Pith. sign in

REVIEW 4 major objections 5 minor 55 references

Towards Personalized Bangla Book Recommendation: A Large-Scale Heterogeneous Book Graph Dataset

T0 review · 4 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read This paper introduces RokomariBG, a large public dataset for Bangla book recommendation that combines user–item interactions, a heterogeneous knowledge graph, and review text, and benchmarks recommenders on it.

desk verdict Useful dataset resource, but internal count inconsistencies make the benchmark untrustworthy until the released artifacts are verified. read the letter →

arxiv 2602.12129 v2 pith:XAAF7F24 submitted 2026-02-12 cs.IR cs.LG

Rahin Arefin Ahmed , Md. Anik Chowdhury , Sakil Ahmed Sheikh Reza , Devnil Bhattacharjee , Muhammad Abdullah Adnan , Julian McAuley , Nafis Sadeq This is my paper
classification cs.IRcs.LG
keywords Banglabookrecommendationheterogeneousknowledgegraphe-commercedatasetlow-resourceNLPtop-Ntwo-towerretrievalcold-start
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 introduces RokomariBG, a large public dataset for personalized book recommendation in Bangla, constructed from Bangladesh's largest online bookstore. It claims to be the first Bangladeshi e-commerce dataset to combine user–book interactions, multi-entity relational knowledge (authors, publishers, categories), and natural-language review text in one graph. To show the dataset is usable, the paper benchmarks a range of recommenders and finds that a neural two-tower model with side features and relational knowledge performs best (NDCG@10 = 0.204). The authors argue this resource fills a gap: prior Bangla datasets were built for sentiment analysis and lacked the interaction and relational signals needed for recommendation. If the dataset is sound, it gives researchers a reproducible testbed for low-resource and multilingual recommendation.

What carries the argument

RokomariBG: a heterogeneous book graph with five entity types (books, authors, categories, publishers, reviews) linked by eight relation types, plus 23 side features including textual metadata and review text. The carrying mechanism for the benchmark is the Neural Two-Tower retrieval model, a dual-encoder architecture that embeds users and items via MLPs and scores relevance by dot product. The item tower concatenates item ID embeddings, pooled author/category embeddings, publisher embedding, numeric metadata, and text embeddings from a multilingual sentence transformer; the user tower combines user ID embedding with pooled interaction history. This design lets the authors ablate interaction

What would settle it

Re-crawl a random sample of the bookstore's public pages and compare the resulting entities, hyperlink-derived edges, and deduplication rate (stated as 13.1%) against the released graph; also reconcile the paper's internal count discrepancies (e.g., the sparsity table sums to 24,208 books versus the stated 127,302). If the released files do not reproduce the reported counts, the central claim collapses.

Watch

Extended reading notes

Core claim

The paper's central claim is that RokomariBG is the first publicly available dataset in the Bangladeshi e-commerce domain that jointly captures user–item interactions, rich multi-relational knowledge, and natural language content in the form of reviews and metadata. The dataset contains 127,302 books, 63,723 users, 16,601 authors, 1,515 categories, 2,757 publishers, and 209,602 reviews, connected through eight relation types and organized as a heterogeneous knowledge graph. A systematic benchmark on Top-N recommendation shows that the Neural Two-Tower model with side features achieves the strongest ranking performance (NDCG@10 = 0.204, NDCG@50 = 0.276), and ablations indicate that relational

Load-bearing premise

The dataset's value rests on the assumption that the crawl-and-link pipeline faithfully reconstructed the true bookstore graph—no silent duplication, dropping, or mislinking of users, books, reviews, or their relationships—because if that fails, all entity counts and all benchmark scores built on the dataset are unreliable.

Editorial extensions

If this is right

  • RokomariBG enables reproducible evaluation of recommendation models in Bangla, a low-resource language with over 230 million speakers.
  • Benchmark results show that combining collaborative signals with relational knowledge and textual side features outperforms interaction-only methods, with NDCG@10 rising from 0.114 (LightGCN) to 0.204 (Neural Two-Tower).
  • Ablations show relational knowledge and side features provide the largest gains in the cold-start setting, suggesting the dataset is well suited for cold-start recommendation research.
  • The polarized rating distribution (65.8% five-star, 17.4% zero-star) and long-tail user activity mean explicit rating prediction performs poorly (NDCG@10 = 0.002), so ranking-based and implicit-feedback models are more appropriate for this domain.
  • Public release of the dataset and code lets other researchers extend the benchmark to sequential, explainable, and cross-lingual recommendation tasks.

