---
title: 'Sub-claim Extraction: Methods and Evaluation'
url: https://www.emergentmind.com/topics/sub-claim-extraction
type: topic
---

# Sub-claim Extraction: Methods and Evaluation

Sub-claim extraction, also referred to as claim decomposition, is the process of mapping a complex, potentially multi-faceted claim or natural language sentence into a finite set of atomic sub-claims or propositions. Each sub-claim should represent a minimal, self-contained assertion that can be independently verified or refuted against evidence. This decomposition is central in fact verification, adversarial claim checking, and fine-grained factuality evaluation, as it allows for targeted validation of each aspect of a statement and robust error localization. The choice of decomposition strategy—methodology, automation, and quality controls—has direct consequences on the reliability, interpretability, and robustness of downstream scoring or verification systems.

## 1. Formal Definitions and Decomposition Frameworks

Sub-claim extraction formalizes complex claim analysis as follows: given an input claim $C$, produce a set of $m$ atomic sub-claims $\{s_j\}_{j=1}^m$, with each $s_j$ constructed to be independently interpretable and directly verifiable. In structured claim verification systems, each $s_j$ is paired with a subset of evidence $E_j \subseteq E_C$ and receives a veracity label $y_j \in \{\mathrm{T}, \mathrm{F}, \mathrm{U}\}$ (True/False/Unverified), with an aggregation function $f$ yielding the final claim-level verdict $Y_C$:
$$
Y_C = f\big(C, \{(s_j, E_j, y_j)\}_{j=1}^m\big)
$$
Alignment of evidence to sub-claims is an essential component; evidence can be associated at the claim level (SRE: repeated claim-level evidence) or explicitly aligned at the sub-claim level (SAE: sub-claim aligned evidence), as shown in [2602.10380].

Sub-claim boundaries are typically defined as contiguous, short sentences or clauses, each corresponding to a single proposition. The guiding principle (e.g., [2403.11903]) is to maximize atomicity—one fact per sub-claim—while maintaining full coverage of the original claim.

Hierarchical and aspect-based models (e.g., ClaimSpect [2506.10728]) generalize this concept, allowing extraction of sub-claim trees with arbitrarily deep levels, where each node is an aspect or finer-grained sub-aspect.

## 2. Annotation Protocols, Datasets, and Sub-claim Characterization

Annotation schema for sub-claims emphasize atomicity, standalone verifiability, and coverage. For example, in [2602.10380], annotation is carried out by expert annotators, who extract and label each sub-claim, ensuring it is check-worthy and does not rely on implicit context. Each sub-claim receives a veracity label and is mapped to a minimal set of supporting or opposing evidence.

Dataset construction details include:
- PHEMEPlus ([2602.10380]): 399 complex claims, 1,169 sub-claims (≈2.93 per claim), expert-labeled with strict temporal evidence alignment.
- ClaimDecomp ([2205.06938]): 1,494 political claims, 6,555 annotated subquestions (≈2.7 per claim), distinguishing literal and implied subquestions.
- Expanded biography/knowledge base domains (“Core” [2407.03572]; “A Closer Look…” [2403.11903]).

Agreement metrics include Bennett’s $S$ (0.81 in [2602.10380]), Fleiss’ $\kappa$ (0.52 in [2205.06938]), and sentence-level overlap via BLEU/BERTScore. Comprehensive coverage of both explicit and implicit aspects is emphasized, and annotation guidelines recommend atomic, non-overlapping, linguistically coherent formulations.

## 3. Methodologies for Sub-claim Extraction

### Prompt-based LLM Decomposition
The dominant paradigm involves LLM prompting, with templates constructed to elicit highly atomic, self-contained sub-claims from complex input ([2602.10380], [2403.11903], [2407.12425], [2506.04583], [2205.06938]). Extraction is typically zero-shot or in-context, optionally enhanced with post-processing for pronoun resolution and format normalization.

Example methodology ([2407.12425]):
1. Prepare in-context examples: demonstrate atomic splitting, pronoun removal, preservation of qualifiers.
2. Prompt LLM with instruction to dissect into numbered, standalone statements.
3. Parse model outputs into a list of sub-claims.

Pseudo-algorithm (EACon [2407.12425]):
```python
def DeconstructClaim(claim):
    prompt = build_prompt_with_examples(claim)
    raw_output = LLM.generate(prompt, temperature=0.05)
    subclaims = parse_numbered_lines(raw_output)
    return subclaims
```

### Structured Reasoning and RL Optimization
Reinforcement learning frameworks, such as Distill-and-Align Decomposition (DAD [2602.21857]), combine supervised fine-tuning on teacher-distilled decompositions with policy optimization for multi-objective rewards. The DAD reward incorporates: format compliance, alignment with verifier preferences, and decomposition quality as judged by an LLM-based rubric. Sequential prompts enforce claim detection, decontextualization, relationship identification, and atomic extraction.

GRPO (Group Relative Policy Optimization) [2602.21857] provides variance reduction for policy updates, allowing simultaneous optimization across formatting, faithfulness, and task performance.

