Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Rule2Text: A Framework for Generating and Evaluating Natural Language Explanations of Knowledge Graph Rules

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Rule2Text shows that chain-of-thought prompting with variable entity types lets LLMs produce accurate, clear explanations of mined knowledge-graph rules, and that fine-tuning an open-source model on judge-selected examples substantially imp

desk verdict Genuinely new task and a credible prompt-engineering study, but the fine-tuning headline rests on unvalidated lexical metrics and a partially circular data pipeline. read the letter →

arxiv 2508.10971 v2 pith:MY2XRS2G submitted 2025-08-14 cs.CL cs.AI

Nasim Shirvani-Mahdavi , Chengkai Li This is my paper
classification cs.CLcs.AI
keywords knowledgegraphrulesnaturallanguageexplanationLLM-as-a-judgechain-of-thoughtpromptingvariableentitytypestypeinferencefine-tuningexplainability
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Knowledge-graph rule mining produces logical if-then rules, but the raw predicates (for example, /travel/accommodation/accommodation_type) are opaque to non-experts. Rule2Text asks whether large language models can translate these rules into plain English, and establishes three things: the best prompting recipe is chain-of-thought reasoning combined with explicit variable entity types; an LLM judge built on Gemini 2.0 Flash agrees with human annotators closely enough (Spearman 0.69) to scale evaluation and select pseudo-ground-truth; and fine-tuning the open-source Zephyr-7B on judge-selected examples sharply improves explanation quality, especially on the biomedical ogbl-biokg graph, where ROUGE rises from 0.02 to 0.78. If right, the framework gives a practical pipeline for making KG rules usable by non-experts in domains such as healthcare.

What carries the argument

The load-bearing mechanism is a comparison-style chain-of-thought prompt: the model is asked, step by step, to check that every variable entity and every relation appearing in the rule also appears in the explanation, and to flag anything missing or hallucinated. This same verification prompt, combined with a scoring rubric and few-shot exemplars, is reused as the LLM-as-a-judge, which is what makes scalable evaluation and pseudo-ground-truth construction possible. Variable entity type information—the other half of the recipe—comes from the KG's type system when available, or from a type-inference module that shows the model three random instantiations of the rule.

What would settle it

Take the 100 rule-explanation pairs that humans scored in phases 2-3 and also obtain judge scores for them. Then adversarially edit a set of correct, human-approved explanations so that each contains exactly one error—one missing entity, one missing relation, or one hallucinated relation—and run the judge on the edited versions. If the judge's score fails to drop below the 'correct' threshold on a substantial share of edited explanations, the judge cannot support the pseudo-ground-truth selection, and the fine-tuning results built on it would need re-evaluation.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the obstacle to explaining KG rules is not the LLM's language ability but the model's ignorance of what the variables in a rule denote. Once the rule is given together with the entity types of its variables—either read from the KG's type system or inferred from a few rule instances—and the model is walked through a chain-of-thought prompt that verifies each variable and relation, explanations become substantially more correct and clear: on human evaluation, Gemini 2.0 Flash reaches 4.67/5 correctness and 4.70/5 clarity. The paper also establishes that an LLM judge can stand in for human annotation: using a similar verification-style pro

Load-bearing premise

The load-bearing premise is that the Gemini 2.0 Flash judge's correctness scores are reliable enough to select pseudo-ground-truth explanations and to measure fine-tuning gains; the paper reports moderate agreement with humans (Spearman 0.69, Krippendorff's alpha 0.59), so if the judge systematically prefers fluent but wrong explanations, the 400 training labels are contaminated and the automatic-metric improvements would not reflect real accuracy.

