Pith. sign in

REVIEW 4 major objections 5 minor 31 references

Expanding Vietnamese SentiWordNet to Improve Performance of Vietnamese Sentiment Analysis Models

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read Expanding Vietnamese SentiWordNet lifts review classification F1 to 0.95.

desk verdict The paper's lexicon-plus-PLM combination is plausible and shows a consistent gain on AIVIVN, but the undefined distance metric and the binary-vs-three-class VLSP comparison leave the headline claims unsubstantiated. read the letter →

arxiv 2501.08758 v1 pith:BP66WZYQ submitted 2025-01-15 cs.CL

classification cs.CL
keywords VietnamesesentimentanalysisSentiWordNetexpansionPhoBERT-V2CombViSAlexical-semanticfeaturesnegationpatternsVLSP2016AIVIVN2019
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 Vietnamese sentiment analysis improves when a pre-trained language model is combined with an expanded sentiment lexicon rather than used alone. To that end it grows the 1,017-word Vietnamese SentiWordNet into a larger resource by pulling pure positive and negative words from English SentiWordNet and adding synonyms and antonyms from Vietnamese and WordNet dictionaries, scoring each candidate by the ratio of its average distance to the positive set over the sum of both distances. The combined model, CombViSA, runs PhoBERT-V2 through an RCNN and a feedforward layer to get a language-model vector, concatenates it with sentiment vectors extracted from the expanded lexicon, with negation patterns flipping the scores, and classifies with softmax. On the VLSP 2016 and AIVIVN 2019 review datasets the model reports F1 scores of 0.95 and 0.94, above the phoBERT-V2-only and LSTM baselines.

What carries the argument

The load-bearing object is the expanded Vietnamese SentiWordNet and the score formula living inside it. For a candidate word w, $PosScore = d_{pos}/(d_{pos}+d_{neg})$ with $NegScore = 1 - PosScore$ forces the two scores to sum to one, so every word carries a normalized polarity between purely positive and purely negative. The sets P and N are grown from pure SentiWordNet entries by synonym and antonym relations, and the SentiVector_Extraction algorithm converts the per-word scores into length-128 PosVec and NegVec vectors for each review, reversing the polarity inside negation patterns. These vectors are concatenated and passed through a feedforward layer to make SWVec, which is then concatenated with the LMVec produced by PhoBERT-V2 plus RCNN; the pair goes through another feedforward layer and softmax.

What would settle it

Re-run the pipeline with a named distance function, say shortest path through synonym and antonym relations in WordNet, and recompute the extended lexicon and the CombViSA F1 scores; if the reproduced F1 drops back to the phoBERT-V2-only values of 0.91 and 0.89, or if the distance choice changes which words are scored positive versus negative, the gain claimed for ExtViSentiWordNet does not hold.

Watch

Extended reading notes

Core claim

The paper's claim, stated on its own terms, is that a larger SentiWordNet plus a way to digest it into fixed-size vectors is what closes the gap between a generic Vietnamese language model and a strong review classifier. The extension procedure defines a pure positive set P and a pure negative set N from SentiWordNet scores, expands them with synonym and antonym relations, and assigns a candidate word w the scores $PosScore = d_{pos}/(d_{pos}+d_{neg})$ and $NegScore = 1 - PosScore$, where $d_{pos}$ and $d_{neg}$ are average distances from w to the two sets. The SentiVector_Extraction algorithm builds PosVec and NegVec from these scores, swapping the two under negation patterns such as không and chẳng. Combining these vectors with PhoBERT-V2 features in CombViSA yields the reported F1 of 0.95 on VLSP 2016 and 0.94 on AIVIVN 2019, which would make the expanded lexicon a working complement to contextual embeddings for Vietnamese.

Load-bearing premise

The expanded lexicon is built by averaging 'distances' from each candidate word to the positive and negative seed sets, but the paper never defines that distance, so Equation 5 has no computable meaning until a distance function is supplied.

