Sumerian Networks

Partner:Adam Anderson, UC Berkeley: DH / D-Lab / BIDS, Academic

Overview

Project Description

The goal of the Sumerian Network project has been to build reproducible socio-economic networks from the Ur III textual archives. We applied novel computational methods for name disambiguation, and in the absence of ‘gold-standard’ data, we built an atomized network based on name instances in textual attestations for c. 15,000 documents. This research project has brought together archaeologists, cuneiform specialists, experts in Computational Text Analysis and Natural Language Processing from around the world. Current results for reproducible network models are available in Jupyter Notebooks [https://github.com/niekveldhuis/Sumerian-network] and Gephi Networks [https://github.com/admndrsn/Gephi]. Link: (http://dh.berkeley.edu/sumerian-networks-reconstructing-ur-iii_)

Expected Deliverable

JupyterNotebooks, D3.js code (e.g. Observablehq.com), Networks (e.g. gephi.org)

What would a successful semester look like to you?

In general we work toward new innovations and updates with well-described code and Markdown in GitHub. Specifically, we’re working with geospatial, temporal, and social networks, and we’re performing exploratory data analysis on big data. We have done some Deep Learning methods before, so this semester we will update and implement the latest tools and methods (from Word2Vec/D2V to BERT and RoBERTa).

Data

Models

Conclusion