Pith. sign in

REVIEW 4 major objections 4 minor 93 references

MATCHED: Multimodal Authorship-Attribution To Combat Human Trafficking in Escort-Advertisement Data

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that combining ad text and images identifies the vendor behind an escort ad more accurately than text or images alone, with the multimodal model reaching Macro-F1 0.9802 on the South region dataset.

desk verdict Useful new dataset and benchmark; the headline multimodal-over-text gain is likely inflated by sample duplication and an unclear split unit, so the specific numbers need a matched-sample control before they are credible. read the letter →

arxiv 2412.13794 v1 pith:H6K3DKY2 submitted 2024-12-18 cs.CL cs.AIcs.CY

Vageesh Saxena , Benjamin Bashpole , Gijs Van Dijck , Gerasimos Spanakis This is my paper · ORCID
classification cs.CLcs.AIcs.CY
keywords multimodalauthorshipattributionhumantraffickingescortadvertisementsvendoridentificationverificationsupervisedcontrastivelearningDeCLUTR-ViTBackpagedataset
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 authorship attribution on escort ads—linking ads to the same vendor—is substantially improved when both text and images are used rather than text alone. To do this it introduces MATCHED, a dataset of 27,619 unique text descriptions and 55,115 unique images from Backpage escort ads across seven U.S. cities, and labels vendors by grouping ads that share extracted phone numbers. The headline result is that an end-to-end multimodal model, DeCLUTR-ViT trained with a cross-entropy plus supervised-contrastive objective, reaches Macro-F1 0.9802 on vendor identification, far above the text-only 0.6540 and vision-only 0.6378 baselines. The broader claim is that images add stylistic cues text alone misses, and that end-to-end multimodal training beats pre-trained text-image alignment methods such as CLIP and BLIP2 for this domain. If the paper is right, law-enforcement analysts can link escort ads and map trafficking networks even when explicit identifiers are masked and vendors post few ads.

What carries the argument

The load-bearing object is the DeCLUTR-ViT backbone: the DeCLUTR-small text encoder combined with the ViT-base-patch16 image encoder, fused by mean pooling and trained end-to-end with a joint objective that combines cross-entropy loss for classification with supervised contrastive loss for similarity learning. The dataset construction also does work: each text description is duplicated once per associated image, expanding the per-vendor training signal, and vendor labels are created by extracting phone numbers from ad text and grouping communities of ads with NetworkX. The argument's force comes from the comparison between this end-to-end joint training and the pre-trained text-image alignment strategies, which the paper shows underperform because escort-ad images and text have low semantic overlap.

What would settle it

Take a random sample of ads from the South region, strip all phone numbers and explicit identifiers, and ask two independent annotators to cluster them by textual and visual style; compare those clusters with the phone-number communities and with the model's predictions. If the model tracks the phone-number labels almost perfectly while the style-based clusters diverge, the central authorship-attribution claim is not supported.

Watch

Extended reading notes

Core claim

The central discovery is that multimodal authorship attribution, pairing the text of an escort ad with the images attached to it, identifies the vendor who posted the ad more reliably than either modality alone. On the South region dataset the multimodal DeCLUTR-ViT backbone with mean pooling and the joint CE+SupCon objective reaches Macro-F1 0.9802, compared with 0.6540 for the best text-only model and 0.6378 for the best vision-only model. The same backbone also outperforms a BLIP2-aligned multimodal variant (0.9420). Text remains the dominant modality, but visual features complement it, and the multimodal model's text representations are the strongest retrieval embeddings; the paper interprets this as evidence that images supply stylistic cues that enrich the learned vendor profile.

Load-bearing premise

The ground-truth vendor labels are created by extracting phone numbers from ad text and grouping ads into communities, and the paper assumes each community is a single vendor; if multiple vendors share a phone number or one vendor posts under many numbers, the reported performance numbers do not cleanly measure authorship attribution.

Editorial extensions

If this is right

  • Vendors with very few text ads become tractable: pairing each text ad with its images multiplies the available training samples and improves attribution for low-frequency vendors.
  • One trained model can serve both closed-set vendor identification, finding known vendors, and open-set vendor verification, linking emerging vendors across out-of-distribution ads, because the joint objective trains for both.
  • The out-of-distribution retrieval results across the Midwest, West, and Northeast datasets indicate the representations transfer across cities and regions, including to vendors never seen in training.
  • CLIP-style and BLIP2-style text-image alignment is not the right recipe for escort-ad data; future multimodal authorship-attribution systems should invest in end-to-end multitask training instead.

