Pith. sign in

REVIEW 4 major objections 5 minor 129 references

Verified Language Processing with Hybrid Explainability: A Technical Report

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a fully rule-based pipeline translating sentences into first-order logic can distinguish logical implication, inconsistency, and indifference more reliably than pretrained language models, with perfect accuracy on…

desk verdict A carefully engineered logic-based NLP pipeline whose headline empirical claim is not supported by the in-sample, author-constructed evaluation. read the letter →

arxiv 2507.05017 v1 pith:SQVHOWME submitted 2025-07-07 cs.CL cs.SC

Oliver Robert Fox , Giacomo Bergami , Graham Morgan This is my paper
classification cs.CLcs.SC
keywords first-orderlogicMontagueGrammarparaconsistentreasoningtextentailmentclassificationexplainableAIfull-textsimilaritylogicalconfidencescorenaturallanguage
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

The paper tries to establish that a transparent, logic-based pipeline can classify the logical relationship between two sentences—implication, inconsistency, or indifference—more accurately than pretrained language models, and that this failure of transformers is not incidental. It builds LaSSI, which rewrites each sentence into first-order logic via Montague Grammar, and compares formulas through a paraconsistent truth-table confidence score. On three small controlled datasets testing propositional connectives, active/passive structure, and spatiotemporal reasoning, the logical method scores 1.00 accuracy while six transformer baselines score between 0.19 and 0.50. The authors conclude that natural-language understanding cannot be easily generalised from large corpora alone, and that their pipeline offers a verifiable and explainable alternative for full-text similarity and reasoning.

What carries the argument

The central object is the confidence score over possible worlds: each sentence's first-order logic formula is decomposed into atoms, a truth table is built over all assignments to those atoms, the two tables are equi-joined, and the score is computed as the average truth value of the consequent over the worlds where the premise holds. This asymmetric score is what lets the system separate entailment (score 1), inconsistency (score 0), and indifference (intermediate values), while the Parmenides upper ontology supplies the expansion rules that rewrite and relate atoms before the tables are joined.

What would settle it

Take a sentence pair with a known entailment relation, have the pipeline output its logical formulas, and compare those formulas against a manually annotated gold-standard first-order-logic translation; any mismatch—such as a missed negation, a wrong quantifier binding, or an incorrect entity grouping—that changes the confidence score away from 1 or 0 would show the score measures the rewritten formulas rather than the original sentences.

Watch

Extended reading notes

Core claim

The paper proposes LaSSI, a modular pipeline that rewrites factoid sentences into first-order logic formulas through Universal Dependency parsing, graph rewriting, and a curated upper ontology (Parmenides) that assigns logical functions to entities and adverbial phrases. Sentence pairs are compared by enumerating the possible worlds over the atoms in their formulas, equi-joining the two truth tables, and computing a confidence score |W(A)∩W(B)|/|W(A)|; worlds that make an atom and its negation both true are discarded, which the authors call paraconsistent reasoning. The authors report perfect accuracy (1.00) on three author-built datasets distinguishing implication, inconsistency, and indifference, against 0.19–0.50 accuracy for six transformer baselines, and they argue that this shows training on large corpora does not by itself yield sound natural-language reasoning.

Load-bearing premise

The load-bearing premise is that the hand-built rewriting chain from natural language to first-order logic produces a semantically correct formula for every input sentence, so the truth-table confidence score measures genuine logical entailment rather than artifacts of the rewriting.

Editorial extensions

If this is right

  • If the reported accuracy holds on broader data, a no-training pipeline can match or beat large language models on controlled logical sentence classification.
  • Symmetric similarity metrics such as cosine similarity cannot express directed entailment, so any system relying on them for equivalence or similarity will misclassify implication versus indifference.
  • The three-way distinction between implication, inconsistency, and indifference can be decided by truth tables rather than requiring labeled training data.
  • The pipeline yields human-readable logical forms and traceable rewrite steps, so classification decisions can be audited without a separate explainer model.
  • The reported failure of pre-trained models on propositional calculus suggests they cannot be relied on for sound first-order-logic reasoning on arbitrary sentences.

Reading between the lines

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

  • If the soundness assumption holds, the same confidence score could serve as a graded inconsistency measure for fact-checking: pairs with scores near 0 would flag contradictions, and the asymmetry would identify which sentence is the stronger claim.
  • The three author-built datasets are small and systematic; an immediate testable extension is to measure accuracy on naturally occurring web sentences with paraphrase and contradiction annotations, where parser errors and entity ambiguity are more frequent.
  • The pipeline's modular rule sets mean the ontology is the main engineering bottleneck: adding new logical functions such as causal or modal expressions would require extending the Parmenides rules rather than retraining, which is both an advantage and a maintenance cost.
  • A testable prediction follows from the paper's negative results: if transformers are given the same first-order-logic formulas rather than raw text, their classification accuracy should rise toward the logical baseline, isolating representation failure from reasoning failure.
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 presents LaSSI, a modular NLP pipeline that converts factoid sentences into First-Order Logic via Stanford CoreNLP dependency parsing, GSM graph rewriting, the Parmenides upper ontology, and Montague-style logical rewriting, and then classifies sentence pairs as implication, inconsistency, or indifference using a truth-table confidence score with paraconsistent semantics. The authors prove small lemmas about symmetric similarity metrics versus their confidence metric (Lemmas 10–13), and report empirical comparisons against six transformer-based baselines on three small author-constructed datasets (8, 6, and 13 sentences), where LaSSI achieves 1.00 accuracy while baselines score 0.19–0.50. The paper concludes that pre-trained language models cannot easily generalise natural language understanding from large corpora.

Significance. If the central claims were established, the paper would contribute a transparent, rule-based alternative to learned sentence representations, with a genuinely explainable pipeline and a paraconsistent confidence score. The theoretical lemmas about symmetry and confidence are correct, and the public code and data are valuable for reproducibility. The negative results for transformer baselines on the controlled datasets are suggestive, and the detailed ablation and explainability study are useful engineering contributions. However, the empirical evidence does not yet support the broad conclusions about state-of-the-art performance or the impossibility of learning NLU from large corpora, because the perfect scores rest on in-sample threshold fitting, hand-built rules and ontologies, and tiny author-constructed datasets with no external validation.