Editorial extensions

If this is right

  • On both evaluation sets, adding the expanded-lexicon SentiVec to PhoBERT-V2 raises F1 by 2 points, from 0.91 to 0.93 on VLSP 2016 and from 0.89 to 0.93 on AIVIVN 2019, and the full CombViSA reaches 0.95 and 0.94.
  • Lexical-semantic features and contextual embeddings are complementary for this task, since the lexicon contributes signal that the language-model-only vector does not already contain.
  • The negation-pattern handling in SentiVector_Extraction is a concrete mechanism that turns explicit negation into flipped polarity, which matters for review text where phrases like 'not good' are negative.
  • The expansion procedure, if it works as described, gives a recipe for enlarging a small sentiment lexicon for Vietnamese without manual annotation, using SentiWordNet seeds and synonym and antonym relations.

Reading between the lines

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

  • A reproducibility check follows directly from the missing distance definition: any concrete metric substituted into Equation 5 could change which words get which scores, so the reported gains are only meaningful once that function is specified. This is an editorial inference about the paper's reproducibility, not a claim the paper makes.
  • The improvement attributed to the lexicon might partly come from the extra RCNN and MLP parameters; an ablation that swaps SentiWordNet vectors for random fixed vectors of the same dimension would isolate the lexicon's contribution. This is a suggested extension, not in the paper.
  • Since P and N are initialized from English SentiWordNet and expanded through dictionary relations, the method's success likely depends on how well those relations cover informal, typo-heavy e-commerce Vietnamese; testing on out-of-vocabulary slang would show the boundary. This is an inference beyond the paper's experiments.
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

4 major / 5 minor

Summary. The paper proposes a Vietnamese sentiment analysis system (CombViSA) that combines PhoBERT-V2 representations with lexical-semantic features from an expanded Vietnamese SentiWordNet. The extension procedure (Section 4.1) selects positive and negative seed sets from SentiWordNet using a threshold T, adds synonyms and antonyms from Vietnamese dictionaries, and computes PosScore/NegScore for each new word via an average-distance formula. The resulting scores are used by Algorithm 1 to build PosVec/NegVec vectors, which are concatenated with PhoBERT-V2 features and passed through RCNN and feedforward layers for classification. Experiments on VLSP 2016 and AIVIVN 2019 report F1 scores of 0.95 and 0.94 for CombViSA, outperforming the cited baselines.

Significance. If the claims are correct, the paper makes an incremental but potentially useful contribution: it demonstrates that lexicon-based sentiment features can add modest gains over a strong pretrained language model on Vietnamese review classification, and it attempts to address the small coverage of an existing Vietnamese SentiWordNet. The proposed resource could be of value to downstream Vietnamese sentiment analysis. However, the paper's strengths are limited by serious reproducibility gaps: the distance function in the core extension step is undefined, the threshold T is not reported, and the evaluation setup is inconsistent with the described task. There is no released code or resource, so the claimed gains cannot be verified without additional detail.