Reading between the lines

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

  • Because the dataset mixes Bangla, English, and code-mixed review text, a natural next step—not tested in the paper—is to probe whether language-aware text encoders designed specifically for Bangla outperform generic multilingual encoders on this benchmark.
  • The paper's own reported counts are not fully consistent (e.g., the sparsity table sums to 24,208 books while the stated catalog is 127,302), so a quick re-count of the released files would be a cheap validity check before building on the dataset.
  • If the graph construction is faithful, the author–category and publisher–category links could also be used for tasks beyond recommendation, such as studying the structure of the Bangladeshi publishing market or training knowledge-graph embeddings.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces RokomariBG, a heterogeneous book graph dataset crawled from Rokomari.com, containing 127,302 books, 63,723 users, 16,601 authors, 1,515 categories, 2,757 publishers, and 209,602 reviews, linked by eight relation types. The authors benchmark top-N recommendation models on this dataset and report that a Neural Two-Tower model with side features achieves the best NDCG@10 (0.204). Ablations indicate that graph relations and side features improve performance, especially for cold-start users. The dataset and code are released publicly.

Significance. If the dataset counts and released artifacts are accurate, RokomariBG is a valuable resource for Bangla and low-resource recommendation research, supplying interaction data, multi-relational structure, and multilingual review text in a combination absent from prior Bangladeshi e-commerce datasets. The paper's main empirical claim—that side and relational signals help—is supported by ablations, not by circular reasoning. However, the manuscript contains serious internal inconsistencies in core entity counts and benchmark tables, so the reliability of the dataset and the rankings cannot be assessed without correction and verification.

major comments (4)
  1. [Section 3, entity counts] The abstract and Section 3 state 403,714 total nodes, but summing the listed entities gives 127,302 + 16,601 + 1,515 + 2,757 + 209,602 + 63,723 = 421,500. The discrepancy is not explained. Since entity counts are the foundation of the dataset contribution, the node total and the per-entity counts must be made consistent and aligned with the released files.
  2. [Table 6, book interaction sparsity] Table 6 reports a Total of 24,208 books, yet the dataset claims 127,302 unique books. If Table 6 is restricted to books that appear in a filtered interaction table, that subset and its relationship to the 127,302 books must be explicitly defined. If the interaction table used in Section 5 indeed contains only ~24,208 distinct books, then the training/test splits and all benchmark results rest on a different item universe than the headline dataset count. This mismatch is load-bearing.
  3. [Table 8, Hybrid: MF + Side row] The Hybrid: MF + Side row reports Hit@5=0.281, Hit@10=0.118, Hit@50=0.204, MRR@10=0.070, NDCG@10=0.064. These values violate the monotonicity of Hit@K (Hit@10 cannot be lower than Hit@5; Hit@50 cannot be lower than Hit@5). The table appears to have misaligned or corrupted values. Because this table is the basis for the benchmarking conclusions, the entire table must be regenerated and checked against the original experimental outputs.
  4. [Appendix B, final paragraph; Table 8/Table 9] Appendix B states 'our Neural method that reached NDCG@10 of 0.2368', but Tables 8, 9, and 10 all report NDCG@10 = 0.204 for the full two-tower model. This unexplained value undermines confidence in the numerical consistency of the experimental reporting. The correct value and the experimental setting that produced it should be clarified.
minor comments (5)
  1. [Section 3.2, Table 14] Table 14 sums to 209,597 reviews, while the paper states 209,602 reviews. The 5-row difference should be reconciled or explained.
  2. [Appendix B, Table 17] The 'Books' column in Table 17 sums to 100,321, not 127,302. If page-range counts are computed on a subset, state that subset explicitly.
  3. [Section 2] The opening sentence 'Research is Bangla e-commerce domain...' contains a typo ('is' for 'in'). The paper would benefit from a careful proofread.
  4. [Appendix A, Table 11] The text refers to USER544691 and USER502420, while Table 11 uses USER44691 and USER02420. These identifiers should be aligned.
  5. [Table 8, formatting] Some rows (Pure Content-Based, Hybrid: MF + Side) contain extra trailing zeros in the metric columns, and the row alignment appears visually inconsistent. Please reformat the table to a single standard.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: dataset construction and benchmarking are empirical, with independent ablations and external baselines.

