If all you have is a hammer, everything looks like a nail. Towards best practice guidelines for network science in archaeology (reblogged)

March 31, 2016

(reblogged from post of 3/10/15 and modified)

Let’s be honest: networks aren’t really the right answer for many research problems. But how can we evaluate this and do network science properly. This is a discussion post about this topics with a preliminary list of best practice guidelines. I will give a presentation on this topic at the CAA in Oslo on Friday at 8:30AM session S16 but would like to stimulate discussion online as well, so engage!!!!! Get in touch, blog, tweet, comment.

It is uninformative for archaeologists and historians to use formal network methods as a hammer to hit every nail we can find with, just because we can. Specific formal network methods should preferably be selected in light of their ability to lead to insights that other approaches cannot offer. But what determines this usefulness of particular formal network methods for those studying the past? Is it the convenient representation of entities such as humans, islands, ports and the past interactions between them as dots and lines? Or is it the good fit between the past phenomena of trade, transportation and communication, with their abstraction as network concepts? Although these reasons might be sufficient to lead scholars to consider using formal network methods for addressing their research aims, they are not sufficient to motivate the adoption of specific network techniques.

I strongly believe that network science has something new to add to our disciplines, but a lot of work still needs to be done to leverage this promise and make it productive. To help us in this I think four things are needed:

  • Communities and events that provide a discussion platform for exploring this potential. The Connected Past and the Historical Networks Research communities have been providing venues for this task, and hopefully many more will follow.
  • Good practical examples should be published to give scholars an idea of how network science techniques could be beneficial in their own work, and to stimulate them to think creatively about applying these formal methods. The Connected Past publications alongside many others aim to serve this purpose.
  • These early examples should not just be accepted at face value but should be critiqued. To enable this, training should be provided to archaeologists and historians. Annual workshops have been provided by The Connected Past and the Historical Networks Research communities and at the CAA conference.
  • Finally, a community of archaeologists and historians should develop guidelines to best scientific practice in using these techniques. This could follow the format of the ADS guides to good practice.

I would like to call upon everyone interested in the use of network science for the study of the past to contribute to the development of these guidelines. Get in touch, blog, tweet, comment. Here is my attempt to develop a few very broad guidelines to good practice:

  • Network science techniques are methodological tools with clear rules and limitations.
  • Archaeologists could be provided with guides to good practice and archaeological examples, making them able to understand what kinds of questions different network science techniques are designed to answer and to evaluate whether it allows them to achieve their research aims. To do this hardly any familiarity with mathematical and computational techniques is required, only a willingness to explore the potential of a scientific method.
  • An evaluation of the potential contribution of network methods to addressing a particular research problem might be enhanced by working explicitly through the network science research process (Brandes et al. 2013), which again does not require much technical skills.
  • However, once archaeologists have decided to apply a specific network science technique, then a thorough understanding of the technical underpinnings of this technique is not an option but a prerequisite for a critical interpretation of its results. Archaeologists could be aided in this process by multi-disciplinary engagement and collaboration where possible.
  • Network concepts developed in network science are associated with specified data requirements, which should be acknowledged by the archaeologists adopting them. If the data requirements cannot be identified in empirical or simulated data then the network concepts loose all explanatory value.
  • When developing new network concepts, one should formulate network data specifications such that it becomes clear how the concept differs from exisiting concepts.
  • Formulating specifications of how network concepts are represented in network data allows for different conceptualisations of the same past phenomenon to be compared and possibly falsified.
  • A shift in perspective from the study of static structures to the emergence of empirical observations and past phenomena might be needed.
  • Confirmatory network science techniques offer archaeologists an approach to understanding how large-scale patterns emerge through the particular interactions of individual agents or relationships.
  • Confirmatory network science techniques can only be usefully applied when specifications are formulated of how the network concepts used should be represented as network data.
  • Confirmatory network science techniques require one to explicitly acknowledge the dynamic nature of past processes and the dynamic assumptions underlying the definition of ties. Because of this, I believe these techniques reveal the potential contribution of network science for archaeology far more than the exploratory network techniques.
  • The past systems we study were governed by dynamic phenomena and the network approach used to understand these phenomena should reflect their changeable nature.
  • Only in cases with a small number of nodes and where dependence assumptions gave rise to specific easily visually identifiable patterns, were network visualisations preferable over other types of data representation for communication purposes.
  • Even in cases where network science techniques do not offer additional functionality compared to other more common archaeological techniques, it could still lead to interesting insights by forcing one to explore a dataset or hypothesis through the lens of one’s assumptions about why and which relationships matter.
  • If a method is needed where the boundaries of entities are ill-defined and fluid, and where one argues these can not under any circumstances be tied down for analytical purposes, then network science does not offer the solution.
  • Network science can never be separated from the archaeological theoretical motivations of how and why certain archaeological evidence allows one to better understand a past phenomenon.
  • Some past processes are unknowable, due to our current techniques and datasets. All archaeological approaches suffer from this disadvantage and network science is no exception.

