Uncovering Latent Connections in Indigenous Heritage: Semantic Pipelines for Cultural Preservation in Brazil

Luis Vitor Zerkowski; Nina S. T. Hirata

REVIEW 4 major objections 5 minor 42 references

Uncovering Latent Connections in Indigenous Heritage: Semantic Pipelines for Cultural Preservation in Brazil

T0 review · 4 major / 5 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 The paper claims that embedding-based similarity, computed separately from images and text, can surface latent connections and inconsistencies inside a large digital heritage collection that its own metadata misses.

desk verdict A credible application of embedding pipelines to an Indigenous heritage collection, with a real circularity in the domain-specific benchmark but enough substance—external STS-B gains and a working tool—to deserve peer review. read the letter →

arxiv 2508.10911 v1 pith:BHIRPA76 submitted 2025-07-31 cs.HC cs.CYcs.LG

Luis Vitor Zerkowski , Nina S. T. Hirata This is my paper · ORCID

classification cs.HCcs.CYcs.LG MSC 68T4568T50

keywords IndigenousheritagepreservationsemanticembeddingscontrastivelearningvisualsimilaritytextualinteractivevisualizationBrazilianPortugueseculturalcollections

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 embedding-based similarity, computed separately from images and text, can surface latent connections and inconsistencies inside a large digital heritage collection that the collection's own metadata misses. Working with the Museu Nacional dos Povos Indígenas's collection of roughly 21,000 Indigenous Brazilian objects, the authors build visual and textual semantic spaces, fine-tune them on domain labels and contrastive pairs, and embed them in an interactive tool. The payoff if true: curators gain a practical device for spotting mislabeled items and for discovering unsuspected material or stylistic relationships between communities, while the public gains a similarity-based way to explore the collection. The paper's core evidence is qualitative — clusters that align with function or style across metadata boundaries and outlier items whose embedding places them with their true kind rather than their recorded label.

What carries the argument

The load-bearing machinery is a pair of embedding pipelines. The visual pipeline removes image backgrounds with an off-the-shelf segmentation model, extracts features with ViT-Base and DINOv2-Base transformers, and fine-tunes them with linear classification heads on the museum's povo and categoria labels, with class-rebalancing and, in the multi-head variant, joint supervision on both targets. The textual pipeline summarizes raw descriptions to 64 tokens with a large language model, then fine-tunes BERTimbau and Albertina encoders either with unsupervised SimCSE dropout-based contrastive learning or with supervised InfoNCE contrastive learning whose positives are LLM-generated paraphrases and negatives are items from other categories; a custom benchmark (In-Context STS-B) built from the same LLM judgments measures domain-specific performance. Both spaces are projected to two dimensions with UMAP and presented in an interactive point-cloud tool with temporal and geographic views, so the clusters become inspectable evidence rather than just numbers.

What would settle it

Have a curator or community expert independently judge whether the ceramic piece recorded as 'trançado' is genuinely ceramic and whether the Mayongong–Kamayurá–Kuikuro necklace cluster corresponds to real shared material culture; if expert judgments systematically disagree with the embeddings' groupings, the claim that the spaces reveal latent connections is not supported. A quantitative check would compare human similarity ratings on a sample of item pairs against the model's cosine similarities and the In-Context STS-B scores, asking whether the synthetic labels agree with human judgment better than chance.

Watch

Extended reading notes

Core claim

The central discovery is that domain-adapted embeddings behave as a curatorial lens: after fine-tuning vision transformers on community (povo) and category (categoria) labels, and sentence encoders on contrastive pairs built from LLM-generated paraphrases of item descriptions, the resulting spaces organize around visual and functional semantics rather than only recorded labels. The paper argues this lens finds things metadata does not say, such as Mayongong, Kamayurá, and Kuikuro necklaces clustering by shared vibrant blue palettes, and a ceramic piece recorded as 'trançado' that nevertheless lands inside the ceramic cluster. The same spaces also organize conceptually related text despite lexical differences, e.g., descriptions using 'quadrangular' versus 'geometrizantes' grouping together, indicating semantic rather than surface-level text similarity. The claim is explicitly framed as a complement to expert curation rather than a replacement.