full rationale

RokomariBG is presented as a crawled, externally sourced artifact, not as the output of a derivation chain. The benchmark claims are measured on a fixed 70/15/15 split (Section 5.1), and the statement that relational knowledge and side features drive performance is supported by controlled ablations (Table 9) that remove each signal and observe degradation. No parameter is fitted to the test labels, and the model gains are not defined in terms of the dataset's headline counts. The 'first publicly available dataset' claim rests on a literature comparison (Table 1) and is checkable against prior work, not on self-citation. Citations to prior datasets or embedding models are external and not load-bearing for the central results. The internal inconsistency between the 403,714 node count in Section 3 and the 24,208-book total in Table 6, as well as other count mismatches, is a data-quality and reproducibility concern, not a circularity concern. No equation or fitted quantity in the paper reduces to its own input by construction, so no significant circularity is present.

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

No theoretical derivation is attempted; the central results are empirical benchmarks. Standard model-selection hyperparameters (Table 23) are tuned on validation and do not function as free parameters in a derivation. The assumptions listed are the domain premises the benchmark conclusions rely on.

assumptions (5)
  • domain assumption Rokomari.com's public pages are a valid and comprehensive source for Bangla book recommendation data.
    Section 3.1 constructs the dataset by crawling one online bookstore; representativeness of a single platform for the broader Bangla reading ecosystem is assumed.
  • domain assumption Reviews, including 0-rating and unverified-purchase reviews, are valid interaction signals for recommendation.
    Section 5.1 builds the interaction table from all user-review and book-review relations; if unverified or non-rating reviews are low-quality, benchmark conclusions weaken.
  • domain assumption Random 70/15/15 split with training-item masking is an adequate evaluation protocol for Top-N recommendation.
    Section 5.1 defines the split and masking, but no temporal or interaction-order split is used, so sequential effects mentioned in the abstract are not assessed.
  • domain assumption Pretrained multilingual text embeddings capture useful Bangla semantic content.
    The item tower relies on multilingual-e5-large-instruct (Section 4.2 and Table 10); performance depends on this external model's quality for Bangla.
  • domain assumption The existing Bangla datasets characterized in Table 1 are accurately represented and no prior dataset with user-item interactions plus relational structure exists.
    The novelty claim depends on the completeness and correctness of the literature survey in Table 1, which cannot be independently verified from the paper alone.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Personalized Bangla Book Recommendation: A Large-Scale Heterogeneous Book Graph Dataset." pith.science (2026). https://pith.science/paper/XAAF7F24

@misc{pith2026260212129,
  author       = {Pith},
  title        = {Pith review of: Towards Personalized Bangla Book Recommendation: A Large-Scale Heterogeneous Book Graph Dataset},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XAAF7F24}},
  note         = {Machine review of arXiv:2602.12129}
}
read the original abstract

Personalized book recommendation in Bangla literature has been constrained by the lack of structured, large-scale, and publicly available datasets. This work introduces RokomariBG, a large-scale heterogeneous book graph dataset designed to support research on personalized recommendation in a low-resource language setting. The dataset comprises 127,302 books, 63,723 users, 16,601 authors, 1,515 categories, 2,757 publishers, and 209,602 reviews, connected through several relation types and organized as a comprehensive knowledge graph. To demonstrate the utility of the dataset, we present a systematic benchmarking study on the top-N recommendation and sequential recommendation tasks, evaluating a diverse set of representative recommendation models. Through comprehensive benchmarking, we demonstrate that recommendation performance in this domain is strongly influenced by both heterogeneous relational information and code-mixed textual metadata. These findings reveal unique challenges of Bangladeshi e-commerce ecosystems that are largely absent from existing recommendation benchmarks. Overall, this work establishes a foundational benchmark and a publicly available resource for Bangla book recommendation research, enabling reproducible evaluation and future studies on recommendation in low-resource cultural domains. The dataset and code are publicly available at https://github.com/backlashblitz/Bangla-Book-Recommendation-Dataset