major comments (4)
  1. [Definition 8 and Lemma 12] Weak Soundness is stated only as a desideratum—'refers to the preservation of the original semantic meaning'—and no proof or external gold standard is provided that the transformation chain (UD parsing, GSM rewriting, Parmenides rules, final FOL rewriting; Section 3.2 and Algorithms S1–S6) yields semantically correct formulas for arbitrary inputs. Lemma 12 explicitly conditions the confidence metric on the correctness of this transformation, so the 1.00 accuracy in Tables 3–5 is a consistency check on hand-built rules rather than evidence of verified entailment. Please add either a soundness proof for the transformation or an evaluation against a corpus with independently authored logical forms.
  2. [Section 4.2] The classification thresholds θ and ϑ are derived on the same evaluation data: θ is obtained from clustering the expected clusters, and ϑ is taken as the maximal similarity score between manually annotated contradictory pairs. Since these thresholds are fitted to the test labels, the comparison with transformer baselines, which receive no such tuning, is not apples-to-apples, and the logical method's perfect scores are partly a consequence of in-sample calibration. Please report results with thresholds selected on a held-out split or fixed a priori from theory.
  3. [Table 6, Section 5.2] The ablation shows that disabling the a priori entity-recognition phase drops LaSSI's logical accuracy on the spatiotemporal dataset from 1.00 to 0.33, which is close to the transformer baselines (0.28–0.29 in Table 5). This admitted dependence on pre-enumerated entities from GeoNames, ConceptNet, and Parmenides means the results do not support the abstract's claim that 'natural language understanding cannot be easily generalised by training over extensive document corpora.' The paper should substantially weaken this conclusion or demonstrate performance on data whose entities were not pre-enumerated.
  4. [Tables 3–5, Section 4.2] The evaluation uses three author-constructed datasets of 8, 6, and 13 sentences with no NLI baselines beyond DeBERTaV2+AMR-LDA, no multiple runs or error bars, and no significance tests. The transformer accuracies (0.19–0.50) are near or below chance, and the logical method's perfect accuracy is not compared with other symbolic semantic parsers (e.g., CCG or Boxer) or standard entailment benchmarks. Please add established datasets or at least hold-out splits with confidence intervals before making comparative claims.
minor comments (5)
  1. [Section 5.3.4] The text references 'Equation 3.2.4' for the FOL syntax, but the displayed grammar is not numbered; please number the equation and update internal references.
  2. [Table 7 and Table 8] Table 7 contains the typo 'Weigthed' for 'Weighted', and Table 8 uses garbled symbols ('' and 'G #') in the requirement matrix; please fix the encoding and ensure the symbols render correctly.
  3. [Abstract] The abstract claims 'to the best of our knowledge, this is the first approach to differentiate between implication, inconsistency, and indifference'; please provide a more systematic comparison with prior three-way or contradiction-aware entailment work to substantiate this novelty claim.
  4. [Section 3] The term 'full text' is used for individual sentences throughout, which is potentially confusing because 'full text' usually denotes a document; please define this usage explicitly near the start of Section 3.
  5. [Figures 11b–13b] The expected clusters are described as manually defined, but no annotation protocol or inter-annotator agreement is reported; please include a brief description of how the clusters and contradictory pairs were constructed and validated.

Circularity Check

1 steps flagged · score 6.0 of 10

Classification thresholds are fit to the same annotated test data whose labels define the reported accuracy, making the perfect scores an in-sample result.

  1. fitted input called prediction [Section 4.2 (Classification), pp. 23–24; thresholds θ and ϑ used for all rows of Tables 3–5]
    "we then derive an upper threshold value ϑ separating the conflicting sentences from the rest by taking the maximal similarity score between the pair of sentences expected to be contradictory from the manual annotation. If ϑ>θ, we consider only one threshold value separating entailing and contradictory data (ϑ=θ). Thus, we consider all similarity values above θ prediction values for a logical entailment and all values lower than ϑ as predictive of a conflict between the two sentences."

    The classification thresholds are fit to the same test sentences whose labels define the evaluation: ϑ is the maximal similarity among manually annotated contradictory pairs, and θ is derived from the expected clusters (Appendix D). The reported accuracy and F1 scores in Tables 3–5 are then computed on these same sentences using those fitted thresholds. The 'predictions' of implication, contradiction, and indifference are therefore not independent; the thresholds encode the test labels, and the scores measure in-sample separation rather than generalizable prediction.

full rationale

The central empirical claim that LaSSI outperforms state-of-the-art models rests on perfect accuracy (1.00) on three author-built datasets. The concrete circular reduction is in Section 4.2: the decision thresholds θ (from expected clusters) and ϑ (from manually annotated contradictory pairs) are fit directly on the test data, and the same test data are then scored in Tables 3–5. No held-out validation or train/test split is described for these thresholds, so the reported classification is partially determined by the ground-truth labels. This alone justifies a partial-circularity finding. Two further observations, while not independently circular, weaken the external validity of the perfect scores: (i) Table 6 shows logical accuracy on the spatiotemporal dataset drops from 1.00 to 0.33 when the a priori entity-recognition phase is disabled, showing the score depends on the hand-curated Parmenides/GeoNames/ConceptNet entity lists; and (ii) Definition 8 states weak soundness only as a desideratum, and Lemma 12 explicitly conditions the confidence metric on the correctness of the φ transformation, which is never verified against an external gold standard. These do not by themselves reduce the derivation to its inputs, but they show the perfect scores are self-consistent rather than independently validated. Overall, the fitted-threshold step makes the headline result partially circular: score 6.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

The central empirical claim rests on a long chain of hand-written rules and on datasets and thresholds created by the authors, so the ledger shows many ad hoc-to-paper elements rather than externally grounded components.

free parameters (3)
  • Entailment threshold theta = Not reported; varies by dataset and clustering algorithm
    Derived from clustering output on the same sentences used for evaluation (Section 4.2), so it is fitted to the test data.
  • Conflict threshold vartheta = Not reported; maximal similarity among manually annotated contradictory pairs
    Chosen from the same annotated data that is later scored (Section 4.2).
  • GeoNames confidence weight = 0.8
    Fixed multiplier from the authors' previous research (Section 3.1.1), used in multi-word entity matching.
assumptions (5)
  • domain assumption Natural language sentences can be represented in first-order logic via Montague Grammar.
    Invoked in Section 3.2.4 and Supplement II to justify turning text into FOL.
  • ad hoc to paper The Parmenides upper ontology and its TBox rules correctly capture entity types, part-of, is-a, and logical rewriting for the tested domains.
    Parmenides is authored by the same team and its coverage is not externally validated; used throughout Section 3.1 and Algorithm S6.
  • ad hoc to paper Each extracted atom is assumed independent when building truth tables, with no background knowledge beyond the KB expansion.
    Stated in Appendix A.2: 'assuming each proposition to be completely independent of the others without any further background knowledge.'
  • domain assumption Removing contradictory worlds yields a valid paraconsistent semantics.
    Borrowed from paraconsistent logic literature (Carnielli and Coniglio), used in Section 3.3 and Appendix A.1.
  • ad hoc to paper The three author-built datasets are representative enough to support general claims about pre-trained models' reasoning abilities.
    Datasets are 8, 6, and 13 sentences, constructed by the authors (Section 4.2), yet used to conclude that 'natural language understanding cannot be easily generalised.'
invented entities (2)
  • Parmenides Upper Ontology
    purpose: Provides entity types, logical rewriting rules, and TBox expansion used to derive FOL formulas and similarity judgements.
    No external benchmark validates it; its success is demonstrated only on the authors' own datasets (Section 3.1, Figure 9).
  • meuDB (Multi-Word Entity Unit DataBase)
    purpose: Stores multi-word entity matches aggregating GeoNames, SUTime, Stanza, ConceptNet, and Parmenides for entity resolution.
    It is a custom database built for this pipeline; independent evidence would require evaluation on external entity-recognition benchmarks (Section 3.1.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Verified Language Processing with Hybrid Explainability: A Technical Report." pith.science (2026). https://pith.science/paper/SQVHOWME

@misc{pith2026250705017,
  author       = {Pith},
  title        = {Pith review of: Verified Language Processing with Hybrid Explainability: A Technical Report},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SQVHOWME}},
  note         = {Machine review of arXiv:2507.05017}
}
read the original abstract

The volume and diversity of digital information have led to a growing reliance on Machine Learning techniques, such as Natural Language Processing, for interpreting and accessing appropriate data. While vector and graph embeddings represent data for similarity tasks, current state-of-the-art pipelines lack guaranteed explainability, failing to determine similarity for given full texts accurately. These considerations can also be applied to classifiers exploiting generative language models with logical prompts, which fail to correctly distinguish between logical implication, indifference, and inconsistency, despite being explicitly trained to recognise the first two classes. We present a novel pipeline designed for hybrid explainability to address this. Our methodology combines graphs and logic to produce First-Order Logic representations, creating machine- and human-readable representations through Montague Grammar. Preliminary results indicate the effectiveness of this approach in accurately capturing full text similarity. To the best of our knowledge, this is the first approach to differentiate between implication, inconsistency, and indifference for text classification tasks. To address the limitations of existing approaches, we use three self-contained datasets annotated for the former classification task to determine the suitability of these approaches in capturing sentence structure equivalence, logical connectives, and spatiotemporal reasoning. We also use these data to compare the proposed method with language models pre-trained for detecting sentence entailment. The results show that the proposed method outperforms state-of-the-art models, indicating that natural language understanding cannot be easily generalised by training over extensive document corpora. This work offers a step toward more transparent and reliable Information Retrieval from extensive textual data.