Load-bearing premise

The load-bearing premise is that LLM-generated paraphrases are valid positive pairs and items from different 'categoria' labels are valid negatives for contrastive training, and that the same LLM's judgments define the In-Context STS-B benchmark; no human or community validation is reported.

Editorial extensions

If this is right

The interactive tool can serve as a curatorial aid at the Museu Nacional dos Povos Indígenas, flagging candidate mislabeled items and suggesting new interpretive groupings for review.
Public users can navigate the collection by visual or textual similarity rather than only metadata facets, lowering the barrier to engagement with Indigenous heritage.
Fine-tuned BERTimbau and Albertina models trained on summarized museum descriptions transfer to general STS-B, indicating that domain contrastive training improves general semantic understanding of Brazilian Portuguese.
The released machine-learning-ready datasets, consisting of background-removed images and summarized descriptions, enable reuse of the two pipelines across other collections.
The two-pronged evaluation, standard STS-B plus domain-specific In-Context STS-B, shows that domain-adapted performance gains are larger on the in-domain benchmark, supporting the value of domain-specific semantic evaluation.

Reading between the lines

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

The authors' framing implies a general recipe: any heritage collection with weak metadata and rich images and texts could be processed the same way, and collections beyond the museum could adopt the pipelines with modest retraining.
The lack of human or community validation means the practical deployment claim, that the tool reveals genuine cultural connections, is not yet established; the paper's own qualitative examples are plausible but need curator sign-off.
The mislabeled ceramic example suggests the visual embedding could double as a data-cleaning tool for the Tainacan repository, with automated outlier detection run over the full 20,965-item metadata to produce a prioritized review list.
The textual clustering by function and theme across lexical differences hints that the same contrastive setup could support semantic search in Portuguese for the collection, a feature the authors list as future work but which is a short step from the current 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.

Desk Editor's Note

read the letter

Quick take: this is a useful application-level paper with a real circularity problem in its domain-specific benchmark, but the external STS-B results and the delivered tool give it enough substance to referee.

The genuinely new parts: a complete, documented pipeline applied to the Museu Nacional dos Povos Indígenas Tainacan collection, a new domain benchmark (In-Context STS-B), and a working interactive visualization tool scheduled for deployment on a federal government site. That's real applied DH work. The authors are also transparent about hardware limits, class imbalance, and the fact that they report only the best hyperparameter configurations.

The soft spots are not fatal but they are real. The In-Context STS-B benchmark is built from LLM-generated positives/negatives, and the supervised contrastive models are trained on the same kind of LLM-generated pairs. So the large gains on that benchmark (e.g., 0.75 to 0.86) partly reflect how well the model reproduces the LLM's similarity judgments rather than genuine semantic structure in the collection. This is a classic closed loop. The paper does not report any human or community validation of those synthetic pairs, which matters especially in a cultural-heritage context. However, the consistent gains on the standard Brazilian STS-B (e.g., BERTimbau 0.70 to 0.79) are independent evidence that the contrastive fine-tuning isn't just fitting the LLM; it improves general semantic similarity. That mitigates the circularity but doesn't remove it.

The visual pipeline has a milder circularity: the models are fine-tuned on categoria and povo metadata, so the resulting clusters are expected to align with those labels. Claiming that the embeddings reveal connections 'not captured by existing metadata' is partly an artifact of using metadata as supervision. That said, the mislabeled-ceramic example is a legitimate use: an item whose predicted class disagrees with its label is a plausible label error. That is a valid curatorial signal, not circular.

Other concerns: the code is not yet available (the GitHub link is a stub), and the hyperparameter search is described only as 'broad' with selected final values. That hampers direct reproduction. These are fixable.

Overall: this is a solid systems paper for the digital humanities and cultural-heritage audience. The central claim that embeddings surface latent connections is plausible, but the paper needs to release the code, validate the synthetic labels against a small set of human/curator judgments, and temper the 'not captured by metadata' language for the visual pipeline. I'd send it to serious referees.

Who's it for: DH researchers, museum informatics folks, and anyone applying contrastive learning to low-resource or domain-specific text. It's not a methods breakthrough, but it's a credible application with a useful artifact. I'd bring it to a reading group discussion about evaluation circularity in LLM-generated benchmarks, but I wouldn't cite it in my own work unless the code appears.

