Pith. sign in

REVIEW 3 major objections 9 minor 42 references

A Probabilistic Model for Node Classification in Directed Graphs

T0 review · 3 major / 9 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A generative classifier built from neighbor label counts, degrees, and text can match or beat graph neural networks on directed graphs while keeping predictions interpretable.

desk verdict A solid, interpretable generative model and a useful new dataset, but the 'superior to state-of-the-art' claim is not yet supported because the main comparison uses mismatched features and dated baselines. read the letter →

arxiv 2501.01630 v1 pith:PITVNYZD submitted 2025-01-03 cs.LG cs.SI

Diego Huerta , Gerardo Arizmendi This is my paper · ORCID
classification cs.LGcs.SI
keywords nodeclassificationdirectedgraphsgenerativeclassifierinterpretabilitymaximumaposterioriMathGenealogyProjectogbn-arxivconditionalindependence
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 simple generative probabilistic classifier, rather than a learned neural network, can match or beat graph neural networks for node classification on directed attributed graphs while remaining interpretable. The model predicts an unlabeled node from four signals: the label counts among its predecessors, the label counts among its successors, its in- and out-degrees, and its own text attributes. The authors estimate parameters by maximum likelihood or MAP and test the model on the ogbn-arxiv citation network and on a newly assembled graph built from the Math Genealogy Project. On both benchmarks the model matches or exceeds the reported GCN and GraphSage baselines, and each prediction can be decomposed into per-component discrepancy terms that show why a label was chosen.

What carries the argument

The load-bearing object is the factorized generative likelihood of Eq. (3), $f_{P,S,X|Y} = f_{P|Y}f_{S|Y}f_{X|Y}$, combined with multinomial models for the predecessor and successor label-count vectors and label-conditional degree and text distributions. Each component contributes a separate negative-log discrepancy term in the prediction objective, which is what makes the model interpretable: the winning label is the one whose total discrepancy is smallest, and the per-component terms show which signals supported or opposed it. The degree distributions are modeled parametrically when the data supports it, with a truncated power law with an atom at zero for the Math Genealogy out-degree and discrete log-normals for ogbn-arxiv in- and out-degrees, with empirical frequencies kept as an alternative.

What would settle it

Generate a synthetic directed graph whose true generative process includes correlations among the labels of a node's predecessors beyond the conditioning on the node's own label, then compare the model's MAP predictions with the true conditional probabilities. If prediction accuracy on such graphs stays near the ogbn-arxiv level, the independence assumption is not load-bearing; if it drops sharply, the reported benchmark numbers rely on that assumption.

Watch

Extended reading notes

Core claim

The paper's central claim is that a generative classifier built on a conditional-independence factorization of the joint distribution over a node's predecessor-label counts, successor-label counts, and attributes can serve as a competitive alternative to GNNs. Conditioned on a node's label, the predecessor counts follow a multinomial with class-dependent probabilities, the successor counts follow a multinomial with a possibly different class-dependent matrix, the degrees follow label-specific parametric or empirical distributions, and the text follows a multinomial-Naive-Bayes-style attribute model. Prediction becomes a sum of negative-log terms, one per component, and choosing the label that minimizes the total is the ML or MAP estimate. Reported results: on ogbn-arxiv the MAP version reaches 0.7432 test accuracy versus 0.7174 for GCN and 0.7149 for GraphSage; on the Math Genealogy graph the ML version reaches 0.5705 F1 versus 0.5689 for GCN, and MAP attains 0.7463 accuracy, the highest of the compared methods.

Load-bearing premise

The model assumes that once you know a node's own label, the labels of its predecessors, the labels of its successors, and its text attributes are completely independent of one another; real directed graphs often violate this, and the paper gives no sensitivity analysis showing how much the benchmark scores depend on that assumption.