Figures

Figures reproduced from arXiv: 2507.05017 by the authors.

Figure 1
Figure 1. Visualisation of the differences between Part of Speech (POS) tagging, AMR graphs, and UDs, providing more explicit relationships between words. AMRs were generated through AMREager (https://bollin.inf.ed.ac.uk, Accessed on 24 April 2025), while UDs were generated using StanfordNLP [42]. Graphs are highlighted to word correspondences, represented as graph nodes [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Detailed view of [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. offers a birds-eye view of the entire pipeline as narrated in the present paper [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Continuing the example from [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Continuing the example from [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: A broad view of the ad hoc phase of the pipeline: with the intermediate graph generated, we iterate over each topologically sorted root node ID and create the final sentence representation. The transformation function, τ, takes our full text enriched with semantic info…
Figure 7
Figure 7. Figure 7: Transforming the GGG rewriting of a UD graph into a graph Intermediate Representation for “Newcastle and Brighton have traffic”. 3.2.1. Graph Rewriting with the Generalised Semistructured Model (GSM) This step employs the proposed GSM [49] to refine the initial graph a…
Figure 8
Figure 8. Figure 8: GGG rewritten graph for the sentence “become able to answer more questions” from Example 2: node IDs are overlayed in the middle of each node. Each node represents a word in the sentence, and each edge represents a UD. Unlike the previous simplistic example, most real-…
Figure 9
Figure 9. Figure 9: Fragments of the Parmenides Upper Ontology encoding rules for capturing logical functions. 1 The previous phase provided a preliminary rewriting, where a new relationship is derived from each verb occurring within the pipeline and connecting the agents performing and r…
Figure 10
Figure 10. Figure 10: summarises the contribution of Algorithm S6 (Supplement V): for each sentence node (either SetOfSingletons or Singleton), we obtain its properties and test all the rules as stored in Parmenides in declaration order. Upon the satisfaction of the premises, the rule dete…
Figure 11
Figure 11. Figure 11: Sentences and expected clusters for RQ №2(a), where no sentences are clustered together, as no sentence is perfectly similar to another in both directions. In [PITH_FULL_IMAGE:figures/full_fig_p024_11.png]
Figure 12
Figure 12. Figure 12: Sentences and expected clusters for RQ №2(b). tokenisation. Differently from the previous set of experiments, subdividing the text encoding into multiple different sentences proved to be ineffective for the ColBERTv2+RAGatouille approach, as subdividing short sentence…
Figure 13
Figure 13. Figure 13: Sentences and expected clusters for RQ №2(c). The clustering results now seem in line with the classification scores from [PITH_FULL_IMAGE:figures/full_fig_p027_13.png]
Figure 14
Figure 14. Figure 14: LIME and SHAP explanations comparing “There is traffic but not in the Newcastle city centre” against itself [PITH_FULL_IMAGE:figures/full_fig_p033_14.png]
Figure 15
Figure 15. Figure 15: LaSSI explanations for indifference and implication between sentence 2 (“There is traffic but not in the Newcastle city centre”) and sentence 11 (“Newcastle has traffic but not in the city centre”). Columns in yellow provide the truth values for the sentences given th…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

129 extracted references · 33 canonical work pages

  1. [1]

    GAIA - A Multi-media Multi-lingual Knowledge Extraction and Hypothesis Generation System

    Zhang, T.; et al. GAIA - A Multi-media Multi-lingual Knowledge Extraction and Hypothesis Generation System. In Proceedings of the 2018 Text Analysis Conference, TAC 2018, Gaithersburg, Maryland, USA. NIST, 2018. http://www.kianasun.com/publication/tac2018/tac2018.pdf, Accessed on 21 May 2025

    2018

  2. [2]

    Man-Assisted Machine Construction Of A Semantic Dictionary For Natural Language Processing

    Yoshida, S.; Tsurumaru, H.; Hitaka, T. Man-Assisted Machine Construction Of A Semantic Dictionary For Natural Language Processing. In Proceedings of the Proceedings of the 9th International Conference on Computational Linguistics, COLING ’82, Prague, Czechoslovakia, July 5-10, 1982. ACADEMIA, Publishing House of the Czechoslovak Academy of Sciences, 1982,...

    1982

  3. [3]

    Extraction of Semantic Word Relations in Turkish from Dictionary Definitions

    Ay¸ se, ¸ S.; Zeynep, O.;˙Ilknur, P . Extraction of Semantic Word Relations in Turkish from Dictionary Definitions. In Proceedings of the Proceedings of the ACL 2011 Workshop on Relational Models of Semantics; Kim, S.N.; Kozareva, Z.; Nakov, P .; Ó Séaghdha, D.; Padó, S.; Szpakowicz, S., Eds., Portland, Oregon, USA, 2011; pp. 11–18

    2011

  4. [4]

    ConceptNet 5.5: An Open Multilingual Graph of General Knowledge

    Speer, R.; Chin, J.; Havasi, C. ConceptNet 5.5: An Open Multilingual Graph of General Knowledge. Association for the Advancement of Artificial Intelligence (AAAI), 2017, Vol. 31. https://doi.org/10.1609/ aaai.v31i1.11164

    2017

  5. [6]

    DBpedia: A Multilingual Cross-domain Knowledge Base

    Mendes, P .; Jakob, M.; Bizer, C. DBpedia: A Multilingual Cross-domain Knowledge Base. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC‘12); Calzolari, N.; Choukri, K.; Declerck, T.; Do˘ gan, M.U.; Maegaard, B.; Mariani, J.; Moreno, A.; Odijk, J.; Piperidis, S., Eds., Istanbul, Turkey, 2012; pp. 1813–1817. h...

    2012

  6. [7]

    A framework supporting imprecise queries and data

    Bergami, G. A framework supporting imprecise queries and data, 2019, [arXiv:cs.DB/1912.12531]. https: //doi.org/10.48550/arXiv.1912.12531

    work page Pith review arXiv doi:10.48550/arxiv.1912.12531 2019

  7. [8]

    CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge

    Talmor, A.; Herzig, J.; Lourie, N.; Berant, J. CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers); Burstein, J.; Doran, C.; Solorio, T., Eds., Minnea...

    doi:10.18653/v1/n19-1421 2019

  8. [9]

    SchenQL: in-depth analysis of a query language for bibliographic metadata.International Journal on Digital Libraries2022,23, 113–132

    Kreutz, C.K.; Wolz, M.; Knack, J.; Weyers, B.; Schenkel, R. SchenQL: in-depth analysis of a query language for bibliographic metadata.International Journal on Digital Libraries2022,23, 113–132. https://doi.org/10.100 7/s00799-021-00317-8

Show all 129 references
  1. [10]

    NaLIR: an interactive natural language interface for querying relational databases

    Li, F.; Jagadish, H.V . NaLIR: an interactive natural language interface for querying relational databases. In Proceedings of the 2014 ACM SIGMOD International Conference on Management of Data, New York, NY, USA; SIGMOD ’14, p. 709–712. https://doi.org/10.1145/2588555.2594519

    2014

  2. [11]

    An Experimental Pipeline for Automated Reasoning in Natural Language (Short Paper)

    Tammet, T.; Järv, P .; Verrev, M.; Draheim, D. An Experimental Pipeline for Automated Reasoning in Natural Language (Short Paper). In Proceedings of the Automated Deduction - CADE 29. Springer Nature Switzerland, 2023, pp. 509–521. https://doi.org/10.1007/978-3-031-38499-8_29

    2023 doi

  3. [13]

    Bender, E.M.; Gebru, T.; McMillan-Major, A.; Shmitchell, S. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, New York, NY, USA, 2021; FAccT ’21, p. 610–623. https://do...