Referee Report

4 major / 5 minor

Summary. The paper describes a dual-pipeline system for exploring the Museu Nacional dos Povos Indígenas digital collection: a visual pipeline that fine-tunes ViT and DINOv2 models on item images using 'categoria' and 'povo' metadata, and a textual pipeline that fine-tunes BERTimbau and Albertina sentence encoders with SimCSE and supervised contrastive learning on LLM-generated paraphrases. The resulting embeddings are projected with UMAP into an interactive Dash/Plotly tool with semantic, temporal, and geographic views. The paper claims that both pipelines surface latent connections and inconsistencies not captured by existing metadata, and reports quantitative gains on external Portuguese STS-B (e.g., BERTimbau supervised 0.79 vs 0.70 vanilla) and larger gains on a self-constructed 'In-Context STS-B' benchmark built from the same LLM used to generate training pairs.

Significance. If the central claim holds, the system would be a valuable, open, and culturally situated contribution: it ships a complete data-processing pipeline, machine-learning-ready datasets, a modular framework, and an interactive tool deployed in collaboration with a national museum, with code and documentation in a public repository. The external STS-B results give some independent evidence that the contrastive fine-tuning improves general semantic representation for Brazilian Portuguese. However, the domain-specific evaluation is circular—training and evaluation labels both come from the same LLM—and the visual pipeline's claimed discoveries are selected examples rather than measured outcomes. The significance therefore depends on replacing or supplementing the synthetic supervision and anecdotal evidence with independent validation, which is a fixable but load-bearing gap.

major comments (4)

[Data Gathering and Processing; Text-Based Semantic Spaces (Evaluation)] The In-Context STS-B benchmark is constructed with the same LLM (Llama-4-Maverick) that generates the paraphrase positives used in supervised contrastive training. Consequently, the large gains on this benchmark (e.g., Table 7: vanilla 0.75 vs. supervised 0.86) largely measure how well the fine-tuned model reproduces the LLM's own similarity judgments, not the model's ability to capture genuine semantic relationships among Indigenous artifacts. Without human or community-curated annotations, the domain-specific evaluation cannot support the paper's central claim. See also the self-reported Reproducibility Checklist item 3.3, which states that no data appendix for the new benchmark is included.
[Text-Based Semantic Spaces (Latent Observations)] The external Brazilian STS-B results (Tables 7–8) show modest but consistent gains (BERTimbau 0.70→0.79, Albertina 0.69→0.74), which support improved general semantic representation. Yet the paper's specific cultural claims—such as clusters of vases or basket-related items—rest on qualitative inspection of UMAP projections (Figures 5–6) and chosen examples. No quantitative cluster-quality metric, retrieval evaluation, or independent annotation is reported, so the reader cannot distinguish genuine latent structure from projection artifacts or LLM-aligned noise.
[Image-Based Semantic Spaces (Fine-Tuning Strategy; Latent Observations)] The visual models are fine-tuned on 'categoria' and 'povo' metadata, and the paper then presents as 'latent' the discovery of a ceramic piece that embeds among ceramics despite a 'trançado' label (Figure 4). Since the model was trained to reproduce categoria structure, this example is not evidence of connections beyond existing metadata; it is evidence that the classifier generalizes to a mislabeled instance, which is a different and weaker claim. There is also no quantitative comparison between the fine-tuned and the pretrained visual embeddings (e.g., retrieval, clustering, or label-consistency metrics), so the gain in semantic structure is asserted rather than demonstrated.
[Reproducibility Checklist (3.3, 4.3, 4.4, 4.12)] The paper's own checklist reports that the new In-Context STS-B dataset is not included in a data appendix (3.3: no), preprocessing code is not included (4.3: no), source code is not in a code appendix (4.4: no), and no statistical significance tests are used (4.12: no). These are not merely administrative details: without the benchmark and the preprocessing scripts, the circularity concern cannot be independently inspected, and without significance tests, the reported improvements over vanilla models—several of which are within 0.01–0.02 standard deviation units of each other—are not established as reliable.

minor comments (5)