major comments (4)
  1. [Section 4.1, Step 3, Equation (5)] The method computes dpos and dneg as the 'average distance' from w to the words in sets P and N, but no distance function over words is ever defined. Because these distances directly determine PosScore and NegScore, and hence the SWVec features used by the model, the core contribution cannot be reproduced or independently validated until the metric is specified (for example, WordNet path similarity, graph distance in VCL, or cosine distance in a pretrained embedding space). Please define the metric explicitly and state how it is computed.
  2. [Section 5.1, Table 1, Tables 3-4] The VLSP 2016 dataset is described in Section 5.1 as three-class ('negative', 'positive', and 'neutral'), but Table 1 and all reported results in Tables 3 and 4 are binary. If neutral reviews were discarded, the comparison to the cited baselines SA1/SA2 [31] and the results from [30] is not valid because those baselines were evaluated on the official three-class VLSP 2016 task. Moreover, the counts in Table 1 sum to 50,706, not the stated 50,760. Please state exactly how neutral labels were handled, correct the arithmetic, and either report three-class results or compare against binary-class baselines on the same subset.
  3. [Section 4.1, Step 1; Section 5.4] The threshold T controls the construction of the seed sets P and N, which in turn generate the features used for classification. The text says T is 'a constant determined through experimentation,' but its value is never reported, and no separation between validation and test data is described. If T was tuned on the same datasets used for evaluation, the reported F1 scores are optimistically biased. Please report the chosen value of T and describe a validation procedure that keeps the test folds untouched during tuning.
  4. [Section 5.4, Tables 3 and 4] The discussion paragraph states that 'the best result of the proposed model is 0.92 in terms of F1-score,' but Table 3 reports CombViSA F1 = 0.95 on VLSP 2016 and Table 4 reports F1 = 0.94 on AIVIVN 2019. This inconsistency, together with the missing precision and recall values for SA1 and SA2 in Table 3, undermines confidence in the numerical results. Please correct the text or the tables and report complete metrics for all baselines.
minor comments (5)
  1. [Section 4.1, Equation (5)] Equation (5) is typeset ambiguously: it should be PosScore = dpos / (dpos + dneg). Please fix the equation and add parentheses.
  2. [Section 4, first paragraph] The sentence 'The Error! Reference source not found. depicts the architecture of our proposed model' contains an unresolved cross-reference; please insert the correct figure number.
  3. [Section 5.4] The phrase 'As shown in Table and Table' uses unresolved cross-references, and the text states the average comment length for VLSP 2016 is 23.39 while Table 2 reports 25.34. Please correct these inconsistencies.
  4. [Algorithm 1 and Section 4.2] The algorithm relies on VSNO and DVSNO, and on the notion of a 'reverse pattern,' but none of these is formally defined. Please define the normalization operators and give a finite specification of the reversal patterns used in Step 2.
  5. [References] Reference [31] is malformed: it begins with a raw author list and does not follow standard citation formatting. Also, the related-work section contains minor typos such as 'Native Bayes' for 'Naive Bayes' and 'the evelopment' for 'the development.'

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the SentiWordNet expansion and PhoBERT combination are external-feature inputs, and the central empirical claim rests on benchmark comparisons rather than on reproducing its own inputs.

full rationale

The paper's derivation chain is not circular at the equation level. Section 4.1 builds an expanded lexicon by selecting positive/negative seeds from English SentiWordNet using a threshold T, propagating labels through synonyms/antonyms, and assigning PosScore/NegScore via Equations 5–6 from distances to those seed sets. Algorithm 1 then converts those scores into PosVec/NegVec features, which are concatenated with PhoBERT-V2 representations in CombViSA. The reported improvements on VLSP 2016 and AIVIVN 2019 are comparisons against external baselines and PhoBERT-only configurations. None of these steps defines the evaluation metric, the test labels, or the baseline F1 values in terms of the proposed model's own outputs, so there is no self-definitional reduction. The free threshold T is described only as 'determined through experimentation' without stating whether it was tuned on the test folds; inferring that the evaluation is a forced fit would require speculation beyond the text. The undefined distance function in Step 3 and the mismatch between the stated three-class VLSP 2016 task and the reported two-class statistics are serious reproducibility and validity concerns, but they are not instances of the model predicting its own inputs. No load-bearing self-citations or imported uniqueness theorems appear; the cited baselines are independent published systems. Therefore the paper is not circular, and the appropriate score is 0.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on an undefined distance metric, a hand-tuned threshold, and the assumption that synonym/antonym relations and negation patterns transfer sentiment polarity in Vietnamese. No new physical or formal entities are introduced; the expanded lexicon is an artifact rather than an invented entity.

