Pith. sign in

REVIEW 4 major objections 6 minor 37 references

TopicImpact: Improving Customer Feedback Analysis with Opinion Units for Topic Modeling and Star-Rating Prediction

T0 review · 4 major / 6 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 Clustering LLM-extracted opinion units instead of whole reviews yields more coherent topics, and splitting those units by sentiment before clustering predicts star ratings with an R2 of 0.726.

desk verdict A useful applied pipeline with an honest limitations section, but the central comparative claim is untested for lack of whole-review or LDA baselines, and the R2 result is weakened by clustering before the CV split. read the letter →

arxiv 2507.13392 v1 pith:CGU5WVF5 submitted 2025-07-16 cs.CL cs.AI

Emil Häglund , Johanna Björklund This is my paper · ORCID
classification cs.CLcs.AI
keywords opinionunitstopicmodelingaspect-basedsentimentanalysisstar-ratingpredictionLLMpreprocessingBERTopicsentiment-awareembeddingscustomerreviews
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

This paper claims that topic modeling of customer reviews improves when the clustering unit is a single opinion—a short aspect label paired with a supporting excerpt and a 1-to-10 sentiment score—rather than the full review. The authors build TopicImpact, which uses an LLM to extract these opinion units, embeds and clusters them into topics, and then regresses topic-level sentiment scores on star ratings. They report that clustering opinion units yields more coherent and interpretable topics than clustering whole reviews, and that pre-splitting opinions by sentiment before clustering gives the best star-rating prediction, with an average R2 of 0.726 across three restaurant datasets. If these results hold, businesses can read off, for each topic, its frequency, its prevailing sentiment, and its measured contribution to a business metric such as star rating.

What carries the argument

The central object is the opinion unit: a triple consisting of a short aspect label, a supporting excerpt from the review, and a sentiment score on a 1-10 scale, generated by an LLM in a preprocessing pass. The load-bearing move is to treat each opinion unit as its own document for embedding and clustering, so a multi-aspect review is distributed across several topics, and to use the unit's sentiment score both as a splitting variable (Method 3 separates positive from negative units before clustering) and as the regression feature that links each topic to the star rating.

What would settle it

Build a gold-standard set of opinion units by having independent annotators extract aspect labels, excerpts, and 1-10 sentiment scores from a random subset of the same Yelp reviews. Run the TopicImpact pipeline twice on that subset, once using LLM-extracted units and once using gold-standard units, and compare topic precision and star-rating R2; a large difference would show that extraction error, not clustering, is the limiting factor. A simpler check is to regress the LLM's opinion-level sentiment scores on the review's star rating: if the scores are nearly a deterministic function of the star rating, the regression is rediscovering the label rather than measuring per-aspect sentiment.

Watch

Extended reading notes

Core claim

TopicImpact's central claim is that restructuring the topic-modeling pipeline to operate on LLM-extracted opinion units—each a triple of an aspect label, a supporting excerpt, and a 1-10 sentiment score—produces more coherent and interpretable topic clusters than clustering entire reviews, and that these units can be connected to business metrics through regression. On three Yelp restaurant datasets, general-purpose embeddings gave average topic precision of 86-92%, while the sentiment-aware SentiCSE embedding lowered topic precision but raised sentiment precision. The best star-rating prediction came from Method 3, which splits opinion units into positive and negative groups using their sentiment scores, clusters each group separately, and feeds per-topic average sentiment scores into multiple linear regression; this achieved an average R2 of 0.726 and RMSE of 0.731. The paper presents these results as evidence that the opinion-unit structure itself—not the choice of embedding—is what delivers both interpretable topics and accurate star-rating prediction.

Load-bearing premise

The pipeline assumes that the LLM's opinion-unit extraction—the aspect labels, excerpts, and 1-10 sentiment scores—faithfully captures the true aspects and sentiments in the reviews; the paper's Section 8 concedes that extraction sometimes misses opinions, creates excerpts lacking full context, and that 'some errors persist, which can undermine the accuracy of subsequent opinion analysis.'