Editorial extensions

If this is right

  • The chain-of-thought plus variable-types prompt is a directly reusable recipe: any KG with readable relation labels and known or inferable entity types can get explanations without task-specific training.
  • The validated LLM judge can replace most human annotation effort; only low-scoring or judge/human-disagreement examples need expert review, making ground-truth dataset construction scale to thousands of rules.
  • Fine-tuning a 7B open model on 400 judge-selected pairs is enough to move explanation quality from near-zero content overlap to high overlap on a specialized biomedical KG, suggesting domain-specific fine-tuning is worthwhile for specialized graphs.
  • The type-inference module extends the recipe to KGs without explicit type information, at the cost of occasional over-specific type guesses when the sampled instances are narrow, such as inferring 'tennis player' instead of 'professional athlete'.
  • Because the framework treats rule mining as a pluggable component, the same explanation pipeline can be attached to other rule miners, not just AMIE.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not test whether its judge is robust to adversarial corruptions; a direct stress-test would edit correct explanations to drop one relation or insert a hallucinated entity and check whether the judge's scores drop as humans' would.
  • The tennis-player subtype error points to an untested improvement: choosing rule instances that maximize entity-type diversity before type inference could eliminate over-specific guesses.
  • The authors only check self-enhancement bias for the proprietary models; a natural extension is to use the fine-tuned Zephyr as its own judge after a similar bias check, turning the whole pipeline into an open, self-sustaining loop.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper introduces Rule2Text, an LLM-based framework for generating natural-language explanations of knowledge graph rules mined by AMIE. It compares prompting strategies (zero-shot, few-shot, variable-type augmentation, and Chain-of-Thought) across GPT-3.5 Turbo, GPT-4o mini, and Gemini 2.0 Flash; evaluates outputs by three human annotators on correctness and clarity; develops an LLM-as-a-judge protocol; constructs fine-tuning datasets for Freebase and ogbl-biokg; fine-tunes Zephyr-7B-beta; and adds a type-inference module for KGs lacking explicit types. The central claims are that CoT combined with variable-type information substantially improves explanation quality, that Gemini 2.0 Flash performs best, that the LLM judge agrees strongly with humans, and that fine-tuning significantly improves explanation quality, especially on biomedical rules.

Significance. If the claims hold, the paper would be the first comprehensive study of LLM explanation generation for KG rules, offering a practical recipe (CoT + variable types), a validated judge, public code/data, and a path to fine-tune small open models. The authors do ship public assets and use independent three-annotator human evaluation for the prompt-engineering phases, with the fine-tuning test set drawn from human-annotated examples. These are concrete strengths. However, the load-bearing evidence for the fine-tuning and judge claims is incomplete: the reported judge agreement is moderate, and the headline fine-tuning result rests on unvalidated lexical metrics with no human evaluation of the fine-tuned model. The stress-test concern about the fine-tuning evaluation therefore lands on reading the paper.