Reading between the lines

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

  • Because the vendor labels are derived from phone-number communities, the superb Macro-F1 may partly reflect the model learning surface patterns that correlate with phone-number groups rather than true authorship; independently annotated style clusters would separate those two explanations.
  • The same recipe—two unimodal encoders, mean pooling, and a CE+SupCon objective—should transfer to other paired text-plus-image marketplaces such as darknet vendor shops or fraud listings, where identity cues are split across modalities.
  • The five-month collection window cannot test whether vendors evade attribution by deliberately shifting their writing style over time; recollecting ads from a later period would be a natural stress test of the approach.
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 / 4 minor

Summary. The paper introduces MATCHED, a multimodal dataset of Backpage escort advertisements (27,619 unique text descriptions and 55,115 images linked to 3,549 phone-number-derived vendor labels across four U.S. regions), and benchmarks text-only, vision-only, and multimodal models on vendor identification (closed-set classification) and vendor verification (open-set retrieval) tasks. The central claim is that integrating text and images via an end-to-end DeCLUTR-ViT backbone trained with a joint cross-entropy and supervised-contrastive objective substantially improves authorship-attribution performance, with the headline result being Macro-F1 0.9802 on the South region compared with 0.6540 for text-only and 0.6378 for vision-only baselines (Table 2). The paper also reports that pre-trained text-image alignment methods such as CLIP and BLIP2 underperform end-to-end multimodal training, and it provides qualitative knowledge-graph examples intended for law-enforcement use.

Significance. If the central claim holds, the work would be a meaningful step for multimodal authorship attribution in a sensitive domain, with practical value for human-trafficking investigations. The strengths are the construction of a new restricted-access multimodal dataset, a thorough datasheet, extensive benchmarking across twelve model variants and four fusion strategies, and explicit attention to privacy and ethics. The paper also ships code and a datasheet, which aids reproducibility. However, the significance is conditional on resolving the experimental confounds described below: the reported multimodal advantage is not currently a clean measure of modality fusion, and the label-generation procedure may introduce noise that weakens the attribution claim.

major comments (4)
  1. [Appendix A.2 and Section 1 contribution (iii)] The claimed multimodal advantage is confounded with a roughly twofold increase in training samples. Appendix A.2 states: "Since ads may contain multiple images, we duplicate the text sequence for each associated image to prepare the dataset for multimodal training." Thus the DeCLUTR-ViT model is trained on about 27,423 (text, image) pairs in South, whereas the text-only baseline sees 13,661 unique texts and the vision-only baseline sees 27,423 images. Table 2's comparison (0.9802 vs 0.6540 and 0.6378) therefore reflects both modality fusion and dataset-size / text-augmentation effects. The paper even lists this sample expansion as a benefit (Section 1, contribution iii), but this makes it impossible to attribute the gain to "integrating multimodal features." A matched control is needed, e.g., a text-only model trained on the same duplicated text sequences, or a multimodal model trained with one image per ad, to isolate the effect of cross-modal fusion.
  2. [Appendix A.4 and Section 4(iv)] The split unit is unspecified, creating a potential train/test leakage that can inflate Macro-F1. Appendix A.4 gives only a 0.75/0.05/0.20 ratio with seed 1111, and Section 4(iv) defines the split as "training ("documents") and test ("queries") sets" without stating whether the unit is an ad, a unique text description, a (text, image) pair, or a vendor community. If the split is performed at the (text, image) pair level, the same text description can appear in both training and test splits paired with different images, allowing the model to memorize text-level patterns and report artificially high numbers. The paper should specify the split unit, and ideally perform a text-overlap audit (e.g., percent of test texts that appear in training) and report results under an ad-level or vendor-level split to confirm the headline numbers.
  3. [Section 3 and Appendix A.3.2] The ground-truth labels are derived from phone-number communities using NetworkX, under the assumption that each community is a single vendor. Section 8 explicitly acknowledges that "our qualitative analysis identifies cases where the trained classifier misclassifies ads, likely due to similarities in writing style and content, suggesting the possibility that multiple vendors might belong to the same entity" and that "we lack definitive ground truth to confirm this hypothesis." This label noise is not just a performance ceiling — it directly affects whether the task being benchmarked is authorship attribution at all, since ads from distinct authors may share a label and ads from one vendor may be split across labels. The paper should at least quantify label noise (e.g., via manual audit of a sample of communities) or provide a robustness analysis showing that the relative ordering of text-only, vision-only, and multimodal methods is stable under label perturbations.
  4. [Section 3, Figure 2a, and Table 12] The out-of-distribution generalization claim is weakened by the substantial vendor overlap across regions. Section 3 notes that "many vendors in our dataset appear across multiple geographic regions," and Appendix A.2 acknowledges that the OOD experiments "may not fully capture real-world conditions." While Table 12 reports shared-versus-unique vendor results for the main multimodal model, the OOD averages in Tables 5-11 and Figure 1 pool shared and unique vendors, so the headline OOD numbers partly measure re-identification of already-seen vendors rather than generalization to never-before-seen authors. The OOD claims should be re-reported separately for vendors unique to each region across all baselines, not only for the multimodal model.
minor comments (4)
  1. [Section 4(iii)(C)] The notation "ViT-base-patch16-244" is used in Section 4 but the appendix sometimes writes "ViT-base-patch16-224" (e.g., Figure 4 caption); please standardize the image resolution notation.
  2. [Table 4] The table header says "Param Loss Fusion Epochs" but the Style-Embedding row and some vision rows do not clearly indicate the loss used; for readability, place the loss in a dedicated column without merging into the model name.
  3. [Section 5(i)] The sentence "The multimodal DeCLUTR-ViT backbone, trained end-to-end with mean pooling as the fusion technique, achieves the highest macro-F1 score (0.9670)" appears before the CE+SupCon result (0.9802) is discussed; consider reordering so the best configuration is stated first consistently.
  4. [Appendix A.1.5] The AI-assistance disclosure says ChatGPT was used for paraphrasing and code comments; please ensure the final camera-ready version also states this in the main text or a dedicated statement, as some venues require the disclosure to appear in the paper itself.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the multimodal authorship-attribution results are empirical benchmark measurements, vendor labels come from external phone-number extraction, and no claimed prediction reduces by construction to the paper's inputs.