    2021

  4. [14]

    GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models, 2024, [arXiv:cs.LG/2410.05229]

    Mirzadeh, I.; Alizadeh, K.; Shahrokhi, H.; Tuzel, O.; Bengio, S.; Farajtabar, M. GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models, 2024, [arXiv:cs.LG/2410.05229]

    2024 arXiv

  5. [15]

    Intentional Biases in LLM Responses, 2023, [arXiv:cs.CL/2311.07611]

    Badyal, N.; Jacoby, D.; Coady, Y. Intentional Biases in LLM Responses, 2023, [arXiv:cs.CL/2311.07611]

    2023 arXiv

  6. [16]

    https: //doi.org/10.1017/CBO9780511576430

    Harrison, J.Handbook of Practical Logic and Automated Reasoning; Cambridge University Press, 2009. https: //doi.org/10.1017/CBO9780511576430. 51 of 88

    2009 doi

  7. [17]

    InLogic and philosophy for linguists; De Gruyter Mouton: Berlin, Boston, 1975; pp

    Montague, R., English as a Formal Language. InLogic and philosophy for linguists; De Gruyter Mouton: Berlin, Boston, 1975; pp. 94–121. https://doi.org/10.1515/9783111546216-007

    1975 doi

  8. [18]

    Inconsistency measures and paraconsistent consequence

    Brown, B. Inconsistency measures and paraconsistent consequence. InMeasuring Inconsistency in Information; Grant, J.; Martinez, M.V ., Eds.; College Press, 2018; chapter 8, pp. 219–234

    2018

  9. [19]

    https://doi.org/10.1007/978-3-319-33205-5

    Carnielli, W.; Esteban Coniglio, M.Paraconsistent Logic: Consistency, Contradiction and Negation; Springer: Switzerland, 2016. https://doi.org/10.1007/978-3-319-33205-5

    2016 doi

  10. [20]

    LaSSI: Logical, Structural, and Semantic text Interpretation

    Fox, O.R.; Bergami, G.; Morgan, G. LaSSI: Logical, Structural, and Semantic text Interpretation. In Proceedings of the Database Engineered Applications: 28th International Symposium, IDEAS 2024, Bayonne, France. Springer-Verlag, 2025, pp. 106–121. https://doi.org/10.1007/978-3...

    2024 doi

  11. [21]

    Toward verified artificial intelligence.Commun

    Seshia, S.A.; Sadigh, D.; Sastry, S.S. Toward verified artificial intelligence.Commun. ACM2022,65, 46–55. https://doi.org/10.1145/3503914

    doi

  12. [22]

    InExplainable Artificial Intelligence for Trustworthy Decisions in Smart Applications; Springer; chapter 2

    Bergami, G.; Fox, O.R.; Morgan, G., Extracting Specifications through Verified and Explainable AI: Inter- pretability, Interoperabiliy, and Trade-offs (In Press). InExplainable Artificial Intelligence for Trustworthy Decisions in Smart Applications; Springer; chapter 2

  13. [23]

    Single-Domain Generalized Predictor for Neural Architecture Search System.IEEE Transactions on Computers2024,73, 1400–1413

    Ma, L.; Kang, H.; Yu, G.; Li, Q.; He, Q. Single-Domain Generalized Predictor for Neural Architecture Search System.IEEE Transactions on Computers2024,73, 1400–1413. https://doi.org/10.1109/TC.2024.3365949

    2024

  14. [24]

    On the Explainability of Natural Language Processing Deep Models.ACM Comput

    Zini, J.E.; Awad, M. On the Explainability of Natural Language Processing Deep Models.ACM Comput. Surv.2023,55, 103:1–103:31. https://doi.org/10.1145/3529755

    2023 doi

  15. [25]

    Combat COVID-19 infodemic using explainable natural language processing models.Information Processing & Management2021,58, 102569

    Ayoub, J.; Yang, X.J.; Zhou, F. Combat COVID-19 infodemic using explainable natural language processing models.Information Processing & Management2021,58, 102569. https://doi.org/10.1016/j.ipm.2021.102569

    2021

  16. [26]

    MPNet: Masked and Permuted Pre-training for Language Under- standing, 2020

    Song, K.; Tan, X.; Qin, T.; Lu, J.; Liu, T. MPNet: Masked and Permuted Pre-training for Language Under- standing, 2020. https://doi.org/10.48550/arXiv.2004.09297

    2020 doi arXiv

  17. [27]

    RoBERTa: A Robustly Optimized BERT Pretraining Approach, 2019

    Liu, Y.; Ott, M.; Goyal, N.; Du, J.; Joshi, M.; Chen, D.; Levy, O.; Lewis, M.; Zettlemoyer, L.; Stoyanov, V . RoBERTa: A Robustly Optimized BERT Pretraining Approach, 2019. https://doi.org/10.48550/arXiv.1907.1 1692

    2019 doi

  18. [28]

    MiniLMv2: Multi-Head Self-Attention Relation Distillation for Compressing Pretrained Transformers

    Wang, W.; Bao, H.; Huang, S.; Dong, L.; Wei, F. MiniLMv2: Multi-Head Self-Attention Relation Distillation for Compressing Pretrained Transformers. In Proceedings of the Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021; Zong, C.; Xia, F.; Li, W.; Navig...

    2021 doi

  19. [29]

    ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction

    Santhanam, K.; Khattab, O.; Saad-Falcon, J.; Potts, C.; Zaharia, M. ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua...

    2022

  20. [30]

    Flocks of Stochastic Parrots: Differentially Private Prompt Learning for Large Language Models, 2023, [arXiv:cs.LG/2305.15594]

    Duan, H.; Dziedzic, A.; Papernot, N.; Boenisch, F. Flocks of Stochastic Parrots: Differentially Private Prompt Learning for Large Language Models, 2023, [arXiv:cs.LG/2305.15594]

    2023 arXiv

  21. [31]

    Does Liking Yellow Imply Driving a School Bus? Semantic Leakage in Language Models

    Gonen, H.; Blevins, T.; Liu, A.; Zettlemoyer, L.; Smith, N.A. Does Liking Yellow Imply Driving a School Bus? Semantic Leakage in Language Models. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Huma...

    2025 doi arXiv

  22. [32]

    Ask the experts: sourcing a high-quality nutrition counseling dataset through Human-AI collaboration

    Balloccu, S.; Reiter, E.; Li, K.J.H.; Sargsyan, R.; Kumar, V .; Reforgiato, D.; Riboni, D.; Dusek, O. Ask the experts: sourcing a high-quality nutrition counseling dataset through Human-AI collaboration. In Proceedings of the Findings of the Association for Computational Lingu...

    2024 doi

  23. [33]

    Automating Bias Testing of LLMs

    Morales, S.; Clarisó, R.; Cabot, J. Automating Bias Testing of LLMs. In Proceedings of the 2023 38th IEEE/ACM International Conference on Automated Software Engineering (ASE), 2023, pp. 1705–1707. https://doi.org/10.1109/ASE56229.2023.00018

    2023

  24. [34]

    Neural-Symbolic Integration: A Compositional Perspective.AAAI Conference on Artificial Intelligence2021,35, 5051–5060

    Tsamoura, E.; Hospedales, T.; Michael, L. Neural-Symbolic Integration: A Compositional Perspective.AAAI Conference on Artificial Intelligence2021,35, 5051–5060. https://doi.org/10.1609/aaai.v35i6.16639

    doi

  25. [35]

    Attention is All you Need

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A.N.; Kaiser, L.u.; Polosukhin, I. Attention is All you Need. In Proceedings of the Advances in Neural Information Processing Systems; 52 of 88 Guyon, I.; Luxburg, U.V .; Bengio, S.; Wallach, H.; Fergus, R....