Editorial extensions

If this is right

  • Topic clusters built from opinion units are more interpretable than clusters of whole reviews because every cluster member is a short, aspect-specific statement carrying a quote and a sentiment score.
  • Splitting opinion units by sentiment before clustering (Method 3) outperforms both general-purpose clustering and sentiment-aware embeddings for star-rating prediction, reaching an average R2 of 0.726.
  • Because the LLM preprocessing step runs once, analysts can iterate cheaply over topic granularity, cluster counts, and seed words without reclassifying the corpus.
  • Combining cluster frequency with regression coefficients yields a priority matrix that ranks customer concerns by both how often they are voiced and how strongly they affect the business metric.
  • The same pipeline can be applied to other numeric outcomes such as employee satisfaction, purchase likelihood, or customer churn by swapping the regression target.

Reading between the lines

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

  • A test not run in the paper: on a subsample with human-annotated gold-standard opinion units, compare star-prediction R2 from clustering gold units versus LLM-extracted units; if the gap is large, the extraction errors conceded in Section 8 are the binding constraint.
  • The regression treats 1-10 sentiment scores as an interval scale; repeating the analysis with ordinal regression would show whether coefficient magnitudes but not topic rankings change.
  • Because the LLM sets the aspect taxonomy, the interpretability claim is tied to the LLM's granularity choices; a user study comparing the LLM's splits to a fixed human taxonomy would test this in other domains.
  • Cross-domain transfer is unproven: the paper notes other domains bring longer opinions, more nuance, and more non-opinion content, so prompt customization will likely be needed for retail product reviews or employee surveys.
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 / 6 minor

Summary. The paper proposes TopicImpact, a topic-modeling pipeline that first uses an LLM to decompose customer reviews into opinion units (aspect label, supporting excerpt, and a 1-10 sentiment score), then embeds and clusters these units with BERTopic, and finally uses multiple linear regression on topic memberships and sentiment scores to predict star ratings. The authors evaluate two embedding models (all-mpnet-base-v2 and SentiCSE) and three integration methods (M1: general embeddings; M2: sentiment-aware embeddings; M3: sentiment-based splitting before clustering) on Yelp restaurant reviews filtered by cuisine. They report high human-evaluated topic precision (86.3--91.7% for the general embedding) and an R2 of 0.726 for star-rating prediction with Method 3.

Significance. The work addresses a practical and timely problem: making topic models more interpretable and more aligned with downstream business metrics by leveraging LLM-extracted aspect-sentiment units. The manuscript has several strengths: it uses authentic full-length reviews rather than artificially short benchmark entries; it includes a human evaluation of cluster coherence; it compares general-purpose and sentiment-aware embeddings; and it provides detailed per-dataset and per-K results in the appendix. If the central claims were properly supported, the system would be a useful addition to applied aspect-based sentiment analysis. However, the claims as currently stated are not adequately tested: no comparison is made against clustering whole reviews or against classical topic models, and the evaluation design contains a cross-validation leakage issue and a circularity in the sentiment-precision metric. These issues are load-bearing for the paper's main contribution.