full rationale

The paper's derivation chain is empirical benchmarking rather than a derivation that reduces to its inputs. Vendor labels are constructed by extracting phone numbers from ad text using TJBatchExtractor, the CNN-LSTM-CRF classifier of Chambers et al. (2019), and NetworkX community detection, following Saxena et al. (2023a); this ground truth is external to the neural models being evaluated and is not produced by the model or by the paper's equations. The headline Macro-F1 scores in Tables 2 and 4 are measured on held-out test data, and no parameter is fitted to the multimodal test set and then renamed as a prediction. The self-citations to Saxena et al. (2023a) for label construction, masking, and text-baseline choices are not load-bearing for the multimodal claim, because the cited phone-number grouping is independently reproducible and externally falsifiable. The acknowledged duplication of text sequences per image (Appendix A.2) and the possible overlap of shared vendors into the OOD sets (Section 3, Appendix A.2) are experimental confounds that affect the interpretation of the multimodal advantage, but they are not circularity: the multimodal model's outcome is not forced by definition or by a self-citation chain. Section 8's caveat that phone-number communities may merge multiple vendors is a label-noise limitation, not a case of the target being built into the input. No uniqueness theorem or ansatz is imported from the authors' prior work to foreclose alternatives, so the central claim has independent empirical content.

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

The central claim rests on the assumption that phone-number-derived communities are accurate vendor labels, and on the standard transfer-learning premise that pre-trained backbones can be fine-tuned. The free parameters are standard hyperparameters and do not by themselves constitute a derivation. No new entities are postulated.

free parameters (4)
  • Learning rate = 0.001
    Selected by trying 0.01, 0.001, 0.0001; best on validation. Not load-bearing for the central claim.
  • Batch size = 32
    Maximum feasible on available GPU; fixed for all baselines for fairness.
  • Temperature for NT-XENT loss = 0.1
    Chosen among 0.01, 0.1, 0.3 for the image-text alignment pre-training.
  • Number of in-batch negatives = 5
    Used for contrastive objectives; increasing did not improve performance.
assumptions (4)
  • domain assumption Phone numbers extracted by TJBatchExtractor and CNN-LSTM-CRF (Chambers et al., 2019) reliably identify the posting vendor.
    Ground-truth vendor labels are built from phone-number communities (Section 3, Appendix A.3.2).
  • domain assumption Ads sharing a phone number are authored by the same vendor or entity.
    This is the core labeling assumption, acknowledged as imperfect in Section 8.
  • standard math Pre-trained DeCLUTR-small and ViT-base-patch16 checkpoints provide suitable initializations for this domain.
    Standard transfer-learning assumption referenced in Sections 4 and 5.
  • domain assumption The 0.75/0.05/0.20 split is performed in a way that prevents text or image duplication across train and test.
    The paper does not specify whether the split is at the ad level or at the duplicated text-image pair level; if pair-level, leakage occurs. This is an unflagged assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MATCHED: Multimodal Authorship-Attribution To Combat Human Trafficking in Escort-Advertisement Data." pith.science (2026). https://pith.science/paper/H6K3DKY2

@misc{pith2026241213794,
  author       = {Pith},
  title        = {Pith review of: MATCHED: Multimodal Authorship-Attribution To Combat Human Trafficking in Escort-Advertisement Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H6K3DKY2}},
  note         = {Machine review of arXiv:2412.13794}
}
read the original abstract

Human trafficking (HT) remains a critical issue, with traffickers increasingly leveraging online escort advertisements (ads) to advertise victims anonymously. Existing detection methods, including Authorship Attribution (AA), often center on text-based analyses and neglect the multimodal nature of online escort ads, which typically pair text with images. To address this gap, we introduce MATCHED, a multimodal dataset of 27,619 unique text descriptions and 55,115 unique images collected from the Backpage escort platform across seven U.S. cities in four geographical regions. Our study extensively benchmarks text-only, vision-only, and multimodal baselines for vendor identification and verification tasks, employing multitask (joint) training objectives that achieve superior classification and retrieval performance on in-distribution and out-of-distribution (OOD) datasets. Integrating multimodal features further enhances this performance, capturing complementary patterns across text and images. While text remains the dominant modality, visual data adds stylistic cues that enrich model performance. Moreover, text-image alignment strategies like CLIP and BLIP2 struggle due to low semantic overlap and vague connections between the modalities of escort ads, with end-to-end multimodal training proving more robust. Our findings emphasize the potential of multimodal AA (MAA) to combat HT, providing LEAs with robust tools to link ads and disrupt trafficking networks.

Figures

Figures reproduced from arXiv: 2412.13794 by the authors.