    2017

  26. [36]

    What Formal Languages Can Transformers Express? A Survey.Transactions of the Association for Computational Linguistics2024,12, 543–561

    Strobl, L.; Merrill, W.; Weiss, G.; Chiang, D.; Angluin, D. What Formal Languages Can Transformers Express? A Survey.Transactions of the Association for Computational Linguistics2024,12, 543–561. https: //doi.org/10.1162/tacl_a_00663

    doi

  27. [37]

    Manning, C.D.; Raghavan, P .; Schütze, H.Introduction to Information Retrieval; Cambridge University Press,

  28. [38]

    ChatGPT is bullshit.Ethics and Information Technology2024,26, 38

    Hicks, M.T.; Humphries, J.; Slater, J. ChatGPT is bullshit.Ethics and Information Technology2024,26, 38. https://doi.org/10.1007/s10676-024-09775-5

    doi

  29. [39]

    Knowledge expansion over probabilistic knowledge bases

    Chen, Y.; Wang, D.Z. Knowledge expansion over probabilistic knowledge bases. In Proceedings of the International Conference on Management of Data, SIGMOD 2014, Snowbird, UT, USA; Dyreson, C.E.; Li, F.; Özsu, M.T., Eds. ACM, pp. 649–660. https://doi.org/10.1145/2588555.2610516

    2014

  30. [40]

    Kyburg, H.E.Probability and the Logic of Rational Belief; Wesleyan University Press: Middletown, CT, USA,

  31. [41]

    NTRS - NASA Technical Reports Server

    Graydon, M.S.; Lehman, S.M.Examining Proposed Uses of LLMs to Produce or Assess Assurance Arguments. NTRS - NASA Technical Reports Server. https://ntrs.nasa.gov/citations/20250001849, Accessed on 21 May 2025

    2025

  32. [42]

    A Fast and Accurate Dependency Parser using Neural Networks

    Chen, D.; Manning, C.D. A Fast and Accurate Dependency Parser using Neural Networks. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing, EMNLP 2014, Doha, Qatar, A meeting of SIGDAT, a Special Interest Group of the ACL; Moschitti, A.; Pan...

    2014 doi

  33. [43]

    Assessment of the accuracy of GeoNames gazetteer data

    Ahlers, D. Assessment of the accuracy of GeoNames gazetteer data. In Proceedings of the 7th Workshop on Geographic Information Retrieval, New York, NY, USA, 2013; GIR ’13, p. 74–81. https://doi.org/10.1145/25 33888.2533938

    2013

  34. [44]

    SUTime: A library for recognizing and normalizing time expressions

    Chang, A.X.; Manning, C. SUTime: A library for recognizing and normalizing time expressions. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC‘12); Calzolari, N.; Choukri, K.; Declerck, T.; Do˘ gan, M.U.; Maegaard, B.; Mariani, J....

    2012

  35. [45]

    Stanza: A Python Natural Language Processing Toolkit for Many Human Languages

    Qi, P .; Zhang, Y.; Zhang, Y.; Bolton, J.; Manning, C.D. Stanza: A Python Natural Language Processing Toolkit for Many Human Languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, 2020. https://doi.org/10.4...

    2020 doi arXiv

  36. [46]

    PostgreSQL: Documentation: 17: F.16

    Group, T.P .G.D. PostgreSQL: Documentation: 17: F.16. fuzzystrmatch — determine string similarities and distanceAppendix F. Additional Supplied Modules and Extensions. https://www.postgresql.org/docs/ current/fuzzystrmatch.html, Accessed on 18 February 2025

    2025

  37. [47]

    Distributed Grouping of Property Graphs with Gradoop

    Junghanns, M.; Petermann, A.; Rahm, E. Distributed Grouping of Property Graphs with Gradoop. In Proceedings of the Datenbanksysteme für Business, Technologie und Web (BTW 2017), 17. Fachtagung des GI-Fachbereichs „Datenbanken und Informationssysteme" (DBIS), 6.-10. März 2017, ...

    2017

  38. [48]

    Matching and Rewriting Rules in Object-Oriented Databases.Mathematics 2024,12

    Bergami, G.; Fox, O.R.; Morgan, G. Matching and Rewriting Rules in Object-Oriented Databases.Mathematics 2024,12. https://doi.org/10.3390/math12172677

    2024 doi

  39. [49]

    Towards a Generalised Semistructured Data Model and Query Language.SIGWEB Newsl.2023,2023

    Bergami, G.; Zegadło, W. Towards a Generalised Semistructured Data Model and Query Language.SIGWEB Newsl.2023,2023. https://doi.org/10.1145/3609429.3609433

    2023

  40. [50]

    Transforming Property Graphs, 2024

    Bonifati, A.; Murlak, F.; Ramusat, Y. Transforming Property Graphs, 2024. https://doi.org/10.48550/arXiv. 2406.13062

    2024 doi arXiv

  41. [51]

    The Faculty of Language: What Is It, Who Has It, and How Did It Evolve?Science2002,298, 1569–1579

    Hauser, M.D.; Chomsky, N.; Fitch, W.T. The Faculty of Language: What Is It, Who Has It, and How Did It Evolve?Science2002,298, 1569–1579. https://doi.org/10.1126/science.298.5598.1569

  42. [52]

    Nivre, J.; et al. case. https://universaldependencies.org/en/dep/case.html, Accessed on 05 March 2025

    2025

  43. [53]

    Nivre, J.; et al. nmod. https://universaldependencies.org/en/dep/nmod.html, Accessed on 05 March 2025. 53 of 88

    2025

  44. [54]

    Towards a standard upper ontology

    Niles, I.; Pease, A. Towards a standard upper ontology. In Proceedings of the 2nd International Conference on Formal Ontology in Information Systems, FOIS 2001, Ogunquit, Maine, USA. ACM, 2001, pp. 2–9. https://doi.org/10.1145/505168.505170

    2001

  45. [55]

    Winstanley, P . Is An Upper Ontology Useful? In Proceedings of the Proceedings of the 7th ISKO UK Biennial Conference 2023, University of Strathclyde, Glasgow, Scotland, July 24-25, 2023; Svarre, T.; Davies, S.; Slavic, A.; Vernau, J.; Brown, N., Eds. CEUR-WS.org, 2023, Vol. 3...

    2023

  46. [56]

    Visualizing Association Rules for Text Mining

    Wong, P .C.; Whitney, P .; Thomas, J. Visualizing Association Rules for Text Mining. In Proceedings of the 1999 IEEE Symposium on Information Visualization, USA, 1999; INFOVIS ’99, p. 120. https: //doi.org/10.5555/857189.857670

    1999

  47. [57]

    Hinman, P .G.Fundamentals of Mathematical Logic; A K Peters/CRC Press, 2005

    2005

  48. [58]

    Sentence Transformers

    Hugging Face. Sentence Transformers. https://huggingface.co/sentence-transformers, Accessed on 24 February 2025

    2025

  49. [59]

    The Science of Design: Creating the Artificial.Design Issues1988,4, 67–82

    Simon, H.A. The Science of Design: Creating the Artificial.Design Issues1988,4, 67–82. http://www.jstor. org/stable/1511391, Accessed on 10 May 2025

    2025

  50. [60]

    https://doi.org/10.1007/978-3- 030-78132-3

    Johannesson, P .; Perjons, E.An Introduction to Design Science; Springer, 2021. https://doi.org/10.1007/978-3- 030-78132-3

    2021 doi

  51. [61]

    From local explanations to global understanding with explainable AI for trees.Nature Machine Intelligence2020,2, 56–67

    Lundberg, S.M.; Erion, G.; Chen, H.; DeGrave, A.; Prutkin, J.M.; Nair, B.; Katz, R.; Himmelfarb, J.; Bansal, N.; Lee, S.I. From local explanations to global understanding with explainable AI for trees.Nature Machine Intelligence2020,2, 56–67. https://doi.org/10.1038/s42256-019-0138-9

    doi

  52. [62]

    Shapley Additive Explanations for Text Classification and Sentiment Analysis of Internet Movie Database

    Dewi, C.; Tsai, B.J.; Chen, R.C. Shapley Additive Explanations for Text Classification and Sentiment Analysis of Internet Movie Database. In Proceedings of the Recent Challenges in Intelligent Information and Database Systems; Szczerbicki, E.; Wojtkiewicz, K.; Nguyen, S.V .; P...

