Archaeology specific BERT models for English, German, and Dutch” (Brandsen 2024) explores the use of BERT-based models for Named Entity Recognition (NER) in archaeology across three languages: English, German, and Dutch. It introduces six models trained and fine-tuned on archaeological literature, followed by the presentation and evaluation of three models specifically tailored for NER tasks. The focus on multilingualism enhances the applicability of the research, while the meticulous evaluation using standard metrics demonstrates a rigorous methodology.

The introduction of NER for extracting concepts from literature is intriguing, while the provision of a method for others to contribute to BERT model pre-training enhances collaborative research efforts. The innovative use of BERT models to contextualize archaeological data is a notable strength, bridging the gap between digitized information and computational models.

Additionally, the paper's release of fine-tuned models and consideration of environmental implications add further value.

In summary, the paper contributes significantly to the task of NER in archaeology, filling a crucial gap and providing foundational tools for data mining and reevaluating legacy archaeological materials and archives.

Reference

Brandsen, A. (2024). Archaeology specific BERT models for English, German, and Dutch. Zenodo, 8296920, ver. 5 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.8296920

Cultural taxonomies are an essential tool for archaeologists working with prehistoric material cultures as they have historically been used to create the basic analytical units for studying cultural evolution through time (de Mortillet, 1883 ; Breuil, 1913). This approach has its limits as the taxonomic units are essentially etic constructions, i.e., they are defined in a cultural context exterior to the one that produced the material culture on which they are based (e.g., Pesesse, 2019). But to approach questions related to cultural evolution, one has to define archaeological units with clear geographic and chronological delineations in order to be compared synchronically and diachronically (e.g., Willey and Philips, 1958). In « A meta-analysis of Final Palaeolitic/Earliest Mesolithic cultural taxonomy and evolution in Europe », F. Riede and colleagues propose a novel and interesting approach to question the end of the Palaeolithic and beginning of the Mesolithic’s « named archaeological cultures » (NACs) analytical pertinence (Riede et al., 2023). In this particular context, NACs are indeed very numerous (n = 86) and result from complex and regional research histories. It seems thus pertinent to question the extent to which the said NACs chronological and geographic patterns result from past cultural diversity and evolution, and are not artefacts of research. 

To do so, the authors adopted a data-driven approach that they describe in detail in the paper. First, they gathered an European data base of lithic tool-kit composition, blade and bladelet technology and armature morphology at 350 key sites considered representative of NACs, dated between 15 and 11 ka (Hussain et al., 2023). These data were then analyzed using geometric morphometrics and a set of statisticaal tests in order to 1) test the coherence of these taxonomic units, and 2) test the chronological change in artefact shape variation. The authors conclude that the data set is partially biased by reasearch practices and histories, as their data-driven approach has only partially replicated traditional NACs for the european Late Palaeolithic/Early Mesolithic. However, their analysis of armature shape evolution has shown a tendency to diversification overtime, a pattern that was already observed in more « traditional » approaches. 

This study is, in my opinion, an excellent contribution for a significant step in macro-regional approaches to the archaeological record: defining discrete archaeological units that serve as a basis for subsequent analyses aimed at delineating cultural evolutionary processes. The authors propose a carefully designed and statistically grounded procedure in order to achieve these definitions in the most replicable and explicit possible manner. Taking advantage of drawings as a primary source of information is also very original despite several limitations of this approach (such as the necessary selection of most typical artefacts to be represented, the incompleteness of data publication or the difficulty to access all published work across such a large geographic area). The results of the study are convincing enough to allow the authors to discuss the pertinence of European Late Paleo/Early Mesolithic NACs, the potential epistemological and historical factors that could affect this taxonomic framework, as well as to give more weight to the traditional hypothesis of lithic cultural diversification towards the end of the Pleistocene/beginning of the Holocene in Europe. 

I would also like to underline the authors’ important efforts to ensure transparence and replicability of their study, as well as the accessibility of the data, thanks to extensive supplementary data and a data paper describing their data set in detail.

