HNR Lunch Lectures

From January 2021 onwards, monthly online lectures will shed a spotlight on recent research and ideas from the field of historical network analysis to promote discussion among the HNR community. If possible, the lectures will be recorded and made available to the HNR Youtube channel.

New lectures will be announced over the HNR Newsletters with infos on speakers, topics and registration. It is possible to register for individual lectures as well as for the whole lecture series.

Program:

Date Speaker
Nov 24th, 2022 Anastasia Glawion, Retrieving online memory practices: an network analysis pipeline
Sep 15th, 2022 Sébastien de Valeriola, Can historians trust centrality?
Jun 16th, 2022 Bernardo Buarque, Roberto Lalli, Malte Vogl and Dirk Wintergrün, Modelling socio-epistemic networks: Tool-oriented proposals for strengthening the community building in HNR
Mar 10th, 2022 John R. Ladd, Comparing Networks of Biography: Six Degrees of Francis Bacon and John Aubrey’s Brief Lives
Jan 20th, 2022 Erik Hornung & Allard Mees, Roman Transport Network Connectivity and Economic Integration
Nov 18th, 2021 Mehwish Alam, Graph Representation Learning
Oct 7th, 2021 Cindarella Petz, Network models and configurations for conviction
May 20th, 2021 Rachel Midura, Early Modern Digital Itineraries (EmDigIt): Networks of European Space, 1545-1761
Apr 15th, 2021 Ramona Roller, Modeling time in letter correspondence networks of the European Reformation: beyond snapshots towards temporal paths
Mar 18th, 2021 Diane Cline, People, Places, and Things in Contexts: Field Maps and the Ancient Greeks at Sea
Febr 18th, 2021 Henrike Rudolph, (Re)Constructing Historical Narratives from Collective Biographies
Jan 21st, 2021 Demival Vasques Filho, Extracting large-scale network data from archives

 

Abstracts

Nov 24, 2022: Retrieving online memory practices: an network analysis pipeline

by Anastasia Glawion

In this talk, I present a pipeline of network analysis applications to the studies of user communities on online forums. In my dissertation, this pipeline was applied to the study of a historical online forum dedicated to the memory of World War II, resulting in the extraction of three large groups of memory practices – empirical, conversational and conservational practices. Each group of practices contained several more specific subgroups. For example, empirical practices were divided into those aimed at memory objects and those aimed at places of memory. Conversational practices were characterized by varying degrees of sentiment and ranged from neutral military-history discussions to heated debates on national memory. Finally, conservational practices were only possible because of the actions of brokers: users who had access to archives were sharing that access with the forum community.

The methodological pipeline consists of three steps. First, interactional data is extracted and a modularity clustering algorithm is applied, thus identifying dense subgroups within the network. In a second step, for exploration purposes, a textual corpus is created on the basis of the dense subgroups. A topic modeling algorithm is applied to the corpus, whereby the resulting topic model can be presented as a term-overlap network to demonstrate dense subgroups that also have strong thematical overlap. The third step contains the analysis of the discussion level while considering each discussion’s role in the dense subgroup extracted in the first step with the help of structural equivalence in networks.

Register for this event via EventBrite

 

Sep 15, 2022: Can historians trust centrality?

by Sébastien de Valeriola

Figure from Sébastien de Valeriola, ‘Can historians trust centrality? Historical network analysis and centrality metrics robustness’, Journal of Historical Network Research 6 (2021), 119

In this talk, we consider four measures of centrality in their use in the analysis of historical networks. Since the sources used by historians to construct such networks are by nature incomplete and imperfect, it is necessary to take into account as much as possible the robustness of these metrics, i.e., their stability with respect to the hazards that time has inflicted on historical documents. To study this, we apply a battery of tests to three networks constructed from medieval history data. These tests are designed to simulate the processes of disappearance and degradation of the information contained in sources by imitating as closely as possible the situations historians face when manipulating graphs. Our results allow us to assess the general relevance of the use of centrality in historical network analysis, to compare the four metrics studied in terms of robustness, and to identify a set of methodological points to which the historian applying such techniques must pay particular attention.

 

Jun 16, 2022: Modelling socio-epistemic networks: Tool-oriented proposals for strengthening the community building in HNR

by Bernardo Buarque, Roberto Lalli, Malte Vogl and Dirk Wintergrün

The talk gives an overview of the basic theoretical framework and the operational approaches of the project ModelSEN-Socio-epistemic networks: Modelling historical knowledge processes. Started in April 2021, the project aims at developing methods of longitudinal multi-layered network analysis of the dynamics of knowledge creation and diffusion. During the talk we will especially focus on the operational developments in three areas: first, the conceptualization and creation of ontologies for the general description of networks that could in principle be used in a variety of projects; second, the development of software packages for the creation and analysis of multi-layered networks; and third, the agent-based modeling approaches developed to study socio-epistemic processes of diffusion of innovation and consensus creation. While these tools have been developed on the basis of one case study (the history of general relativity), the central goal of the ModelSEN project is to promote and support dialogue among practitioners in the HNR community regarding the identification of standards and best practices in the various stages of historical network research projects. The talk is then meant to open the floor to the discussion about future steps for the elaboration of tools that can address common challenges and for the creation of frameworks that allow a greater flow of data and tools among projects.