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

55 extracted references · 3 canonical work pages

  1. [1]

    Tuhin Akter, Manoara Begum, and Rashed Mustafa

    Mst. Tuhin Akter, Manoara Begum, and Rashed Mustafa. 2021. Bengali Sentiment Analysis of E-commerce Product Reviews using K-Nearest Neighbors. In2021 In- ternational Conference on Information and Communication Technology for Sustain- able Development (ICICT4SD). 40–44. doi:10.1109/ICICT4SD50815.2021.9396910

    arXiv 2021

  2. [2]

    Fahad Hossain, Shaon Bhatta Shuvo, and Ahmed Al Marouf

    Hasmot Ali, Md. Fahad Hossain, Shaon Bhatta Shuvo, and Ahmed Al Marouf

  3. [3]

    Bobadilla, F

    J. Bobadilla, F. Ortega, A. Hernando, and A. GutiéRrez. 2013. Recommender systems survey.Know.-Based Syst.46 (July 2013), 109–132. doi:10.1016/j.knosys. 2013.03.012

    doi:10.1016/j.knosys 2013

  4. [4]

    Hugging Face contributors. 2026. MTEB Leaderboard. https://huggingface.co/ spaces/mteb/leaderboard Accessed: 2026-01-31

    2026

  5. [5]

    Wikipedia contributors. 2026. Wikipedia. https://en.wikipedia.org/wiki/List_of_ languages_by_number_of_native_speakers Accessed: 2026-01-31

    2026

  6. [6]

    Paul Covington, Jay Adams, and Emre Sargin. 2016. Deep Neural Networks for YouTube Recommendations. InProceedings of the 10th ACM Conference on Recommender Systems, Boston, MA, USA, September 15-19, 2016, Shilad Sen, Werner Geyer, Jill Freyne, and Pablo Castells (Eds.). ACM, New York, NY, USA, 191–198. doi:10.1145/2959100.2959190

    arXiv 2016

  7. [7]

    Maxwell Harper and Joseph A

    F. Maxwell Harper and Joseph A. Konstan. 2015. The MovieLens Datasets: History and Context. 5, 4, Article 19 (Dec. 2015), 19 pages. doi:10.1145/2827872

    doi:10.1145/2827872 2015

  8. [8]

    Ruining He and Julian J. McAuley. 2016. Ups and Downs: Modeling the Vi- sual Evolution of Fashion Trends with One-Class Collaborative Filtering. In Proceedings of the 25th International Conference on World Wide Web, WWW 2016, Montreal, Canada, April 11 - 15, 2016, Jacqueline Bourdeau, Jim Hendler, Roger Nkambou, Ian Horrocks, and Ben Y. Zhao (Eds.). ACM, ...

    arXiv 2016

Show all 55 references
  1. [9]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yong-Dong Zhang, and Meng Wang. 2020. LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation. InProceedings of the 43rd International ACM SIGIR con- ference on research and development in Information Retrieva...

    2020

  2. [10]

    Iqbal Hossain, Maqsudur Rahman, Md

    Md. Iqbal Hossain, Maqsudur Rahman, Md. Tofael Ahmed, Md. Saifur Rahman, and A Z M Touhidul Islam. 2021. Rating Prediction of Product Reviews of Bangla Language using Machine Learning Algorithms. In2021 International Conference on Artificial Intelligence and Mechatronics Syste...

    2021

  3. [11]

    Jahed Hossain, Dabasish Das Joy, Sowmitra Das, and Rashed Mustafa

    Md. Jahed Hossain, Dabasish Das Joy, Sowmitra Das, and Rashed Mustafa. 2022. Sentiment Analysis on Reviews of E-commerce Sites Using Machine Learning Algorithms. In2022 International Conference on Innovations in Science, Engineering and Technology (ICISET). 522–527. doi:10.110...