Anaïs L. Vignoles 

References

Breuil, H. (1913). Les subdivisions du paléolithique supérieur et leur signification. In Congrès international d’anthropologie et d’archéologie préhistoriques - compte-rendu de la XIVème session, tome 1:165‑238. Genève: Imprimerie Albert Kündig.

Hussain, S. T., Riede, F., Matzig, D. N., Biard, M., Crombé, P., Fernández-Lopéz de Pablo, J., Fontana, F., Groß, D., Hess, T., Langlais, M., Mevel, L., Mills, W., Moník, M., Naudinot, N., Posch, C., Rimkus, T., Stefański, D. and Vandendriessche, H. (2023). A Pan-European Dataset Revealing Variability in Lithic Technology, Toolkits, and Artefact Shapes ~15-11 Kya. Scientific Data 10 (1): 593. https://doi.org/10.1038/s41597-023-02500-9.

Mortillet, G. (1883). Le Préhistorique, antiquité de l’homme. Reinwald. Paris.

Pesesse, D. (2019). Analyser un silex, le façonner à nouveau ? Sur certains usages de la chaîne opératoire au Paléolithique supérieur. Techniques & culture, no 71: 74‑77. https://doi.org/10.4000/tc.11321.

Riede, F., Matzig, D. N., Biard, M., Crombé, P., Fernández-Lopéz de Pablo, J., Fontana, F., Groß, D., Hess, T., Langlais, M., Mevel, L., Mills, W., Moník, M., Naudinot, N., Posch, C., Rimkus, T., Stefański, D., Vandendriessche, H. and Hussain, S. T. (2023). A meta-analysis of Final Palaeolithic/earliest Mesolithic cultural taxonomy and evolution in Europe, Zenodo, 8195587., ver. 3 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.8195587

Willey, G. R. and Phillips, P. (1958). Method and Theory in American Archaeology. Chicago, IL: The University of Chicago Press.

Diamanti et al. (2024) is a significant contribution to the field of underwater robotics and their use in archaeology, with an innovative approach to some major problems in the deployment of said technologies. It identifies issues when it comes to approaching Underwater Cultural Heritage (UCH) sites and does so through an interest in the combination of data, maneuverability, and the interpretation provided by the instruments that archaeologists operate. The article's motives are clear: it is not enough to find the means to reach these sites, but rather is fundamental to take a step forward in methodology and how we can safeguard certain aspects of data recovery with robust mission planning.

To this end, the article does not fail to highlight previous contributions, in an intertwined web of references that demonstrate the marked evolution of the use of Unmanned Underwater Vehicles (UUVs), Remote Operated Vehicles (ROVs), Autonomous Underwater Vehicles (AUVs) and Autonomous Surface Vehicles (ASVs), which are growing exponentially in use (see Kapetanović et al. 2020). It should be emphasized that the notion of ‘aquatic environment’ used here is quite broad and is not limited to oceanic or maritime environments, which allows for a larger perspective on distinct technologies that proliferate in underwater archaeology. There is also a relevant discussion on the typologies of sensors and how these autonomous vehicles obtain their data, where are debated Inertial Measurement Units (IMU) and LiDAR systems. 

Thus, the authors of this article propose the creation of a model that acquires data through simulations, which allows for a better understanding of what a real mission presupposes in the field. Their tripartite method - pre-mission planning; mission plan and post-mission plan - offers a performing algorithm that simplifies and provides reliability to all the parts of the intervention. The use of real cases to create simulation models allows for a substantial approximation to common practice in underwater environments. And yet, the article is at its most innovative status when it combines all the elements it sets out to explore. It could simply focus on the methodological or planning component, on obtaining data, or on theoretical problems. But it goes further, which makes this approach more complete and of interest to the archaeological community. By not taking any part as isolated, the problems and possible solutions arising from the course of the mission are carried over from one parameter to another, where details are worked upon and efficiency goals are set.

