IEEE Workshop on Big Data Metadata and Management (BDMM ’2017)
IEEE Workshop on Big Data Metadata and Management
Boston, MA, USA
Dec 11-12, 2017
Dec 11 -- Hackathon
Dec 12 -- Workshop
In conjuction to IEEE Big Data 2017
Sponsored by IEEE Big Data Initiative (BDI)
Hackathon Registration: CLOSED
This workshop is aligned with the effort from the IEEE Big Data Initiative (BDI) on Standardization (see http://bigdata.ieee.org/). The BDI standards research group is studying on where there is a need and opportunity for developing IEEE Standards for Big Data Metadata, its management, and governance. BDMM 2017 follows the successful BDMM workshop held at IEEE BigData 2016 (http://cci.drexel.edu/bigdata/bigdata2016).
Big Data is a collection of data so large, so complex, so distributed, and growing so fast (or 5Vs- volume, variety, velocity, veracity, and value). It has been known for unlocking new sources of economic values, providing fresh insights into sciences, and assisting on policy making. However, Big Data is not practically consumable until it can be aggregated and integrated into a manner that a computer system can process. For instance, in the Internet of Things (IoT) environment, there is a great deal of variation in the hardware, software, coding methods, terminologies and nomenclatures used among the data generation systems. Given the variety of data locations, formats, structures and access policies, data aggregation has been extremely complex and difficult. More specifically, a health researcher was interested in finding answers to a series of questions, such as “How is the gene ‘myosin light chain 2’ associated with the chamber type hypertrophic cardiomyopathy? What is the similarity to a subset of the genes’ features? What are the potential connections among pairs of genes”? To answer these questions, one may retrieve information from databases he knows, such as the NCBI Gene database or PubMed database. In the Big Data era, it is highly likely that there are other repositories also storing the relevant data. Thus, we are wondering
- Is there an approach to manage such big data, so that a single search engine available to obtain all relevant information drawn from a variety of data sources and to act as a whole?
- How do we know if the data provided is related to the information contained in our study?
TopicsTopics include, but are not limited to:
- Metadata standard(s) development for Big Data management
- Methodologies, architecture and tools for metadata annotation, discovery, and interpretation
- Case study on metadata standard development and application
- Metadata interoperability (crosswalk)
- Metadata and Data Privacy
- Metadata for Semantic Webs
- Human Factors on Metadata
- Innovations in Big Data management
- Opportunities in standardizing Big Data management
- Digital object architectures and infrastructures for Big Data management
- Best practices and standard based persistent identifiers, data types registry structures and representations for Big Data management
- Query languages and ontology in Big Data
- NoSQL databases and Schema-less data modeling
- Multimodal resource and workload management
- Availability, reliability and Fault tolerance
- Frameworks for parallel and distributed information retrieval
- Domain standardization for Big Data management
- Big Data governance for data integrity, quality, provenance, retention, asset management, and business intelligence
Paper submission instructions
This workshop will only accept for review original papers that have not been previously published. Papers should be formatted based on the IEEE Transactions journals and conferences style; maximum allowed camera-ready paper length is ten (10) pages. Submissions must use the followiing formatting instructions:
8.5" x 11" x 2 (DOC, PDF, LaTex Formatting Macros)
Please use this submission site to submit your paper(s).
Accepted papers will be published in the IEEE BigData2017 proceedings (EI indexed). For further information please see IEEE BigData2017 @ http://cci.drexel.edu/bigdata/bigdata2017.
All submitted papers will be reviewed by 3 international program committees.
Hackathon: 24 hours on Data Mashup (Varieties Problem) Big Data Analytics
Hackathon Registration: https://bigdatawg.nist.gov/bdmm_reg.php
Problem – Healthcare Fraud Detection
Large amount healthcare data is produced continually and store in different databases. With the wide adoption of electronic health records that has increased the amount of data available exponentially. Nevertheless, the healthcare providers have been slow to leverage the vast amount of data to improve healthcare system or use data to improve efficiency to reduce overall cost of healthcare.
Health care data has the potential to innovate the procedure of healthcare delivery in the US and inform healthcare providers about the most efficient and effective treatments. Value-based healthcare programs will provide incentives to both healthcare providers and insurers to explore new ways to leverage healthcare data to measure the quality and efficiency of care.