Editorial extensions

If this is right

  • Node classification on directed attributed graphs can be done without training a neural network, using parameters that have direct frequency-based estimates.
  • Predictions come with an explanation: the six discrepancy terms, for predecessor labels, successor labels, predecessor count, successor count, attributes, and prior, show which evidence supports or opposes each candidate label.
  • The method is inductive, since a node not present during training can be classified once the global parameters are estimated.
  • The iterative refinement scheme allows predictions to be updated in parallel, and the stopping iteration is chosen by validation performance.
  • The new Math Genealogy graph gives the community a benchmark where graph-aware methods clearly beat text-only methods, with BERT at 0.5028 F1 falling below GCN at 0.5689 F1 and the proposed model at 0.5705 F1.

Reading between the lines

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

  • The ML/MAP choice effectively tunes a prior over class frequencies; on imbalanced labels, MAP's preference for frequent classes explains its higher accuracy and lower F1, and the same knob could be tuned per application cost.
  • The discrepancy decomposition suggests a natural active-learning rule: label next the node whose top-two total discrepancies are closest, since that margin is a readymade uncertainty estimate.
  • Because the transition matrices $\Theta$ and $\Xi$ encode cross-label edge propensities, the generative model is not inherently limited to homophily and could be used on heterophilous graphs in the same iterative refinement loop.
  • A stronger comparison would give the GNN baselines the same raw-text features the model uses; the reported ogbn-arxiv baselines rely on precomputed 128-dimensional embeddings and an undirected version of the graph, so part of the gap may reflect feature choice rather than model class.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 9 minor

Summary. The paper proposes a probabilistic generative model for node classification in directed attributed graphs. Labels are predicted by an ML or MAP score that combines multinomial terms for predecessor and successor label counts, per-label degree distributions, and text-attribute likelihoods under a conditional-independence assumption. Parameters are estimated from labeled training nodes, and predictions for unlabeled nodes are produced by an iterative pseudo-labeling procedure with validation-based selection of the iteration. The model is evaluated on a newly scraped Math Genealogy Project dataset and on ogbn-arxiv, with reported F1/accuracy values that are competitive with or higher than several baselines, including GCN.

Significance. Interpretable, non-neural baselines for node classification are useful, and the new Math Genealogy dataset with benchmark numbers is a genuine asset to the community. The model is computationally cheap and the code is public. If the comparisons were feature-matched and repeated over seeds, the paper could support a credible claim that a simple generative classifier is competitive with GNNs. As it stands, however, the central comparative claim is not yet supported, and the definitional inconsistency between the stated event-level parameters and the edge-level estimators used in the likelihood must be resolved before the model is fully coherent.