Figure 1
Figure 1. Comparison of retrieval performance across multiple baselines for text-to-text, image-to-image, and [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 3
Figure 3. Comparison of model performance among text-only, vision-only, and multimodal classifiers trained [PITH_FULL_IMAGE:figures/full_fig_p033_3.png] view at source ↗
Figure 4
Figure 4. Comparison of retrieval performance on the South region test datasets. Text, vision, and multimodal [PITH_FULL_IMAGE:figures/full_fig_p035_4.png] view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Comparison of retrieval performance on the Midwest region test datasets. Text, vision, and multimodal [PITH_FULL_IMAGE:figures/full_fig_p036_5.png]
Figure 6
Figure 6. Figure 6: Comparison of retrieval performance on the West region test datasets. Text, vision, and multimodal [PITH_FULL_IMAGE:figures/full_fig_p037_6.png]
Figure 7
Figure 7. Figure 7: Comparison of retrieval performance on the Northeast region test datasets. Text, vision, and multimodal [PITH_FULL_IMAGE:figures/full_fig_p038_7.png]
Figure 8
Figure 8. Figure 8: Knowledge graph representation generated using AA retrieval for Vendor labels 784 and 1101 from the [PITH_FULL_IMAGE:figures/full_fig_p040_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

93 extracted references · 38 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Bo Ai, Yuchen Wang, Yugin Tan, and Samson Tan. 2022. https://aclanthology.org/2022.aacl-main.84 Whodunit? learning to contrast for authorship attribution . In Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: L...

    2022

  4. [4]

    Mohamad Alansari, Oussama Abdul Hay, Sajid Javed, Abdulhadi Shoufan, Yahya Zweiri, and Naoufel Werghi. 2023. https://doi.org/10.1109/ACCESS.2023.3266068 Ghostfacenets: Lightweight face recognition model from cheap operations . IEEE Access, 11:35429--35446

    work page arXiv 2023

  5. [5]

    Hamidreza Alvari, Paulo Shakarian, and J. E. Kelly Snyder. 2017. https://arxiv.org/abs/1705.10786 Semi-supervised learning for detecting human trafficking . Preprint, arXiv:1705.10786

    work page Pith review arXiv 2017

  6. [6]

    Kelly Snyder

    Hamidreza Alvari, Paulo Shakarian, and J.E. Kelly Snyder. 2016. https://doi.org/10.1109/ISI.2016.7745456 A non-parametric learning approach to identify online human trafficking . In 2016 IEEE Conference on Intelligence and Security Informatics (ISI), pages 133--138

    work page arXiv 2016

  7. [7]

    Hassan Marzoughi Ardakani. 2020. Identifying Human Trafficking Networks in Louisiana by Using Authorship Attribution and Network Modeling. Louisiana State University and Agricultural & Mechanical College

    2020

  8. [8]

    Mudit Bhargava, Pulkit Mehndiratta, and Krishna Asawa. 2013. Stylometric analysis for authorship attribution on twitter. In Big Data Analytics: Second International Conference, BDA 2013, Mysore, India, December 16-18, 2013, Proceedings 2, pages 37--47. Springer

    2013

Show all 93 references
  1. [9]

    Parkhi, and Andrew Zisserman

    Qiong Cao, Li Shen, Weidi Xie, Omkar M. Parkhi, and Andrew Zisserman. 2018. https://arxiv.org/abs/1710.08092 Vggface2: A dataset for recognising faces across pose and age . Preprint, arXiv:1710.08092

    2018 arXiv

  2. [10]

    Pelham Carter, Matt Gee, Hollie McIlhone, Harkeeret Lally, and Robert Lawson. 2021. https://doi.org/10.1016/j.metip.2021.100065 Comparing manual and computational approaches to theme identification in online forums: A case study of a sex work special interest community . Metho...

    2021

  3. [11]

    Nathanael Chambers, Timothy Forman, Catherine Griswold, Kevin Lu, Yogaish Khastgir, and Stephen Steckler. 2019. https://doi.org/10.18653/v1/D19-5507 Character-based models for adversarial phone extraction: Preventing human sex trafficking . In Proceedings of the 5th Workshop o...

    2019 doi

  4. [12]

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. https://arxiv.org/abs/2002.05709 A simple framework for contrastive learning of visual representations . Preprint, arXiv:2002.05709

    2020 arXiv

  5. [13]

    Leshem Choshen, Dan Eldad, Daniel Hershcovich, Elior Sulem, and Omri Abend. 2019. https://doi.org/10.18653/v1/P19-1419 The language of legal and illegal activity on the D arknet . In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages...

    2019 doi

  6. [14]

    Marco Cotogni, Marco Arazzi, and Claudio Cusano. 2024. https://doi.org/10.1109/TMM.2024.3408683 Photostyle60: A photographic style dataset for photo authorship attribution and photographic style transfer . IEEE Transactions on Multimedia, pages 1--12

    2024

  7. [15]

    Arun Das and Paul Rad. 2020. https://arxiv.org/abs/2006.11371 Opportunities and challenges in explainable artificial intelligence (xai): A survey . Preprint, arXiv:2006.11371

    2020 arXiv

  8. [16]

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. 2021. https://arxiv.org/abs/2010.11929 An image is worth 16x16 words...