[Text-Based Semantic Spaces (Evaluation)] The paragraph beginning 'Differently from the image-based pipeline' is duplicated nearly verbatim in consecutive sentences of the Evaluation subsection; one copy should be removed.
[Introduction and Conclusion] The code availability statement is inconsistent: the abstract and repository links claim code is public at github.com/Luizerko/indigenous clusters and communities, while the Introduction says 'code, documentation, and dataset references available at will be available on camera-ready.' Please reconcile.
[Figure 4 caption] The caption contains a typo: 'trançadolabel' should be 'trançado label'.
[References] The paper cites the Llama-4-Maverick model as (Touvron et al. 2023), which points to the original LLaMA paper; the correct reference for Llama-4-Maverick should be provided, or the text should acknowledge that the model is accessed via Groq and cite the appropriate documentation.
[Data Gathering and Processing] The description of In-Context STS-B mentions that 'chunks were originally designed for description summarization' and that false negatives occur; this caveat is important but not reflected in the benchmark's later use as a success metric. A more detailed description of the negative-pair construction and how false negatives were handled would strengthen the paper.

Circularity Check

2 steps flagged · score 6.0 of 10

The domain-specific text benchmark is generated by the same LLM that supplies the supervised training pairs, so In-Context STS-B gains partly measure reproduction of the LLM's similarity judgments; the visual mislabel observation likewise restates the categoria supervision used to train the model.

fitted input called prediction [Data Gathering and Processing (contrastive dataset construction) and Text-Based Semantic Spaces (Evaluation, In-Context STS-B)]
"For the supervised setup, each anchor was paired with a paraphrased positive, generated via the aforementioned large language model (LLM), and ten negative examples randomly sampled from different categoria labels... For each item, we used the LLM-generated paraphrase as a positive example, but additionally asked the LLM to select the most semantically distinct description of a chunk as a negative example."

The supervised InfoNCE objective is trained to pull anchors toward LLM paraphrases and push away cross-categoria negatives. The In-Context STS-B benchmark computes Pearson correlation on exactly the same pattern: positive = LLM-generated paraphrase, negative = LLM-selected distinct description. High scores on this benchmark therefore largely measure how well the encoder has fit the LLM's pair-generation behavior, not an independent human or community judgment of cultural semantics. The standard extraGLUE STS-B results provide external grounding, but the paper's emphasized domain-specific gains are partly a closed loop between the label generator and the evaluation set it also defines.
fitted input called prediction [Image-Based Semantic Spaces (Fine-Tuning Strategy and Latent Observations)]
"To achieve this, we used supervised training with the povo and categoria attributes as targets... Outlier detection also exposed mislabeled items, such as a ceramic piece incorrectly tagged as tranc ¸ado, yet embedded correctly among ceramics."

The visual encoder is fine-tuned with categoria as a classification target, so placing the ceramic image in the ceramics cluster is precisely the training objective. Presenting this placement as a latent inconsistency 'exposed' by the embedding renames the supervision signal as an independent discovery: the embedding is not an external ground truth but a learned function of the same metadata labels used for training. This particular qualitative finding is therefore circular, although the other visual-similarity observations (e.g., necklace colors) are less directly forced.

full rationale

The clearest circularity is in the textual pipeline. Both the supervised contrastive training pairs and the In-Context STS-B evaluation pairs are produced by the same LLM, so the large domain-specific gains are at least partly a measure of how well the models reproduce the label generator's notion of similarity. This does not fully invalidate the paper: the standard extraGLUE STS-B benchmark is external and all fine-tuned models improve on it, which gives an independent, if modest, signal; the qualitative visual and textual clusters are selected examples rather than measured claims. No load-bearing self-citation chain or imported uniqueness theorem appears. The visual mislabel example, however, is also partially circular because the model was trained on the very categoria attribute it is used to 'correct'. Overall the central quantitative evaluation in the text domain reduces to fitting the LLM's own labels, warranting a 6 rather than a higher score because of the external STS-B results and non-circular qualitative observations.