major comments (3)
  1. [Section 3, Eqs. (5)-(6); Section 4; Section 5.1.1, Eqs. (17)-(18)] There is a mismatch between the parameters as defined and as used. Equations (5)-(6) define Theta_{i,j} and Xi_{i,j} as probabilities of a node's label given the existence of a predecessor or successor with a given label. In the likelihood, however, the multinomial factors g(p; d_in, xi_i) and g(s; d_out, theta_i) use xi_i and theta_i as per-edge categorical distributions over the labels of a single predecessor or successor conditioned on the node's own label. The estimators in Section 4 are edge-frequency ratios, e.g., \hat{Theta}_{i,j} = #{(u,v): y_u=i, y_v=j} / #{(u,v): y_u=i}, which estimate the edge-level transition P(Y_v=j | Y_u=i), not the event-level probability in Eq. (5). These quantities are generally different in directed graphs. The paper should either define Theta and Xi directly as the per-edge transition probabilities used by the multinomial model, or derive the likelihood from the event-level definitions; otherwise the score being optimized is not the likelihood of the stated generative model.
  2. [Section 7.3, Table 7; Abstract] The comparative claim in the abstract is not supported by the evidence on ogbn-arxiv. The baselines reported from Hu et al. consume the standard 128-dimensional average word embeddings and an undirected graph, whereas the proposed model consumes raw title+abstract text expanded to unigrams/bigrams with TF-IDF. The accuracy gap (0.7432 vs 0.7174 for GCN) could be due entirely to the richer input features or to the different graph convention; it does not isolate the proposed model's mechanism. On the Math Genealogy dataset (Table 3), the ML F1 of 0.5705 exceeds GCN's 0.5689 by only 0.0016, which is within plausible run-to-run variability, and no repeated-seed statistics or confidence intervals are reported on either dataset. A matched-feature comparison with the same text representation and graph directionality for all methods, across multiple seeds, is needed before the paper can claim superiority to state-of-the-art methods.
  3. [Section 3, Eq. (3); Section 5.1.1] The model's likelihood rests on the conditional-independence and i.i.d. multinomial assumptions: given a node's label, the predecessor-label counts, successor-label counts, degrees, and attributes are independent, and each neighbor label is drawn independently from a single label-specific categorical distribution. In real citation and advisor graphs, neighbor labels are typically correlated beyond the ego node's label (e.g., topic communities and co-citation patterns), and text and degrees may be correlated with neighbor-label composition. Because the ML and MAP scores are products of these likelihood factors, a violation changes the ranking among candidate labels. The paper provides no diagnostic or ablation testing the sensitivity of the results to this assumption; I would like to see either a synthetic-data validation of the generative model or an ablation comparing the full model against its text-only, degree-only, and neighbor-only components.
minor comments (9)
  1. [Eq. (13)] The product index appears as \prod_{j=i}^{\tau}, which should presumably be \prod_{j=1}^{\tau}.
  2. [Section 4] The sentence 'These estimators correspond to the maximum likelihood estimators and provide unbiased estimates' is inaccurate after additive smoothing is introduced; smoothed estimators are not MLEs, and ratios of counts need not be unbiased.
  3. [Algorithm 1] The update line is written as 'Update \hat y_v^{(t)} \leftarrow \hat y_v', which is missing the iteration index; it should state that \hat y_v^{(t)} is computed using the previous iteration's labels for neighbors.
  4. [Section 6.2.1] When TF-IDF vectorization is used, the vector \bar x_v contains real-valued weights, so Equations (12)-(14), which are based on integer counts, no longer define a valid multinomial likelihood; this should be acknowledged explicitly as a heuristic approximation.
  5. [Section 6.5 and Table 4] The first example is given as node ID 1526 in the text but as ID 12408 in the Table 4 caption; please make these consistent.
  6. [Section 6.4.2] The claim that ML achieves better F1 'due to balanced predictions across classes' is asserted without supporting evidence; a brief explanation of the mechanism would help the reader.
  7. [Section 7.2.1] For out-degree on ogbn-arxiv, only 23 of 40 labels pass the goodness-of-fit test, yet Equation (21) is retained as one option; since the frequency-based alternative is selected by validation, this is acceptable, but the text should state more explicitly that the parametric form is an approximation for many labels.
  8. [Throughout] Minor typos include 'lustrates' (Section 3), 'Algoritm 1', 'to to additive smoothing' (Section 2.2.1), and 't he' (Section 7.3).
  9. [Section 6/Data availability] The paper introduces a new dataset but provides no direct data release or download link; the GitHub repository appears to contain code only. Please state data availability explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the model is fitted on training labels and evaluated on held-out nodes, with no fitted constant renamed as a prediction and no load-bearing self-citation.

full rationale

The paper's derivation chain is standard supervised learning: assume conditional independence (Eq. 3), estimate the parameters (π, Θ, Ξ, ψ_i, φ_i, ω_i) from labeled training nodes via MLE with additive smoothing, and then classify test nodes by maximizing the conditional likelihood or posterior (Section 5.1). Predictions are not used to define the parameters; parameters are estimated first and then applied to unseen nodes. The iterative pseudo-labeling procedure in Algorithm 1 is a transductive/self-training scheme, and the best iteration is selected on a validation set, which is model selection rather than circular reasoning. The text-attribute component is explicitly acknowledged as an extension of multinomial Naive Bayes, reducing to it only in the degenerate case of a graph with no edges, while the graph structure contributes separate transition matrices Θ and Ξ estimated from edges. No self-citations appear in the reference list, so there is no load-bearing self-citation chain and no uniqueness theorem imported from the authors' prior work. The unmatched comparison in Table 7 (raw text versus 128-dimensional embeddings) and the conditional-independence assumption in Eq. (3) are empirical validity or benchmarking concerns, not circular reductions: the reported predictions are not equivalent to the model's inputs by construction.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The model's predictions depend on several smoothing hyperparameters, per-label parametric degree distributions fitted by MLE, and a validation-selected iteration count. The two load-bearing assumptions are the conditional independence in Eq. (3) and the i.i.d. multinomial neighbor-label model; both are explicit but unverified. No invented entities are introduced.