 

Mar 10, 2022: Six Degrees of Francis Bacon and John Aubrey’s Brief Lives

by John R. Ladd

A portion of the SDFB network, using role detection for node color

A portion of the SDFB network, using role detection for node color

John Aubrey devotes a distinct subset of the biographical sketches in Brief Lives to famous authors of his recent past, helping to cement a developing early modern canon. By contrast, Six Degrees of Francis Bacon contributes to a tradition of expanding canons by reorganizing its data—drawn from the Oxford Dictionary of National Biography—as a network of early modern relationships. Using role detection, in contrast to traditional community detection techniques, I consider the status of authors within Aubrey’s single-source network of biographies and the much larger secondary-source-based SDFB network. I argue that Aubrey presents authorship as a series of social interactions and collaborations, and likewise early modern social networks reveal that the work of authorship is distributed across a system that includes writers and a diverse group of other professionals. Drawn from my book project, Network Poetics, this talk will cover the generation of networks from biographical sources, the uses and limitations of the SDFB data set, and the application of role detection to small and large historical networks.

Watch this lecture on the HNR YouTube channel

Jan 20, 2022: Roman Transport Network Connectivity and Economic Integration

by Erik Hornung and Allard Mees

The figure shows bilateral trade flows in Terra Sigillata during Roman Era. It maps trade flow shares of terra sigillata between grid-cell pairs. Each colour is specific to one origin grid cell.

Bilateral trade flows in Terra Sigillata during Roman Era Figure maps trade flow shares of terra sigillata between grid-cell pairs. Each colour is specific to one origin grid cell.

We show that the creation of the first integrated multi-modal pan-European transport network during Roman times influences economic integration over two millennia. Drawing on spatially highly disaggregated data on excavated Roman ceramics, we document that contemporary interregional trade was influenced by connectivity within the network. Today, these connectivity differentials continue to influence integration as approximated by cross-regional firm investment behaviour. Continuity is partly explained by selective infrastructure routing and cultural integration due to bilateral convergence in preferences and values. We show that our results are Roman-connectivity specific and do not reflect pre-existing patterns of exchange using pre-Roman trade data.

*Based on our paper:  Flückiger, Matthias, Erik Hornung, Mario Larch, Markus Ludwig, and Allard Mees. “Roman transport network connectivity and economic integration.” (forthcoming). Review of Economic Studies. https://doi.org/10.1093/restud/rdab036

Watch this lecture on the HNR YouTube channel

 

Nov 18, 2021: Graph Representation Learning

by Mehwish Alam

The Linked Open Data Cloud

This talk focuses on ways to perform representation learning on graphs such as citation network, author collaboration network, etc. It then moves on to the graphs with relational information, known as Knowledge Graphs. As they are typically used in an open-world setting, Knowledge Graphs can almost never assumed to be complete, i.e., some information will typically be missing. In order to address this problem, different Knowledge Graph embedding models have been proposed for automated Knowledge Graph completion. These models are mostly based on the tasks such as link prediction, triple classification, and entity classification/typing. This talk will also target the topic of Knowledge Graph embedding techniques. Finally, various applications of Knowledge Graphs and Knowledge Graph embeddings will be discussed.

Watch this lecture on the HNR YouTube channel

 

 

Oct 7th, 2021: Network models and configurations for conviction

by Cindarella Petz

Figure of a multimodal sample network of 88 court proceedings, featuring 201 defendants in 204 cases with verdicts, 55 judges and 27 prosecutors. An edge equals a conviction. On the right: Figure focusing on one defendant in a random single case’s court trial interpreted as a network. The suspect shown here was a twenty year old metal worker and member of the Socialist Youth, who was – individually charged alongside fifteen others – accused of forbidden gathering, discrimination against the administration and general societal values, and of spreading rumors. The suspect was sentenced in June 1935 due to the discrimination charges, and was imprisoned until June 1936. Layout algorithm used: Fruchterman-Rheingold in R::igraph.

Figure of a multimodal sample network of 88 court proceedings (edge equals a conviction)

In the case study* presented in this talk, we evaluate the extend of political judiciary during the Corporate State of Austria based on a dataset of over 1,800 court cases tried at the provincial courts of Vienna I and II in 1935. This dataset was originally curated in the process of a project on political repression by the University of Vienna (Wenniger/Mesner/Ardelt 2015–17).

We propose to model these court trials as networks, thus utilizing networks as epistemic tools for structural analysis.

We show how a quantitative-based approach combined with a qualitative evaluation and contextualisation can further historical insights on the judicial practice and structural forms of political judiciary in these courts. Additionally, we point out to chances and pitfalls of the proposed mixed methods approach to provide a critical perspective to modeling big data in digital legal history, and address limitations of working with judicial records.

*Based on our paper: Petz, Cindarella & Pfeffer, Jürgen (2021). Configuration to Conviction. Network Structures of Political Judiciary in the Austrian Corporate State. Social Networks, vol. 66, July 2021, pp. 185–201. DOI= 10.1016/j.socnet.2021.03.001.