    2021 arXiv

  9. [17]

    Alaaeldin El-Nouby, Natalia Neverova, Ivan Laptev, and Hervé Jégou. 2021. https://arxiv.org/abs/2102.05644 Training vision transformers for image retrieval . Preprint, arXiv:2102.05644

    2021 arXiv

  10. [18]

    EUROPOL. 2020. https://www.europol.europa.eu/publications-events/publications/challenges-of-countering-human-trafficking-in-digital-era The challenges of countering human trafficking in the digital era

    2020

  11. [19]

    Ma \"e l Fabien, Esau Villatoro-Tello, Petr Motlicek, and Shantipriya Parida. 2020. https://aclanthology.org/2020.icon-main.16 B ert AA : BERT fine-tuning for authorship attribution . In Proceedings of the 17th International Conference on Natural Language Processing (ICON), pa...

    2020

  12. [20]

    William Falcon and The PyTorch Lightning team . 2019. https://doi.org/10.5281/zenodo.3828935 PyTorch Lightning

    2019 doi

  13. [21]

    Andrian Firmansyah, Tien Fabrianti Kusumasari, and Ekky Novriza Alam. 2023. https://doi.org/10.1109/ICCoSITE57641.2023.10127799 Comparison of face recognition accuracy of arcface, facenet and facenet512 models on deepface framework . In 2023 International Conference on Compute...

    2023

  14. [22]

    Eilika Fobbe. 2021. Text-linguistic analysis in forensic authorship attribution

    2021

  15. [23]

    Gallo, A

    I. Gallo, A. Calefati, S. Nawaz, and M. K. Janjua. 2018. https://doi.org/10.1109/DICTA.2018.8615789 Image and encoded text fusion for multi-modal classification . In 2018 Digital Image Computing: Techniques and Applications (DICTA), pages 1--7

    2018

  16. [24]

    Chenquan Gan, Xiang Fu, Qingdong Feng, Qingyi Zhu, Yang Cao, and Ye Zhu. 2024. https://doi.org/10.1016/j.eswa.2023.122731 A multimodal fusion network with attention mechanisms for visual–textual sentiment analysis . Expert Systems with Applications, 242:122731

    2024

  17. [25]

    Luyu Gao, Yunyi Zhang, Jiawei Han, and Jamie Callan. 2021. https://doi.org/10.18653/v1/2021.repl4nlp-1.31 Scaling deep contrastive learning batch size under memory limited setup . In Proceedings of the 6th Workshop on Representation Learning for NLP (RepL4NLP-2021), pages 316-...

    2021 doi

  18. [26]

    Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna Wallach, Hal Daumé III au2, and Kate Crawford. 2021. https://arxiv.org/abs/1803.09010 Datasheets for datasets . Preprint, arXiv:1803.09010

    2021 arXiv

  19. [27]

    John Giorgi, Osvald Nitski, Bo Wang, and Gary Bader. 2021. https://doi.org/10.18653/v1/2021.acl-long.72 D e CLUTR : Deep contrastive learning for unsupervised textual representations . In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics a...

    2021 doi

  20. [28]

    Chatzichristofis

    Socratis Gkelios, Yiannis Boutalis, and Savvas A. Chatzichristofis. 2021. https://arxiv.org/abs/2101.03771 Investigating the vision transformer model for image retrieval tasks . Preprint, arXiv:2101.03771

    2021 arXiv

  21. [29]

    Hagberg, Daniel A

    Aric A. Hagberg, Daniel A. Schult, and Pieter J. Swart. 2008. Exploring network structure, dynamics, and function using networkx. In Proceedings of the 7th Python in Science Conference, pages 11 -- 15, Pasadena, CA USA

    2008

  22. [30]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2015. https://arxiv.org/abs/1512.03385 Deep residual learning for image recognition . Preprint, arXiv:1512.03385

    2015 arXiv

  23. [31]

    Zhiqiang Hu, Roy Ka-Wei Lee, Lei Wang, Ee-peng Lim, and Bo Dai. 2020. Deepstyle: User style embedding for authorship attribution of short texts. In Web and Big Data: 4th International Joint Conference, APWeb-WAIM 2020, Tianjin, China, September 18-20, 2020, Proceedings, Part I...

    2020

  24. [32]

    Weinberger

    Gao Huang, Zhuang Liu, Laurens van der Maaten, and Kilian Q. Weinberger. 2018. https://arxiv.org/abs/1608.06993 Densely connected convolutional networks . Preprint, arXiv:1608.06993

    2018 arXiv

  25. [33]

    Javier Huertas-Tato, Alejandro Mart \' n, and David Camacho. 2024. Understanding writing style in social media with a supervised contrastively pre-trained transformer. Knowledge-Based Systems, 296:111867

    2024

  26. [34]

    John D. Hunter. 2007. https://doi.org/10.1109/MCSE.2007.55 Matplotlib: A 2d graphics environment . Computing in Science & Engineering, 9(3):90--95

    2007 doi

  27. [35]

    Michelle Ibanez and Rich Gazan. 2016. https://doi.org/10.1109/ASONAM.2016.7752332 Virtual indicators of sex trafficking to identify potential victims in online advertisements . In 2016 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM...

    2016

  28. [36]

    Michelle Ibanez and Daniel D. Suthers. 2014. https://doi.org/10.1109/HICSS.2014.200 Detection of domestic human trafficking indicators and movement trends using content available on open internet sources . In 2014 47th Hawaii International Conference on System Sciences, pages ...

