Basel Shbita, Ph.D.

scientist, engineer, and artist

Contact at basel921 at gmail dot com

Basel Shbita is a scientist, engineer, and artist, currently working as a Research Scientist at IBM Research - Silicon Valley in San Jose. He focuses on Generative AI (GenAI), large language models (LLMs), and data-centric AI as part of the Granite project, contributing to the development of production-grade foundation models.

In 2024, Basel earned his Ph.D. in Computer Science, advised by Prof. Craig Knoblock, at the Center on Knowledge Graphs within the University of Southern California (USC) Information Sciences Institute (ISI). Basel's research focuses on techniques for structuring, integrating, and leveraging data for machine learning systems, and on combining these techniques with knowledge graphs to solve complex information integration problems, with recent work applying these ideas to LLMs and GenAI. He has participated in several projects at USC, including Linked Maps, Critical-MAAS, Model Integration (MINT), Table Understanding, and the Alexa Prize Socialbot Grand Challenge 4. For more about Basel's published academic research, please visit this page.

Basel has over 10 years of experience building high-performance and scalable systems, with a strong background in software engineering, programming, and debugging across various languages, having assumed several tech lead roles. Before pursuing his PhD at USC, he worked for Mellanox Technologies (acquired by Nvidia) from 2011 to 2017 as a Senior Firmware Engineer and Team Leader in the Switch Silicon Core Department, and for Apple in 2018 as an Embedded Software Engineer in the Flash Storage Software Department. Basel received his B.Sc. degree in Electrical Engineering from Tel Aviv University in 2015.

Besides his research and academic work, Basel practices plastic and visual arts, exploring various mediums including hand drawing, painting, and AI-driven artworks.

Selected Publications

Teaching

TA (Teaching Assistant) for DSCI 558: Building Knowledge Graphs (Graduate). Fall 2020 @ USC

TA (Teaching Assistant) for CSCI 563/INF 558: Building Knowledge Graphs (Graduate). Spring 2020 @ USC

News

2026/03: Our paper on Adaptive Decoding via Test-Time Policy Learning for Self-Improving Generation has been accepted to the 14th International Conference on Learning Representations (ICLR' 2026) Workshop on AI with Recursive Self-Improvement which will be held in Rio de Janeiro, Brazil.

2026/03: Our paper on Evaluating Ill-Defined Tasks in Large Language Models has been accepted to the 14th International Conference on Learning Representations (ICLR' 2026) Workshop on I Can't Believe It's Not Better: Where Large Language Models Need to Improve which will be held in Rio de Janeiro, Brazil.

2026/03: Served as a reviewer for the CVPR 2026 Workshop on Emerging Directions in Data for Multimodal Foundation Models (DataMFM).

2026/02: Our paper on Context-Aware Visual Multi-Turn Conversation Generation from Wikipedia and Wikidata has been accepted to the 23rd Extended Semantic Web Conference (ESWC' 26) which will be held in Dubrovnik, Croatia.

2025/12: Presented & published a paper on MermaidSeqBench: An Evaluation Benchmark for LLM-to-Mermaid Sequence Diagram Generation at the LLM Evaluation Workshop at the 39th Annual Conference on Neural Information Processing Systems (NeurIPS 2025).

2025/12: Received a Corporate O-level (Outstanding) Accomplishment Award at IBM Research for AI Governance Innovations to watsonx.governance and Open Source, recognizing impactful contributions to enterprise AI governance and open innovation.

2025/12: Served as a Program Committee (PC) member for the 2026 ACM Web Conference (formerly WWW).

2025/11: Our work on Exploiting LLMs and Semantic Technologies to Build a Knowledge Graph of Historical Mining Data was recognized as Best paper (In-Use Track) runner-up at the 24th International Semantic Web Conference (ISWC'25). This was accompanied by a demo paper at ISWC'25 titled Demonstrating MinMod: A Large-scale Knowledge Graph of Historical Mining Data. This work is part of DARPA's CriticalMAAS program. Here's a short video about this project.

2025/11: Our work at the ACM SIGSPATIAL 2025 conference, Exploiting Polygon Metadata to Colorize Draft Maps, received the Best short paper award.

2025/10: Attended BayLearn 2025, the Machine Learning Symposium in the San Francisco Bay Area, hosted this year at Santa Clara University.

2025/08: Served as a reviewer for the NeurIPS 2025 Workshop on Evaluating the Evolving LLM Lifecycle: Benchmarks, Emergent Abilities, and Scaling.

2025/08: Served as a Program Committee (PC) member for the AAAI 2026 main track.

2025/07: Served as program committee member (Industry track) in the 24th International Semantic Web Conference (ISWC' 25).

2025/04: Presented & published a paper on Embedding Spatial and Semantic Contexts for Geo-Entity Typing in Smart City Applications at the 11th International Smart City Workshop co-located with the 2025 ACM Web Conference (formerly WWW).

2025/01: Spotlighted in "The new generation of ISI Ph.D.s", reflecting on my time & research at ISI.

2025/01: Published my first patent, Implementing Personal Knowledge Graphs (Publication No: 20250007723); The patent introduces systems and methods for creating and managing Personal Knowledge Graphs to organize user information into an integrated graph-based structure.

2024/12: Received a Corporate O-level (Outstanding) Accomplishment Award at IBM Research for Granite 3.0 Language Models: Open, state-of-the-art enterprise models, recognizing contributions to IBM's open and high-performing enterprise language models (LLMs).