Watch this lecture on the HNR YouTube channel

 

May 20th, 2021: Early Modern Digital Itineraries (EmDigIt): Networks of European Space, 1545-1761

by Rachel Midura

Due to the time difference to the US, the lecture will take place as an afternoon tea lecture at 4 p.m. CET

From dictionaries to gazetteers, the early modern period was the age of the reference book, with more than 1,600 guides published between 1470 to 1599 alone. Itinerary books structured European conceptualization and navigation of space through the eighteenth century, yet have rarely been studied as a text technology and international knowledge project. The EmDigIt database represents the first attempt at a comprehensive bibliography, and currently consists of eighty-four itinerary books representing over 3,500 unique routes, connecting over 1,500 locations, published and republished over the course of two centuries. The talk will cover applications of Social Network Analysis in a forthcoming article for the Journal of Social History, as well as next steps for the project using the Transkribus text recognition software.

Watch this lecture on the HNR YouTube channel

 

Apr 15th, 2021: Modeling time in letter correspondence networks of the European Reformation: beyond snapshots towards temporal paths

by Ramona Roller

Converting time stamped edges to temporal paths

Interactions between individuals are inherently temporal. For example, letters are sent and received at a specific date. Despite this temporality, we commonly model interactions with aggregated networks where temporal relations are ignored. As a consequence, interactions at a later point in time can affect those at an earlier point: chronology is broken leading to flawed network statistics, such as centrality measures. I will address this problem by using network models that are based on paths of interactions rather than on classic interaction dyads. By applying these paths to a letter correspondence network of European reformers from the 16th century, I show that central reformers become less important when accounting for time and the spread of ideas gets restricted to specific routes.

Watch this lecture on the HNR YouTube channel

 

Mar 18th, 2021: People, Places, and Things in Contexts: Field Maps and the Ancient Greeks at Sea

by Diane Harris Cline, Ph.D.

Field map for the ecosystem for Greek shipping in the ancient Mediterranean

The topic for discussion is the introduction of ‘field maps’, which combine sociograms with an interpretive layer of fields (e.g. economic, social, technological, infrastructure, religious) to analyse the ways humans and things are entangled. ‘Field map’ is a term which I have coined for this kind of data visualization. These fields are layered over the sociogram to show  the larger chains of interactions that tie people, their activities, places, and non-human things in the social puzzle.  To demonstrate the application of field maps, we will use as a case study the ties between olive oil, wine, and the sea for the ancient Greeks, and in this way we can model the entanglement inside the complex systems of an ancient Greek merchant ship in its contexts at sea and in ports. Using this adaptation of sociograms, we explore the role of material things in maintaining social relations, inspired by materiality studies, entanglement, Actor-Network Theory, sensory experience studies, and social network analysis.

Watch this lecture on the HNR YouTube channel

 

Febr 18th, 2021: (Re)Constructing Historical Narratives from Collective Biographies

by Henrike Rudolph

Collective biographies follow a particular logic of compilation: they depict individuals deemed either exceptional or representative for a certain period or group and emphasize such exceptionality or representativeness in the life stories. The resulting biases and gaps render collective biographies a challenging resource granting only limited insights into universal traits or biographical details. As I will argue in my talk, however, once we employ network analysis, collective biographies can become objects of study in their own right as markers of dominant historical narratives. Based on a case study of collective biographies of women activists in twentieth-century China, I will explore the possibilities of tracing recurring tropes, reflect on the importance of the narrative construction of indirect social ties, and critically examine temporal aspects in biographical sketches. The proposed analytical framework, which uses both one- and two-mode graphs, can be applied to revisit collective biographies or other biographical corpora in different historical contexts.

Watch this lecture on the HNR YouTube channel

 

Jan 21st, 2021: Extracting large-scale network data from archives

by Demival Vasques Filho

Network of documents of administrative correspondence of the Portuguese Empire

Network of documents of administrative correspondence of the Portuguese Empire

The paper Networks from archives: Reconstructing networks of official correspondence in the early modern Portuguese empire–recently published in the journal Social Networks, with my co-authors Agata Bloch and Michal Bojanowski–addresses the problem of processing a large corpus of unstructured textual information to extract network data. Our corpus consists of almost 170,000 documents of administrative correspondence of the Portuguese empire with its Atlantic overseas territories, from 1610 to 1833, cataloged in the Portuguese Overseas Archives of Lisbon. In this HNR lunch-lecture, I will first talk over the methods we used to extract network data from natural language corpora, with detailed technical information about the six steps for creating network data as we described in the paper. These steps include manually annotating texts, developing a named entity recognition model, using regular expressions to identify relational (sender/recipient) data, and avoiding duplicates among the entities extracted. Second, focusing on networks analysis, I will discuss the methods we envision to perform (future work) with this dataset in order to engage with the historiography of the early modern Portuguese empire and the discussion about the multi-continental monarchy theory. This discussion will cover the limitations of correspondence data and the reconstruction of multilayer networks of actors and institutions.

Watch this lecture on the HNR YouTube channel