    2014 doi

  29. [37]

    ILO. 2012. http://www.ilo.org/wcmsp5/groups/public/---ed_norm/---declaration/documents/publication/wcms_182004.pdf Ilo global estimate of forced labour

    2012

  30. [38]

    Plotly Technologies Inc. 2015. https://plot.ly Collaborative data science

    2015

  31. [39]

    Fung, and Mourad Debbabi

    Farkhund Iqbal, Rachid Hadjidj, Benjamin C.M. Fung, and Mourad Debbabi. 2008. https://doi.org/10.1016/j.diin.2008.05.001 A novel approach of mining write-prints for authorship attribution in e-mail forensics . Digital Investigation, 5:S42--S51. The Proceedings of the Eighth An...

    2008 doi

  32. [40]

    Jeff Johnson, Matthijs Douze, and Herv \'e J \'e gou. 2019. Billion-scale similarity search with GPUs . IEEE Transactions on Big Data, 7(3):535--547

    2019

  33. [41]

    Patrick Juola and R Harald Baayen. 2005. A controlled-corpus experiment in authorship identification by cross-entropy. Literary and Linguistic Computing, 20(Suppl):59--67

    2005

  34. [42]

    Kimmo Karkkainen and Jungseock Joo. 2021. Fairface: Face attribute dataset for balanced race, gender, and age for bias measurement and mitigation. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 1548--1558

    2021

  35. [43]

    Mahmut Kaya and Hasan S akir Bilge. 2019. Deep metric learning: A survey. Symmetry, 11(9):1066

    2019

  36. [44]

    Douwe Kiela, Suvrat Bhooshan, Hamed Firooz, Ethan Perez, and Davide Testuggine. 2020. https://arxiv.org/abs/1909.02950 Supervised multimodal bitransformers for classifying images and text . Preprint, arXiv:1909.02950

    2020 arXiv

  37. [45]

    Wonjae Kim, Bokyung Son, and Ildoo Kim. 2021. https://arxiv.org/abs/2102.03334 Vilt: Vision-and-language transformer without convolution or region supervision . Preprint, arXiv:2102.03334

    2021 arXiv

  38. [46]

    Narine Kokhlikyan, Vivek Miglani, Miguel Martin, Edward Wang, Bilal Alsallakh, Jonathan Reynolds, Alexander Melnikov, Natalia Kliushkina, Carlos Araya, Siqi Yan, and Orion Reblitz-Richardson. 2020. https://arxiv.org/abs/2009.07896 Captum: A unified and generic model interpreta...

    2020 arXiv

  39. [47]

    Simon Kornblith, Mohammad Norouzi, Honglak Lee, and Geoffrey Hinton. 2019. https://arxiv.org/abs/1905.00414 Similarity of neural network representations revisited . Preprint, arXiv:1905.00414

    2019 arXiv

  40. [48]

    Satyapriya Krishna, Tessa Han, Alex Gu, Javin Pombra, Shahin Jabbari, Steven Wu, and Himabindu Lakkaraju. 2022. https://arxiv.org/abs/2202.01602 The disagreement problem in explainable machine learning: A practitioner's perspective . Preprint, arXiv:2202.01602

    2022 arXiv

  41. [49]

    Vlad Krotov, Leigh Johnson, and Leiser Silva. 2020. Tutorial: Legality and ethics of web scraping

    2020

  42. [50]

    Alexandre Lacoste, Alexandra Luccioni, Victor Schmidt, and Thomas Dandres. 2019. Quantifying the carbon emissions of machine learning. arXiv preprint arXiv:1910.09700

    2019 arXiv

  43. [51]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023. https://arxiv.org/abs/2301.12597 Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models . Preprint, arXiv:2301.12597

    2023 arXiv

  44. [52]

    Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, and Kai-Wei Chang. 2019. https://arxiv.org/abs/1908.03557 Visualbert: A simple and performant baseline for vision and language . Preprint, arXiv:1908.03557

    2019 arXiv

  45. [54]

    Yifei Li, Pratheeksha Nair, Kellin Pelrine, and Reihaneh Rabbany. 2022 b . https://doi.org/10.18653/v1/2022.findings-acl.225 Extracting person names from user generated text: Named-entity recognition for combating human trafficking . In Findings of the Association for Computat...

    2022 doi

  46. [55]

    Yushi Li, Xin Zheng, Ming Zhu, Jie Mei, Ziwen Chen, and Yunfei Tao. 2024. Compact bilinear pooling and multi-loss network for social media multimodal classification. Signal, Image and Video Processing, 18(11):8403--8412

    2024

  47. [56]

    Javin Liu, Hao Yu, Vidya Sujaya, Pratheeksha Nair, Kellin Pelrine, and Reihaneh Rabbany. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.219 SWEET - weakly supervised person name extraction for fighting human trafficking . In Findings of the Association for Computational...