It is estimated that in the US healthcare spending approximately, $75B to $265B is lost each year to healthcare fraud1. With the amount of healthcare fraud, the importance of identifying fraud and abuse in healthcare cannot be ignored; healthcare providers must develop automated systems to identify fraud, waste and abuse to reduce its harmful impact on their business.
1 White SE. Predictive modelling 101. How CMS’s newest fraud prevention tool works and what it means for providers. J AHIMA. 2011;82(9): 46–47.
Develop data mashup scheme to cross reference different healthcare datasets    and apply statistical analysis, visualization, and machine learning tools to statistically analyze and develop predictive models for healthcare payment data and possibly detect irregularities and prevent healthcare payment fraud. Think outside the box and come up with innovative ideas that bring more value out of the data, or choose one or more of the following to solve:
- How many physicians from each state?
- How many specializations out of how many physicians?
- Map anomalies or missing data across the country or within states or counties or electoral districts
- Correlate anomalies with research funding of the respective conditions
- Identify counties/ hospitals/ suppliers etc. with most or least anomalies
- List top 5 anomalies with probability ranking and wrong charges statistics
All participants must be registered via the IEEE Big Data main conference website and attend physically. You may register as a team (up to four per team) or an individual (we will place you on a team). Each participant brings his/her own laptop with all the necessary computing tools. No remote computing resources are allowed. All implementation must be based on the original work. Participating teams are encouraged to submit implementation approach as a white paper which will be published as part of the IEEE Big Data Governance and Metadata Management publication three months after the hackathon event.
- Medicare Provider Utilization and Payment Data - Physician and Other Supplier from Centers for Medicare & Medicaid Services (CMS) - Physician’s Billing:
https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Physician-and-Other-Supplier.html (9 million records; ~500MB compressed; ~1.7 GB uncompressed)
- National Physician Identifiers (NPI) from CMS - Physician Identifiers:
http://download.cms.gov/nppes/NPI_Files.html (~600 MB)
- Health Care Provider Taxonomy Code Set CSV from National Uniform Claim Committee, American Medical Association - Physician Specialization:
http://www.nucc.org/index.php/code-sets-mainmenu-41/provider-taxonomy-mainmenu-40/csv-mainmenu-57 (~400 KB)
- David Belanger, Chair of IEEE Big Data Technical Community, Stevens Institute of Technology
- Mahmoud Daneshmand, Vice-Chair of BDGMM, Steven Institute of Technology
- Kathy Grise, Senior Program Director, Future Directions, IEEE
- Joan Woolery, Senior Project Manager, Industry Connections, IEEE Standards Association, IEEE
- Cherry Tom, Emerging Technologies Initiatives Manager, IEEE Standards Association, IEEE
- Robby Robson, Member of IEEE Standards Association Standards Board, CEO, Eduworks Corporation
Evaluation CriteriaTechnical Approach (40 pts)
- Data mashup (20)
- Big Data analytics (20)
Novelty (40 pts)
- Creativity (20)
- Efficiency (20)
Results (20 pts)
- Output content (10)
- Output format (10)
- 1st Place: $2,000*
- 2nd Place: $1,000*
- 3rd Place: $500*
- All team members win a t-shirt
'*' - at the discretion from the Evaluation Team
Nov 10, 2017: Due date for full workshop paper submission
Nov 15, 2017: Notification of paper acceptance to authors
Nov 20, 2017: Camera-ready of accepted papers
Dec 1, 2017: Deadline for hackathon sign-up
Dec 11, 2017: Hackathon
Dec 12, 2017: Workshop
Mar 12, 2018: Due date for Hackathon White Paper
Day-1: December 11, 2017
|08:00 – 08:10||Welcome, Wo Chang, Chair of IEEE BDGMM, NIST||08:10 – 08:20||Opening Remark, David Belanger, Chair of IEEE Big Data Technical Community, Stevens Institute of Technology||08:20 – 10:00||Briefing about the use case, datasets, challenges, Q/As, Wo Chang||10:00 – till next day 08:00||Solving hackathon challenges||Next day 08:00 – 09:00||Evaluation, See Team & Criteria||Day-2 Late Afternoon||Award Ceremony|