One of the most significant cases is the tuning of ocean optics in aquatic environments according to the idiosyncracies of real cases (Diamanti et al. 2024: 8), a complex endeavor but absolutely necessary in order to increase the informative potential of the simulation. The exploration of various data capture models is also welcome, for the purposes of comparison and adaptation on a case-by-case basis. The brief theoretical reflection offered at the end of the article dwells in all these points and problematizes the difference between terrestrial and aquatic archaeology. In fact, the distinction does not only exist in the technical component, as although it draws in theoretical elements from archaeology that is carried out on land (see Krieger 2012 for this matter), the problems and interpretations are shaped by different factors and therefore become unique (Diamanti et al 2024: 15). The future, according to the authors, lies in increasing the autonomy of these vehicles so that the human element does not have to make decisions in a systematic way. It is in that note, and in order for that path to become closer to reality, that we strongly recommend this article for publication, in conjunction with the comments of the reviewers. We hope that its integrated approach, which brings together methods, theories and reflections, can become a broader modus operandi within the field of underwater robotics applied to archaeology.

References:

Diamanti, E., Yip, M., Stahl, A. and Ødegård, Ø. (2024). Advancing data quality of marine archaeological documentation using underwater robotics: from simulation environments to real-world scenarios, Zenodo, 8305098, ver. 4 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.8305098

Kapetanović, N., Vasilijević, A., Nađ, Đ., Zubčić, K., and Mišković, N. (2020). Marine Robots Mapping the Present and the Past: Unraveling the Secrets of the Deep. Remote Sensing, 12(23), 3902. MDPI AG. http://dx.doi.org/10.3390/rs12233902

Krieger, W. H. (2012). Theory, Locality, and Methodology in Archaeology: Just Add Water? HOPOS: The Journal of the International Society for the History of Philosophy of Science, 2(2), 243–257. https://doi.org/10.1086/666956

The paper “A closer look at an eroded dune landscape: first functional insights into the Federmessergruppen site of Lommel-Maatheide” [1] focuses on the final Palaeolithic (Federmesser) site of Lommel-Maatheide. Federmesser sites from northern Belgium such as Lommel-Maatheide, Meer and Rekem, show evidence for dense human occupation of specific areas located on top of Tardiglacial dunes nearby water bodies [2]. Preserved spatial distribution of finds at the sites suggest different activity areas and the presence of habitat structures [2]. However, because of the low organic preservation at the sites, functional analyses of lithic assemblages have the potential to significantly contribute to the spatial organisation of activities at these sites. This study by Tomasso et al. [1], represents an excellent example of a large-scale integrated approach to the study of lithic industries. The article undoubtedly demonstrates the potential of the proposed methodology and the reliability of the results obtained.

The article explores two different aspects (linked and excellently interconnected here): the possibility to apply use wear, residue and fracture analyses, on lithic assemblages affected by taphonomical alterations and to study lithic assemblages from dune landscapes. 

The study allows to answer differentiated questions: what is the influence of taphonomical alterations on use wear analysis? How do excavation methods impact the formation of use wear and the preservation of residues? Can we recognize distinct domestic activities? The article also provides an interesting hypothesis about hunting activities and propulsion methods.

The applied methodology is effectively interdisciplinary and innovative. It demonstrates how a truly integrated and articulated approach can represent the turning point for going beyond a mainly descriptive dimension to move towards a real understanding of the sites.

Studies dedicated to the analysis of the propulsion mode are not very frequent, but they are surely very important to better understand human behaviour [3]. Here, the methodology developed for the evaluation of the propulsion mode represent an important starting point for the definition of a new approach. Morphological and morphometrical analysis are integrated to the evaluation of the mechanical stress, to fracture delineations and to the hafting system (the latter defined on experimental basis).

This article therefore underlines the potential of combining different approaches to functional analysis associated with a ‘tailored’ reference collection and applying them to a high number of artefacts for reconstructing past human activities involving materials that are otherwise not preserved in these contexts. 

[1] Tomasso, S., Cnuts, D., Coppe, J., Geerts, F., Gils, M.V., Bie, M.D., Rots, V. (2021). A closer look at an eroded dune landscape: first functional insights into the Federmessergruppen site of Lommel-Maatheide. https://doi.org/10.31219/osf.io/pf3sm, ver 3 peer-reviewed and recommended by PCI Archaeology.