major comments (4)
  1. [Section 5.2 / Section 6.1] The central claim that clustering opinion units 'instead of entire reviews' yields more coherent and interpretable topics is not tested. All three evaluated methods (M1, M2, M3) cluster opinion units; there is no baseline that clusters whole reviews with the same embedding pipeline, nor an LDA baseline, nor a BERTopic-on-whole-reviews baseline. Without such a comparison, the reported topic precision values (Table 1) cannot be attributed to the opinion-unit representation. The limitation section mentions the lack of a benchmark dataset, but that does not substitute for an internal baseline against whole-review clustering on the same Yelp data.
  2. [Section 5.4 / Section 6.2] The cross-validation procedure is vulnerable to information leakage. Section 5.2 states that the topic model is fit on the full dataset (including clustering and dimensionality reduction), and Section 5.4 states that prediction performance is then evaluated with 5-fold cross-validation on a hold-out sample. Because clustering is performed before the data is split, the topic assignment of a test review depends on the test review itself through the UMAP/HDBSCAN fit. This can optimistically bias the R2 and RMSE values in Tables 2 and 4--6. The regression should be evaluated with clustering performed inside each training fold (nested or repeated per-fold clustering).
  3. [Section 5.3 / Table 1] The sentiment-precision metric in Table 1 is not an external validation. It is computed as the percentage of opinion units in a cluster whose LLM-assigned sentiment score falls on the dominant side of the positive/negative threshold. These are the same LLM scores that were used to construct the opinion units in Step A and to split the data in Method 3. The metric therefore measures the self-consistency of the LLM's own scores rather than the quality of the discovered clusters. Independent sentiment labels, human annotation of sentiment, or a separately trained sentiment classifier are needed to support the claim that the clusters are coherent with respect to sentiment.
  4. [Section 6.2 / Table 2] The R2 gain attributed to Method 3 may be largely explained by the LLM sentiment scores themselves, rather than by topic modeling. Since the opinion-unit sentiment scores are derived from the same text that determines the star rating, a simple baseline that regresses star ratings on the average (or per-aspect) LLM sentiment scores, without any topic modeling, should be reported. Without such a baseline, the R2 value of 0.726 cannot be interpreted as evidence that the topic clusters add predictive value beyond the sentiment scores. The high R2 values for M3 'without' sentiment scores (0.641--0.678 in Tables 4--6) further suggest that the sentiment split itself, not the topics, carries most of the predictive signal.
minor comments (6)
  1. [Table 2 vs Appendix A3] Table 2 reports R2 = 0.383 for M1 with sentiment scores at K = 20, but the per-dataset values in Appendix A3 are 0.380 (Italian), 0.364 (Mexican), and 0.326 (Japanese), whose average is 0.357. The discrepancy should be reconciled or explained.
  2. [Appendix A3, Table 4] In the Italian dataset row for M3 with K = 20 and sentiment scores, the RMSE value is printed as '712' and should presumably be '0.712'.
  3. [References] The same LDA paper is listed twice as Blei et al. 2003a and 2003b; these duplicate entries should be merged.
  4. [Section 1] There is a typo in the sentence 'by applying regression to the the combination of topics and sentiments' — 'the the' should be 'the'.
  5. [Section 5.4] The sentence 'It the latter case, we let sk(r) = 1...' should read 'In the latter case'.
  6. [Throughout] The embedding model name is inconsistently capitalized as 'SentiCSE' in Section 5.2 and as 'sentiCSE' in Section 6.1 and Table 1; please standardize.

Circularity Check

1 steps flagged · score 3.0 of 10

Sentiment coherence metric is self-referential, but topic coherence and star-rating prediction rest on independent human/external labels.

  1. self definitional [Section 5.3 (Evaluation of Topic Modeling) and Section 6.1 (Topic and Sentiment Coherence)]
    "Sentiment precision is calculated based on the distribution of LLM-assigned sentiment scores. We define it as the percentage of opinions with the dominant sentiment (positive or negative). For example, if 18/20 opinion units in a cluster have a sentiment score > 5 (positive), the precision is 90%. The same applies if 18/20 opinion units are negative (sentiment score ≤ 5)."