free parameters (4)
  • Smoothing hyperparameters alpha_pi, alpha_Theta, alpha_Xi, alpha_psi, alpha_omega = Math Genealogy: alpha_omega=0.03, alpha_psi=0.1, alpha_pi=0, alpha_Theta=1, alpha_Xi=1
    Additive smoothing constants tuned on the validation set for each dataset. They affect the estimated probabilities and final predictions.
  • Per-label out-degree parameters beta_i, kappa_i, lambda_i (Math Genealogy)
    Maximum likelihood fits of the zero-inflated truncated power law for each label's student-count distribution. 5 of 63 labels fail the chi-square goodness-of-fit test but are still used with the parametric form.
  • Per-label in/out-degree parameters beta_i, mu_i, sigma_i (ogbn-arxiv)
    Maximum likelihood fits of the zero-inflated discrete log-normal for citation counts and reference counts. Only 23 of 40 out-degree fits pass the goodness-of-fit test, yet the parametric form is retained as a hyperparameter option.
  • Best iteration T for iterative prediction = Math Genealogy: 5; ogbn-arxiv: 2
    The iteration that gives the best validation performance is selected. This is a hyperparameter of Algorithm 1 and is tuned on the validation set.
assumptions (5)
  • ad hoc to paper Conditional independence f_{P,S,X|Y} = f_{P|Y} f_{S|Y} f_{X|Y}
    Core assumption in Section 3, Eq. (3), allowing the likelihood to factor into attribute, in-degree, and neighbor-label terms. It is stated without justification and is likely violated in real graphs.
  • ad hoc to paper Predecessor and successor labels are i.i.d. draws from label-specific categorical distributions (multinomial)
    Introduced in Section 3 after Eq. (8), this leads to P_v ~ Multinomial(d_in, xi_i) and S_v ~ Multinomial(d_out, theta_i), which are used in the likelihood equations (17)-(18).
  • domain assumption All nodes obey the same conditional distributions (stationarity)
    Stated in Section 3: 'we assume that these random variables follow the same distribution for every node v'. This ignores node-specific heterogeneity beyond label and degree.
  • domain assumption The truncated power law (Math Genealogy) and discrete log-normal (ogbn-arxiv) are valid models for degree distributions
    Used in Sections 6.2.2 and 7.2.1. Only 58/63 and 23/40 out-degree samples pass the chi-square test, yet the parametric forms are used in the experiments.
  • domain assumption Multinomial naive Bayes model for text attributes
    Section 4, Eqs. (13)-(14), treats words as independent given the label. This is a standard but strong assumption, acknowledged by the authors in the context of text classification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Probabilistic Model for Node Classification in Directed Graphs." pith.science (2026). https://pith.science/paper/PITVNYZD

@misc{pith2026250101630,
  author       = {Pith},
  title        = {Pith review of: A Probabilistic Model for Node Classification in Directed Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PITVNYZD}},
  note         = {Machine review of arXiv:2501.01630}
}
read the original abstract

In this work, we present a probabilistic model for directed graphs where nodes have attributes and labels. This model serves as a generative classifier capable of predicting the labels of unseen nodes using either maximum likelihood or maximum a posteriori estimations. The predictions made by this model are highly interpretable, contrasting with some common methods for node classification, such as graph neural networks. We applied the model to two datasets, demonstrating predictive performance that is competitive with, and even superior to, state-of-the-art methods. One of the datasets considered is adapted from the Math Genealogy Project, which has not previously been utilized for this purpose. Consequently, we evaluated several classification algorithms on this dataset to compare the performance of our model and provide benchmarks for this new resource.