    2022 doi

  53. [63]

    Chain-of- Thought Prompting Elicits Reasoning in Large Language Models

    Wei, J.; Wang, X.; Schuurmans, D.; Bosma, M.; Ichter, B.; Xia, F.; Chi, E.H.; Le, Q.V .; Zhou, D. Chain-of- Thought Prompting Elicits Reasoning in Large Language Models. In Proceedings of the Advances in Neural Information Processing Systems 35: Annual Conference on Neural Inf...

    2022 doi arXiv

  54. [64]

    Why Should I Trust You?

    Ribeiro, M.T.; Singh, S.; Guestrin, C. "Why Should I Trust You?": Explaining the Predictions of Any Classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA; Krishnapuram, B.; Shah, M.; Smola, A.J...

    2016 doi arXiv

  55. [65]

    Anchors: High-Precision Model-Agnostic Explanations.AAAI Conference on Artificial Intelligence2018,32

    Ribeiro, M.T.; Singh, S.; Guestrin, C. Anchors: High-Precision Model-Agnostic Explanations.AAAI Conference on Artificial Intelligence2018,32. https://doi.org/10.1609/aaai.v32i1.11491

    doi

  56. [66]

    OptiLIME: Optimized LIME Explanations for Diagnostic Computer Algorithms

    Visani, G.; Bagli, E.; Chesani, F. OptiLIME: Optimized LIME Explanations for Diagnostic Computer Algorithms. In Proceedings of the CIKM 2020 Workshops co-located with 29th ACM International Conference on Information and Knowledge Management (CIKM 2020), Galway, Ireland; Conrad...

    2020 doi arXiv

  57. [67]

    Explaining Predictive Uncertainty with Information Theoretic Shapley Values

    Watson, D.S.; O’Hara, J.; Tax, N.; Mudd, R.; Guy, I. Explaining Predictive Uncertainty with Information Theoretic Shapley Values. In Proceedings of the Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIP...

    2023 doi arXiv

  58. [68]

    A unified approach to interpreting model predictions

    Lundberg, S.M.; Lee, S.I. A unified approach to interpreting model predictions. In Proceedings of the 31st International Conference on Neural Information Processing Systems, Red Hook, NY, USA, 2017; NIPS’17, p. 4768–4777. https://doi.org/10.48550/arXiv.1705.07874

    2017 doi arXiv

  59. [69]

    https://doi.org/10.5555/3285754

    Bengfort, B.; Bilbro, R.; Ojeda, T.Applied Text Analysis with Python: Enabling Language-Aware Data Products with Machine Learning, 1st ed.; O’Reilly Media, Inc., 2018. https://doi.org/10.5555/3285754

    2018 doi

  60. [70]

    DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter.CoRR2019,abs/1910.01108, [1910.01108]

    Sanh, V .; Debut, L.; Chaumond, J.; Wolf, T. DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter.CoRR2019,abs/1910.01108, [1910.01108]

    1910 arXiv

  61. [71]

    Construction of Domain-Specific DistilBERT Model by Using Fine-Tuning

    Bai, J.; Cao, R.; Ma, W.; Shinnou, H. Construction of Domain-Specific DistilBERT Model by Using Fine-Tuning. In Proceedings of the International Conference on Technologies and Applications of Artificial Intelligence, TAAI 2020, Taipei, Taiwan. IEEE, 2020, pp. 237–241. https://...

    2020 arXiv

  62. [72]

    Evaluating the robustness of interpretability methods through explanation invariance and equivariance

    Crabbé, J.; van der Schaar, M. Evaluating the robustness of interpretability methods through explanation invariance and equivariance. In Proceedings of the 37th International Conference on Neural Information Processing Systems, Red Hook, NY, USA, 2023; NIPS ’23. https://doi.or...

    2023

  63. [73]

    Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods

    Slack, D.; Hilgard, S.; Jia, E.; Singh, S.; Lakkaraju, H. Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, New York, NY, USA, 2020; AIES ’20, p. 180–186. https://doi.org/10.1145/33...

    2020

  64. [74]

    Towards Complex Ontology Alignment Using Large Language Models

    Amini, R.; Norouzi, S.S.; Hitzler, P .; Amini, R. Towards Complex Ontology Alignment Using Large Language Models. In Proceedings of the Knowledge Graphs and Semantic Web; Tiwari, S.; Villazón-Terrazas, B.; Ortiz-Rodríguez, F.; Sahri, S., Eds., Cham, 2025; pp. 17–31

    2025

  65. [75]

    Logical scalability and efficiency of relational learning algorithms

    Picado, J.; Termehchy, A.; Fern, A.; Ataei, P . Logical scalability and efficiency of relational learning algorithms. The VLDB Journal2019,28, 147–171. https://doi.org/10.1007/s00778-018-0523-8

    doi

  66. [76]

    Learning Over Dirty Data Without Cleaning

    Picado, J.; Davis, J.; Termehchy, A.; Lee, G.Y. Learning Over Dirty Data Without Cleaning. In Proceedings of the 2020 International Conference on Management of Data, SIGMOD Conference 2020, online conference [Portland, OR, USA]; Maier, D.; Pottinger, R.; Doan, A.; Tan, W.; Ala...

    2020

  67. [77]

    https://doi.org/10.4135/9781412985130

    Kruskal, J.B.; Wish, M.Multidimensional Scaling; Quantitative Applications in the Social Sciences, SAGE Publications, Inc. https://doi.org/10.4135/9781412985130

    doi

  68. [78]

    Review of the Development of Multidimensional Scaling Methods.Journal of the Royal Statistical Society

    Mead, A. Review of the Development of Multidimensional Scaling Methods.Journal of the Royal Statistical Society. Series D (The Statistician)1992,41, 27–39

    1992

  69. [79]

    Generalized Non-metric Multidimensional Scaling

    Agarwal, S.; Wills, J.; Cayton, L.; Lanckriet, G.R.G.; Kriegman, D.J.; Belongie, S.J. Generalized Non-metric Multidimensional Scaling. In Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, AISTATS 2007, San Juan, Puerto Rico, March ...