    The metric used to answer RQ1's sentiment-coherence question is defined directly from the LLM's own 1-10 sentiment scores, which are produced in preprocessing Step A and are an input to the pipeline. Reporting that clusters have high sentiment precision is therefore only a statement that the clusters separate the LLM's labels; it does not validate that the extracted topics capture true sentiment. Any method that splits opinion units at the same >5 / <=5 threshold used to define positive/negative (as M3 does) would attain near-perfect sentiment precision by construction. The 'sentiment coherence' result is thus a renaming of the input label distribution rather than an independent derivation.

full rationale

The paper's principal empirical claims are not circular: topic coherence is assessed by human evaluators (Section 5.3, Appendix A2) rather than by the model's own labels, and star-rating prediction is evaluated against the external Yelp star ratings using 5-fold cross-validation of the regression (Section 5.4). The sentiment-splitting method's R2=0.726 is therefore a genuine out-of-sample regression result with respect to the rating outcome, not a quantity forced by definition. One evaluation step is self-referential: the 'sentiment precision' metric in Section 5.3 is defined as agreement with the LLM's own 1-10 sentiment scores, which are an input produced in Step A. Consequently, the RQ1 claim about sentiment coherence only measures how well clusters separate the LLM's labels, not true sentiment; M3's split at the same >5/<=5 threshold would make such precision high by construction. This is a secondary evaluation circularity, not a circular derivation of the central topic-coherence or star-prediction results. The paper also relies on the authors' prior work (Häglund and Björklund, 2025) for opinion-unit extraction accuracy, but that is separate published evidence rather than a deduction from the present results. Reported limitations about missing standardized benchmarks, residual extraction errors, and the absence of LDA/whole-review baselines are relevant to correctness and generalizability but are not themselves circularity. Overall score 3.

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

The main claims rest on five assumptions listed above and on five evaluation-design parameters. No new physical or conceptual entities are introduced; opinion units are reused from the authors' earlier work. The free parameters are all choices about experimental setup, not constants derived from theory, so the paper's contribution is empirical and pipeline-level rather than theoretical.

free parameters (5)
  • Number of topic clusters K = 20 for main analysis; 10, 30, 40 in appendix
    Chosen by authors for manageable human evaluation; directly determines cluster granularity and regression feature count.
  • Minimum topic size = 50 opinion units
    Set to provide sufficient data for regression significance; affects how many clusters are formed and which units are outliers.
  • Sentiment split threshold = 5 (negative <= 5, positive > 5)
    Arbitrary cut used in Method 3; changes how sentiment and topic information are separated and weighted in regression.
  • Review subsample size per cuisine = 5000 reviews
    Subsampled from the Yelp dataset with no specified random seed or stratification procedure; affects all statistics.
  • Outlier policy = outliers excluded from clusters
    17-32% of opinion units are left unclustered; these units contribute zero topic features in the regression, and the impact is not tested.
assumptions (5)
  • domain assumption LLM-generated opinion units faithfully represent the aspects, excerpts, and sentiment polarities in the reviews.
    The pipeline and all conclusions rest on extraction quality; Section 8 admits extraction errors, and fidelity is delegated to prior work (Haglund and Bjorklund, 2025).
  • domain assumption Human inclusion-based precision on 20 sampled opinion units per topic is a valid measure of topic coherence.
    Section 5.3 and Appendix A2; one of the three raters is an author, only 5 of 20 units overlap for inter-rater agreement, and no full annotation data are released.
  • standard math Multiple linear regression is an acceptable model for ordinal star ratings with these topic-sentiment features.
    Section 3 explicitly acknowledges the ordinal-scale criticism but adopts MLR as standard; linearity, homoscedasticity, and multicollinearity are not checked.
  • domain assumption BERTopic with HDBSCAN and UMAP, using only the two selected embedding models, represents the relevant space of neural topic modeling choices.
    Section 5.2; Section 8 notes that only two embedding models were compared, so conclusions may not generalize.
  • domain assumption Outlier opinion units (17-32% of units, not assigned to any cluster) can be excluded without materially biasing topic or regression conclusions.
    Section 6.1 allows outliers; their content and influence on R2 and coefficients are not analyzed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TopicImpact: Improving Customer Feedback Analysis with Opinion Units for Topic Modeling and Star-Rating Prediction." pith.science (2026). https://pith.science/paper/CGU5WVF5