Day-2: December 12, 2017
|14:00 – 14:10||Welcome, Wo Chang, Chair of IEEE BDGMM, NIST||14:10 – 14:30||Opening Remark, David Belanger, Chair of IEEE Big Data Technical Community, Stevens Institute of Technology||14:30 – 15:00|| Keynote Speaker: Digital Object Architecture
Larry Lannom, Vice President of Corporation for National Research Initiatives (CNRI)
|15:00 – 15:30||Invited Speaker: Managing Big Time Series & Text Data for Unsupervised Feature Representation Learning
Linqfei Wu, Research Staff Member of IBM AI Foundations Lab, IBM T. J. Watson Research Center, US
|15:30 – 15:50||Invited Speaker: Towards FAIR Open Science with PID Kernel Information: The RPID Testbed
Yu Luo, Research Assistant of Data To Insight Center, Indiana University Bloomington. Bloomington, Indiana, US
|15:50 – 16:05||Why-Diff: Explaining Differences amongst Similar Workflow Runs by exploiting Scientific Metadata
Priyaa Thavasimani, Jacek Cala, and Paolo Missie
|16:05 – 16:25||Coffee Break||16:25 – 16:40||Case: Big Geosciences Data Validation Challenges and Achievements
|16:40 – 16:55||Deep Learning for Big Data Analytics: A Review from Fog and Edge Computing Perspective
Swarnava Dey and Arijit Mukherjee
|16:55 – 17:20||Hackathon Ceremony
David Belanger and Kathy Grise
|17:20 – 17:30||Announcement for next BDGMM Event, Wo Chang|
Mr. Larry Lannom is Director of Information Services and Vice President at the Corporation for National Research Initiatives (CNRI), where he works with organizations in both the public and private sectors to develop experimental and pilot applications of advanced networking and information management technologies.
In addition to his activities at CNRI, Mr. Lannom serves as Co-chair of the U.S. Branch of the Research Data Alliance, as a member of the RDA Technical Advisory Board, as a member of the National Data Service Technical Advisory Council, and as a member of the U. S. Treasury Office of Financial Research (OFR) Financial Research Advisory Committee.
Mr. Lannom joined CNRI in September of 1996. Prior to that, he was a Technical Director at DynCorp, Inc., where he served as an advisor on digital library research for the ISTO, CSTO, and ITO offices of the U.S. Defense Advanced Research Projects Agency (DARPA), including initiating the Computer Science Technical Reports (CS-TR) project, DARPA's first effort in the digital library area. In addition, he managed the development of internal information systems for DARPA.
Abstract: Learning effective representation is a key foundation for numerous machine learning and data mining techniques in time-series and NLP applications. Despite a number of feature representation methods including kernel methods and deep learning approaches have been proposed in each domain, the effectiveness and efficiency of most methods are still challenged by either limited number of labeled data or high computational complexity. In this talk, I will introduce a generic framework to generate vector representation of time-series and text. To this end, we first construct a family of positive definite (p.d.) alignment-aware time series or text kernels, guided by a new methodology for transforming a distance metric to a positive-definite kernel. Then we present a novel time-series and text embeddings (RWS and WME), a random features method for these proposed p.d. kernels to learn an unsupervised representation for time-series and text data. Extensive experiments on real-world time-series and text classification tasks demonstrate that RWS and WME can outperform or match current state-of-the-art methods in terms of both testing accuracy and runtime in each domain.
Dr. Lingfei Wu is a Research Staff Member of IBM AI Foundations Lab at IBM T. J. Watson Research Center. He earned his Ph.D. degree in computer science from College of William and Mary in August 2016, under the supervision of Prof. Andreas Stathopoulos. His research interests mainly span in large-scale machine learning, scalable data mining, big data analytics, numerical linear algebra, and high-performance mathematical software.