    2022

  4. [12]

    Yifan Hu, Yehuda Koren, and Chris Volinsky. 2008. Collaborative Filtering for Implicit Feedback Datasets. InProceedings of the 8th IEEE International Conference on Data Mining (ICDM 2008), December 15-19, 2008, Pisa, Italy. IEEE Computer Society, 263–272. doi:10.1109/ICDM.2008.22

    2008 doi

  5. [13]

    Bell, and Chris Volinsky

    Yehuda Koren, Robert M. Bell, and Chris Volinsky. 2009. Matrix Factorization Techniques for Recommender Systems.Computer42, 8 (2009), 30–37. doi:10. 1109/MC.2009.263

    2009

  6. [14]

    Julian McAuley, Christopher Targett, Qinfeng Shi, and Anton van den Hengel

  7. [15]

    Mohammad Mukit, Md Moshiul Huq Rabbi, Mohammad Masud Ahmed, Subhenur Latif, and Sajib Saha. 2023. Sentiment Analysis on Bangla and Phonetic Bangla Reviews: A Product Rating Procedure using NLP and Machine Learning. In2023 IEEE 9th International Women in Engineering (WIE) Confe...

    2023

  8. [16]

    Mohammad Rashid, Kazi Hasan, Rakibul Hasan, Aritra Das, Mithila Sultana, and Mahamudul Hasan. 2024. A comprehensive dataset for sentiment and emotion classification from Bangladesh e-commerce reviews.Data in Brief53 (04 2024), 110052. doi:10.1016/j.dib.2024.110052

    2024

  9. [17]

    Nils Reimers and Iryna Gurevych. 2019. Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. InProceedings of the 2019 Conference on Em- pirical Methods in Natural Language Processing. Association for Computational Linguistics. http://arxiv.org/abs/1908.10084

    2019 arXiv

  10. [18]

    Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme

  11. [19]

    Francesco Ricci, Lior Rokach, and Bracha Shapira. 2011. Introduction to Recom- mender Systems Handbook. InRecommender Systems Handbook, Francesco Ricci, Lior Rokach, Bracha Shapira, and Paul B. Kantor (Eds.). Springer, Boston, MA, USA, 1–35. doi:10.1007/978-0-387-85820-3_1

    2011 doi

  12. [21]

    Golam Sarowar, Mushfiqur Rahman, Md

    Md. Golam Sarowar, Mushfiqur Rahman, Md. Nawab Yousuf Ali, and Omor Faruk Rakib. 2019. An Automated Machine Learning Approach for Sentiment Classifi- cation of Bengali E-Commerce Sites. In2019 IEEE 5th International Conference for Convergence in Technology (I2CT). 1–5. doi:10....

    2019

  13. [22]

    Badrul Sarwar, George Karypis, Joseph Konstan, and John Riedl. 2001. Item- based collaborative filtering recommendation algorithms. InProceedings of the 10th International Conference on World Wide Web(Hong Kong, Hong Kong) (WWW ’01). Association for Computing Machinery, New Yo...

    2001

  14. [23]

    Mehedi Hasan, Md

    Minhajul Abedin Shafin, Md. Mehedi Hasan, Md. Rejaul Alam, Mosaddek Ali Mithu, Arafat Ulllah Nur, and Md. Omar Faruk. 2020. Product Review Sentiment Analysis by Using NLP and Machine Learning in Bangla Language. In2020 23rd International Conference on Computer and Information ...

    2020

  15. [24]

    Mohammad Nazmush Shamael, Sabila Nawshin, Swakkhar Shatabda, and Salekul Islam. 2024. BanglishRev: A Large-Scale Bangla-English and Code-mixed Dataset of Product Reviews in E-Commerce.CoRRabs/2412.13161 (2024), 1–

    2024 arXiv

  16. [25]

    Shakib Shanto, Zishan Ahmed, Nisma Hossain, Auditi Roy, and Akinul Jony. 2023. Binary vs. Multiclass Sentiment Classification for Bangla E-commerce Product Reviews: A Comparative Analysis of Machine Learning Models.International Journal of Information Engineering and Electroni...

