Timetable: Friday morning 0930-1110 in the Theresa Singleton Zoom Room
Format: Standard
Organisers: Daniel Carvalho
Contact: danielcarvalho1@campus.ul.pt
Our present lives are profoundly impacted by the new Technological Revolution. Data has become the new currency, with a plethora of applications on contemporary problems. But what about the Past? Current investigations in Artificial Intelligence and Simulation are not only providing new information in colossal quantity that are useful for Archaeology but also rising disruptive questions: are archaeologists building more accurate narratives of past societies? Is Archaeology augmenting its thematic reach with more advanced tools? Will we be able to experiment what the Past was like, with Virtual Reality? Can archaeological excavation – the unrepeatable experience – be recorded in a way that increases its repeatability, using gradually innovative technological techniques? Although the uses of Artificial Intelligence in Archaeology are relativity known, an emerging global interest in its capacities suggests a wider debate, not only on method but in theoretical terms, from epistemological turns to ethical issues. To answer all these questions, this session welcomes papers (no longer than 10 minutes) regarding those topics, as well as those focusing on the application of these areas in archaeological research and their theoretical frameworks and reflecting in the implications of Simulation for Archaeology and its near future.
Papers
0930 – The Artificial Theorist: can (and should) robots think about Archaeology?
Daniel Carvalho, University of Lisbon, Autonomous University of Barcelona
Robotics is now a constant presence in our daily lives. Their scope and objectives vary greatly: some are tools, other companions and even weapons. Their versatility is commonly described in many forms of media. Archaeology is no strange to robotic activities: drones soar through the sky, offering unique perspectives of archaeological sites; AI is used intensively for the recognition of artifacts; unmanned vehicles are sent to places that humans cannot reach.
However, this is but a mere fraction of their potential. Could a robot engage with archaeological thought? Could a robot make theories about the Past? Many scientific areas already use robots to aid them in discoveries beyond our grasp: shouldn’t Archaeology do the same?
In this presentation, we will discuss the possibilities of an artificial theorist, their limitations and even the dangers that can rise with it. By introducing project T.A.L.O.S – Theoretical Archaeology Learning Operating Systems – we intend to produce debate about these aspects, as they are, in our view, of paramount importance not only to Archaeology, but for contemporary society as well.
0945 – Integrating Machine Learning, Computer Vision and Survey Practice: A Theoretical Framework-In-Progress
Lucy Killoran, University of Glasgow & Historic Environment Scotland
The applications of Machine Learning (ML) and Computer Vision (CV) techniques for automating elements of archaeological survey and prospection have been demonstrated by multiple studies. These studies show an emerging approach to survey automation in which workflow processes traditionally essential to archaeological survey practice are not explicitly included, such as the use of contextual information to help decide on the classification of an object. Several new software tools which facilitate the use of ML/CV for geospatial data analysis in a relatively accessible way have recently become available, however there is not yet consensus on the way that these technologies can or should integrate with current archaeological workflows and interpretive practices.
This research project aims to understand the impacts of ML/CV on workflows for archaeological survey. With these technologies evolving at a rapid pace, it is important to clearly define the current and future needs of archaeological survey practitioners in relation to these tools. This project uses participatory design methods to work with stakeholders to better define these needs, such as engaging survey practitioners in the testing of a lo-fi prototype of a proposed automated survey system. This presentation will discuss the developing theoretical framework being used to understand the impacts of ML/CV on archaeological survey practice.
1000 – Break
1020 – Archaeological drawing and the new technologies: Replacement or complement?
Paula do Nascimento, University of Lisbon
With the advance of new technologies, our era has become mostly digital. All this progress has entered Archaeology as well, in the form of new ways to divulgate to the public its advancements, with 3D modelling and visual reconstitutions of artifacts and archaeological sites. As this trend becomes gradually the new status quo, a question arises: is archaeological drawing and illustration in the brink of disappearing? We argue that although traditional means of engaging with archaeological data may be constantly updated and revised, they must not be forgotten, as with would result in a destruction of pages in the history of archaeology and its participants.
However, the complexity of this problem is elevated when new software and automatic machinery can independently draw, with unparalleled speed and, in some cases, accuracy. Is the substitution of the traditional methods an inevitability? Or the solution is to build a bridge between the “old” and the “new”, building complementary methods for the benefit of both worlds? In this presentation we shall present these ideas and defend that the essence of Archaeology, human interpretation of artifacts and their drawings is not reproducible by a machine, while we must embrace new technologies for their potential to benefit the discipline.
1035 – AmphoraeFinder – Merging Neural Networks with Artifact Identification
Joel Santos, NOVA University of Lisbon
How do archaeologists identify the vast quantities of artifacts of Human Past? This question is the base of AmphoraeFinder, a project that intends to provide archaeologists reliable and quick methodology for classifying fragments of roman amphorae.
By using Deep Learning and CNN (Convolutional Neural Networks) with a data base, AmphoraeFinder is a tool that may well represent the future of methodology in Archaeology. The first results shall be presented, as well as the perspectives of further uses of this technology in other archaeological problems.
1050 – Discussion and debate
TAG_2021 