Jun 2010 – Aug 2010 · 2 mos · Singapore

• Analyzed and identified a major drawback in extended Bayesian multiple imputation for missing data
• Achievement: Proposed a novel approach reducing mean absolute error of prediction in MovieLens dataset by 7%

May 2009 – Jul 2009 · 2 mos · Saarbrücken, Saarland, Germany

• Objective: Reasoning a large knowledge base, namely YAGO using an automated theorem prover SPASS
• Conceptualized and developed an automated translator from the YAGO ontology to SPASS input syntax
• Achievement: Restructured the YAGO formal semantics which significantly improved its efficiency by 400%

May 2008 – Jul 2008 · 2 mos · Saarbrücken, Saarland, Germany

• Objective: Enabling SPASS to overcome copyright issue from UC Berkeley by moving into Apache style license
• Developed a finite state automaton based command line parser module which replaced the earlier module
• Achievement: This research work was published at the Conference on Automated Deduction’22, Canada