2024/11: Our team's work on Learning Semantic Descriptions of Tables was recently published as full paper at the 23rd International Semantic Web Conference (ISWC' 24).

2024/08: Served as program committee member (Industry track) in the 23rd International Semantic Web Conference (ISWC' 24).

2024/06: Joined IBM Research in Almaden, San Jose as a Research Scientist at the Enterprise Data & Governance for AI Models as part of IBM's Granite project. I will be working on knowledge graphs and data integration for LLMs in the domain of business transformation.

2024/05: Presented & published a paper on Constructing a Knowledge Graph of Historical Mining Data at the 6th International Workshop on Geospatial Linked Data (GeoLD) co-located with the 21st Extended Semantic Web Conference (ESWC' 2024).

2024/05: Featured in "Parting Thoughts as Eight ISI Graduate Research Assistants Earn their Doctoral Degrees", celebrating the 2024 ISI GRA graduates.

2024/05: Successfully defended my PhD dissertation. I'm very grateful for the invaluable support & encouragement of my committee (Prof. Craig A. Knoblock [chair], Prof. Cyrus Shahabi, Prof. John P. Wilson, Prof. Jay Pujara, Prof. Yao-Yi Chiang).

2024/03: Guest on the AI/nsiders Podcast, discussing my views on AI and human knowledge in EPISODE 17 of the show.

2024/03: Served as a reviewer (Research track) in the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD' 24).

2024/02: Presented & published a paper on Automatically Constructing Geospatial Feature Taxonomies from OpenStreetMap Data at the 18th IEEE International Conference on Semantic Computing (ICSC' 24).

2023/12: We authored a comprehensive chapter titled GeoAI for the Digitization of Historical Maps in collaboration with UMN's Knowledge Computing Lab. This chapter is featured in the Handbook of Geospatial Artificial Intelligence.

2023/11: Our team's work on polygonal feature extraction for raster map understanding was recently published as full paper at the prestigious 31st ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL' 23).

2023/08: Served as program committee member (Industry track) and reviewer (Research & In-use tracks) in the 22nd International Semantic Web Conference (ISWC' 23).

2023/05: My work last summer at IBM Research just got published at the 20th Extended Semantic Web Conference (ESWC' 23) which will be held in Hersonissos, Greece.

2023/04: Published journal paper Building Spatio-Temporal Knowledge Graphs from Vectorized Topographic Historical Maps in the Semantic Web journal!

2022/11: Served as program committee member (Industry track) in the The 2023 Web Conference (formerly known as International World Wide Web Conference or WWW).

2022/09: Featured on USC Viterbi web article alongside my colleagues at USC ISI talking about my research internship experience at IBM this summer.

2022/07: Served as reviewer (Research track) & program committee member (Industry track) in the 21st International Semantic Web Conference (ISWC' 22).

2022/05: Joined IBM Research in Almaden, San Jose. I will be working on knowledge graphs and data integration in the domain of business transformation!

2022/05: My work last summer at GE Research just got published at the 16th International Symposium on Functional and Logic Programming (FLOPS' 22).

2022/04: Received the University Outstanding Teaching Assistant Award at USC's 41st Annual Academic Honors Convocation. This is the first time since 2017 that this university-wide award is given to a student in Viterbi & the first since the start of these awards in 1991 that the award goes to a student in computer science. So grateful!

2021/10: Served as reviewer in the 10th International Joint Conference on Knowledge Graphs (IJCKG' 21).

2021/09: Featured (my favorite is @ 17:37) in the behind-the-scenes video of the Amazon Science's Alexa Prize Socialbot Grand Challenge 4.

2021/08: The Alexa Prize Socialbot Grand Challenge 4 has concluded. Although our socialbot Viola was not part of the finalists, we have come far as first-timer semi-finalists. Very proud of the team! Check out our system paper.

2021/06: We published 2 papers at the 2nd International Workshop on Knowledge Graph Construction at ESWC 2021: A short paper, and a demo paper. The projects were led by 2 excellent groups of CS MSc students at USC that I've had the honor to TA.

2021/05: Joined GE Research as a research fellow intern for 12 weeks this summer. I will be working at the Artificial Intelligence Technical Domain on a project at the intersection of Machine Learning, Knowledge Discovery, Semantic Web, and NLP.

2021/05: Passed my Ph.D. qualifying exam!

2021/01: Featured on VoyageLA Magazine! I was really honored and excited to share my story.

2020/11: Happy to announce that our team from USC will be competing in the Alexa Prize Socialbot Grand Challenge 4, a multimillion-dollar university challenge focused on advancing human-computer interaction!

2020/06: Presented paper (talk) on Building Linked Spatio-Temporal Data from Vectorized Historical Maps at the 17th Extended Semantic Web Conference (ESWC' 20) which was held online (originally planned in Heraklion, Crete, Greece).

2019/09: Presented paper (talk) on Creating a FAIR Data Catalog to Support Scientific Modeling at AKTS 2019. A workshop co-located with the IEEE eScience Conference in San Diego, CA. Our work was referred to as "Science Fiction" by none other than Wikidata's founder; Read about it here!

2019/05: Presented paper (talk) and poster on Parsing, Representing and Transforming Units of Measure at MOMACS's 2019 Modeling the World's Systems Conference in Washington, D.C.

2018/12: Delivered a talk at Tira Talks 4 (a TEDx event held in my hometown). I talked about the importance of challenges, change, globalization, diversity and AI. I couldn't be more honored to stand in front of almost 500 people and contribute to the place where I grew-up.