    2023 doi

  17. [26]

    Shakib Shanto, Zishan Ahmed, and Akinul Jony. 2023. Mining User Opinions: A Balanced Bangla Sentiment Analysis Dataset for E-Commerce.Malaysian Journal of Science and Advanced Technology(11 2023), 272–279. doi:10.56532/mjsat.v3i4. 200

    2023 doi

  18. [27]

    Nusrath Tabassum and Muhammad Ibrahim Khan. 2019. Design an Empirical Framework for Sentiment Analysis from Bangla Text using Machine Learning. In2019 International Conference on Electrical, Computer and Communication Engineering (ECCE). 1–5. doi:10.1109/ECACE.2019.8679347

    2019

  19. [28]

    arXiv:2412.13161 doi:10.48550/ARXIV.2412.13161

    doi arXiv

  20. [29]

    Mengting Wan and Julian McAuley. 2018. Item recommendation on monotonic behavior chains. InProceedings of the 12th ACM Conference on Recommender Systems(Vancouver, British Columbia, Canada)(RecSys ’18). Association for Computing Machinery, New York, NY, USA, 86–94. doi:10.1145...

    2018

  21. [30]

    Hongwei Wang, Fuzheng Zhang, Xing Xie, and Minyi Guo. 2018. DKN: Deep Knowledge-Aware Network for News Recommendation. InProceedings of the 2018 World Wide Web Conference(Lyon, France)(WWW ’18). International World Wide Web Conferences Steering Committee, Republic and Canton o...

    2018

  22. [31]

    Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2024. Multilingual E5 Text Embeddings: A Technical Report. arXiv:2402.05672 [cs.CL] https://arxiv.org/abs/2402.05672

    2024 arXiv

  23. [32]

    Shihab Uddin, Mohd Ariful Haque, Rakib Hossain Rifat, Marufa Kamal, Kishor Datta Gupta, and Roy George

    Md. Shihab Uddin, Mohd Ariful Haque, Rakib Hossain Rifat, Marufa Kamal, Kishor Datta Gupta, and Roy George. 2024. Bangla SBERT - Sentence Embedding Using Multilingual Knowledge Distillation. In2024 IEEE 15th Annual Ubiquitous Computing, Electronics & Mobile Communication Confe...

    2024

  24. [34]

    Xiang Wang, Xiangnan He, Meng Wang, Fuli Feng, and Tat-Seng Chua. 2019. Neural Graph Collaborative Filtering. InProceedings of the 42nd International Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Trovato et al. ACM SIGIR Conference on Research and Development in Info...

    2019

  25. [35]

    Ji Yang, Xinyang Yi, Derek Zhiyuan Cheng, Lichan Hong, Yang Li, Simon Xi- aoming Wang, Taibai Xu, and Ed H. Chi. 2020. Mixed Negative Sampling for Learning Two-tower Neural Networks in Recommendations. InCompanion of The 2020 Web Conference 2020, Taipei, Taiwan, April 20-24, 2...

    2020

  26. [36]

    Xiao Wang, Deyu Bo, Chuan Shi, Shaohua Fan, Yanfang Ye, and Philip S. Yu

  27. [37]

    arXiv:2011.14867 https://arxiv.org/abs/2011.14867

    A Survey on Heterogeneous Graph Embedding: Methods, Techniques, Applications and Sources.CoRRabs/2011.14867 (2020), 1–28. arXiv:2011.14867 https://arxiv.org/abs/2011.14867

    2011 arXiv

  28. [38]

    Chuxu Zhang, Dongjin Song, Chao Huang, Ananthram Swami, and Nitesh V. Chawla. 2019. Heterogeneous Graph Neural Network. InProceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2019, Anchorage, AK, USA, August 4-8, 2019, Ankur Ter...

    2019

  29. [39]

    Fuzheng Zhang, Nicholas Jing Yuan, Defu Lian, Xing Xie, and Wei-Ying Ma. 2016. Collaborative Knowledge Base Embedding for Recommender Systems. InProceed- ings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(San Francisco, California, USA)...