@misc{pith2026250713392,
  author       = {Pith},
  title        = {Pith review of: TopicImpact: Improving Customer Feedback Analysis with Opinion Units for Topic Modeling and Star-Rating Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CGU5WVF5}},
  note         = {Machine review of arXiv:2507.13392}
}
read the original abstract

We improve the extraction of insights from customer reviews by restructuring the topic modelling pipeline to operate on opinion units - distinct statements that include relevant text excerpts and associated sentiment scores. Prior work has demonstrated that such units can be reliably extracted using large language models. The result is a heightened performance of the subsequent topic modeling, leading to coherent and interpretable topics while also capturing the sentiment associated with each topic. By correlating the topics and sentiments with business metrics, such as star ratings, we can gain insights on how specific customer concerns impact business outcomes. We present our system's implementation, use cases, and advantages over other topic modeling and classification solutions. We also evaluate its effectiveness in creating coherent topics and assess methods for integrating topic and sentiment modalities for accurate star-rating prediction.

Figures

Figures reproduced from arXiv: 2507.13392 by the authors.

Figure 1
Figure 1. Overview of TopicImpact. A: LLM processes reviews to opinion units: label: excerpt {sentiment 1-10}. B: Embed and cluster opinion units through topic modeling. C: Regress topics and sentiments onto star ratings. D: Analyze sentiment, frequency, and topic impact. distributions (Blei et al., 2003a). A core principle of LDA is that documents are modeled as mixtures of topics, meaning each document is assumed to cover m… view at source ↗
Figure 2
Figure 2. Prompt template for generating opinion units. [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 32 canonical work pages

  1. [1]

    Markus Binder, Bernd Heinrich, Mathias Klier, Andreas Alexander Obermeier, and Alexander Schiller. 2019. Explaining the stars: Aspect-based sentiment analysis of online customer reviews. In Proceedings of the 27th European Conference on Information Systems (ECIS), Stockholm & Uppsala, Sweden. Research Papers

    work page 2019

  2. [2]

    David Blei, Andrew Ng, and Michael Jordan. 2003 a . https://doi.org/10.1162/jmlr.2003.3.4-5.993 Latent dirichlet allocation . Journal of Machine Learning Research, 3:993--1022

    doi:10.1162/jmlr.2003.3.4-5.993 2003

  3. [3]

    David M Blei, Andrew Y Ng, and Michael I Jordan. 2003 b . Latent dirichlet allocation. Journal of machine Learning research, 3(Jan):993--1022

    work page 2003

  4. [4]

    Ricardo JGB Campello, Davoud Moulavi, and J \"o rg Sander. 2013. Density-based clustering based on hierarchical density estimates. In Pacific-Asia conference on knowledge discovery and data mining, pages 160--172. Springer

    2013

  5. [5]

    Rob Churchill and Lisa Singh. 2022. The evolution of topic modeling. ACM Computing Surveys, 54(10s):1--35

    work page 2022

  6. [6]

    Stefan Debortoli, Oliver M \"u ller, Iris Junglas, and Jan Vom Brocke. 2016. Text mining for information systems researchers: An annotated topic modeling tutorial. Communications of the Association for Information Systems (CAIS), 39(1):7

    work page 2016

  7. [7]

    Anton Eklund and Mona Forsman. 2022. https://doi.org/10.18653/v1/2022.emnlp-industry.65 Topic modeling by clustering language model embeddings: Human validation on an industry dataset . In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 635--643, Abu Dhabi, UAE. Association for Computational Li...

    work page doi:10.18653/v1/2022.emnlp-industry.65 2022

  8. [8]

    Anton Eklund, Mona Forsman, and Frank Drewes. 2024. https://doi.org/10.18653/v1/2024.nlp4dh-1.52 CIPHE : A framework for document cluster interpretation and precision from human exploration . In Proceedings of the 4th International Conference on Natural Language Processing for Digital Humanities, pages 536--548, Miami, USA. Association for Computational L...

    work page doi:10.18653/v1/2024.nlp4dh-1.52 2024

Show all 37 references
  1. [9]

    Shuai Fan, Chen Lin, Haonan Li, Zhenghao Lin, Jinsong Su, Hang Zhang, Yeyun Gong, JIan Guo, and Nan Duan. 2022. https://doi.org/10.18653/v1/2022.emnlp-main.332 Sentiment-aware word and sentence level pre-training for sentiment analysis . In Proceedings of the 2022 Conference o...

