Sunday, 19 November 2017

Automatic Interpretation and Documentation of Ceramics

The ArchaAIDE Project is run by a consortium of institutions and professionals (Universities of Pisa, Barcelona, York and Cologne, Italian National Research Council, School of Computer Science at the University of Tel Aviv and IT and archaeology professionals) and it is funded by the European Union. 

The goal of ArchAIDE is to create "new system for the automatic recognition of archaeological pottery", which is usually the most common artifact found in archaeological sites. Pottery analysis yield a great amount of socio-economic information, but the interpretation of ceramics requires high skills, experience and time. 

In order to streamline the process and make knowledge accessible wherever, the ArchAIDE project will develop an app for smartphones and tablets. The app will allow archaeologists to take pictures of potsherds and, connceting to a database, it will recognize the sherd and retrieve information about the specific ceramic type. In addition, recognized sherds will be stored and shared. The app will also enable archaeologists to generate an electronic document about the artifact, reducing the amount of paperwork.

Such an app might revolutionize fieldwork, but is it possible to replace skilled scholars with an app? Does a streamlined interpretation and classification process have only advantages or do we still need to be reflexive when we are studying artifacts? Could this app reduce the interpretation of material culture to a mere "labelling" process?


kayla lausanne said...

In theory, this app looks like it'd be beneficial in speeding up the in field process and for having information right there instead of waiting till lab analysis. However, as you said, does it replace us? When looking at the webpage, there is a step for the validation of the automatic classification. This I believe is key as it allows for the archaeologist to confirm the classification is correct and still speeds up the process. This step still keeps the archeologist relevant and necessary as a ceramic could meet multiple criteria for different categories, but the archaeologist must decide what is more correct. That being said, I believe we would still need to be cautious. As Caraher discusses, by increasing the efficiency, we are also creating new categories of interpretation and experience. The labels that are being created are generic categories that may not completely encompass the specific ceramic, so the software would place it into a category it saw as most correct, even if that is not the case. So, although this software looks good on paper, it may have consequences that effect our interpretation of ceramics.

Joanna said...

I share concerns about this automation, as it seems overly reliant on typology. While artifact typologies are certainly used by archaeologists, I think we could all come up with messy examples of artifacts not fitting into our neat categories. I am curious to know how would the system handle a stylistic transition or aberrant sherd.

While I am not against automation wholesale, I find automation that attempts to replace analytical work rather than quantitative or repetitive tasks to be disconcerting. While I am somewhat comforted by the review process that Kayla points out, I do wonder if this could encourage an entrenchment in pre-determined typologies that might not reflect the actual archaeological materials recovered.

Hillary Kiazyk said...

Hey all,

I find this discussion really interesting. I read this post a few weeks ago but after watching this ( TED talk last night I had something new to contribute to this conversation.

For those of you that aren't going to watch it the TLDR is basically NASA launched the Kepler space telescope to monitor stars and came across one that didn't meet any of their expectations. They were using computers to do this work but they also launched the zooniverse project (citizen science that we talked about earlier in the semester (Trevor's Lecture). This star (still no solid conclusions on what makes it so unique) was found only with the help of all the citizen scientists.

Now back to archaeology, I certainly think there is some validity of this argument. It is frightening to think about being replaced by a machine. The point that I always come back to is machines can only be as smart as we program them to be (unless they start programming themselves) and because of this (I think Joanna said it best) "I think we could all come up with messy examples of artifacts not fitting into our neat categories"

Every site is different and there are so many things about the past that we still have yet to rediscover.