Benjamin P Segal  Benjamin P Segal photo         

contact information

IBM Z Research - Master Inventor
Thomas J. Watson Research Center, Yorktown Heights, NY USA




Ben Segal is an IBM Z Researcher that works closely with clients and IBM product development in support of new Z technologies in the areas of security, compliance, and new workload enablement.


Prior to IBM Z Research, Ben was the chief architect of Insights for Patient Data (IPD) within Watson Health, a cognitive service founded from Watson research, that analyzes longitudinal patient data in EMRs and provides problem-oriented summarization and key medical insights to help physicians assess their patients' wellbeing. Ben was responsible for end-to-end product architecture of IPD and routinely part of architecture and execution. He worked collaboratively with research, IPD product development, Cloud platform, and application teams within IBM, and with IBM's clients and partners, to help enable pilots and ensure requirements are factored into a scalable architecture and overall product roadmap.


Prior to joining Watson, he worked in IBM Systems, designing, developing, and/or supporting various z/OS operating system components, Communications Server, hardware management consoles, and new growth initiates for the System Z platform. Ben is an IBM master inventor and author or coauthor of several patents and publications ranging from server-oriented high performance computing models across heterogeneous clusters, high speed networking protocols, to NLP, machine learning and passage scoring analytics. He has advised IP legal, served on the Watson Research IDT review board, and actively mentors other colleagues of varying experience.


Parsing and Domain Adaptation:
Method for Augmenting Unstructured Domain Terms with Syntactic Information (91799581)

Method for extracting kernel lexical chunks (91799517)

Method for Guiding a Rule-Based Parser With Lexical Collocational Information (91128693, 91128694)

Improving Watson Q&A performance:
System and method for answering polar questions (91051439, 91051440, 91051441)
Multi Instance Learning for Question Answering Systems (89819526)
Using Paraphrase Metrics For Answering Questions (91755837)
Evaluating Passages in Question Answering by Aggregation of Relation Matches (91715790)

Improving social media applications with Watson:
Targeted marketing & purchasing algorithm utilizing social media constructs to streamline online retail transactions and to additionally feed into cognitive analysis for smarter commerce (92621097)

System Z and High Performance Computing:
Scientific Kernel Execution for Distributed OpenCL Runtimes (90435876)

Distributed Function Execution for Hybrid Systems (90419179)
Surrogate Execution for Hybrid Systems (90378820, 90378826)
Unified, Workload-Optimized, Adaptive RAS for Hybrid Systems (90377373, 90377386)


High-speed networking:
Shared Memory Communications Protocol (90881477)

RDMA multilink (90881751)