### Modular and Semantic Filtering
Core ([2407.03572]) augments any decomposition procedure with a selection mechanism that scores sub-claims for faithfulness (entailment by context), informativeness (conditional surprisal via UNLI), and non-redundancy (textual entailment between sub-claims). An integer linear program selects a maximally informative, non-overlapping subset subject to a faithfulness threshold.

### Hierarchical/Aspect Extraction
ClaimSpect ([2506.10728]) recursively applies an LLM and a discriminative retrieval module to construct a tree of aspects and sub-aspects, each node annotated with its own keywords and retrieved supporting segments. Discriminativeness scores guide the selection of corpus evidence most representative of a specific aspect versus distractors.

### Shallow/Parse-based Methods
Earlier approaches involve dependency parsing (PredPatt), semantic role labeling, or shallow segmenters with later LLM re-writing. However, these approaches show lower sub-claim cohesion, atomicity, or coverage compared to in-context LLM decomposition ([2403.11903]).

## 4. Evaluation Methodology, Metrics, and Results

Unlike typical extraction tasks, explicit extraction F1/precision/recall is rarely measured; sub-claim extraction is treated either as an oracle (fixed step), or its quality is assessed by impact on downstream tasks ([2602.10380], [2407.12425], [2506.04583]).

Key evaluation metrics include:
- **DecompScore** ([2403.11903]): Measures the number and coherence of sub-claims entailed by the original claim.
  $$
  \text{DecompScore}(P; D) = \frac{1}{n} \sum_{j=1}^n \sum_{c\in D(C_j)} \mathbf{1}\{ M(c; C_j)=1\}
  $$
  This quantifies both atomicity and coverage.

- **FActScore** ([2403.11903]): Fraction of sub-claims supported by external evidence.
  
- **Human/Automatic Recovery** ([2205.06938]): Human-matched recall rate for subquestions (e.g., 0.74 for literal, 0.18 for implied with T5-3B model).

- **Downstream Gains**: Removal of sub-claim extraction drops claim verification F1 by 4–7 points on standard datasets ([2407.12425], [2506.04583]).

- **Robustness**: Core filtering eliminates inflation of factual precision by trivial or repetitive sub-claims, restoring robustness to adversarial generations ([2407.03572]).

- **Human Judgments**: Structured decomposition (DAD [2602.21857]) achieves high marks (>0.75) on verifiability, coherence, clarity, and uniqueness.

## 5. Error Analysis, Robustness, and Limitations

Explicit error analysis is often focused not on the extraction spans themselves but on the propagation of label errors and alignment mismatches to downstream metrics ([2602.10380]).

Known failure modes include:
- **Granularity Mismatch**: Too coarse or too fine granularity can harm verification accuracy ([2602.21857], [2506.04583]).
- **Implicit Fact Omission**: Standard prompts often miss implied or contextually licensed sub-claims ([2205.06938], [2403.11903]).
- **Hallucinations**: LLMs may invent unsupported sub-claims, especially without coverage checks or strict coherence validation ([2403.11903], [2506.04583]).
- **Non-redundancy**: Repetitive or paraphrased sub-claims can game factual precision metrics unless filtered ([2407.03572]).

Robustness strategies:
- **Coverage + Coherence Filtering** ([2403.11903], [2407.03572]): Use validators to remove unsupported sub-claims.
- **Abstention in Labeling** ([2602.10380]): Aggressive forced prediction of T/F labels propagates more error than conservative "unknown" predictions.

Limitations:
- No standard, reliable reference metric for extraction quality beyond DecompScore.
- Extraction models are primarily tested in English and domain-general or entity-centric settings.
- Many pipelines require extensive in-context example engineering or hand-crafted rules for maximal performance.

## 6. Practical Recommendations and Future Directions

Empirical findings suggest several best practices:
1. **Prioritize atomicity and coverage**: Use decomposition prompts or retrieval-augmented strategies that enforce splitting on conjunctions, explicit qualifiers, and all relevant event arguments ([2403.11903]).
2. **Ground decomposition with retrieval**: Iterative, discriminative retrieval (ClaimSpect) helps align aspect trees to corpus evidence and suppress hallucinations ([2506.10728]).
3. **Filter and select**: Deploy selection layers (Core) to enforce informativeness and eliminate redundancy before verifying sub-claims ([2407.03572]).
4. **Optimize jointly**: Align extraction and verifier behavior via RL or other feedback approaches to tune sub-claim granularity ([2602.21857]).
5. **Validate extraction steps**: Use LLM validators or dedicated coverage checks to ensure both completeness and faithfulness ([2403.11903], [2407.03572]).

Future work is identifying more robust evaluation metrics, extending extraction methods and metrics to multilingual and multi-modal settings, integrating domain ontology constraints, learning adaptive or context-sensitive granularity, and exploring truly end-to-end decomposer-verifier-retriever pipelines ([2506.10728], [2602.21857], [2602.10380]).

Source: https://www.emergentmind.com/topics/sub-claim-extraction