Yu Luo, Research Assistant of Data To Insight Center, Indiana University Bloomington. Bloomington, Indiana, US
Abstract: Using persistent identifiers (PIDs) to identify digital data products whether a product is a collection, a file, or an object of some types is a good practice in open science. A persistent identifier ensures that a 1:1 relationship between identifier and data product persists into the future. Naming solutions for digital data products eventually resolve a PID down to the digital object it identifies, but the current landscape is limited by multiple solutions with weak interoperability, and inconsistent protocols for getting from PID to data object. In a world of increasing PID use, we will soon be awash with billions of PIDs that all resolve to digital objects using various inconsistent and unpredictable approaches, making it difficult to build higher level services that cross the various approaches. In this talk, I will introduce the RPID Testbed, which is using Handle System and Data Type Registry Service, generating PIDs for digital objects. The PIDs will be assigned as Handles of specified data types in Data Type Registry (DTR) and relative values. Data types in DTR provide a way of easily registering detailed and structured descriptions of data, and reuse them in different PIDs. PIDs contain limited Information (PID Kernel Information) connected to FAIR Principles: findable, accessible, interoperable and reusable. The RPID Testbed aims to the exploration driven by identifying and evaluating minimal information that can go into Kernel Information that can help make Data Objects FAIR and less dependent on the repository system to enforce FAIRness.
Dr. Yu Luo is a Research Assistant of Data To Insight Center at Indiana University Bloomington. He is working on his Ph.D. degree in computer science from Indiana University, under the supervision of Prof. Beth Plale. His research interests mainly span in Persistent Identifier, Provenance, and Data Management.
General Co-ChairsAlex Mu-Hsing Kuo (PhD)
University of Victoria, Canada
Leader, IEEE Big Data Education Tracks
Co-chair, IEEE BDI - Big Data Management Standardization
Mahmoud Daneshmand (PhD)
Professor, Stevens Institute of Technology, USA
Co-Chair, IEEE Big Data Governance and Metadata Management
Co-founder, IEEE BDIs
Digital Data Advisor
National Institute of Standards and Technology, USA
Convenor, ISO/IEC JTC 1/WG 9 Working Group on Big Data
Chair, IEEE Big Data Governance and Metadata Management
Program Co-ChairsKathy Grise
Senior Program Director, Future Directions, IEEE Technical Activities, USA
Yinglong Xia (PhD)
Huawei Research America, USA
Co-chair, IEEE BDI - Big Data Management Standardization
Publicity ChairsCherry Tom
Emerging Technologies Intelligence Manager
IEEE Standards Association
445 Hoes Lane, Piscataway, NJ 08854-4141
Technical Program Committee
|Name||Organization||Country||Paventhan Arumugam||ERNET||India||Claire Austin||S&T Strategies,Environment & Climate Change Canada||Canada||Ismael Caballero||UCLM||Spain||Yue-Shan Chang||National Central University||Taiwan||Periklis Chatzimisios||Department of Informatics, Alexander TEI of Thessaloniki||Greece||Hung-Ming Chen||National Taichung University of Science and Technology||Taiwan||Miyuru Dayarathna||WSO2 Inc.||Sri Lanka||Jacob Dilles||Acuant Corp.||US||Robert Hsu||Chung Hua University||Taiwan||Wei Hu||Nanjing University||China||Carson Leung||University of Manitoba||Canada||Sian Lun Lau||Sunway University||Malaysia||Christian Camilo Urcuqui Lóepz||Icesi University||Colombia||Neil Miller||The bioinformatics for Children's Mercy Hospital||USA||Jinghua Min||China Electronic Cyberspace Great Wall Co., Ltd.||China||Carlos Monroy||Rice University||US||Huansheng Ning||USTB||China||Arindam Pal||TCS Research||India||Lijun Qian||Prairie View A&M University||USA||Weining Qianx||East China Normal University||China||Yufei Ren||IBM||USA||Robby Robson||Eduworks Corporation||US||Angelo Simone Scotto||European Food Safety Authority||Italy||Priyaa Thavasimani||Newcastle University||UK||Alex Thomo||University of Victoria||Canada||Chongang Wang||InterDigital Communications||USA||Jianwu Wang||University of Maryland, Baltimore County||US||Shu-Lin Wang||National Taichung University of Science and Technology||Taiwan||Jens Weber||University of Victoria||Canada||Lingfei Wu||IBM Research||USA||Hao Xu||University of North Carolina at Chapel Hill||US||Godwin Yeboah||University of Warwick||UK||Tim Zimmerlin||Automation Technologies||US|