[2] De Bie, M., Van Gils, M. (2006). Les habitats des groupes à Federmesser (aziliens) dans le Nord de la Belgique. Bulletin de la Société préhistorique française, 103, 781–790.

[3] Coppe, J., Lepers, C., Clarenne, V., Delaunois, E., Pirlot, M. and Rots V. (2019). Ballistic Study Tackles Kinetic Energy Values of Palaeolithic Weaponry. Archaeometry, (61)4, 933-956. https://doi.org/10.1111/arcm.12452

“3Duewelsteene - A website for the 3D visualization of the megalithic passage grave Düwelsteene near Heiden in Westphalia, Germany” (Tharandt 2024) presents several 3-dimensional models of the Düwelsteene monument, along with contextual information about the grave and the process of creating the models. The website (https://3duewelsteene.github.io/) includes English and German versions, making it accessible to a wide audience. The website itself serves as the primary means of presenting the data, rather than as a supplement to a written text. This is an innovative and engaging way to present the research to a wider public.

Düwelsteene (“Devil’s Stones”) is a megalithic passage grave from the Funnel Beaker culture, dating to approximately 3300 BC. to 2600 BC. that was excavated in 1932. The website displays three separate 3-dimensional models. They ares shown in the 3D viewer software 3DHOP, which enables viewers to interact with the models in several ways, Annotations on the models display further information.

The first model was created by image-based modeling and shows the monument as it appears today.

A second model uses historical photographs and excavation data to reconstruct the grave as it appeared prior to the 1932 archaeological excavation. Restoration work following the excavation relocated many of the stones. Pre-1932 photographs collected from residents of the nearby town of Heiden were then used to create a model showing what the tomb looked like before the restoration work. It is commendable that a “certainty view” of the model shows the certainty with which the stones can be put at the reconstructed place. Gaps in the 3D models of stones that were caused by overlap with other stones have been filled with a rough mesh and marked as such, thereby differentiating between known and unknown parts of the stones.

The third model is the most imaginative and most interesting. As it shows as the grave as it might have appeared in approximately 3000 B.C., many aspects of this model are necessarily somewhat speculative. There is no direct evidence for exact size and shape of the capstones, the height of the mound, and other details. But enough is known about other similar constructions to allow these details to be inferred with some confidence. Again, care was taken to enable viewers to distinguish between the stones that are still in existence and those that were reconstructed.

A video on the home page of the website adds a nice touch. It starts with the model of the Düwelsteene as it currently appears then shows, in reverse order, the changes to the grave, ending with the inferred original state.

The 3D reconstructions are convincing and the methods well described. This project follows an open science approach and the FAIR principles, which is commendable and cutting edge in the field of Digital Archaeology. The preprint of the website hosted on zenodo includes all the photos, text, html files, and nine individual 3D model (.ply) files that are combined in the reconstructions exhibited on the website. A “readme.md” file includes details about building the models using CloudCompare and Blender, and modifications to the 3D viewer software (3DHOP) to get the website to improve the display of the reconstructions. We have to note that the link between the reconstructed models and the html page does not work when the files are downloaded from zenodo and opened offline. The html pages open in the browser, and the individual ply files work fine, but the 3D models do not display on the browser page when the html files are opened offline. The online version of the website is working perfectly.

The 3Düwelsteene website combines the presentation of archaeological domain knowledge to a lay audience as well as in-depths information on the reconstruction process to make it an interesting contribution for researchers. By providing data and code for the website it also models an Open Science approach, which enables other researchers to re-use these materials. We congratulate the author on a successful reconstruction of the megalithic tomb, an admirable presentation of the archaeological work and the thoughtful outreach to a broad audience.

Bibliography
Tharandt, L., 3Duewelsteene - A website for the 3D visualization of the megalithic passage grave Düwelsteene near Heiden in Westphalia, Germany, https://3duewelsteene.github.io/, Zenodo, 7948379, ver. 4 peer-reviewed and recommended by Peer Community in Archaeology. https://doi.org/10.5281/zenodo.7948379