About Me

Xiang Hu is a research scientist from Ant Research. He graduated with a master’s degree from Shanghai Jiao Tong University, where he studied programming languages under the supervision of Prof. Jianjun Zhao. During his career at Ant Group, he developed an interest in natural language processing, started self-studying, and began publishing academic papers. His main interest lies in exploring the inductive biases that facilitate human-like language comprehension, including hierarchical language structures, long-term randomly accessible semantic memory, and the relationship between language and symbols. Currently, his research is primarily conducted along two lines. One line focuses on human-like hierarchical language comprehension by mapping utterance-level representations to ontological concepts, and then linking these concepts to basic neural predicates. Another newly initiated line of research involves ultra-long range language modeling to explore human-like long-term randomly accessible semantic memory. He built a library for building self-supervised hierarchical text representation at ant-research/StructuredLM_RTDT.