Figures

Figures reproduced from arXiv: 2501.01630 by the authors.

Figure 1
Figure 1. Example of notation. Let the label of a node denote its color, and [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Induced subgraph of size 500 of the resulted directed graph represent [PITH_FULL_IMAGE:figures/full_fig_p017_2.png] view at source ↗
Figure 4
Figure 4. Chi-squared goodness-of-fit for the out degree of nodes with label 58 [PITH_FULL_IMAGE:figures/full_fig_p020_4.png] view at source ↗
Figures from the paper (2 more)
Figure 3
Figure 3. Figure 3: Chi-squared goodness-of-fit for the out degree of nodes with label [PITH_FULL_IMAGE:figures/full_fig_p020_3.png]
Figure 5
Figure 5. Figure 5: Chi-squared goodness-of-fit for the in degree of nodes with label 2. [PITH_FULL_IMAGE:figures/full_fig_p027_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 36 canonical work pages

  1. [1]

    Interpretable predictions for crime categories using log loss approach for imbalanced target feature

    Manasa A and Snigdha Sen. Interpretable predictions for crime categories using log loss approach for imbalanced target feature. In 2024 IEEE In- ternational Conference on Contemporary Computing and Communications (InC4), volume 1, pages 1–6, 2024

    work page 2024

  2. [2]

    A survey on bert and its applications

    Sulaiman Aftan and Habib Shah. A survey on bert and its applications. In 2023 20th Learning and Technology Conference (L&T) , pages 161–166, 2023

    work page 2023

  3. [3]

    A gentle introduction to deep learning for graphs

    Davide Bacciu, Federico Errica, Alessio Micheli, and Marco Podda. A gentle introduction to deep learning for graphs. Neural Networks , 129:203–221, 2020

    work page 2020

  4. [4]

    Belle and I

    V. Belle and I. Papantonis. Principles and practice of explainable machine learning. Frontiers in Big Data , 4:688969, 2021

    work page 2021

  5. [5]

    Interpretability in deep learning for finance: a case study for the heston model, 2021

    Damiano Brigo, Xiaoshan Huang, Andrea Pallavicini, and Haitz Saez de Ocariz Borde. Interpretability in deep learning for finance: a case study for the heston model, 2021

    work page 2021

  6. [6]

    Ap- plications of graph convolutional networks in computer vision.Neural Com- puting and Applications , 34(16):13387–13405, 2022

    Pingping Cao, Zeqi Zhu, Ziyuan Wang, Yanping Zhu, and Qiang Niu. Ap- plications of graph convolutional networks in computer vision.Neural Com- puting and Applications , 34(16):13387–13405, 2022

    work page 2022

  7. [7]

    Carvalho, Eduardo M

    Diogo V. Carvalho, Eduardo M. Pereira, and Jaime S. Cardoso. Machine learning interpretability: A survey on methods and metrics. Electronics, 8(8), 2019

    work page 2019

  8. [8]

    Feature selection for text classification with na ¨ ıve bayes.Expert Systems with Ap- plications, 36(3, Part 1):5432–5435, 2009

    Jingnian Chen, Houkuan Huang, Shengfeng Tian, and Youli Qu. Feature selection for text classification with na ¨ ıve bayes.Expert Systems with Ap- plications, 36(3, Part 1):5432–5435, 2009

    work page 2009

Show all 42 references
  1. [9]

    Jackknifing documents and additive smoothing for naive bayes with scarce data

    Vinay Deolalikar. Jackknifing documents and additive smoothing for naive bayes with scarce data. In 2015 IEEE International Conference on Data Mining, pages 91–100, 2015. 30

    2015

  2. [10]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. In North American Chapter of the Association for Computational Linguis- tics, 2019

    2019

  3. [11]

    A convo- lutional encoder model for neural machine translation

    Jonas Gehring, Michael Auli, David Grangier, and Yann Dauphin. A convo- lutional encoder model for neural machine translation. In Regina Barzilay and Min-Yen Kan, editors, Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long P...