major comments (4)
  1. [Section 7, Table 5 (Zephyr Performance)] The abstract's headline claim — 'significant improvements in explanation quality after fine-tuning' — is supported only by BLEU, ROUGE, and METEOR on 50 test examples per dataset. The paper defines correctness and clarity as the evaluation criteria (Section 5.1) and uses human ratings for the prompt-engineering phases, but no human evaluation of the fine-tuned Zephyr outputs is reported, and no evidence is given that these lexical metrics track the human rubric. The ROUGE jump from 0.02 to 0.78 on ogbl-biokg is especially hard to interpret: a base model with 0.02 ROUGE shares essentially no n-grams with the reference, and a large gain could reflect imitation of the Gemini-derived reference style rather than improved semantic correctness. Please add human evaluation of the fine-tuned model, or at minimum validate the automatic metrics against human scores on the same test instances, and r
  2. [Section 5.3 and Section 4.3] The LLM-as-a-judge is load-bearing for the fine-tuning pipeline because it selects the 400 pseudo-ground-truth training examples (Section 6.2). Yet the reported agreement with humans is Spearman 0.69 and Krippendorff's alpha = 0.59 on 100 validation instances. That is moderate agreement, not 'strong' as claimed in the Abstract and Section 5.3. A judge with this level of consensus can systematically favor fluent but incorrect explanations; if so, the training labels are contaminated. Please report judge error patterns (e.g., a confusion matrix over score categories), examples where the judge disagrees with humans, and a human check on a sample of judge-selected training instances. Also report human-human inter-annotator agreement to calibrate what alpha=0.59 means in this task.
  3. [Section 7, Tables 2-4] The prompt-engineering conclusions (e.g., variable-type prompt improves correctness from 3.94 to 4.21 in Table 3; CoT improves Gemini to 4.67 in Table 4) are based on means over 100 rules with three annotators, with no standard deviations, significance tests, or inter-annotator reliability statistics. Without these, 'substantial improvements' and 'significant improvements' are not established; the differences are small relative to the granularity of a 1-5 scale and could be annotation noise. Please add per-condition distributions, paired tests across the 100 rules, and human-human agreement.
  4. [Section 6.2 and Table 5] The fine-tuning test set has only 50 instances per dataset and uses a single reference per rule. The references are Gemini-generated explanations that were human-annotated and possibly edited (Section 4.3). Fine-tuning on Gemini-generated pseudo-ground truth and then evaluating against Gemini-derived references creates a risk that high scores reflect lexical closeness to a particular style rather than correctness. Please report multi-reference evaluation, human evaluation, or both, and clarify exactly which of the 100 human-annotated examples fall into the validation split versus the test split.
minor comments (5)
  1. [Abstract and Section 5.3] 'Strong agreement' is inconsistent with the reported Krippendorff's alpha of 0.59; suggest 'moderate agreement'.
  2. [Section 6.1] The sentence 'the conversion process has resulted in a higher number of rules in these two datasets compared to those in FB+CVT-REV' is ambiguous; clarify which datasets are being compared.
  3. [Section 6.2] Calling judge-selected training instances 'ground-truth datasets' is misleading; use 'pseudo-ground truth' consistently.
  4. [Section 5.3] The 'triple evaluation' of each explanation is not described; specify whether repeated judge scores were averaged and how intra-model consistency was measured.
  5. [Section 7, Tables 2-4] Report inter-annotator agreement and confidence intervals; currently only mean scores are shown, which makes it impossible to judge the reliability of the reported differences.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the fine-tuning evaluation rests on human-annotated test references disjoint from the LLM-judge-filtered training data, and the central claims are supported by independent human evaluation within the paper.

full rationale

The paper's derivation chain is not circular in the sense defined by the analysis. The central claim that fine-tuning improves explanation quality is tested on a 50-example test set whose references are human-annotated, not LLM-judge-selected pseudo-ground-truth: Section 6.2 states 'The 100 examples that were directly annotated by human evaluators were used exclusively for the validation and test sets.' The 400 training examples are filtered by the LLM-as-a-judge, but the test set is explicitly disjoint and human-annotated, so the reported BLEU/ROUGE/METEOR gains are not forced by construction. The LLM judge itself is validated against human judgments in Section 5.3 and Section 7 (Spearman 0.69, Krippendorff's alpha 0.59), which is independent evidence rather than a self-referential definition. The choice of Gemini 2.0 Flash as both generator and judge is a potential validity concern, but the paper directly addresses self-enhancement bias through a preliminary study and through the judge's agreement with humans; the final fine-tuning evaluation does not use the judge as the arbiter. The self-citations to the authors' prior short paper [26] for CoT prompt details and the self-enhancement bias test are not load-bearing: the paper's own experiments (Tables 3 and 4) independently demonstrate the benefit of variable type information and CoT, and the judge-human agreement is measured in this paper. The skeptical concern that BLEU/ROUGE/METEOR are unvalidated lexical metrics is a validity/robustness issue, not a circularity issue, and the paper does not claim those metrics substitute for human correctness scoring of the fine-tuned model. No equation or fitted parameter is repackaged as a prediction; no uniqueness theorem is imported from the authors' prior work; no ansatz is smuggled in solely via citation. The pipeline is an empirically grounded teacher-student setup with human validation on the evaluation split, so it does not reduce to its own inputs by construction.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The paper is an empirical systems paper rather than a formal derivation, so the ledger records the hand-chosen thresholds and hyperparameters that set the experimental scope, plus the domain assumptions needed to interpret the results as evidence for the framework. No new conceptual entities (particles, forces, dimensions) are introduced.