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

The paper introduces a new benchmark (In-Context STS-B) and a visualization tool, but no new physical or conceptual entity such as a particle, mediator, force, or dimension. The benchmark is a dataset, not an entity in the sense used by this ledger.

free parameters (5)

LLM summarization token limit = 64
Set to fit 8GB GPU memory; changes all text inputs.
Number of contrastive negatives per anchor = 10
Chosen during dataset construction; affects training signal.
Contrastive temperature = 0.05 and 0.2 (selected)
Reported as best-performing values; other values were presumably tried but not shown.
Learning rate = 1e-6 to 3e-5 depending on model
Selected by validation; only best values reported.
Class filtering threshold for povo rebalancing = Not specified
A rebalancing strategy combining class filtering, augmentation, and weighted loss; threshold value is not given, affecting comparability.

assumptions (4)

ad hoc to paper LLM-generated paraphrases are positive pairs and items from different categoria labels are negative pairs for contrastive learning.
Used to construct the supervised contrastive dataset; no human or community validation is reported (Data Gathering and Processing).
ad hoc to paper The In-Context STS-B benchmark, with LLM-chosen positives and negatives, measures domain-specific semantic similarity.
Used as the main domain evaluation in Text-Based Semantic Spaces; because the same LLM generated training data, this is at risk of circularity.
domain assumption UMAP projections preserve the semantic structure enough to support qualitative interpretation of clusters.
Used throughout to claim coherent clusters and latent observations (Constructing Semantic Spaces).
domain assumption Museum metadata attributes povo and categoria are meaningful and reliable semantic targets for visual fine-tuning.
Used as classification targets in the image pipeline; mislabeling in the data is later used as evidence of the model's usefulness.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Uncovering Latent Connections in Indigenous Heritage: Semantic Pipelines for Cultural Preservation in Brazil." pith.science (2026). https://pith.science/paper/BHIRPA76

@misc{pith2026250810911,
  author       = {Pith},
  title        = {Pith review of: Uncovering Latent Connections in Indigenous Heritage: Semantic Pipelines for Cultural Preservation in Brazil},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BHIRPA76}},
  note         = {Machine review of arXiv:2508.10911}
}

read the original abstract

Indigenous communities face ongoing challenges in preserving their cultural heritage, particularly in the face of systemic marginalization and urban development. In Brazil, the Museu Nacional dos Povos Indigenas through the Tainacan platform hosts the country's largest online collection of Indigenous objects and iconographies, providing a critical resource for cultural engagement. Using publicly available data from this repository, we present a data-driven initiative that applies artificial intelligence to enhance accessibility, interpretation, and exploration. We develop two semantic pipelines: a visual pipeline that models image-based similarity and a textual pipeline that captures semantic relationships from item descriptions. These embedding spaces are projected into two dimensions and integrated into an interactive visualization tool we also developed. In addition to similarity-based navigation, users can explore the collection through temporal and geographic lenses, enabling both semantic and contextualized perspectives. The system supports curatorial tasks, aids public engagement, and reveals latent connections within the collection. This work demonstrates how AI can ethically contribute to cultural preservation practices.

Figures

Figures reproduced from arXiv: 2508.10911 by the authors.

**Figure 1.** Embedding spaces generated by the ViT model finetuned on povo (left) and categoria (right). The povo projection shows limited global structure due to the huge amount of classes and class imbalance. The categoria projection displays well-defined clusters and smoother transitions, indicating a coherent underlying manifold. Multi-head training, in turn, generally provided numeric gains, interestingly showing that jo… view at source ↗

**Figure 2.** Embedding space of the DINOv2 multi-head model, colored by categoria. Despite joint supervision, categoria clusters remain interpretable. Latent Observations: We highlight a few latent patterns surfaced by a ViT model fine-tuned on categoria. Embeddings revealed that some necklaces from the Mayongong community featured vibrant blue palettes, closely aligned with those of Kamayura´ and Kuikuro, suggesting shared ma… view at source ↗

**Figure 3.** Visual similarities between necklaces from the Mayongong (left), Kamayura´ (center), and Kuikuro (right) communities [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Typical ceramic item (left) and a misclassified outlier (right), embedded correctly by the model despite its incorrect tranc¸ado label. Text-Based Semantic Spaces Building on the image-based framework, we developed a parallel pipeline for textual data using summarized…

**Figure 5.** Figure 5: Zoomed-in views of UMAP projections for BERTimbau trained with supervised contrastive learning. Both regions reveal coherent and interpretable semantic clusters. Visualization Tool Having described the technical groundwork, we conclude with the final component of our s…

