A neogeo approach to Classical resources.
First and foremost, I think I may have had my head in the sand, as I completed missed Google’s Digital Humanities Research Awards! The DH awards are supporting 12 university research groups with unrestricted grants for one year as well as receiving access to Google tools, technologies and expertise. How cool is that!? The awarded projects are clearly places to look to for exciting text based stuff soon.
Anyway… Today I attended the CASPAR seminar over at the Institute of Archaeology. Leif Isaksen told us about Google Ancient Places (GAP): Discovering historic geographical entities in the Google Books corpus.
What I like about Gap is it is about experimenting. The aim is to create a search-facility to help the discoveryof data of ancient world locations, and then followed by experimenting with lots of different ways of visualizing the results.
It all started with HESTIA project which investigates the Greek historian Herodotus texts for places that are mentioned and the stories that he tells about them. The aim was to break open the idea of space and locality. How do you predict the histories spatially?
GAP builds on HESTIA and the Open Context project which hosts data for a range of archaeological sites. The end goal for GAP is to discover all references to a particular ancient location, and then visualize the results in GoogleEarth or Google Maps to show the geographic spread of the locations in ancient texts.
My quick notes from the seminar:
- TEI used to markup text. All the places were marked up in XML. Where are the places Herodotus is talking about and where is the focus. Visualised as a huge clustering around the Aegean and the Nile Valley.
- Analysed the relationship between places using Network analysis. Number of times places are co referenced. Identifiers used to locate the references in texts. Global common identifiers.
- Google books have Limited metadata, lots of digitising but limited info.
- Geo automated processing, Tag toponyms in classical texts. Identify locations and then Save as metadata (not dynamic searching)
- Gazetters and place names – use multilingual gazetteers, accessing Latin Wikipedia.
- URI’s Geoname identifiers to help. Geolocating alternative possibilities e.g. Athens Greece vs Athens Georgia
- Leverage contextual clustering at multiple scales, weight by possible number of alternatives. Use higher scales for undecidable cases and then derive URIs
- Indexing and clustering of Alexandria archive.
- Generate additional properties – frequencies of concept identification.
- SIP association of toponyms
- Mapping between translations
- Evolving project open to exploring alternative solutions
- Looking for Useful not looking for perfect
- More applicable to narratives than databases
Projects to look at:
PELAGIOS: enabled lined ancient geodata in open systems funded by JISC
Digital memory institute (AIT)
3 thoughts on “CASPAR Seminar Notes: Google ancient places project.”
I literally didn’t understand a word of this.
Fear that we’ve lost you to academia. Oh noes! 😉
Frankie, why do you think my notes are bullet points!? I didnt have a clue. Leif was very good at explaining, but I dont really understand texts, network analysis, or how you actually do visualizations. I took it to mean it has something to do with a clever thing that reads texts to find locations and then puts it into google maps so you can see it. I think. I dont know what the clever thing is, or how you get it into google maps. hence the bullet points of words I don’t really understand. fear not, I still be the odd tin miner girl at heart.
Ah, well that’s a relief!
What’s CASPAR? A friendly ghost I thought…