    2017

  4. [12]

    node2vec: Scalable feature learning for networks, 2016

    Aditya Grover and Jure Leskovec. node2vec: Scalable feature learning for networks, 2016

    2016

  5. [13]

    Hamilton

    William L. Hamilton. Graph representation learning. Synthesis Lectures on Artificial Intelligence and Machine Learning , 14(3):1–159, 2020

    2020

  6. [14]

    Hamilton, Rex Ying, and Jure Leskovec

    William L. Hamilton, Rex Ying, and Jure Leskovec. Inductive representa- tion learning on large graphs, 2018

    2018

  7. [15]

    Open graph benchmark: Datasets for machine learning on graphs, 2021

    Weihua Hu, Matthias Fey, Marinka Zitnik, Yuxiao Dong, Hongyu Ren, Bowen Liu, Michele Catasta, and Jure Leskovec. Open graph benchmark: Datasets for machine learning on graphs, 2021

    2021

  8. [16]

    Ben- son

    Qian Huang, Horace He, Abhay Singh, Ser-Nam Lim, and Austin R. Ben- son. Combining label propagation and simple models out-performs graph neural networks, 2020

    2020

  9. [17]

    Toward online node classification on streaming networks

    Ling Jian, Jundong Li, and Huan Liu. Toward online node classification on streaming networks. Data Mining and Knowledge Discovery , 32(1):231– 257, 2018

    2018

  10. [18]

    Anomaly detection with graph convolutional networks for insider threat and fraud detection

    Jianguo Jiang, Jiuming Chen, Tianbo Gu, Kim-Kwang Raymond Choo, Chao Liu, Min Yu, Weiqing Huang, and Prasant Mohapatra. Anomaly detection with graph convolutional networks for insider threat and fraud detection. In MILCOM 2019 - 2019 IEEE Military Communications Con- ference (...

    2019

  11. [19]

    Dan Jurafsky and James H. Martin. Speech and language processing : an introduction to natural language processing, computational linguistics, and speech recognition. Pearson Prentice Hall, Upper Saddle River, N.J., 2009

    2009

  12. [20]

    Semi-supervised classification with graph convolutional networks

    Thomas Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. ArXiv, abs/1609.02907, 2016

    2016 arXiv

  13. [21]

    Thabet, and Bernard Ghanem

    Guohao Li, Matthias M¨ uller, Ali K. Thabet, and Bernard Ghanem. Can gcns go as deep as cnns? CoRR, abs/1904.03751, 2019. 31

    1904 arXiv

  14. [22]

    Deep Learning on Graphs

    Yao Ma and Jiliang Tang. Deep Learning on Graphs. Cambridge University Press, 2021

    2021

  15. [23]

    Abusive language detection with graph convolutional networks, 2019

    Pushkar Mishra, Marco Del Tredici, Helen Yannakoudakis, and Ekaterina Shutova. Abusive language detection with graph convolutional networks, 2019

    2019

  16. [24]

    Fast and Accurate Sen- timent Classification Using an Enhanced Naive Bayes Model, page 194–201

    Vivek Narayanan, Ishan Arora, and Arjun Bhatia. Fast and Accurate Sen- timent Classification Using an Enhanced Naive Bayes Model, page 194–201. Springer Berlin Heidelberg, 2013

    2013

  17. [25]

    Networks

    Mark Newman. Networks. Oxford University Press, 07 2018

    2018

  18. [26]

    Denish Omondi Otieno, Akbar Siami Namin, and Keith S. Jones. The ap- plication of the bert transformer model for phishing email classification. In 2023 IEEE 47th Annual Computers, Software, and Applications Conference (COMPSAC), pages 1303–1310, 2023

    2023

  19. [27]