    2022 doi

  2. [10]

    Bin Fu, Jialiu Lin, Lei Li, Christos Faloutsos, Jason Hong, and Norman Sadeh. 2013. Why people hate your app: Making sense of user feedback in a mobile app store. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 1276--1284

    2013

  3. [11]

    Gayatree Ganu, Yogesh Kakodkar, and Am \'e Lie Marian. 2013. Improving the quality of predictions using textual information in online user reviews. Information Systems, 38(1):1--15

    2013

  4. [12]

    Lei Gao, Yulong Wang, Tongcun Liu, Jingyu Wang, Lei Zhang, and Jianxin Liao. 2021. Question-driven span labeling model for aspect--opinion pair extraction. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 12875--12883

    2021

  5. [13]

    Vahid Ghafouri, Jose Such, and Guillermo Suarez-Tangil. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.1171 I love pineapple on pizza != I hate pineapple on pizza: Stance-aware sentence transformers for opinion mining . In Proceedings of the 2024 Conference on Empirical Met...

    2024 doi

  6. [14]

    Rachel A Gordon. 2015. Regression analysis for the social sciences. Routledge

    2015

  7. [16]

    Maarten Grootendorst. 2022 b . https://arxiv.org/abs/2203.05794 Bertopic: Neural topic modeling with a class-based tf-idf procedure . Preprint, arXiv:2203.05794

    2022 arXiv

  8. [17]

    Nan Hu, Ting Zhang, Baojun Gao, and Indranil Bose. 2019. What do hotel customers complain about? text analysis using structural topic model. Tourism Management, 72:417--426

    2019

  9. [18]

    Emil Häglund and Johanna Björklund. 2025. Opinion units: Concise and contextualized representations for aspect-based sentiment analysis. In Proceedings of the Joint 25th Nordic Conference on Computational Linguistics and 11th Baltic Conference on Human Language Technologies

    2025

  10. [19]

    Jaemin Kim, Yohan Na, Kangmin Kim, Sang-Rak Lee, and Dong-Kyu Chae. 2024. https://aclanthology.org/2024.lrec-main.1280/ S enti CSE : A sentiment-aware contrastive sentence embedding framework with sentiment-guided textual similarity . In Proceedings of the 2024 Joint Internati...

    2024

  11. [20]

    Anusuya Krishnan. 2023. Exploring the power of topic modeling techniques in analyzing customer reviews: a comparative analysis. arXiv preprint arXiv:2308.11520

    2023 arXiv

  12. [21]

    Xin Li and Wai Lam. 2017. Deep multi-task learning for aspect term extraction with memory interaction. In Proceedings of the 2017 conference on empirical methods in natural language processing, pages 2886--2892

    2017

  13. [22]

    Zongxia Li, Lorena Calvo-Bartolomé, Alexander Hoyle, Paiheng Xu, Alden Dima, Juan Francisco Fung, and Jordan Boyd-Graber. 2025. https://arxiv.org/abs/2502.14748 Large language models struggle to describe the haystack without human help: Human-in-the-loop evaluation of llms . P...

