Foundation Models
Reasoning, generation, and domain adaptation for complex IP knowledge.
Joint Research Initiative · Est. 2024
Intelligence for IP.
Advancing language models, agentic systems, and information retrieval for the creation, management, and examination of intellectual property.
01 / RESEARCH QUESTIONS
How can AI support creation, patentability, examination, and enforcement across the IP lifecycle?
02 / REPRESENTATIVE WORK
AutoPatent · PatRe · FlowPIE · IPBench
03 / OPEN RESOURCES
SIAT-NLP×DUT-IR
SHENZHEN · DALIAN · CHINA
01 / Mission
From foundational research
to professional impact.
IP Intelligence is a joint research initiative connecting SIAT-NLP at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, and DUT-IR at Dalian University of Technology.
Agentic Workflows
Reliable systems that support real-world patent and IP procedures.
Information Retrieval
Evidence-grounded search and discovery across technical documents.
02 / News
Latest from the lab.
Two papers accepted to ACL 2026
IPBench and Beyond Quantity were accepted to ACL 2026 Findings.
View publications →RELEASE
PatRe joins the IPIntelligenceArena
A benchmark for Office Action and rebuttal generation in patent examination.
RESEARCH
IPBench released
10,374 instances and 20 tasks for evaluating IP knowledge and practice.
SYSTEM
AutoPatent introduced
A multi-agent framework for automatic patent generation.