free parameters (1)
  • Threshold T = not reported
    Used in Step 1 to select positive and negative seed word sets from SentiWordNet; its value affects the entire lexicon and the resulting sentiment features, and it is 'determined through experimentation' on the same benchmarks.
assumptions (4)
  • domain assumption Synonyms and antonyms from VCL and WordNet preserve or invert sentiment polarity, respectively
    Step 2 expands the seed sets on the basis of these relations; if the relation does not hold for Vietnamese, the expansion is invalid.
  • ad hoc to paper The average distance from a word to the positive and negative seed sets (dpos, dneg) can be computed and predicts sentiment orientation
    Equation 5 defines the sentiment score as a ratio of these distances, but the distance metric is never specified, making this the load-bearing modeling choice.
  • domain assumption PhoBERT-V2 provides contextual representations that are useful for Vietnamese sentiment analysis
    The model relies on PhoBERT-V2 as the distributional feature source; this is an accepted assumption given published results.
  • domain assumption Negation patterns flip sentiment polarity
    Algorithm 1 swaps PosScore and NegScore for words in manually defined negation patterns; the pattern list is not exhaustive and the swap rule is a modeling assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Expanding Vietnamese SentiWordNet to Improve Performance of Vietnamese Sentiment Analysis Models." pith.science (2026). https://pith.science/paper/BP66WZYQ

@misc{pith2026250108758,
  author       = {Pith},
  title        = {Pith review of: Expanding Vietnamese SentiWordNet to Improve Performance of Vietnamese Sentiment Analysis Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BP66WZYQ}},
  note         = {Machine review of arXiv:2501.08758}
}
read the original abstract

Sentiment analysis is one of the most crucial tasks in Natural Language Processing (NLP), involving the training of machine learning models to classify text based on the polarity of opinions. Pre-trained Language Models (PLMs) can be applied to downstream tasks through fine-tuning, eliminating the need to train the model from scratch. Specifically, PLMs have been employed for Sentiment Analysis, a process that involves detecting, analyzing, and extracting the polarity of text sentiments. Numerous models have been proposed to address this task, with pre-trained PhoBERT-V2 models standing out as the state-of-the-art language models for Vietnamese. The PhoBERT-V2 pre-training approach is based on RoBERTa, optimizing the BERT pre-training method for more robust performance. In this paper, we introduce a novel approach that combines PhoBERT-V2 and SentiWordnet for Sentiment Analysis of Vietnamese reviews. Our proposed model utilizes PhoBERT-V2 for Vietnamese, offering a robust optimization for the prominent BERT model in the context of Vietnamese language, and leverages SentiWordNet, a lexical resource explicitly designed to support sentiment classification applications. Experimental results on the VLSP 2016 and AIVIVN 2019 datasets demonstrate that our sentiment analysis system has achieved excellent performance in comparison to other models.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 29 canonical work pages

  1. [30]

    Fine-Tuning BERT for Sentiment Analysis of Vietnamese Reviews

    Q. T. Nguyen, T. L. Nguyen, N. H. Luong, and Q. H. Ngo, “Finetuning bert for sentiment analysis of vietnamese reviews”, CoRR, vol. abs/2011.10426, 2020

  2. [31]

    Vlsp shared task sen timent analysis

    N. T. Q. V. X. L. T. M. V. N. X. B. L. A. C. NGUYEN THI MINH HUYEN, NGUYEN VIET HUNG, “Vlsp shared task sen timent analysis”, Journal of Computer Science and Cybernetics, vol. V.34, N.4, 2018

  3. [1]

    PhoBERT: Pre-trained language models for Vietnamese

    D. Q. Nguyen and A. Tuan Nguyen, “PhoBERT: Pre-trained language models for Vietnamese”, in Findings of the Association for Computational Linguistics: EMNLP 2020, T. Cohn, Y. He, and Y. Liu, Eds. Online: Association for Computational Linguistics, Nov. 2020, pp. 1037– 1042

  4. [2]