    2007

  70. [80]

    Distributional Scaling: An Algorithm for Structure-Preserving Embedding of Metric and Nonmetric Spaces.J

    Quist, M.; Yona, G. Distributional Scaling: An Algorithm for Structure-Preserving Embedding of Metric and Nonmetric Spaces.J. Mach. Learn. Res.2004,5, 399–420

    2004

  71. [81]

    Diverse nearest neighbors queries using linear skylines

    Costa, C.F.; Nascimento, M.A.; Schubert, M. Diverse nearest neighbors queries using linear skylines. GeoInformatica2018,22, 815–844. https://doi.org/10.1007/S10707-018-0332-7

    doi

  72. [82]

    PIST: An Efficient and Practical Indexing Technique for Historical Spatio-Temporal Point Data.GeoInformatica2008,12, 143–168

    Botea, V .; Mallett, D.; Nascimento, M.A.; Sander, J. PIST: An Efficient and Practical Indexing Technique for Historical Spatio-Temporal Point Data.GeoInformatica2008,12, 143–168. https://doi.org/10.1007/S10707- 007-0030-3

    doi

  73. [83]

    Hopcroft, J.E.; Ullman, J.D.Introduction to Automata Theory, Languages and Computation; Addison-Wesley,

  74. [84]

    Asperti, A.; Ciabattoni, A.Logica a informatica; Collana di istruzione scientifica, McGraw-Hill Companies,

  75. [85]

    Machine Teaching: A New Paradigm for Building Machine Learning Systems.CoRR 2017,abs/1707.06742, [1707.06742]

    Simard, P .Y.; Amershi, S.; Chickering, D.M.; Pelton, A.E.; Ghorashi, S.; Meek, C.; Ramos, G.A.; Suh, J.; Verwey, J.; Wang, M.; et al. Machine Teaching: A New Paradigm for Building Machine Learning Systems.CoRR 2017,abs/1707.06742, [1707.06742]

    2017 arXiv

  76. [86]

    Interactive machine teaching: a human-centered approach to building machine-learned models.Human–Computer Interaction2020,35, 413–451

    Ramos, G.; Meek, C.; Simard, P .; Suh, J.; and, S.G. Interactive machine teaching: a human-centered approach to building machine-learned models.Human–Computer Interaction2020,35, 413–451. https: //doi.org/10.1080/07370024.2020.1734931

  77. [87]

    Human- in-the-loop machine learning: a state of the art.Artificial Intelligence Review2023,56, 3005–3054

    Mosqueira-Rey, E.; Hernández-Pereira, E.; Alonso-Ríos, D.; Bobes-Bascarán, J.; Fernández-Leal, Á. Human- in-the-loop machine learning: a state of the art.Artificial Intelligence Review2023,56, 3005–3054. https: //doi.org/10.1007/s10462-022-10246-w

    doi

  78. [88]

    A new Nested Graph Model for Data Integration

    Bergami, G. A new Nested Graph Model for Data Integration. PhD thesis, University of Bologna, Italy, 2018. https://doi.org/10.6092/UNIBO/AMSDOTTORATO/8348

    2018 doi

  79. [89]

    Matita Tutorial.Journal of Formalized Reasoning2014,7, 91–199

    Asperti, A.; Ricciotti, W.; Sacerdoti Coen, C. Matita Tutorial.Journal of Formalized Reasoning2014,7, 91–199. https://doi.org/10.6092/issn.1972-5787/4651

    1972 doi

  80. [90]

    The LRU-K Page Replacement Algorithm For Database Disk Buffering

    O’Neil, E.J.; O’Neil, P .E.; Weikum, G. The LRU-K Page Replacement Algorithm For Database Disk Buffering. In Proceedings of the 1993 ACM SIGMOD International Conference on Management of Data, Washington, DC, USA; Buneman, P .; Jajodia, S., Eds. ACM Press, 1993, pp. 297–306. ht...

    1993 doi

  81. [91]

    2Q: A Low Overhead High Performance Buffer Management Replacement Algorithm

    Johnson, T.; Shasha, D.E. 2Q: A Low Overhead High Performance Buffer Management Replacement Algorithm. In Proceedings of the 20th International Conference on Very Large Data Bases, Santiago de Chile, Chile; Bocca, J.B.; Jarke, M.; Zaniolo, C., Eds. Morgan Kaufmann, 1994, pp. 4...

    1994

  82. [92]

    ArchimedesOne: query processing over probabilistic knowledge bases.Proc

    Zhou, X.; Chen, Y.; Wang, D.Z. ArchimedesOne: query processing over probabilistic knowledge bases.Proc. VLDB Endow.2016,9, 1461–1464. https://doi.org/10.14778/3007263.3007284

    2016

  83. [93]

    https://dataminingbook.info/book_html/, Accessed on 21 May 2025

    Zaki, M.J.; Meira, Jr, W.Data Mining and Machine Learning: Fundamental Concepts and Algorithms, 2 ed.; Cambridge University Press, 2020. https://dataminingbook.info/book_html/, Accessed on 21 May 2025

    2020

  84. [94]

    k-means++: the advantages of careful seeding

    Arthur, D.; Vassilvitskii, S. k-means++: the advantages of careful seeding. In Proceedings of the Eighteenth Annual ACM-SIAM Symposium on Discrete Algorithms, SODA 2007, New Orleans, Louisiana, USA; Bansal, N.; Pruhs, K.; Stein, C., Eds. SIAM, 2007, pp. 1027–1035. https://doi....

    2007

  85. [95]

    https://doi.org/10.1007/978-3-540-68856-3

    Logical and Relational Learning; Springer Berlin Heidelberg, 2008. https://doi.org/10.1007/978-3-540-68856-3

    2008 doi

  86. [96]

    https://archive.org/details/BubliothecaMathematicaStephenColeKleeneIntroductionToMetamathematicsWoltersNoordhoffPublishing1 971/page/n3/mode/2up, Accessed on 21 May 2025

    Kleene, S.Introduction to Metamathematics; Bibliotheca Mathematica, a Series of Monographs on Pure and, Wolters-Noordhoff, 1971. https://archive.org/details/BubliothecaMathematicaStephenColeKleeneIntroductionToMetamathematicsWoltersNoordhoffPublishing1 971/page/n3/mode/2up, Ac...

    1971

  87. [97]

    Silhouettes: A graphical aid to the interpretation and validation of cluster analysis.Journal of Computational and Applied Mathematics1987,20, 53–65

    Rousseeuw, P .J. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis.Journal of Computational and Applied Mathematics1987,20, 53–65. https://doi.org/10.1016/0377-0427(87)90125-7

    doi

  88. [98]

    Information Theoretic Measures for Clusterings Comparison: Variants, Properties, Normalization and Correction for Chance.J

    Nguyen, X.V .; Epps, J.; Bailey, J. Information Theoretic Measures for Clusterings Comparison: Variants, Properties, Normalization and Correction for Chance.J. Mach. Learn. Res.2010,11, 2837–2854. https: //doi.org/10.5555/1756006.1953024

    2010

  89. [99]

    Structure Regularization for Structured Prediction

    Sun, X. Structure Regularization for Structured Prediction. In Proceedings of the Advances in Neural Information Processing Systems 27: Annual Conference on Neural Information Processing Systems 2014, Montreal, Quebec, Canada; Ghahramani, Z.; Welling, M.; Cortes, C.; Lawrence,...

    2014

  90. [100]

    The Stanford CoreNLP Natural Language Processing Toolkit

    "Manning, C.; Surdeanu, M.; Bauer, J.; Finkel, J.; Bethard, S.; McClosky, D. The Stanford CoreNLP Natural Language Processing Toolkit. In Proceedings of the 52nd Annual Meeting of the Association for Compu- tational Linguistics: System Demonstrations; Bontcheva, K.; Zhu, J., E...

    2014 doi

  91. [101]

    Online manuscript released January 12, 2025

    Jurafsky, D.; Martin, J.H.Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition with Language Models, 3rd ed.; 2025. Online manuscript released January 12, 2025. https://web.stanford.edu/~jurafsky/slp3...

    2025

  92. [102]

    Nivre, J.; et al. conj. https://universaldependencies.org/en/dep/conj.html, Accessed on 13 February 2025

    2025

  93. [103]

    Nivre, J.; et al. cc. https://universaldependencies.org/en/dep/cc.html, Accessed on 13 February 2025

    2025

  94. [104]

    Noise reduction and targeted exploration in imitation learning for Abstract Meaning Representation parsing

    Goodman, J.; Vlachos, A.; Naradowsky, J. Noise reduction and targeted exploration in imitation learning for Abstract Meaning Representation parsing. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers); Erk, K.; Smi...

