Publications

PCA-OS overview figure showing evidence inputs, world model, adaptation intervention ledger, causal effect atlas, decision layer, and outcomes

Enlarge figure

KDD 2026 Accepted

PCA-OS: A Planetary Climate Adaptation Operating System

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Blue Sky Ideas Track work framing climate adaptation as an operating-system problem.

Preprints.org Project site GitHub

ESGenius pipeline figure showing ESGenius corpus construction, question generation, zero-shot evaluation, and RAG evaluation flow

Enlarge figure

EMNLP 2025 Oral

ESGenius: Benchmarking LLMs on ESG and Sustainability Knowledge

Chaoyue He, Xin Zhou, Yi Wu, Xinjia Yu, Yan Zhang, Lei Zhang, Di Wang, Shengfei Lyu, Hong Xu, Xiaoqiao Wang, Wei Liu, Chunyan Miao

Strong accept with Resource Award and Theme Paper Award nominations.

ACL Anthology Code

MMESGBench pipeline figure showing ESG document sources, MLLM-based QA generation, automated filtering, expert calibration, and benchmark statistics

Enlarge figure

ACM MM 2025 Dataset Track

MMESGBench: Multimodal Understanding and Complex Reasoning for ESG Tasks

Lei Zhang, Xin Zhou, Chaoyue He, Di Wang, Yi Wu, Hong Xu, Wei Liu, Chunyan Miao

Datasets Track work connecting multimodal reasoning with ESG evaluation.

ACM DL alphaXiv

Conceptual overview of a domain expert editing a knowledge graph and dynamic subgraph to guide an LLM in producing a professional document, with an iterative review loop

Enlarge figure

HCII 2026 Accepted

An Expert-in-the-Loop Design Utilising Knowledge Graphs to Prompt LLMs in Professional Writing

Xinjia Yu, Chaoyue He, Xin Zhou, Lei Zhang, Chunyan Miao

Knowledge-graph-guided design for LLM-assisted professional writing.

ACM DL

OpenClaw AERO framework figure organizing public agent ecosystem works across grounding, action, transfer, and exchange with delegated-force dimensions

Enlarge figure

EMNLP 2026 Under review

OpenClaw as Language Infrastructure: A Case-Centered Survey of a Public Agent Ecosystem in the Wild

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

EMNLP 2026 ARR review: Strong Findings, with recommendation to Borderline Main Conference. Case-centered analysis of a public agent ecosystem, framing agentic language use as executable, persistent infrastructure.

Preprints.org

Harness Engineering overview figure showing software, prompt, and context engineering widening into the harness layer across control, agency, and runtime

Enlarge figure

Apr 2026 Submitted

Harness Engineering for Language Agents: The Harness Layer as Control, Agency, and Runtime

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to EMNLP 2026 (ARR May). Framework for the extra-model layer that governs language-agent control, action interfaces, state, and runtime reliability.

Preprints.org

SP2DPO overview figure comparing static DPO temperature with semantic per-pair temperature schedules guided by an upstream LLM annotator

Enlarge figure

Jan 2026 Submitted

SP^2DPO: An LLM-assisted Semantic Per-Pair DPO Generalization

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to ICML 2026. Preference-optimization work replacing a single global DPO temperature with auditable semantic per-pair schedules.

arXiv

Human-AI productivity time-to-acceptance figure comparing human-only and AI plus human workflows across production, verification, governance, rework, and acceptance

Enlarge figure

2026 Submitted

Human-AI productivity claims should be reported as time-to-acceptance under explicit acceptance tests

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to ICML 2026 Position Track. Work reframing AI productivity evidence around explicit acceptance tests and time-to-acceptance reporting.

TechRxiv

KG4ESG overview figure showing ESG data sources, social, governance, holistic, and environmental topics, Data-to-KG pipelines, expert review, and KG applications

Enlarge figure

Mar 2026 Submitted

KG4ESG: The ESG Knowledge Graph Atlas

Chaoyue He, Xin Zhou, Di Wang, Xinjia Yu, Lei Xiao, Langyue Li, Hong Xu, Wei Liu, Chunyan Miao

Submitted to EMNLP 2026. Atlas-style survey of ESG knowledge graph research, organizing the field around Data-to-KG and KG-to-App pipelines.