    2023 doi

  48. [57]

    Indicators of sex trafficking in online escort ads

    Kristina Lugo-Graulich. Indicators of sex trafficking in online escort ads. https://www.ojp.gov/pdffiles1/nij/grants/305453.pdf

  49. [58]

    Kristina Lugo-Graulich and Leah F. Meyer. 2021. https://www.jrsa.org/projects/escorts-resources/jrsa-le-guide-on-indicators-of-sex-trafficking-in-online-escort-ads.pdf Law enforcement guide on indicators of sex trafficking in online escort ads . Justice Research and Statistics...

    2021

  50. [59]

    Scott Lundberg and Su-In Lee. 2017. https://arxiv.org/abs/1705.07874 A unified approach to interpreting model predictions . Preprint, arXiv:1705.07874

    2017 arXiv

  51. [60]

    Andrei Manolache, Florin Brad, Antonio Barbalau, Radu Tudor Ionescu, and Marius Popescu. 2022. https://arxiv.org/abs/2207.03477 Veridark: A large-scale benchmark for authorship verification on the dark web . Preprint, arXiv:2207.03477

    2022 arXiv

  52. [61]

    Chirag Nagpal, Kyle Miller, Benedikt Boecking, and Artur Dubrawski. 2017. https://arxiv.org/abs/1509.06659 An entity resolution approach to isolate instances of human trafficking online . Preprint, arXiv:1509.06659

    2017 arXiv

  53. [62]

    Dubrawski

    Chirag Nagpal, Kyle Miller, Benedikt Boecking, and Artur W. Dubrawski. 2015. https://api.semanticscholar.org/CorpusID:11047665 An entity resolution approach to isolate instances of human trafficking online . In NUT@EMNLP

    2015

  54. [63]

    Pratheeksha Nair, Yifei Li, Catalina Vajiac, Andreas Olligschlaeger, Meng-Chieh Lee, Namyong Park, Duen Horng Chau, Christos Faloutsos, and Reihaneh Rabbany. 2022. https://doi.org/10.1145/3487553.3524263 Vispad: Visualization and pattern discovery for fighting human traffickin...

    2022

  55. [64]

    Smita Nirkhi and Dr. R.V. Dharaskar. 2013. https://doi.org/10.14569/IJACSA.2013.040505 Comparative study of authorship identification techniques for cyber forensics analysis . International Journal of Advanced Computer Science and Applications, 4(5)

    2013

  56. [65]

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Kopf, Edward Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu ...

    2019

  57. [66]

    Fabian Pedregosa, Ga \"e l Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al. 2011. Scikit-learn: Machine learning in python. Journal of machine learning research, 12(Oct):2825--2830

    2011

  58. [67]

    POLARIS. 2018. https://wiki.preventconnect.org/wp-content/uploads/2018/08/Human-Trafficking-Statistics-from-the-Polaris-Project.pdf Human trafficking statistics

    2018

  59. [68]

    POLARIS. 2020. https://polarisproject.org/wp-content/uploads/2022/01/Polaris-Analysis-of-2020-Data-from-the-National-Human-Trafficking-Hotline.pdf Polaris analysis of 2020 data from the national human trafficking hotline

    2020

  60. [69]

    Portnoff, Danny Yuxing Huang, Periwinkle Doerfler, Sadia Afroz, and Damon McCoy

    Rebecca S. Portnoff, Danny Yuxing Huang, Periwinkle Doerfler, Sadia Afroz, and Damon McCoy. 2017. https://doi.org/10.1145/3097983.3098082 Backpage and bitcoin: Uncovering human traffickers . In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery ...

    2017

  61. [70]

    Reihaneh Rabbany, David Bayani, and Artur Dubrawski. 2018. Active search of connections for case building and combating human trafficking. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 2120--2129

    2018

  62. [71]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. https://arxiv.org/abs/2103.00020 Learning transferable visual models from natural lan...

    2021 arXiv

  63. [72]

    Hoshiladevi Ramnial, Shireen Panchoo, and Sameerchand Pudaruth. 2016. Authorship attribution using stylometry and machine learning techniques. In Intelligent Systems Technologies and Applications: Volume 1, pages 113--125. Springer

    2016

  64. [73]

    why should i trust you?

    Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. 2016. https://arxiv.org/abs/1602.04938 "why should i trust you?": Explaining the predictions of any classifier . Preprint, arXiv:1602.04938

    2016 arXiv

  65. [74]

    Gaurav Sahu and Olga Vechtomova. 2021. https://doi.org/10.18653/v1/2021.eacl-main.275 Adaptive fusion techniques for multimodal data . In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 3156--3166,...

    2021 doi

  66. [75]

    Vageesh Saxena, Benjamin Ashpole, Gijs van Dijck, and Gerasimos Spanakis. 2023 a . https://doi.org/10.18653/v1/2023.emnlp-main.524 IDT raffickers: An authorship attribution dataset to link and connect potential human-trafficking operations on text escort advertisements . In Pr...

    2023 doi

  67. [76]

    Vageesh Saxena, Nils Rethmeier, Gijs van Dijck, and Gerasimos Spanakis. 2023 b . https://doi.org/10.18653/v1/2023.acl-long.481 V endor L ink: An NLP approach for identifying & linking vendor migrants & potential aliases on D arknet markets . In Proceedings of the 61st Annual M...