**Figure 6.** Figure 6: Two items from the basket-related cluster of the righthand image of [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

**Figure 7.** Figure 7: Primary interface tabs: collection semantic space (topleft), temporal view (top-right), geographic view (bottom). Collection Semantic Spaces The core of the visualization tool is the semantic point-cloud interface, where each item is displayed as a 2D point using eit…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 28 canonical work pages

[1]

2025. Groq . https://groq.com/. Accessed: 2025-07-09

work page 2025
[2]

Museu do Índio

2025. Museu do Índio. https://www.gov.br/museudoindio/pt-br. Accessed: 2025-07-18

work page 2025
[3]

Socioambiental Map of Indigenous Lands

2025. Socioambiental Map of Indigenous Lands . https://mapa.socioambiental.org/pages/?lang=pt-br. Accessed: 2025-07-09

work page 2025
[4]

Tainacan – Museu do Índio Collection

2025. Tainacan – Museu do Índio Collection. https://tainacan.museudoindio.gov.br/. Accessed: 2025-07-18

work page 2025
[5]

Agirre, E.; Cer, D.; Diab, M.; and Gonzalez-Agirre, A. 2012. S em E val-2012 Task 6: A Pilot on Semantic Textual Similarity. In Agirre, E.; Bos, J.; Diab, M.; Manandhar, S.; Marton, Y.; and Yuret, D., eds., * SEM 2012: The First Joint Conference on Lexical and Computational Semantics – Volume 1: Proceedings of the main conference and the shared task, and ...

work page 2012
[6]

Z.; Talib, A

Alma’aitah, W. Z.; Talib, A. Z.; and Osman, M. A. 2020. Opportunities and Challenges in Enhancing Access to Metadata of Cultural Heritage Collections: A Survey. Artificial Intelligence Review, 53: 3621--3646

work page 2020
[7]

Amid, E.; and Warmuth, M. K. 2019. TriMap: Large-scale Dimensionality Reduction Using Triplets. CoRR, abs/1910.00204

arXiv 2019
[8]

Bentley, J. L. 1975. Multidimensional binary search trees used for associative searching. Commun. ACM, 18(9): 509--517

work page 1975

Show all 42 references

[9]

Carter, S.; Armstrong, Z.; Schubert, L.; Johnson, I.; and Olah, C. 2019. Activation Atlas. Distill, 4(3): e15. https://distill.pub/2019/activation-atlas

2019
[10]

Center for Digital Scholarship and Curation, Washington State University . 2025. Mukurtu CMS: The free, mobile and open source platform built with Indigenous communities to manage and share digital cultural heritage. https://mukurtu.org/. Accessed: 2025-07-18

2025
[11]

Chen, T.; Kornblith, S.; Norouzi, M.; and Hinton, G. 2020. A Simple Framework for Contrastive Learning of Visual Representations. arXiv preprint arXiv:2002.05709

2020 arXiv
[12]

Dosovitskiy, A.; Beyer, L.; Kolesnikov, A.; Weissenborn, D.; Zhai, X.; Unterthiner, T.; Dehghani, M.; Minderer, M.; Heigold, G.; Gelly, S.; Uszkoreit, J.; and Houlsby, N. 2021. An Image is Worth 16×16 Words: Transformers for Image Recognition at Scale. In International Confere...

2021
[13]

Ferreira, E. F. 2011. Brazilian Indigenous Peoples and the Debate on Authenticity and Cultural Change. INDIANA - Estudios Antropol \'o gicos sobre Am \'e rica Latina y el Caribe , 28: 395--408

2011
[14]

Gao, T.; Yao, X.; and Chen, D. 2021. S im CSE : Simple Contrastive Learning of Sentence Embeddings. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, 6894--6910. Online and P...