Preprints.org

SSKG Hub overview figure showing standards-aware ingestion, Neo4j graph storage, platform modules, expert review, audit logs, and web app screens

Enlarge figure

Mar 2026 Submitted

SSKG Hub: An Expert-Guided Platform for LLM-Empowered Sustainability Standards Knowledge Graphs

Chaoyue He, Xin Zhou, Xinjia Yu, Lei Zhang, Yan Zhang, Yi Wu, Lei Xiao, Liangyue Li, Di Wang, Hong Xu, Xiaoqiao Wang, Wei Liu, Chunyan Miao

Submitted to EMNLP 2026 Demo Track. Interactive platform for transforming sustainability standards into auditable knowledge graphs with expert review and certification.

arXiv Platform

AutoResearch timeline figure showing research automation milestones from AutoML-Zero and AlphaTensor through AI Scientist, AI Research Agents, FARS, and AutoResearch

Enlarge figure

Mar 2026 Submitted

The AutoResearch Moment: From Experimenter to Research Director

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to EMNLP 2026 (ARR May). Position paper on agentic research automation, claim governance, and the shift from experiment execution to research direction.

Preprints.org

ESGlass overview figure showing observation inputs, evidence support, derived computations, provenance and policy, uncertainty, replay, and stakeholder views

Enlarge figure

ACMMM 2026 Submitted

ESGlass: Glass-Box ESG and Sustainability Reports

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to ACMMM 2026 BNI Track. Glass-box reporting work for ESG and sustainability evidence, traceability, and report-generation support.

Preprints.org

Synthetic Media Exchange overview figure showing generation and orchestration, provenance graph, rights and licensing, trust and evidence, derivative settlement, and consumption governance

Enlarge figure

ACMMM 2026 Submitted

The Synthetic Media Exchange: When Lineage Becomes Currency

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to ACMMM 2026 BNI Track. Multimedia-systems proposal treating synthetic-media lineage, provenance, and exchange as core infrastructure.

Preprints.org

Agentic Multimedia Firm overview figure showing an executive planner and orchestrator coordinating creative, editorial, compliance, distribution, ledger, oversight, and external-resource agents

Enlarge figure

ACMMM 2026 Submitted

From Prompts to Portfolios: AI Agents as Agentic Multimedia Firms

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submitted to ACMMM 2026 BNI Track. Agentic multimedia work framing AI agents as firms that compose prompts, assets, workflows, and portfolios.

Preprints.org

Recommender systems toward agents figure comparing RSTH, agentic recommender systems toward humans, and recommender systems toward agents paradigms

Enlarge figure

Apr 2026 Submitted

Recommender Systems Should Now Be Designed Towards Agents

Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao

Submission-stage position paper arguing for agent-facing recommendation layers that rank tools, interventions, oversight actions, and downstream trajectories.

Preprints.org

Multi-scale Quasi-RNN recommender architecture figure showing convolutional gating, dynamic average pooling, and single-scale Quasi-RNN components

Enlarge figure

2019 Preprint

Multi-Scale Quasi-RNN for Next Item Recommendation

Chaoyue He, Yong Liu, Qingyu Guo, Chunyan Miao

Sequential-recommendation work exploring multi-scale quasi-recurrent neural networks for next-item prediction.

arXiv

Selected / Accepted

PCA-OS: A Planetary Climate Adaptation Operating System

ESGenius: Benchmarking LLMs on ESG and Sustainability Knowledge

MMESGBench: Multimodal Understanding and Complex Reasoning for ESG Tasks

An Expert-in-the-Loop Design Utilising Knowledge Graphs to Prompt LLMs in Professional Writing

Preprints & Under Review

OpenClaw as Language Infrastructure: A Case-Centered Survey of a Public Agent Ecosystem in the Wild

Harness Engineering for Language Agents: The Harness Layer as Control, Agency, and Runtime

SP^2DPO: An LLM-assisted Semantic Per-Pair DPO Generalization

Human-AI productivity claims should be reported as time-to-acceptance under explicit acceptance tests

KG4ESG: The ESG Knowledge Graph Atlas

SSKG Hub: An Expert-Guided Platform for LLM-Empowered Sustainability Standards Knowledge Graphs

The AutoResearch Moment: From Experimenter to Research Director

ESGlass: Glass-Box ESG and Sustainability Reports

The Synthetic Media Exchange: When Lineage Becomes Currency

From Prompts to Portfolios: AI Agents as Agentic Multimedia Firms

Recommender Systems Should Now Be Designed Towards Agents

Earlier Work

Multi-Scale Quasi-RNN for Next Item Recommendation