free parameters (7)
  • AMIE minimum head coverage threshold = 0.1
    Hand-chosen threshold to filter mined rules; affects the rule set used throughout the study.
  • AMIE minimum standard confidence threshold = 0.1
    Hand-chosen threshold for rule confidence; influences which rules are explained and evaluated.
  • AMIE maximum number of atoms = 3
    Limits rule complexity; the study never evaluates rules with more than three atoms, bounding the findings.
  • Number of random rule instances for type inference = 3
    Few-shot examples shown to the LLM for type inference; affects specificity of inferred types (e.g., tennis player vs pro athlete).
  • Fine-tuning learning rate = 5e-5
    Hand-chosen hyperparameter for Zephyr fine-tuning; standard value but not justified by experiments.
  • Fine-tuning epochs = 2
    Hand-chosen training duration; no early stopping or epoch search is reported.
  • Ground-truth dataset size per domain = 500 (400 train / 50 val / 50 test)
    Hand-chosen size; the 50-example test sets limit reliability of the reported automatic metric gains.
assumptions (5)
  • domain assumption Mined rules via AMIE 3.5.1 accurately represent the logical content of the KG
    Section 3.2; the framework takes AMIE rules as given and does not evaluate whether the rules are semantically correct.
  • domain assumption Annotator correctness and clarity ratings are reliable
    Section 5.2; three expert annotators with training are used, but no inter-annotator agreement statistic is reported.
  • domain assumption The LLM-as-a-judge scores correlate with human judgments sufficiently for pseudo-labeling
    Section 5.3; Spearman 0.69 and Krippendorff alpha 0.59 are reported on 100 instances, but the judge is applied to a broader 400-example set.
  • domain assumption Automatic metrics (BLEU, ROUGE, METEOR, perplexity) proxy explanation quality
    Sections 5.1 and 7; these metrics measure surface similarity or fluency, not correctness or clarity as defined by the paper.
  • domain assumption Freebase variants and ogbl-biokg are representative of KG rule explanation settings
    Section 6.1; only two KG families are used, with specific labeling conventions (Freebase concatenated labels, ogbl-biokg typed IDs).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rule2Text: A Framework for Generating and Evaluating Natural Language Explanations of Knowledge Graph Rules." pith.science (2026). https://pith.science/paper/MY2XRS2G

@misc{pith2026250810971,
  author       = {Pith},
  title        = {Pith review of: Rule2Text: A Framework for Generating and Evaluating Natural Language Explanations of Knowledge Graph Rules},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MY2XRS2G}},
  note         = {Machine review of arXiv:2508.10971}
}
read the original abstract

Knowledge graphs (KGs) can be enhanced through rule mining; however, the resulting logical rules are often difficult for humans to interpret due to their inherent complexity and the idiosyncratic labeling conventions of individual KGs. This work presents Rule2Text, a comprehensive framework that leverages large language models (LLMs) to generate natural language explanations for mined logical rules, thereby improving KG accessibility and usability. We conduct extensive experiments using multiple datasets, including Freebase variants (FB-CVT-REV, FB+CVT-REV, and FB15k-237) as well as the ogbl-biokg dataset, with rules mined using AMIE 3.5.1. We systematically evaluate several LLMs across a comprehensive range of prompting strategies, including zero-shot, few-shot, variable type incorporation, and Chain-of-Thought reasoning. To systematically assess models' performance, we conduct a human evaluation of generated explanations on correctness and clarity. To address evaluation scalability, we develop and validate an LLM-as-a-judge framework that demonstrates strong agreement with human evaluators. Leveraging the best-performing model (Gemini 2.0 Flash), LLM judge, and human-in-the-loop feedback, we construct high-quality ground truth datasets, which we use to fine-tune the open-source Zephyr model. Our results demonstrate significant improvements in explanation quality after fine-tuning, with particularly strong gains in the domain-specific dataset. Additionally, we integrate a type inference module to support KGs lacking explicit type information. All code and data are publicly available at https://github.com/idirlab/KGRule2NL.

Figures

Figures reproduced from arXiv: 2508.10971 by the authors.