2021
[15]

Guimar\ aes, S. 2015. The Teaching of Afro-Brazilian and Indigenous Culture and History in Brazilian Basic Education in the 21st Century. Policy Futures in Education, 13(8)

2015
[16]

Hisayasu, L. M. 2019. Mediated Memory and Internet Indigenous Protagonism in Brazil. Unpublished manuscript

2019
[17]

Huang, Z. 1998. Extensions to the k-Means Algorithm for Clustering Large Data Sets with Categorical Values. Data Mining and Knowledge Discovery, 2(3): 283--304

1998
[18]

Kerbl, B.; Kopanas, G.; Leimk \"u hler, T.; and Drettakis, G. 2023. 3D Gaussian Splatting for Real‑Time Radiance Field Rendering. ACM Transactions on Graphics, 42(4)

2023
[19]

Kirillov, A.; Mintun, E.; Ravi, N.; Mao, H.; Rolland, A.; Fu, T.; and Girshick, R. 2023. Segment Anything. arXiv:2304.02643

2023 arXiv
[20]

Kokhlikyan, N.; Miglani, V.; Martin, M.; Wang, E.; Alsallakh, B.; Reynolds, J.; Melnikov, A.; Kliushkina, N.; Araya, C.; Yan, S.; and Reblitz‐Richardson, O. 2020. Captum: A unified and generic model interpretability library for PyTorch. arXiv preprint arXiv:2009.07896

2020 arXiv
[21]

Liao, D. 2021. Sentence Embeddings using Supervised Contrastive Learning. CoRR, abs/2106.04791

2021 arXiv
[22]

McInnes, L.; Healy, J.; and Melville, J. 2018. UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction. arXiv preprint arXiv:1802.03426

2018 arXiv
[23]

Mu \ n oz, P.; and Tisi, J. 2020. Digital Preservation and Indigenous Knowledge: Strategies and Case Studies. Heritage Science, 8: 45

2020
[24]

Oquab, M.; Darcet, T.; Moutakanni, T.; Vo, H.; Szafraniec, M.; Khalidov, V.; Fernandez, P.; Haziza, D.; Massa, F.; El-Nouby, A.; Assran, M.; Ballas, N.; Galuba, W.; Howes, R.; Huang, P.-Y.; Li, S.-W.; Misra, I.; Rabbat, M.; Sharma, V.; Synnaeve, G.; Xu, H.; Jegou, H.; Mairal, ...

2023 arXiv
[25]

F.; Leite, B.; Lopes Cardoso, H.; Gomes, L.; Rodrigues, J.; Santos, R.; and Branco, A

Os \'o rio, T. F.; Leite, B.; Lopes Cardoso, H.; Gomes, L.; Rodrigues, J.; Santos, R.; and Branco, A. 2024. PORTULAN E xtra GLUE Datasets and Models: Kick-starting a Benchmark for the Neural Processing of Portuguese. In Proceedings of the 17th Workshop on Building and Using Co...

2024
[26]

Plotly Technologies Inc. 2015. Plotly : Collaborative Data Science

2015
[27]

Plotly Technologies Inc. 2024. Dash : A data and analytics web app framework for Python, no JavaScript required

2024
[28]

W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; Krueger, G.; and Sutskever, I

Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; Krueger, G.; and Sutskever, I. 2021. Learning Transferable Visual Models From Natural Language Supervision. In Proceedings of the 38th International Conferen...

2021
[29]

Rebellato, L. 2025. Digital repatriation of cultural heritage in the Global South: A model for open access to museum collections empowering indigenous communities in the Brazilian Amazon. https://www.digarv.se/en/participating-research-projects/digital-repatriering-av-kulturar...

2025
[30]

Rodrigues, J.; Gomes, L.; Silva, J.; Branco, A.; Santos, R.; Lopes Cardoso, H.; and Osório, T. 2023. Advancing Neural Encoding of Portuguese with Transformer Albertina PT-* . arXiv:2305.06721

2023 arXiv
[31]

Ronneberger, O.; Fischer, P.; and Brox, T. 2015. U-Net: Convolutional Networks for Biomedical Image Segmentation. In Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2015, 234--241. Springer

2015
[32]

B.; and Wattenberg, M

Smilkov, D.; Thorat, N.; Nicholson, C.; Reif, E.; Viégas, F. B.; and Wattenberg, M. 2016. Embedding Projector: Interactive Visualization and Interpretation of Embeddings. arXiv preprint arXiv:1611.05469

2016 arXiv
[33]

