CST: Lightweight Risk Triage with Source-Grounded Evidence for Web Discussions

Zhihua Wang, Yifeng Gao, Yunhao Feng, Rui Pan, Baiheng Zhao, Zixing Chen, Nan LiarXiv 2026 · arXiv

TL;DR

CST (Character-Semantic Triage) is a lightweight moderation pipeline that moves beyond simple risk scores to provide "source-grounded evidence" for web discussions. By combining character-level patterns with semantic embeddings and grouping related messages, CST helps moderators find more harmful content and make decisions 27% faster than traditional methods—all while running efficiently on a standard CPU.

What Problem Does This Paper Address?

Online moderation is a massive, high-stakes game of "needle in a haystack." Volunteer moderators on platforms like Reddit work thousands of hours daily, yet they can’t possibly read every message. Current AI tools typically assign a "risk score" to individual messages, but these scores lack context.

As the authors point out, meaning is context-dependent: a simple "I agree" could be a harmless affirmation or an endorsement of a vicious personal attack, depending on the parent message.

Context matters: "I agree" can support or oppose removal based on the thread.

Moderators don't just need to know where to look; they need to know what to read and how to verify it without losing the thread of the conversation.

Methods

The CST framework is designed to be "lightweight," meaning it avoids the heavy computational cost of fine-tuning massive Transformer models. It consists of three main layers:

  1. Shared Representations: It pairs a frozen multilingual MiniLM (for sentence meaning) with character n-gram hashing (to capture exact toxic wording). This hybrid approach allows it to handle multiple languages and run on a CPU.
  2. Discussion Ranking: Instead of just scoring messages, CST aggregates message risks, semantic summaries, and reply relations to rank entire discussions, helping moderators prioritize which threads to open first.
  3. Independent Organization: This is the "evidence" part. CST clusters related messages into groups and names them using phrases extracted directly from the text. It then uses a group round-robin selection strategy to ensure a small reading budget covers diverse harms rather than repeating similar messages.
The CST framework: from raw discussion to source-linked review.

Datasets

The authors evaluated CST across six tasks from five diverse datasets:

  • SCCD: A session-based Chinese cyberbullying dataset from Weibo (the primary benchmark).
  • CONDA: In-game toxicity from Dota 2 chats.
  • ToxiCN: Chinese toxic language.
  • HateBR: Brazilian Instagram comments (Portuguese).
  • ScamSpot: Instagram spam and scam comments.

Key Insights and Ideas

  • Risk Scores as Queues, Not Reading Orders: The authors argue that simply reading messages in descending order of risk score is inefficient. It often leads to "redundancy," where a moderator reads five variations of the same insult. By selecting messages across different semantic groups (round-robin), CST recovers more unique harmful instances within the same "reading budget."
  • The Power of "Shallow" Features: Surprisingly, the combination of character-level features and frozen embeddings outperformed fully fine-tuned models like XLM-R. Character features are particularly good at catching exact slurs or "leetspeak" that semantic models might smooth over.
  • Source-Groundedness is Key: By keeping original message IDs and reply links, CST ensures that every piece of evidence is "traceable." Moderators can click back to the original thread to see the full context, which is vital for maintaining high decision accuracy.
Comparing selection strategies: Group round-robin covers more ground than simple risk ordering.

Key Results

  • Better Accuracy: CST achieved higher Average Precision (AP) than a fully fine-tuned XLM-R model across all six message classification tasks.
  • Faster Review: In a user study, moderators using CST’s grouped evidence needed 27.3% less time to review discussions while maintaining over 94% decision accuracy.
  • Efficiency: The complete pipeline runs 2.6x faster than an XLM-R service on a standard CPU. It processes a discussion in less than a second on average.
  • Higher Recall: Under a fixed budget (e.g., reading only 3 messages per discussion), the group round-robin strategy found significantly more harmful messages than picking by risk score alone.

Limitations and Future Work

The authors acknowledge several limitations:

  • Dataset Bias: Much of the evidence-selection and LLM-judging data comes from a single Chinese platform (SCCD).
  • Confounded Variables: In cross-language transfer tests, the source platform and the language change simultaneously, making it hard to isolate which factor impacts performance more.
  • Stance Rules: The specific lexical rules used to protect stance cues (like negation) currently only cover Chinese, English, and Portuguese.
  • Live vs. Controlled: The user study was conducted in a controlled environment; real-world moderation involves different workloads and incentives that might affect how these tools are used in practice.

Summary updated

Abstract

Web communities rely on moderators to catch harmful messages such as insults, hate speech, and scams, but no one can read every message. Existing tools give each message a risk score, which tells moderators where to look but not what to read or how to check it. Yet meaning depends on context: the same reply, "I agree.", can support or oppose removing someone from a group. Moderators therefore need source-grounded evidence, not just scores: a few key messages kept with their original wording and replies. We present CST, a lightweight triage pipeline that first decides where to look, then what to show. It scores risk by pairing character patterns, which capture exact wording, with sentence meaning from a fixed multilingual model, so it handles several languages and runs on an ordinary CPU. It then groups related messages and picks from each group in turn, so a small reading budget covers different parts of a discussion, and links every pick back to the original thread. Across six tasks from five datasets in several languages, CST ranks harmful messages more accurately than a fully fine-tuned multilingual model and, under the same budget, finds more of them than reading by score alone. In a user study, reviewers needed at least 27% less time with CST's grouped evidence while keeping decision accuracy above 94%. CST turns risk scores into evidence that moderators can read quickly and check in context. Code and dataset reconstruction resources will be released at https://github.com/Kirito14IT/CST-MIL.
BibTeX
@misc{wang2026cst,
  author = {Zhihua Wang and Yifeng Gao and Yunhao Feng and Rui Pan and Baiheng Zhao and Zixing Chen and Nan Li},
  title = {{CST: Lightweight Risk Triage with Source-Grounded Evidence for Web Discussions}},
  year = {2026},
  eprint = {2611.01710},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.01710},
}
Figure