    2025 arXiv

  14. [23]

    Jack Linshi. 2014. Personalizing yelp star ratings: a semantic topic modeling approach. Yale University

    2014

  15. [24]

    Pengfei Liu, Shafiq Joty, and Helen Meng. 2015. Fine-grained opinion mining with recurrent neural networks and word embeddings. In Proceedings of the 2015 conference on empirical methods in natural language processing, pages 1433--1443

    2015

  16. [25]

    Ling Luo, Xiang Ao, Yan Song, Jinyao Li, Xiaopeng Yang, Qing He, and Dong Yu. 2019. Unsupervised neural aspect extraction with sememes. In Proceedings of the 28th International Joint Conference on Artificial Intelligence, pages 5123--5129

    2019

  17. [26]

    Nikolaos Pappas and Andrei Popescu-Belis. 2014. Explaining the stars: Weighted multiple-instance learning for aspect-based sentiment analysis. In Proceedings of the 2014 Conference on Empirical Methods In Natural Language Processing (EMNLP), pages 455--466

    2014

  18. [27]

    Haiyun Peng, Lu Xu, Lidong Bing, Fei Huang, Wei Lu, and Luo Si. 2020. Knowing what, how and why: A near complete solution for aspect-based sentiment analysis. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pages 8600--8607

    2020

  19. [28]

    Chau Minh Pham, Alexander Miserlis Hoyle, Simeng Sun, and Mohit Iyyer. 2024. https://api.semanticscholar.org/CorpusID:264935207 Topicgpt: A prompt-based topic modeling framework . In North American Chapter of the Association for Computational Linguistics

    2024

  20. [29]

    Maria Pontiki, Dimitris Galanis, Haris Papageorgiou, Ion Androutsopoulos, Suresh Manandhar, Mohammed AL-Smadi, Mahmoud Al-Ayyoub, Yanyan Zhao, Bing Qin, Orph \'e e De Clercq, et al. 2016. Semeval-2016 task 5: Aspect based sentiment analysis. In ProWorkshop on Semantic Evaluati...

    2016

  21. [30]

    Tijana Radojevic, Nemanja Stanisic, and Nenad Stanic. 2017. Inside the rating scores: A multilevel analysis of the factors influencing customer satisfaction in the hotel industry. Cornell Hospitality Quarterly, 58(2):134--164

    2017

  22. [31]

    Nigel Slack. 1994. The importance-performance matrix as a determinant of improvementpriority. International Journal of Operations & Production Management, 14(5):59--75

    1994

  23. [32]

    Duyu Tang, Furu Wei, Bing Qin, Nan Yang, Ting Liu, and M. Zhou. 2016. https://api.semanticscholar.org/CorpusID:18133407 Sentiment embeddings with applications to sentiment analysis . IEEE Transactions on Knowledge and Data Engineering, 28:496--509

    2016

  24. [33]

    Asin Tavakoli, Holger Harreis, Kayvaun Rowshankish, and Michael Bogobowicz. 2024. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/charting-a-path-to-the-data-and-ai-driven-enterprise-of-2030?utm_source=chatgpt.com Charting a path to the data and ai-driven e...

    2024

  25. [34]

    Sentence Transformers. 2024. Pretrained models. https://www.sbert.net/docs/sentence_transformer/pretrained_models.html. Accessed: 2024-04-20

    2024

  26. [35]

    Yelp. 2015. https://www.yelp.com/dataset Yelp open dataset . Dataset

    2015

  27. [36]

    Robert Lai, and Xuejie Zhang

    Liang-Chih Yu, Jin Wang, K. Robert Lai, and Xuejie Zhang. 2017. https://doi.org/10.18653/v1/D17-1056 Refining word embeddings for sentiment analysis . In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 534--539, Copenhagen, Denmark...

    2017 doi

  28. [37]

    Wenxuan Zhang, Xin Li, Yang Deng, Lidong Bing, and Wai Lam. 2022. A survey on aspect-based sentiment analysis: Tasks, methods, and challenges. IEEE Transactions on Knowledge and Data Engineering

    2022

  29. [38]

    Xinjie Zhou, Xiaojun Wan, and Jianguo Xiao. 2015. Representation learning for aspect category detection in online reviews. In Proceedings of the AAAI conference on artificial intelligence, volume 29

    2015

Pith tools

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