    Qader, Musa M

    Wisam A. Qader, Musa M. Ameen, and Bilal I. Ahmed. An overview of bag of words;importance, implementation, applications, and challenges. In 2019 International Engineering Conference (IEC) , pages 200–204, 2019

    2019

  20. [28]

    Text mining: Use of tf-idf to examine the relevance of words to documents

    Shahzad Qaiser and Ramsha Ali. Text mining: Use of tf-idf to examine the relevance of words to documents. International Journal of Computer Applications, 2018

    2018

  21. [29]

    Universality of citation distributions: Toward an objective measure of scientific impact

    Filippo Radicchi, Santo Fortunato, and Claudio Castellano. Universality of citation distributions: Toward an objective measure of scientific impact. Proceedings of the National Academy of Sciences , 105(45):17268–17272, November 2008

    2008

  22. [30]

    Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

    Cynthia Rudin. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature Machine Intelligence, 1(5):206–215, 2019

    2019

  23. [31]

    En- sembles of bert for depression classification

    Saskia Senn, ML Tlachac, Ricardo Flores, and Elke Rundensteiner. En- sembles of bert for depression classification. In 2022 44th Annual Interna- tional Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), pages 4691–4694, 2022

    2022

  24. [32]

    Interpretability of machine learning-based prediction models in healthcare

    Gregor Stiglic, Primoz Kocbek, Nino Fijacko, Marinka Zitnik, Katrien Ver- bert, and Leona Cilar. Interpretability of machine learning-based prediction models in healthcare. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 10(5):e1379, 2020

    2020

  25. [33]

    Node Classification in Signed Social Networks, pages 54–62

    Jiliang Tang, Charu Aggarwal, and Huan Liu. Node Classification in Signed Social Networks, pages 54–62. Proceedings of the 2016 SIAM International Conference on Data Mining, 2016. 32

    2016

  26. [34]

    Are the discretised lognormal and hooked power law distri- butions plausible for citation data? Journal of Informetrics, 10(2):454–470, 2016

    Mike Thelwall. Are the discretised lognormal and hooked power law distri- butions plausible for citation data? Journal of Informetrics, 10(2):454–470, 2016

    2016

  27. [35]

    The discretised lognormal and hooked power law distribu- tions for complete citation data: Best options for modelling and regression

    Mike Thelwall. The discretised lognormal and hooked power law distribu- tions for complete citation data: Best options for modelling and regression. Journal of Informetrics , 10(2):336–346, 2016

    2016

  28. [36]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. CoRR, abs/1706.03762, 2017

    2017 arXiv

  29. [37]

    Dickerson, and Su-In Lee

    Sahil Verma, Aditya Lahiri, John P. Dickerson, and Su-In Lee. Pitfalls of explainable ml: An industry perspective, 2021

    2021

  30. [38]

    Naive bayes: applications, variations and vulnerabilities: a review of literature with code snippets for implementation

    Indika Wickramasinghe and Harsha Kalutarage. Naive bayes: applications, variations and vulnerabilities: a review of literature with code snippets for implementation. Soft Computing , 25(3):2277–2293, 2021

    2021

  31. [39]

    Research on the application of deep learning-based bert model in sentiment analysis, 2024

    Yichao Wu, Zhengyu Jin, Chenxi Shi, Penghao Liang, and Tong Zhan. Research on the application of deep learning-based bert model in sentiment analysis, 2024

    2024

  32. [40]

    A comprehensive survey on graph neural networks

    Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems , 32(1):4–24, 2020

    2020

  33. [41]

    Bayesian na ¨ ıve bayes classifiers to text classification.Journal of Information Science, 44(1):48–59, 2016

    Shuo Xu. Bayesian na ¨ ıve bayes classifiers to text classification.Journal of Information Science, 44(1):48–59, 2016

    2016

  34. [42]

    Graph con- volutional networks: a comprehensive review

    Si Zhang, Hanghang Tong, Jiejun Xu, and Ross Maciejewski. Graph con- volutional networks: a comprehensive review. Computational Social Net- works, 6(1):11, 2019. 33

    2019

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.