    SENTIWORDNET: A publicly available lexical resource for opinion mining

    A. Esuli and F. Sebastiani, “SENTIWORDNET: A publicly available lexical resource for opinion mining”, in Proceedings of the Fifth International Conference on Language Resources and Evaluation (LREC’06). Genoa, Italy: European Langu age Resources Association (ELRA), May 2006

  5. [3]

    Techniques and applications for sentiment analysis

    R. Feldman, “Techniques and applications for sentiment analysis”, Communications of the ACM, vol. 56, no. 4, pp. 82–89, apr 2013

  6. [4]

    Twitter brand sentiment analysis: A hybrid system using n-gram analysis and dynamic artificial neural network

    M. Ghiassi, J. Skinner, and D. Zimbra, “Twitter brand sentiment analysis: A hybrid system using n-gram analysis and dynamic artificial neural network”, Expert Syst. Appl., vol. 40, no. 16, pp. 6266–6282, 2013

  7. [5]

    Applying authorship analysis to extremistgroup web forum messages

    A. Abbasi and H. Chen, “Applying authorship analysis to extremistgroup web forum messages”, IEEE Intelligent Systems, vol. 20, no. 5, pp. 67 – 75, 2005

  8. [6]

    Neo -nazis sympathizers on the forums of the romanian online publications

    E. Macovei, “Neo -nazis sympathizers on the forums of the romanian online publications”, Styles of Communication , vol. 5, no. 1, pp. 93 – 109, 2013

Show all 31 references
  1. [7]

    Detecting linguistic markers for radi cal violence in social media,

    J. F. K. L. M. J. Cohen, K., “Detecting linguistic markers for radi cal violence in social media,” Terrorism and Political Violence , vol. 26, pp. 246–256, 2014

  2. [8]

    Thumbs up? sentiment classification using machine learning techniques

    B. Pang, L. Lee, and S. Vaithyanathan, “Thumbs up? sentiment classification using machine learning techniques”, in Proceedings of the 2002 Conference on Empiri cal Methods in Natural Language Processing (EMNLP 2002), Jul. 2002, pp. 79–86

  3. [9]

    Sentiment analysis using support vector machine based on feature selection and semantic analysis

    B. M. and V. B., “Sentiment analysis using support vector machine based on feature selection and semantic analysis”, International Journal of Computer Applications , vol. 146, pp. 26–30, 07 2016

  4. [10]

    Sentimental analysis using naive bayes classifier

    P. P. Surya and B. Subbulakshmi, “Sentimental analysis using naive bayes classifier”, in 2019 International Conference on Vision Towards Emerging Trends in Communication and Networking (ViTECoN), 2019, pp. 1–5

  5. [11]

    Lexiconbased methods for sentiment analysis

    M. Taboada, J. Brooke, M. Tofiloski, K. Voll, and M. Stede, “Lexiconbased methods for sentiment analysis”, Computational Linguistics, vol. 37, pp. 267–307, 06 2011

  6. [12]

    Distributed representations of sentences and documents

    Q. Le and T. Mikolov, “Distributed representations of sentences and documents”, International Conference on Machine Learning , ICML 2014, vol. 4, 05 2014. 8

  7. [13]

    Character - level convolutional networks for text classification

    X. Zhang, J. Zhao, and Y. Lecun, “Character - level convolutional networks for text classification”, Proceedings of the 29th International Conference on Neural Information Processing Systems, Vol. 1, 2015

  8. [14]

    Enhancing deep learning sentiment analysis with ensemble techniques in social applications

    O. Araque, I. Corcuera -Platas, J. F. S´anchez - Rada, and C. A. Iglesias, “Enhancing deep learning sentiment analysis with ensemble techniques in social applications” , Expert Systems with Applications, vol. 77, pp. 236–246, 2017

  9. [15]

    Comparison of text classification methods using deep learning neural networks

    M. Amjad, A. F. Gelbukh, I. Voronkov, and A. Saenko, “Comparison of text classification methods using deep learning neural networks” , Lecture Notes in Computer Science , vol. 13452. Springer, 2019, pp. 438–450

