Effective Human Teaming Supported by Social Sensing

Description:

TECHNOLOGY AREA(S): Human Systems 

OBJECTIVE: Develop and deliver a standalone social sensing platform for behavioral analysis of small team interactions providing teamwork diagnostics and intelligent decision-support software capabilities to track, manage, and significantly impact team readiness and performance. 

DESCRIPTION: Advances in information and network technology continue to transform the way human organizations communicate and operate. So much so that networked organizations lie at the core of the political, military, economic, and social fabric. Massive volumes of data that characterize online 'digital behaviors' including communication and collaboration, are increasingly collected and subjected to data mining by companies, governments, and researchers alike [1,2,3,4]. Recent advances in the collection of 'analog behaviors' with the pervasiveness of sensor technology - from smart phones and wearables sensing devices - coupled with methodological developments in data science - have the potential to drive new and innovative behavioral applications that involve tools and platforms for continuous human monitoring. The goal is to develop a robust, intelligent software system that leverages individual level social sensing data to provide network level analytics of behavioral interactions for designing effective teams. The first challenge (manage collection) is to create a social sensing platform to allow non-technical end-users to collect and manage data from a variety of digital and analog sensors by automatically configuring a scalable data system flexible enough to store variable datatypes stemming from a plethora of data sources. This include pre-processing vast amounts of noisy digital and sensor data-streams to distill key human performance meta-data. The second challenge (team-level diagnostics) is to design novel algorithms for data analysis which must include, but are not limited to, social network analysis of single graphs and knowledge-based multi-graphs; flow analysis to assess team collaborations and interactions; and individual-level behavioral pattern analysis with the capability to infer human states that can be aggregated to infer team-level states. Additional capabilities should incorporate the latest analytical capabilities based on language analysis to infer team context from communications and indicators of hierarchical positioning [5]. The third challenge (recommendation engine) should provide a means to run what-if scenarios addressing team management challenges, such as address the family of problems under the scope of team composition and team member replacement, and approaches to link team-level diagnostics to predictive outcome-based models of team performance. The fourth challenge (visualization) should provide team snapshots by creating scalable visualization techniques to allow the user to explore individual and team profile information; fluidly visualize multimodal network graphs over time and in response to key work-directed events or changes to the team context (e.g. attrition, new teammate); drill through graphs to uncover the underlying data; and show how team behavior patterns change over time (e.g. team structure). The implemented system will have a small form factor that is multi-platform, scalable software. The system will provide the ability to choose multiple algorithms for analysis and recommendations based on user needs. Additionally, the graphical-user interface must be turnkey, with an easy to use interface for non-technical end-users. The system will contain a searchable database as well as modular and modifiable transparency with respect to the knowledge-based multi-graphs driving the recommendation engine. All stored data must retain available and relevant behavioral meta-data such as sender, receiver, date, time, geo-tags, and inferred individual and team-level states. The system should provide efficient analytical capabilities including the ability to create and customize diagnostic team profiles. 

PHASE I: The Phase I effort will address the first two challenges (manage collection & team-level diagnostics) by developing and demonstrating a clearly defined approach to a social sensing framework to automatically capture social and collaborative team interactions in a variety of work-directed and organizational settings. A key data management challenge involves the aggregation and pre-processing of vast amounts of heterogeneous multi-scale, multi-level data. The social sensing framework will be designed to aggregate disparate data types from a variety of sources and be modular in design to accommodate frequent updates to APIs. A key requirement for a successful Phase I is an initial pass-through of multi-scale and multi-level data aggregation and the identification of social network metrics combined with algorithmic approaches to address a clearly defined framework of team-level diagnostics. The academic partner will focus on developing algorithmic approaches to model team composition and their interactions. The industry partner will focus on developing potential use-cases as well as a viable business plan to commercialize team analytics using a social sensing platform. This includes understanding customer needs and user requirements for developing a metric framework needed to effectively manage teams. 

PHASE II: The Phase II effort will address the third and fourth challenges (visualization & recommendation engine) by concentrating on the design and development of analytical capabilities to provide a composite picture from multiple data sources and providing informative, scalable visualization capabilities of the underlying team-level analytics. A key consideration includes the design of privacy-preserving organizational social analysis system that uses social sensors to gather, crawl, and mine various types of data sources, potentially including individual email and instant message communications, calendars, the formal social structure (i.e. organizational chart) as well as individual role assignments, and data from wearables technology (from heart-rate to mobility to cameras). A key analytical challenge is that wearables and social sensing platforms not only produce vast data streams collected in natural settings over long periods of time, but the overriding context is dynamic, unpredictable, and perhaps even unknown. Contextual variables will be ascribed such as significant environmental events and changes to team composition. Phase II will develop multi-graph approaches that combine social network analytics with knowledge graphs to predict outcome-based measures of team performance. Specific Phase II milestones include the collection of a minimum of six weeks of longitudinal behavioral data from a variety of sources from a work-directed team and a demonstration of the key functional concepts derived from this dataset. The offeror must demonstrate a clear understanding of analytics relevant to military needs. 

PHASE III: Phase III efforts will be directed toward refining a final deployable design with sophisticated, cross-platform GUI; incorporating design modifications based on results from the tests conducted during Phase II; the system should be hardened for security and protection of personally identifiable information (PII) and results by taking all appropriate measures to incorporate technical security of data collection, aggregation, and analytics; and improving engineering/form factors, equipment hardening, and manufacturing designs to meet U.S. Army CONOPS and end-user requirements. 

REFERENCES: 

1: Navaroli N., & Smyth, P. (2015). "Modeling response time in digital human communication," in Ninth International AAAI Conference on Web and Social Media. Oxford,UK.

2:  Buchler N., Fitzhugh S.M., Marusich L.R., Ungvarsky D.M., Lebiere C., & Gonzalez C. (2016) Mission Command in the Age of Network-Enabled Operations: Social Network Analysis of Information Sharing and Situation Awareness. Frontiers of Psychology, 7:93 doi: 10.3389/fpsyg.20100937

3:  Wuchty, S., Jones, B.F., & Uzzi, B. (2007). The increasing dominance of teams in production of knowledge. Science, 316 (5827), 1036-103

4:  Lin, N. (2008). A network theory of social capital. The Handbook of Social Capital, 50, 6

5:  Kozlowski, S.W., & Bell, B.S. (2003). Work groups and teams in organizations. Handbook of Psychology, Wiley

KEYWORDS: Team Science, Data-Driven Behavioral Analytics, Team Readiness, Human State Estimation Team Performance Assessment, Team Management, Knowledge Management 

CONTACT(S): 

Edward Palazzolo 

(919) 549-4234 

edward.t.palazzolo.civ@mail.mil 

Norbou Buchler 

(410) 278-9403 

Agency Micro-sites

SBA logo
Department of Agriculture logo
Department of Commerce logo
Department of Defense logo
Department of Education logo
Department of Energy logo
Department of Health and Human Services logo
Department of Homeland Security logo
Department of Transportation logo
Environmental Protection Agency logo
National Aeronautics and Space Administration logo
National Science Foundation logo
US Flag An Official Website of the United States Government