Pengkun Jiao

Output—

Output data is insufficient for this profile: the underlying record provides no computed publication counts or venue-level productivity, so total output volume cannot be characterized. This axis would measure the quantity and distribution of works over time and across venues.

Impact—

Impact data is insufficient: citation-based indicators are not reliably available in this record, and the snapshot's reported h-index of 0 reflects missing coverage rather than measured influence. This axis would measure citation uptake and citation-based indices.

Focus0/100

The focus dimension scores 0, and the raw topics explain why: the OpenAlex record spans five labels — Reinforcement Learning in Robotics (2), Robot Manipulation and Learning, AI-based Problem Solving and Planning, Artificial Intelligence in Games, and Explainable AI — while the field distribution splits evenly (0.5/0.5) between reinforcement learning for LLM agents and LLM-based agents. On a thin, name-resolved record this reads as diffuse tagging rather than demonstrated breadth, so the zero should be treated cautiously.

Collaboration71/100

The collaboration dimension scores 71, supported by 35 distinct coauthors in the recent bibliographic window. The recurring names Feida Zhu and Steven Hoi each appear twice in the frequent-coauthor list, alongside Ming Ma, Yi Zhu, Yiran Zhong, Chonghan Liu, Qichao Wang, Yanhao Jia, and Tianming Yang. Both catalog papers sit in middle author positions on multi-author arXiv preprints, a pattern consistent with large industrial research teams rather than small academic groups.

Activity32/100

The activity dimension scores 32, reflecting a thin recent record: recent3yWorks is 2 and the countsByYear series contains only 2026 with 2 works. Both outputs are arXiv preprints from 2026, so measured activity is concentrated in a single year with no earlier publication trail in this snapshot. The low score signals limited observable cadence rather than inactivity outside the indexed window, and it should be re-read as coverage improves.

Reproducibility38/100

The reproducibility dimension scores 38 (reproMean 38 over 2 reviewed papers). ProCredit rates medium: consistent gains across three AppWorld scales with clean ablations, though bounded by verifier coverage. Qwen-Planner-Agent rates low: its headline 77.05% on the self-built MobilePA-Bench lacks significance tests, weakening external verification. The composite suggests work whose evidence quality depends heavily on whether benchmarks are self-constructed or established.