The Connected Past edited volume: now available from Oxford University Press! (discount voucher)

March 28, 2016

IMG_2135The most eagerly awaited book of the last decade is finally here! We are delighted to announce the publication of our The Connected Past volume: 200 pages of pure networky joy! Over the coming weeks I will write a series of blog posts about all the great work in the book. Expect adventures full of romance, frustration, epic struggles and humor. Click here to be redirected to the Oxford University Press website. Consider buying the book for your library or yourself. Click here to download a discount voucher to get the book cheaper.

The Connected Past. Challenges to Network Studies in Archaeology and History

Edited by Tom Brughmans, Anna Collar, and Fiona Coward

  • Features a comprehensive volume introduction which explains what network science is, why it is of interest for studying the past, and outlines the challenges faced when using network science in archaeology and history
  • Provides archaeologists and historians with a selection of the methodological and conceptual tools they need to compare and evaluate the strengths and limitations of different network approaches
  • Features international contributors from a range of disciplines (archaeology, history, physics, and mathematics) who are all pioneers in applying network perspectives to the study of the past
  • Presents a solid set of case studies which demonstrate how the challenges of applying network science perspectives to archaeological and historical datasets are overcome

One of the most exciting recent developments in archaeology and history has been the adoption of new perspectives which see human societies in the past—as in the present—as made up of networks of interlinked individuals. This view of people as always connected through physical and conceptual networks along which resources, information, and disease flow, requires archaeologists and historians to use new methods to understand how these networks form, function, and change over time. The Connected Past provides a constructive methodological and theoretical critique of the growth in research applying network perspectives in archaeology and history, and considers the unique challenges presented by datasets in these disciplines, including the fragmentary and material nature of such data and the functioning and change of social processes over long timespans. An international and multidisciplinary range of scholars debate both the rationale and practicalities of applying network methodologies, addressing the merits and drawbacks of specific techniques of analysis for a range of datasets and research questions, and demonstrating their approaches with concrete case studies and detailed illustrations. As well as revealing the valuable contributions archaeologists and historians can make to network science, the volume represents a crucial step towards the development of best practice in the field, especially in exploring the interactions between social and material elements of networks, and long-term network evolution.

Table of Contents

List of Figures
List of Tables
List of Contributors
Part I: Challenging Network Methods and Theories
1: Tom Brughmans, Anna Collar, Fiona Coward: Introduction: Challenging Network Perspectives on the Past
2: Carl Knappett: Networks in Archaeology: Between Scientific Method and Humanistic Metaphor
3: Astrid Van Oyen: Networks or Work-Nets? Actor-Network Theory and Multiple Social Topologies in the Production of Roman Terra Sigillata
Part II: Challenging Network Analysis of Archaeological and Historical Data
4: Matthew A. Peeples, Barbara J. Mills, W. Randall Haas, Jr., Jeffery J. Clark, and John M. Roberts, Jr.: Analytical Challenges for the Application of Social Network Analysis in Archaeology
5: Marten Düring: How Reliable are Centrality Measures for Data Collected from Fragmentary and Heterogeneous Historical Sources? A Case Study
6: Constantinos Tsirogiannis and Christos Tsirogiannis: Uncovering the Hidden Routes: Algorithms for Identifying Paths and Missing Links in Trade Networks
Part III: Challenging Network Models
7: Ray Rivers: Can Archaeological Models Always Fulfil our Prejudices?
8: Tim Evans: Which Network Model Should I Use? Towards a Quantitative Comparison of Spatial Network Models in Archaeology
9: Anne Kandler and Fabio Caccioli: Networks, Homophily, and the Spread of Innovations

Awesome archaeological networks tutorial using Visone

March 15, 2016

visone_logoTime to learn! If you ever thought about rolling up your sleeves and finally getting to do some actual network analysis then here is your chance: our new archaeological networks tutorial using Visone is out! The tutorial will teach you how to visualise, explore and analyse networks using the free-to-use Visone software. It will guide you through a case study on Maya obsidian networks in Mesoamerica by Golitko and colleagues. The tutorial is created by my colleague Daniel Weidele and I. Check out the tutorial on the resources tab of this blog.

Workshop computational history

March 11, 2016


This workshop will be of interest to many, including archaeologists. And I can definitely recommend going to the Digital Humanities conference, if only to visit awesome Krakow!

