Zhizhuo Kou

Zhizhuo Kou

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2025
Verified
Conference Paper
Open access
PDF file
ZKZhizhuo Kou
Zhizhuo Kou
HYHolam Yu
Holam Yu
JLJunyu Luo
Junyu Luo
JPJingshu Peng
Jingshu Peng
XLXujia Li
Xujia Li
YGYike Guo
Yike Guo
Zhizhuo Kou, Holam Yu, Junyu Luo, Jingshu Peng, Xujia Li, Chengzhong Liu, Juntao Dai, Lei Chen, Sirui Han, Yike Guo. Findings of the Association for Computational Linguistics: EMNLP 2025. 2025.

3 February 2026 in ACM Computing Surveys
Verified
Journal Article
Open access
LWLiping Wang
Liping Wang
JLJiawei Li
Jiawei Li
LZLifan Zhao
Lifan Zhao
ZKZhizhuo Kou
Zhizhuo Kou
XWXiaohan Wang
Xiaohan Wang
LCLei Chen
Lei Chen
Predicting stock prices presents a challenging research problem due to the inherent volatility and non-linear nature of the stock market. In recent years, knowledge-enhanced stock price prediction methods have shown groundbreaking results by utilizing external knowledge to understand the stock market. Despite the importance of these methods, there is a scarcity of scholarly works that systematically synthesize previous studies from the perspective of external knowledge types. Specifically, the external knowledge can be modeled in different data structures, which we group into non-graph-based formats and graph-based formats: (1) non-graph-based knowledge captures contextual information and multimedia descriptions specifically associated with an individual stock; (2) graph-based knowledge captures interconnected and interdependent information in the stock market. This survey article aims to provide a systematic and comprehensive description of methods for acquiring external knowledge from various unstructured data sources and then incorporating it into stock price prediction models. We also explore fusion methods for combining external knowledge with historical price features. Moreover, this article includes a compilation of relevant datasets and delves into potential future research directions in this domain.


9 May 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
ZKZhizhuo Kou
Zhizhuo Kou
ZQZhiqiang Qian
Zhiqiang Qian
ZZZhenghao Zhu
Zhenghao Zhu
JXJiyuan Xin
Jiyuan Xin
YCYakun Cui
Yakun Cui
YGYike Guo
Yike Guo
Credit risk assessment requires both accurate prediction and structured decomposition of how hetero-geneous evidence contributes to each decision. Monolithic Large Language Models can incorporate unstructured evidence and natural-language reasoning into such workflows, but in high-stakes un- derwriting they may be distracted by noisy inputs, miss rare but decisive risk cues, and offer limited control over policy-dependent decision thresholds. We present CREDITAGENT, a hierarchical credit review system with three stages: evidence filtering, specialist risk analysis by agents, and decision fusion. Our central contribution is holding the adapted backbone, specialist-agent outputs, hard-stop rules, and data split fixed, we vary only the final fusion strategy to isolate the effect of hierarchical fusion on underwriting quality. On a held-out set of 6,000 personal credit cases from Chinese financial institution, CREDITAGENT achieves 83.32% accuracy and a Business Efficiency Coefficient of 0.7647 outperform flagship model. We present these findings as an institution-specific case study while identifying which components (hierarchical fusion, GRPO training recipe) are mechanism-portable versus institution-specific (hard-stop rules, cost ratios). To ensure reproducibility, we make code and dataset publicly available at https://github.com/kouzhizhuo/Credit_Agents.

12 February 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
SHSirui Han
Sirui Han
ZKZhizhuo Kou
Zhizhuo Kou
RLRuoxi Li
Ruoxi Li
YZYuyao Zhang
Yuyao Zhang
YZYujin Zhou
Yujin Zhou
YGYike Guo
Yike Guo
As large language models are increasingly used for contract drafting, case research and even judicial work, a central question is how to make their outputs trustworthy. This survey addresses that question through the lens of verified generation for legal AI, focusing on systems that are robust against hallucinations and traceable to authoritative legal sources. First, we propose a unified framework for verified generation in legal AI, linking reasoning, retrieval, and validation around factual reliability. Second, we cast reliability methods into two paradigms of epistemic negotiation, by failure and by conflict, enabling models to recognize and act on their competence limits. Third, we survey the legal-AI landscape and identify challenges for verifiable, governance-native systems. This survey outlines a roadmap for trustworthy legal AI and for reliable reasoning beyond the legal domain.

8 May 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
ZKZhizhuo Kou
Zhizhuo Kou
JYJian Yang
Jian Yang
JLJunyu Luo
Junyu Luo
YZYuyao Zhang
Yuyao Zhang
SHSirui Han
Sirui Han
YGYike Guo
Yike Guo
Modular decision systems expose multiple operating points, but downstream utility can vary by regime. Portfolio construction is a useful setting because routing, aggregation, and allocation can help or hurt depending on market structure. We instantiate a hierarchy with three operating points: direct optimizer, routed consensus, and alpha-augmented optimizer. Across universes, these modes are not uniformly ranked. Routing helps when dispersion/decorrelation is high; direct optimization is safer in low-signal settings; alpha augmentation helps in concentrated signal-rich settings. We identify the market characteristics (cross-sectional dispersion, concentration, forecast signal density) that predict which mode dominates. A rolling adaptive meta-policy that selects among operating points based on recent performance achieves competitive or superior risk-return profiles without foreknowledge of the optimal mode. We validate against classical baselines demonstrate that the operating-point structure persists across time frequencies from 15-minute bars to daily rebalancing, and confirm robustness under realistic transaction costs (0--15 bps). More broadly, our results suggest for hierarchical decision systems: the key is not to find one universally best configuration, but to characterize when each operating mode is most effective. To support future research and ensure reproducibility, we make source code publicly available at https://github.com/kouzhizhuo/Regime-Adaptive-Portfolio-Agents.