Figure 1
Figure 1. Rule2Text Framework then evaluate these generated explanations. If the explanations are not perfectly correct, annotators modify them rather than writing complete explanations from scratch. This approach allows us to generate ground-truth data using a relatively strong model. The resulting dataset is significantly smaller than the complete set of rules extracted from the KG. We can then use this data to fine-tune th… view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. On the Importance and Evaluation of Narrativity in Natural Language AI Explanations

    cs.CL 2026-04 unverdicted novelty 6.0 of 10

    XAI explanations should be narratives with continuous structure, cause-effect, fluency and diversity, and new metrics are needed to evaluate this better than standard NLP scores.

  2. A Two-Stage LLM Framework for Accessible and Verified XAI Explanations

    cs.AI 2026-04 unverdicted novelty 6.0 of 10

    A two-stage LLM explainer-verifier framework with iterative refeed improves faithfulness and accessibility of XAI explanations, as shown in experiments across five techniques and three LLM families, with EPR analysis ...

Reference graph

Works this paper leans on

36 extracted references · 28 canonical work pages · cited by 2 Pith papers

  1. [1]

    Kristoffer Æsøy and Ana Ozaki. 2023. Rule Learning as Machine Translation using the Atomic Knowledge Bank. arXiv:2311.02765 (2023)

    work page Pith review arXiv 2023

  2. [2]

    Satanjeev Banerjee and Alon Lavie. 2005. METEOR: An automatic metric for MT evaluation with improved correlation with human judgments. In ACL. 65–72

    work page 2005

  3. [3]

    Patrick Betz, Luis Galárraga, Simon Ott, Christian Meilicke, Fabian Suchanek, and Heiner Stuckenschmidt. 2023. PyClause-Simple and efficient rule handling for knowledge graphs. In IJCAI. 8610–8613

    work page 2023

  4. [4]

    Kurt Bollacker, Colin Evans, Praveen Paritosh, Tim Sturge, and Jamie Taylor

  5. [5]

    Antoine Bordes, Nicolas Usunier, Alberto Garcia-Durán, Jason Weston, and Ok- sana Yakhnenko. 2013. Translating embeddings for modeling multi-relational data. In NeurIPS. 2787–2795

    work page 2013

  6. [6]

    Zhiyu Chen, Wenhu Chen, Hanwen Zha, Xiyou Zhou, Yunkai Zhang, Sairam Sundaresan, and William Yang Wang. 2020. Logic2Text: High-fidelity natural language generation from logical forms. arXiv:2004.14579 (2020)

    work page Pith review arXiv 2020

  7. [7]

    Cheng-Han Chiang and Hung-yi Lee. 2023. Can large language models be an alternative to human evaluations? arXiv preprint arXiv:2305.01937 (2023)

    arXiv 2023

  8. [8]

    Peter Clark, Oyvind Tafjord, and Kyle Richardson. 2020. Transformers as soft reasoners over language. arXiv preprint arXiv:2002.05867 (2020)

    arXiv 2020

Show all 36 references
  1. [9]

    Suchanek

    Luis Galárraga, Christina Teflioudi, Katja Hose, and Fabian M. Suchanek. 2015. Fast Rule Mining in Ontological Knowledge Bases with AMIE++. VLDB 24, 6 (Dec. 2015), 707–730

    2015

  2. [10]

    Luis Antonio Galárraga, Christina Teflioudi, Katja Hose, and Fabian Suchanek

  3. [11]

    Google DeepMind. 2024. Gemini 2.0 Flash. https://blog.google/technology/ google-deepmind/google-gemini-ai-update-december-2024/

    2024

  4. [12]

    Weihua Hu, Matthias Fey, Marinka Zitnik, Yuxiao Dong, Hongyu Ren, Bowen Liu, Michele Catasta, and Jure Leskovec. 2020. Open graph benchmark: Datasets for machine learning on graphs. Advances in neural information processing systems 33 (2020), 22118–22133

    2020

  5. [13]

    Shaoxiong Ji, Shirui Pan, Erik Cambria, Pekka Marttinen, and S Yu Philip. 2021. A survey on KGs: Representation, acquisition, and applications. TNNLS 33, 2 (2021), 494–514

    2021

  6. [14]

    Daniel Jurafsky and James H. Martin. 2009. Speech and Language Processing . Prentice Hall

    2009

  7. [15]

    Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries . Technical Report SRI-TR-04-019. SRI International