***3rd Workshop on Computational History (HistoInformatics 2016) – 11th July, Krakow, Poland***

Held in conjunction with Digital Humanities 2016, 12-16 July, Krakow, Poland


The HistoInformatics workshop series brings together researchers in the historical disciplines, computer science and associated disciplines as well as the cultural heritage sector. Historians, like other humanists show keen interests in computational approaches to the study and processing of digitized sources (usually text, images, audio). In computer science, experimental tools and methods stand the challenge to be validated regarding their relevance for real-world questions and applications. The HistoInformatics workshop series is designed to bring researchers in both fields together, to discuss best practices as well as possible future collaborations.
Traditionally, historical research is based on the hermeneutic investigation of preserved records and artifacts to provide a reliable account of the past and to discuss different hypotheses. Alongside this hermeneutic approach historians have always been interested to translate primary sources into data and used methods, often borrowed from the social sciences, to analyze them. A new wealth of digitized historical documents have however opened up completely new challenges for the computer-assisted analysis of e.g. large text or image corpora. Historians can greatly benefit from the advances of computer and information sciences which are dedicated to the processing, organization and analysis of such data. New computational techniques can be applied to help verify and validate historical assumptions. We call this approach HistoInformatics, analogous to Bioinformatics and ChemoInformatics which have respectively proposed new research trends in biology and chemistry. The main topics of the workshop are: (1) support for historical research and analysis in general through the application of computer science theories or technologies, (2) analysis and re-use of historical texts, (3) visualisations of historical data, (4) provision of access to historical knowledge.
HistoInformatics workshops took place twice in the past. The first one (http://www.dl.kuis.kyoto-u.ac.jp/histoinformatics2013/) was held in conjunction with the 5th International Conference on Social Informatics in Kyoto, Japan in 2013. The second workshop (http://www.dl.kuis.kyoto-u.ac.jp/histoinformatics2014/) took place at the same conference in the following year in Barcelona.

For our workshop at DH2016 we invite papers from a wide range of topics which are of relevance for history, the cultural heritage sector and the humanities in general. The workshop targets researchers who work on the intersections of history and computer science. We invite papers on the following and related topics:

•    Natural language processing and text analytics applied to historical documents
•    Analysis of longitudinal document collections
•    Search and retrieval in document archives and historical collections, associative search
•    Causal relationship discovery based on historical resources
•    Named entity recognition and disambiguation in historical texts
•    Entity relationship extraction, detecting and resolving historical references in text
•    Finding analogical entities over time
•    Analysis of language change over time
•    Modeling evolution of entities and relationships over time
•    Network Analysis
•    Automatic multimedia document dating
•    Simulating and recreating the past course of actions, social relations, motivations, figurations
•    Handling uncertain and fragmentary text and image data
•    Mining Wikipedia for historical data
•    OCR and transcription old texts
•    Effective interfaces for searching, browsing or visualizing historical data collections
•    Studies on collective memory
•    Studying and modeling forgetting and remembering processes
•    Estimating credibility of historical findings
•    Epistemologies in the Humanities and computer science

**Practical matters**

Submission deadline: 9th May 2016
Notification deadline: 31st May 2016
Camera ready copy deadline: 7th June 2016

Submissions need to be:

•    formatted according to Easychair paper formatting guidelines (http://www.easychair.org/publications/?page=1594225690).
•    original and have not been submitted for publication elsewhere.
•    submitted in English in PDF format
•    at the workshop’s Easychair page: https://easychair.org/conferences/?conf=histoinformatics2016.

Full paper submissions are limited to 10 pages, while short paper submissions should be less than 5 pages. Submissions will be evaluated by at least three different reviewers who come from Computer Science and History backgrounds. The accepted papers will be published on CEUR Workshop Proceedings (http://ceur-ws.org/).

Presenters and participants are expected to cover their travel and accommodation costs.

For any inquiries, please contact the organising committee at histoinformatics2016@easychair.org

**Organising committee**

•    Marten Düring (CVCE Luxembourg)
•    Adam Jatowt (Kyoto University)
•    Antal van den Bosch (Radboud University Nijmegen)
•    Johannes Preiser-Kappeller (Austrian Academy of Sciences)

**Programme committee**

•    Adam Kosto (Columbia University, USA)
•    Andrea Nanetti (Nanyang Technological University, Singapore)
•    Catherine Jones (Centre Virtuel de la Connaissance sur l’Europe (CVCE), Luxemburg)
•    Ching-man Au Yeung (Huawei Noah’s Ark Lab, Hong Kong)
•    Christian Gudehus (University of Bochum, Germany)
•    Daan Odijk (University of Amsterdam, The Netherlands)
•    Frederick Clavert (Paris Sorbonne University, France)
•    Günter Mühlberger (University of Innsbruck, Austria)
•    Lars Wieneke (Centre Virtuel de la Connaissance sur l’Europe (CVCE), Luxemburg)
•    Marc Spaniol (Max Planck Institute for Informatics, Germany)
•    Mike Kestemont (University of Antwerp, Belgium)
•    Nattiya Kanhabua (LS3 Research Center, Germany)
•    Nina Tahmasebi (University of Gothenburg, Sweden)
•    Pim Huijnen (Utrecht University, The Netherlands)
•    Robert Allen (Yonsei University, South Korea)
•    Roger Evans (University of Brighton, United Kingdom)
•    Tom Kenter (University of Amsterdam, The Netherlands)