    2016

  30. [40]

    Sunjare Zulfiker, Ankur Chowdhury, Dip Roy, Shukdev Datta, and Sifat Mo- men. 2022. Bangla E-Commerce Sentiment Analysis Using Machine Learning Approach. In2022 4th International Conference on Sustainable Technologies for Industry 4.0 (STI). 1–5. doi:10.1109/STI56238.2022.1010...

    2022

  31. [41]

    Xinyang Yi, Ji Yang, Lichan Hong, Derek Zhiyuan Cheng, Lukasz Heldt, Aditee Kumthekar, Zhe Zhao, Li Wei, and Ed H. Chi. 2019. Sampling-bias-corrected neural modeling for large corpus item recommendations. InProceedings of the 13th ACM Conference on Recommender Systems, RecSys ...

    2019

  32. [42]

    Hamilton, and Jure Leskovec

    Rex Ying, Ruining He, Kaifeng Chen, Pong Eksombatchai, William L. Hamilton, and Jure Leskovec. 2018. Graph Convolutional Neural Networks for Web-Scale Recommender Systems. InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2...

    2018

  33. [46]

    Collection of 3 Books to Ensure Good Results in JSC & SSC and Speak English Fluently (0.999)

  34. [47]

    Stress Management (1.000)

  35. [48]

    Silent Treasure of Legend (0.999) 3.Collection of 5 Books for Primary Teacher Recruit- ment (0.998)✓

    HSC (0.998) 2. Silent Treasure of Legend (0.999) 3.Collection of 5 Books for Primary Teacher Recruit- ment (0.998)✓

  36. [49]

    Joblit (Free Cadre Book) (0.998)

  37. [50]

    Collection of 7 Books to Get More Than 70% Marks in University Admission Test (0.997) 4.First Lesson of Success (0.998)✓

  38. [51]

    Spoken English for Beginners (0.997)

    IELTS Reading in Bangla (0.996) 5. Spoken English for Beginners (0.997)

  39. [52]

    9 Books for Science Department of Class Eleven–Twelve (0.996)

  40. [53]

    Consummate English Grammar and Composition for HSC (0.996)

  41. [54]

    Road to Success (0.996)

  42. [55]

    Curse (0.996) 9.Collection of 2 Books to Strengthen English Foun- dation (0.994)✓

    Two Books by Monirul Islam Sir (0.995) 8. Curse (0.996) 9.Collection of 2 Books to Strengthen English Foun- dation (0.994)✓

  43. [56]

    Bangla First Paper (Prose, Poetry, Drama & Novel) (0.995)

  44. [57]

    Side” denotes side features (text embeddings + numeric), and “Relations

    Aviation Career (0.994) 10. Prime Textile Engineering Job Solution (0.994) ✓ denotes successfully recommended ground truth items. Values in parentheses represent model confidence scores. Both users achieved perfect hit rates, demonstrating the model’s effectiveness in capturin...

    2007

  45. [2009]

    InUAI 2009, Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence, Montreal, QC, Canada, June 18-21, 2009, Jeff A

    BPR: Bayesian Personalized Ranking from Implicit Feedback. InUAI 2009, Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence, Montreal, QC, Canada, June 18-21, 2009, Jeff A. Bilmes and Andrew Y. Ng (Eds.). AUAI Press, Corvallis, OR, USA, 452–461....

    2009

  46. [2015]

    InProceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval(Santiago, Chile)(SIGIR ’15)

    Image-Based Recommendations on Styles and Substitutes. InProceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval(Santiago, Chile)(SIGIR ’15). Association for Computing Machinery, New York, NY, USA, 43–52. doi:10.1145/2766...

  47. [2020]

    In11th International Conference on Computing, Communication and Networking Technolo- gies, ICCCNT 2020, Kharagpur, India, July 1-3, 2020

    BanglaSenti: A Dataset of Bangla Words for Sentiment Analysis. In11th International Conference on Computing, Communication and Networking Technolo- gies, ICCCNT 2020, Kharagpur, India, July 1-3, 2020. IEEE, Piscataway, NJ, USA, 1–4. doi:10.1109/ICCCNT49239.2020.9225565

    2020

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.