Souza, F.; Nogueira, R.; and Lotufo, R. 2020. BERTimbau: Pretrained BERT Models for Brazilian Portuguese . In Intelligent Systems. BRACIS 2020. Lecture Notes in Computer Science, volume 12319 of Lecture Notes in Computer Science, 403--417. Springer, Cham

2020
[34]

P.; Kolesnikov, A.; Zhai, X.; Wightman, R.; Uszkoreit, J.; and Beyer, L

Steiner, A. P.; Kolesnikov, A.; Zhai, X.; Wightman, R.; Uszkoreit, J.; and Beyer, L. 2022. How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers. Transactions on Machine Learning Research

2022
[35]

Sundararajan, M.; Taly, A.; and Yan, Q. 2017. Axiomatic Attribution for Deep Networks. arXiv preprint arXiv:1703.01365

2017 arXiv
[36]

Touvron, H.; Lavril, T.; Izacard, G.; Martinet, X.; Lachaux, M.-A.; Lacroix, T.; Rozi \`e re, B.; Goyal, N.; Hambro, E.; Azhar, F.; Rodriguez, A.; Joulin, A.; Grave, E.; and Lample, G. 2023. LLaMA : Open and Efficient Foundation Language Models. arXiv:2302.13971

2023 arXiv
[37]

van der Maaten, L.; and Hinton, G. 2008. Visualizing Data using t-SNE. Journal of Machine Learning Research, 9: 2579--2605

2008
[38]

Wang, A.; Singh, A.; Michael, J.; Hill, F.; Levy, O.; and Bowman, S. R. 2018. GLUE : A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding. arXiv:1804.07461

2018 arXiv
[39]

T.; O’Meara, C.; Monti, L.; and Nabhan, G

Wilder, B. T.; O’Meara, C.; Monti, L.; and Nabhan, G. P. 2016. The Importance of Indigenous Knowledge in Curbing the Loss of Language and Biodiversity. BioScience, 66(6): 499--509

2016
[40]

Zheng, P.; Gao, D.; Fan, D.-P.; Liu, L.; Laaksonen, J.; Ouyang, W.; and Sebe, N. 2024. Bilateral Reference for High-Resolution Dichotomous Image Segmentation. CAAI Artificial Intelligence Research

2024
[41]

, " * write output.state after.block = add.period write newline

ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...
[42]

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 gl...

Pith tools

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

[1] [1]

2025. Groq . https://groq.com/. Accessed: 2025-07-09

work page 2025

[2] [2]

Museu do Índio

2025. Museu do Índio. https://www.gov.br/museudoindio/pt-br. Accessed: 2025-07-18

work page 2025

[3] [3]

Socioambiental Map of Indigenous Lands

2025. Socioambiental Map of Indigenous Lands . https://mapa.socioambiental.org/pages/?lang=pt-br. Accessed: 2025-07-09

work page 2025

[4] [4]

Tainacan – Museu do Índio Collection

2025. Tainacan – Museu do Índio Collection. https://tainacan.museudoindio.gov.br/. Accessed: 2025-07-18

work page 2025

[5] [5]

Agirre, E.; Cer, D.; Diab, M.; and Gonzalez-Agirre, A. 2012. S em E val-2012 Task 6: A Pilot on Semantic Textual Similarity. In Agirre, E.; Bos, J.; Diab, M.; Manandhar, S.; Marton, Y.; and Yuret, D., eds., * SEM 2012: The First Joint Conference on Lexical and Computational Semantics – Volume 1: Proceedings of the main conference and the shared task, and ...

work page 2012

[6] [6]

Z.; Talib, A

Alma’aitah, W. Z.; Talib, A. Z.; and Osman, M. A. 2020. Opportunities and Challenges in Enhancing Access to Metadata of Cultural Heritage Collections: A Survey. Artificial Intelligence Review, 53: 3621--3646

work page 2020

[7] [7]

Amid, E.; and Warmuth, M. K. 2019. TriMap: Large-scale Dimensionality Reduction Using Triplets. CoRR, abs/1910.00204

arXiv 2019

[8] [8]

Bentley, J. L. 1975. Multidimensional binary search trees used for associative searching. Commun. ACM, 18(9): 509--517

work page 1975