    2023 doi

  68. [77]

    Sefik Ilkin Serengil and Alper Ozpinar. 2023. https://doi.org/10.33774/coe-2023-18rcn An evaluation of sql and nosql databases for facial recognition pipelines

    2023 doi

  69. [78]

    Karen Simonyan and Andrew Zisserman. 2015. https://arxiv.org/abs/1409.1556 Very deep convolutional networks for large-scale image recognition . Preprint, arXiv:1409.1556

    2015 arXiv

  70. [79]

    Sleeman, Rishabh Kapoor, and Preetam Ghosh

    William C. Sleeman, Rishabh Kapoor, and Preetam Ghosh. 2022. https://doi.org/10.1145/3543848 Multimodal classification: Current landscape, taxonomy and future directions . ACM Comput. Surv., 55(7)

    2022 doi

  71. [80]

    Gage, Daniel Dakota, and Sandra K \"u bler

    Jacob Striebel, Abishek Edikala, Ethan Irby, Alex Rosenfeld, J. Gage, Daniel Dakota, and Sandra K \"u bler. 2024. https://doi.org/10.18653/v1/2024.naacl-industry.24 Scaling up authorship attribution . In Proceedings of the 2024 Conference of the North American Chapter of the A...

    2024 doi

  72. [81]

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens, and Zbigniew Wojna. 2015. https://arxiv.org/abs/1512.00567 Rethinking the inception architecture for computer vision . Preprint, arXiv:1512.00567

    2015 arXiv

  73. [82]

    Kovács Tamás, Atzenhofer-Baumgartner, Florian, Aoun, Sandy, Nicolaou, Anguelos, Luger, Daniel, Decker, Franziska, Lamminger, Florian, Vogeler, and Georg. 2022. https://doi.org/10.57967/hf/0135 langdetect (revision 0215f72)

    2022 doi

  74. [83]

    Mingxing Tan and Quoc V. Le. 2021. https://arxiv.org/abs/2104.00298 Efficientnetv2: Smaller models and faster training . Preprint, arXiv:2104.00298

    2021 arXiv

  75. [84]

    Edmund Tong, Amir Zadeh, Cara Jones, and Louis-Philippe Morency. 2017. https://doi.org/10.18653/v1/P17-1142 Combating human trafficking with multimodal deep models . In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pape...

    2017 doi

  76. [85]

    Jacob Tyo, Bhuwan Dhingra, and Zachary C Lipton. 2021. Siamese bert for authorship verification. In CLEF (Working Notes), pages 2169--2177

    2021

  77. [86]

    UNDOC. 2020. https://www.unodc.org/documents/data-and-analysis/tip/2021/GLOTiP_2020_15jan_web.pdf Global report on trafficking in persons

    2020

  78. [87]

    van Leeuwen

    Nik Vaessen and David A. van Leeuwen. 2024. https://arxiv.org/abs/2402.13723 The effect of batch size on contrastive self-supervised speech representation learning . Preprint, arXiv:2402.13723

    2024 arXiv

  79. [88]

    Catalina Vajiac, Duen Horng Chau, Andreas Olligschlaeger, Rebecca Mackenzie, Pratheeksha Nair, Meng-Chieh Lee, Yifei Li, Namyong Park, Reihaneh Rabbany, and Christos Faloutsos. 2023. https://doi.org/10.1109/TVCG.2022.3209403 Trafficvis: Visualizing organized activity and spati...

    2023

  80. [89]

    Danae Sánchez Villegas, Daniel Preoţiuc-Pietro, and Nikolaos Aletras. 2024. https://arxiv.org/abs/2309.07794 Improving multimodal classification of social media posts by leveraging image-text auxiliary tasks . Preprint, arXiv:2309.07794

    2024 arXiv

  81. [90]

    Xiangwen Wang, Peng Peng, Chun Wang, and Gang Wang. 2018. https://doi.org/10.1145/3196494.3196529 You are your photographs: Detecting multiple identities of vendors in the darknet marketplaces . In Proceedings of the 2018 on Asia Conference on Computer and Communications Secur...

    2018

  82. [91]

    Anna Wegmann, Marijn Schraagen, and Dong Nguyen. 2022. https://doi.org/10.18653/v1/2022.repl4nlp-1.26 Same author or just same topic? towards content-independent style representations . In Proceedings of the 7th Workshop on Representation Learning for NLP, pages 249--268, Dubl...

    2022 doi

  83. [92]

    Sanghyun Woo, Shoubhik Debnath, Ronghang Hu, Xinlei Chen, Zhuang Liu, In So Kweon, and Saining Xie. 2023. https://arxiv.org/abs/2301.00808 Convnext v2: Co-designing and scaling convnets with masked autoencoders . Preprint, arXiv:2301.00808

    2023 arXiv

  84. [93]

    Zhanhong Ye, Changle Zhong, Haoliang Qi, and Yong Han. 2023. Supervised contrastive learning for multi-author writing style analysis. In CLEF (Working Notes), pages 2817--2822

    2023

  85. [94]

    Yiming Zhang, Yujie Fan, Wei Song, Shifu Hou, Yanfang Ye, Xin Li, Liang Zhao, Chuan Shi, Jiabin Wang, and Qi Xiong. 2019. https://doi.org/10.1145/3308558.3313537 Your style your identity: Leveraging writing and photography styles for drug trafficker identification in darknet m...

    2019

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.