  10. [16]

    Twitter-based opinion mining for flight service utilizing machine learning

    P. Tiwari, H. M. Pandey, A. Khamparia, and S. Kumar, “Twitter-based opinion mining for flight service utilizing machine learning” , Informatica (Slovenia), vol. 43, no. 3, 2019

  11. [17]

    Three-way enhanced convolutional neural networks for sentence -level sentimen t classification

    Y. Zhang, Z. Zhang, D. Miao, and J. Wang, “Three-way enhanced convolutional neural networks for sentence -level sentimen t classification”, Inf. Sci. , vol. 477, pp. 55 –64, 2019

  12. [18]

    Sentiment analysis using deep learning architectures: a review

    A. Yadav and D. K. Vishwakarma, “Sentiment analysis using deep learning architectures: a review”, Artif. Intell. Rev. , vol. 53, no. 6, pp. 4335–4385, 2020

  13. [19]

    Sentiment analysis using deep learning approaches: an overview

    O. Habimana, Y. Li, R. Li, X. Gu, and G. X. Yu, “Sentiment analysis using deep learning approaches: an overview”, Science China Information Sciences, vol. 63, 2019

  14. [20]

    Deep learning based techniques for sentiment analysis: A survey

    W. Etaiwi, D. Suleiman, and A. Awajan, “Deep learning based techniques for sentiment analysis: A survey”, Informatica, vol. 45, pp. 89 –96, 08 2021

  15. [21]

    An analysis of emotional tendency under the network public opinion: Deep learning

    J. Li, Y. Wang, and J. Wang, “An analysis of emotional tendency under the network public opinion: Deep learning”, Informatica (Slovenia), vol. 45, no. 1, 2021

  16. [22]

    Finding influential users in social networking using sentiment analysis

    A. Al -Rasheed, “Finding influential users in social networking using sentiment analysis”, Informatica, vol. 46, 03 2022

  17. [23]

    BERT: pre -training of deep bidirectional transformers for language understanding

    J. Devlin, M. Chang, K. Lee, and K. Toutanova, “BERT: pre -training of deep bidirectional transformers for language understanding”, CoRR, vol. abs/1810.04805, 2018

  18. [24]

    Roberta: A robustly optimized bert pretraining approach

    Y. Liu, M. Ott, N. Goyal, J. Du, M. Joshi, D. Chen, O. Levy, M. Lewis, L. Zettlemoyer, and V. Stoyanov, “Roberta: A robustly optimized bert pretraining approach” , ArXiv, vol. abs/1907.11692, 2019

  19. [25]

    Long short- term memory

    S. Hochreiter and J. Schmidhuber, “Long short- term memory”, Neural computation, vol. 9, no. 8, pp. 1735–1780, 1997

  20. [26]

    Recurrent convolutional neural networks for text classification

    S. Lai, L. Xu, K. Liu, and J. Zhao, “Recurrent convolutional neural networks for text classification”, in AAAI , vol. 333, 2015, pp. 2267–2273

  21. [27]

    Wordnet: An on -line lexical database

    G. A. Miller, R. Beckwith, C. Fellbaum, D. Gross, and K. Miller, “Wordnet: An on -line lexical database”, International Journal of Lexicography, vol. 3, pp. 235–244, 1990

  22. [28]

    Measuring praise and criticism: Inference of semantic orientation from association

    P. D. Turney and M. L. Littman, “Measuring praise and criticism: Inference of semantic orientation from association” CoRR, vol. cs.CL/0309034, 2003

  23. [29]

    Building a vietnamese sentiwordnet using vietnamese electronic dictionary and string kernel

    X.-S. Vu, H.-J. Song, and S.-B. Park, “Building a vietnamese sentiwordnet using vietnamese electronic dictionary and string kernel”, Lecture Notes in Computer Science, vol. 8863. Springer, 2014, pp. 223–235

Pith tools

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