    2016 doi

  95. [105]

    The Stanford Natural Language Processing Group

    Stanford NLP Group. The Stanford Natural Language Processing Group. https://nlp.stanford.edu/ software/lex-parser.shtml, Accessed on 24 February 2025

    2025

  96. [106]

    ini – His Work and its Traditions, 2 ed.; Vol

    Cardona, G.P ¯an. ini – His Work and its Traditions, 2 ed.; Vol. 1, Motilal Banarsidass: London, 1997. https: //doi.org/10.2307/415552

    1997 doi

  97. [107]

    Arguments for and against the Idea of Universal Grammar.Leviathan: Interdisciplinary Journal in English2019, p

    Christensen, C.H. Arguments for and against the Idea of Universal Grammar.Leviathan: Interdisciplinary Journal in English2019, p. 12–28. https://doi.org/10.7146/lev.v0i4.112677

    doi

  98. [108]

    Dardano, M.; Trifone, P .Italian grammar with linguistics notions (in Italian); Zanichelli: Milan, 2002

    2002

  99. [109]

    terminology - Syntactic analysis in English: correspondence between Italian complements and English ones

    terdon. terminology - Syntactic analysis in English: correspondence between Italian complements and English ones. https://english.stackexchange.com/questions/628592/syntactic-analysis-in-english- correspondence-between-italian-complements-and/628597#628597, Accessed on 10 Febr...

    2025

  100. [110]

    Modality: forms - Grammar - Cambridge Dictionary

    2025, C.U.P ..A. Modality: forms - Grammar - Cambridge Dictionary. https://dictionary.cambridge.org/ grammar/british-grammar/modality-forms#:~:text=Dare%2C%20need%2C%20ought%20to%20and%20 used%20to%20(semi%2Dmodal%20verbs), Accessed on 17 March 2025

    2025

  101. [111]

    Nivre, J.; et al. dep. https://universaldependencies.org/en/dep/dep.html, Accessed on 05 March 2025. 56 of 88

    2025

  102. [112]

    English Prepositions in the History of English Grammar Writing.AAA: Arbeiten aus Anglis- tik und Amerikanistik2012,37, 227–243

    Weber, D. English Prepositions in the History of English Grammar Writing.AAA: Arbeiten aus Anglis- tik und Amerikanistik2012,37, 227–243. https://www.researchgate.net/publication/286632773_English_ Prepositions_in_the_History_of_English_Grammar_Writing, Accessed on 21 May 2025

    2025

  103. [113]

    An efficient algorithm for a complete link method.The Computer Journal1977,20, 364–366

    Defays, D. An efficient algorithm for a complete link method.The Computer Journal1977,20, 364–366. https://doi.org/10.1093/comjnl/20.4.364

    doi

  104. [114]

    InIntroduction to HPC with MPI for Data Science; Springer International Publishing: Cham, 2016; pp

    Nielsen, F., Hierarchical Clustering. InIntroduction to HPC with MPI for Data Science; Springer International Publishing: Cham, 2016; pp. 195–211. https://doi.org/10.1007/978-3-319-21903-5_8

    2016 doi

  105. [115]

    InFinding Groups in Data; John Wiley & Sons, Ltd, 1990; chapter 2, pp

    Partitioning Around Medoids (Program PAM). InFinding Groups in Data; John Wiley & Sons, Ltd, 1990; chapter 2, pp. 68–125. https://doi.org/https://doi.org/10.1002/9780470316801.ch2

    1990 doi

  106. [116]

    A comparison of machine learning methods to classify radioactive elements using prompt-gamma-ray neutron activation data

    Mathew, J.; Kshirsagar, R.; Abidin, D.; Griffin, J.; Kanarachos, S.; James, J.; Alamaniotis, M.; Fitzpatrick, M. A comparison of machine learning methods to classify radioactive elements using prompt-gamma-ray neutron activation data. https://doi.org/10.21203/rs.3.rs-2518432/v1

    doi

  107. [117]

    Word2Vec Model Analysis for Semantic Similarities in English Words

    Jatnika, D.; Bijaksana, M.A.; Suryani, A.A. Word2Vec Model Analysis for Semantic Similarities in English Words. In Proceedings of the Enabling Collaboration to Escalate Impact of Research Results for Society: The 4th International Conference on Computer Science and Computation...

    2019 doi

  108. [118]

    Efficient Estimation of Word Representations in Vector Space

    Mikolov, T.; Chen, K.; Corrado, G.; Dean, J. Efficient Estimation of Word Representations in Vector Space. In Proceedings of the 1st International Conference on Learning Representations, ICLR 2013, Scottsdale, Arizona, USA; Bengio, Y.; LeCun, Y., Eds., 2013. https://doi.org/10...

    2013 doi arXiv

  109. [119]

    CareerBERT: Matching resumes to ESCO jobs in a shared embedding space for generic job recommendations.Expert Systems with Applications2025, 275, 127043

    Rosenberger, J.; Wolfrum, L.; Weinzierl, S.; Kraus, M.; Zschech, P . CareerBERT: Matching resumes to ESCO jobs in a shared embedding space for generic job recommendations.Expert Systems with Applications2025, 275, 127043. https://doi.org/10.1016/j.eswa.2025.127043

  110. [120]

    Concept Vector for Similarity Measurement Based on Hierarchical Domain Structure

    Liu, H.; Bao, H.; Xu, D. Concept Vector for Similarity Measurement Based on Hierarchical Domain Structure. Comput. Informatics2011,30, 881–900. http://www.cai.sk/ojs/index.php/cai/article/view/201, Accessed on 21 May 2025

    2025

  111. [121]

    Poincaré Embeddings for Learning Hierarchical Representations

    Nickel, M.; Kiela, D. Poincaré Embeddings for Learning Hierarchical Representations. In Proceedings of the Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems 2017, Long Beach, CA, USA; Guyon, I.; von Luxburg, U.; Be...

    2017

  112. [122]

    Approximate querying of RDF graphs via path alignment.Distributed Parallel Databases2015,33, 555–581

    Virgilio, R.D.; Maccioni, A.; Torlone, R. Approximate querying of RDF graphs via path alignment.Distributed Parallel Databases2015,33, 555–581. https://doi.org/10.1007/s10619-014-7142-1

    doi

  113. [123]

    Alice plays football

    Nivre, J.; et al. English Dependency Relations. https://universaldependencies.org/en/dep/, Accessed on 24 February 2025. Disclaimer/Publisher’s Note:The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) ...

    2025

  114. [128]

    There is traffic but not in the Newcastle city centre

    We then check for this in Line 18, where we compare the current and previous kernels in order to determine whether the previous kernel should become the root or be added as a property; this is determined by the positions of each (Line 20). Supplement IV .3.1. Kernel Assignment...

    2025

  115. [129]

    A given node does not contain adetpropertyAND

  116. [130]

    A given node does not contain ‘on’ within a foundcase[52] propertyAND

  117. [131]

    emilia’s short tailed opossum

    One of the following conditions is met: (a) A given node has at least one incoming edgeANDhas acaseproperty; (b) A given node is the last occurring in the graphANDhas arootproperty; (c) A given node is the last occurring in the graphANDhas a parent with a root property which i...

  118. [1961]

    https://www.jstor.org/stable/186504, Accessed on 21 May 2025

    2025

  119. [1979]

    https://doi.org/10.1016/0096-0551(80)90011-9

    doi

  120. [1997]

    https://books.google.co.uk/books?id=R2-4PAAACAAJ, Accessed 21 May 2025

    2025

  121. [2008]

    https://nlp.stanford.edu/IR-book/information-retrieval-book.html, Accessed on 21 May 2025

    2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.