    2004

  8. [16]

    Christian Meilicke, Melisachew Wudage Chekol, Patrick Betz, Manuel Fink, and Heiner Stuckeschmidt. 2024. Anytime bottom-up rule learning for large-scale KGC. VLDB 33, 1 (2024), 131–161

    2024

  9. [17]

    Ndapandula Nakashole, Mauro Sozio, Fabian M Suchanek, and Martin Theobald

  10. [18]

    OpenAI. 2023. OpenAI GPT-3.5 Turbo. https://openai.com/blog/gpt-3-5-turbo

    2023

  11. [19]

    OpenAI. 2024. OpenAI GPT -4o Mini. https://openai.com/index/gpt-4o-mini- advancing-cost-efficient-intelligence/

    2024

  12. [20]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. BLEU: a method for automatic evaluation of machine translation. In ACL. 311–318

    2002

  13. [21]

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language Models are Unsupervised Multitask Learners. OpenAI Blog 1, 8 (2019)

    2019

  14. [22]

    Sergio Servantez, Joe Barrow, Kristian Hammond, and Rajiv Jain. 2024. Chain of Logic: Rule-Based Reasoning with Large Language Models. arXiv:2402.10400 (2024)

    2024 arXiv

  15. [23]

    Xiao Shi, Zhengyuan Zhu, Zeyu Zhang, and Chengkai Li. 2023. Hallucination mit- igation in natural language generation from large-scale open-domain knowledge graphs. In EMNLP. 12506–12521

    2023

  16. [24]

    Nasim Shirvani-Mahdavi, Farahnaz Akrami, and Chengkai Li. 2025. On Large- scale Evaluation of Embedding Models for Knowledge Graph Completion. arXiv:2504.08970 (2025)

    2025 arXiv

  17. [25]

    Nasim Shirvani-Mahdavi, Farahnaz Akrami, Mohammed Samiul Saeef, Xiao Shi, and Chengkai Li. 2023. Comprehensive analysis of Freebase and dataset creation for robust evaluation of knowledge graph link prediction models. In ISWC. Springer, 113–133

    2023

  18. [26]

    Nasim Shirvani-Mahdavi, Devin Wingfield, Amin Ghasemi, and Chengkai Li

  19. [27]

    Rush, and Thomas Wolf

    Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Ra- sul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Four- rier, Nathan Habib, Nathan Sarrazin, Omar Sanseviero, Alexander M. Rush, and Thomas Wolf. 2023. Zephyr: Direct Distillation of LM...

    2023 arXiv

  20. [28]

    Denny Vrandečić and Markus Krötzsch. 2014. Wikidata: a free collaborative knowledge base. CACM 57, 10 (2014), 78–85

    2014

  21. [29]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. NeurIPS 35 (2022), 24824–24837

    2022

  22. [30]

    Xin Wu, Yi Cai, Zetao Lian, Ho-fung Leung, and Tao Wang. 2023. Generating natural language from logic expressions with structural representation. TASLP 31 (2023), 1499–1510

    2023

  23. [31]

    Zonglin Yang, Xinya Du, Rui Mao, Jinjie Ni, and Erik Cambria. 2023. Logi- cal reasoning over natural language as knowledge representation: A survey. arXiv:2303.12023 (2023)

    2023 arXiv

  24. [32]

    Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, et al. 2023. Judging llm-as-a-judge with mt-bench and chatbot arena.NeurIPS 36 (2023), 46595–46623

    2023

  25. [2008]

    In SIGMOD

    Freebase: a collaboratively created graph database for structuring human knowledge. In SIGMOD. 1247–1250

  26. [2012]

    Query-time reasoning in uncertain RDF KBs with soft and hard rules.VLDS 884, 6 (2012), 15–20

    2012

  27. [2013]

    AMIE: association rule mining under incomplete evidence in ontological knowledge bases. In WWW. 413–422

  28. [2025]

    arXiv preprint arXiv:2507.23740 (2025)

    Rule2Text: Natural Language Explanation of Logical Rules in Knowledge Graphs. arXiv preprint arXiv:2507.23740 (2025)

    2025 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.