooperative meetings, email exchanges etc. Such data cannot be used
My Profile
John Doe
Senior Systems Engineer
My Connections
22 in Primary Circles behaviors in large enterprises. in the same way as they are recorded in those systems. For
ITMYYS – Pune – British Petroleum
80 in Secondary Circles
3 Groups
example, both common meetings data and project co-
CDG-Bangalore
“Working on AI, Big Data and Hadoop”
200
Overview All Primary Secondary Groups
Suyash Goyal Richard Bryan Rajiv Sharma
Professional Network Platform Development assignments data can be used to construct relationships
The term enterprise network in our demo paper refers to between employees and project social network, but they are
Assistant Manager Delivery Head Project Manager
ITMYYS – Pune – British Petroleum ITMYYS – Pune – British Petroleum ITMYYS – Pune – British Petroleum
Preeti Mishra Saurabh Hitesh Bisht
Senior Systems Engineer Technology Lead Sr Project Manager
the modeling of all employee professional data as graphs inherently different. Differences in the content of these
wherein nodes are employees and connections between connections should be recognized before combining them to
ITMYYS – Pune – British Petroleum ITMYYS – Pune – British Petroleum ITMYYS – Pune – British Petroleum
Figure 2: Basic View of them represent different types of relationships such as create network between employees.
Employee Profile worked in same project during same time, served same
clients, possess same skills, report to same managers, co-
6
Demonstration CSCW 2017, February 25–March 1, 2017, Portland, OR, USA
In order to best leverage these diverse data, our platform based on different data sources, analyses of those networks
first converts all of them to a unified format and makes it including social distance between employees, and
easy for consumption by other components. In the presenting them for consumption by various applications.
preprocessing stage, we identified three types of
Applications and Visualizations
relationships including profile, matrix, and events. Profiles
In this section, we illustrate three application services that
include data describing entities such as employees, projects,
are enabled by the platform such as knowing connections
departments, customers etc. Matrix are data describing the
and competence of other employees in the enterprise,
relationships between two entities, like employee worked
Figure 3: finding right contacts and shortest paths to reach them and,
for a project, employee served a customer, two employees
Basic View of Employee Profile recommendations for forming new beneficial connections.
graduated from the same university etc. Events involve
data that the system tracks for each entity, like members of
i Knowing Competence and Connections of Others
team B had a meeting in location A, last week.
The knowledge of who knows what and who knows whom
would be useful for employees in different ways. The profile
ii Managing Large and Complex Data
view service of our system enables this knowledge of
The second challenge is regarding the management of large
connections and competence of all employees accessible to
volume of data feed from different operational databases.
everyone in the enterprise. It provides an overview of the
They are dynamic and heterogeneous, which cannot be
skills of an employee, colleagues she has added to her
stored or managed in any single database management
network, members in current team and other employees
system. To address this challenge, we have built a Data
she has worked with during her tenure in the organization.
Service Center, which is a heterogeneous database
A graphical representation of the basic views are presented
functional component equipped with all the features that are
as Fig. 2 and 3.
required and provides interfaces for other components to
access the data easily and quickly. We have also applied ii. Finding Contacts and Shortest Paths
state-of-the-art approach of making the system distributed Finding contacts is a helpful function for employees to
and part of the functionality working as a micro service search for colleagues whom they might be interested to
while they can cooperate together. connect with. The system allows employees to search for
others based on a number of parameters including those
iii Servicing Multiple Users working for a specific client, in a specific project, in a
The third challenge is that results of various analyses should particular location, holding a specific level in hierarchy etc.
serve multiple user’s interests that include managers and
The system also suggests the “The shortest paths” through
employees in different roles and levels. Different users have
which an employee can reach the target contact. The
different requirements and expectations from analytics
‘shortest path view’ of the system enables this by
Figure 4: Shortest path view results range from helping users to find and establish new
computing and rank ordering the social distance between
connections to forming new project teams. To address this
two employees, as presented in Fig. 4.
challenge, our platform perform various computations that
involve construction of employee professional network
7
Demonstration CSCW 2017, February 25–March 1, 2017, Portland, OR, USA
iii. Recommendations for New Connections collaboration challenges in such settings. Future work on
The recommendation feature of the system nudges the network platform development side involve temporal
employees to connect with others who are not necessarily visualizations of network data for employees in different
similar to them but can add significant professional value to hierarchical levels in the context of specific management
them. The recommendation engine suggests new contacts challenges.
for each employee based on different parameters related to
References
their work, role, location, current contacts, and other 1. Balog, Krisztian. People search in the enterprise. In
preferences defined by employees. Proceedings of the 30th Annual International ACM
SIGIR Conference on Research and Development in
The key considerations for recommendation engine are: Information Retrieval. ACM, 2007.
Leveraging data from different sources: Three types of 2. Chen, Jilin, Werner Geyer, Casey Dugan, Michael
data namely, 1) Profile data of employees, 2) project Muller, and Ido Guy. Make new friends, but keep the
social network data and, 3) log data are used to create old: recommending people on social networking sites.
the recommendations. We apply recommendation In Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems. ACM, 2009.
algorithms on these data, which is made available by
the data service center. 3. Demartini, Gianluca, Julien Gaugaz, and Wolfgang
Nejdl. A vector space model for ranking entities and its
Explanation of recommendation and interaction choice:
application to expert search. European Conference on
In addition to suggesting a set of recommended Information Retrieval. Springer Berlin Heidelberg, 2009.
contacts for each employee, the system also highlight
4. Guy, Ido, Sigalit Ur, Inbal Ronen, Adam Perer, and
the rationale underlying the recommendations. Further
Michal Jacovi. Do you want to know? Recommending
to viewing the details of recommended contacts and strangers in the enterprise. In Proceedings of the ACM
recommendation rationale, the system also help users 2011 conference on Computer Supported Cooperative
to connect with new contacts through official email or Work. ACM, 2011.
phone call. 5. Guy, Ido, Inbal Ronen, and Eric Wilcox. Do you know?
Refinement of recommendation logic: As employees Recommending people to invite into your social
move to new projects, new locations, acquire new network. In Proceedings of the 14th International
skills, and form new ties, their preferences of new Conference on Intelligent User Interfaces. ACM, 2009.
contacts change. We have designed the system to 6. Macdonald, Craig, and Iadh Ounis. The influence of the
detect such changes and adapt the recommendations document ranking in expert search. Information
accordingly. Processing & Management, 47.3 (2011): 376-390.
7. Yarosh, Svetlana, Tara Matthews, and Michelle Zhou.
CONCLUSION Asking the right person: supporting expertise selection
We have demonstrated how different types of voluminous in the enterprise. In Proceedings of the SIGCHI
data recorded in different databases is leveraged to build Conference on Human Factors in Computing Systems.
professional network of employees within a large ACM, 2012.
multinational enterprise and help address important