Search

We present MapReader, a free, open-source software library written in Python for analyzing large map collections. MapReader allows users with little computer vision expertise to i) retrieve maps via web-servers; ii) preprocess and divide them into patches; iii) annotate patches; iv) train, fine-tune, and evaluate deep neural network models; and v) create structured data about map content. We demonstrate how MapReader enables historians to interpret a collection of ≈16K nineteenth-century maps of Britain (≈30.5M patches), foregrounding the challenge of translating visual markers into machine-readable data. We present a case study focusing on rail and buildings. We also show how the outputs from the MapReader pipeline can be linked to other, external datasets. We release ≈62K manually annotated patches used here for training and evaluating the models.

This chapter discusses the open access digitisation programme undertaken by Living with Machines, exploring the range of constraints that inform digitisation strategies and selection priorities. Because the landscape of digitised newspaper collections is so complex, and research and digitisation processes operate on different timelines, we have focused on opportunities to make digitisation choices both transparent and pragmatic. Working towards solutions that reflect collaborations between library staff and scholars, we introduce: a) Press Picker, our custom visualisation tool designed to support decision making about digitisation; and b) the Environmental Scan, a process of automatic metadata generation from the Newspaper Press Directories, a contemporaneous record of British newspapers.

This is the second of a two-part lesson introducing deep learning based computer vision methods for humanities research. This lesson digs deeper into the details of training a deep learning based computer vision model. It covers some challenges one may face due to the training data used and the importance of choosing an appropriate metric for your model. It presents some methods for evaluating the performance of a model.

This is the first of a two-part lesson introducing deep learning based computer vision methods for humanities research. Using a dataset of historical newspaper advertisements and the fastai Python library, the lesson walks through the pipeline of training a computer vision model to perform image classification.

Although the Ordnance Survey has itself been the subject of historical research, scholars have not systematically used its maps as primary sources of information. This is partly for disciplinary reasons and partly for the technical reason that high-quality maps have not until recently been available digitally, geo-referenced, and in color. A final, and crucial, addition has been the creation of item-level metadata which allows map collections to become corpora which can for the first time be interrogated en masse as source material. By applying new Computer Vision methods leveraging machine learning, we outline a research pipeline for working with thousands (rather than a handful) of maps at once, which enables new forms of historical inquiry based on spatial analysis. Our ‘patchwork method’ draws on the longstanding desire to adopt an overall or ‘complete’ view of a territory, and in so doing highlights certain parallels between the situation faced by today’s users of digitized maps, and a similar inflexion point faced by their predecessors in the nineteenth century, as the project to map the nation approached a form of completion.

This paper proposes a new approach to animacy detection, the task of determining whether an entity is represented as animate in a text. In particular, this work is focused on atypical animacy and examines the scenario in which typically inanimate objects, specifically machines, are given animate attributes. To address it, we have created the first dataset for atypical animacy detection, based on nineteenth-century sentences in English, with machines represented as either animate or inanimate. Our method builds upon recent innovations in language modeling, specifically BERT contextualized word embeddings, to better capture fine-grained contextual properties of words. We present a fully unsupervised pipeline, which can be easily adapted to different contexts, and report its performance on an established animacy dataset and our newly introduced resource. We show that our method provides a substantially more accurate characterization of atypical animacy, especially when applied to highly complex forms of language use.

In this half-day tutorial, we will review the basic concepts of, methods for, and applications of geographic information retrieval, also showing some possible applications in fields such as the digital humanities. The tutorial is organized in four parts. First we introduce some basic ideas about geography, and demonstrate why text is a powerful way of exploring relevant questions. We then introduce a basic end-to-end pipeline discussing geographic information in documents, spatial and multi-dimensional indexing, and spatial retrieval and spatial filtering. After showing a range of possible applications, we conclude with suggestions for future work in the area.

In this half-day tutorial, we will review the basic concepts of, methods for, and applications of geographic information retrieval, also showing some possible applications in fields such as the digital humanities. The tutorial is organized in four parts. First we introduce some basic ideas about geography, and demonstrate why text is a powerful way of exploring relevant questions. We then introduce a basic end-to-end pipeline discussing geographic information in documents, spatial and multi-dimensional indexing [19], and spatial retrieval and spatial filtering. After showing a range of possible applications, we conclude with suggestions for future work in the area.

In this paper we use network analysis to identify qualitative “neighbors” for toponyms in an eighteenth-century French encyclopedia, but could apply to any entry-based text with annotated toponyms. This method draws on relations in a corpus of articles, which improves disambiguation at a later stage with an external resource. We suggest the network as an alternative to geospatial representation, a useful proxy when no historical gazetteer exists for the source material’s period. Our first experiments have shown that this approach goes beyond a simple text analysis and is able to find relations between toponyms that are not co-occurring in the same documents. Network relations are also usefully compared with disambiguated toponyms to evaluate geographical coverage, and the ways that geographical discourse is expressed, in historical texts.

Historians investigating evidence with spatial significance increasingly rely on gazetteers to identify the location of geographical features/places. Existing digital gazetteers cater to twenty-first century or discrete historical geographies (the classical world, for example). Early modernists (ca. 1450-1750), particularly those who work on non-Anglophone cultures, represent a major scholarly community with no temporally-appropriate gazetteers available. In this paper, we introduce a project that fills this research infrastructure gap. Mapping place names in the canonical eighteenth-century Encyclopédie is a case study for semi-automating the identification, classification, and location of places and spatial relations in historical geographic reference works printed in French. We demonstrate the challenges of using existing geoparsers and introduce our plan for new tools and protocols for working with historical French texts.

I capture a turning point in territorial development when public works outgrew medieval organizational methods, and provincial leaders sought to establish new controls over land and labor. In their approach to public works, leaders of the French province of Brittany took older compulsory labor and estate management practices and shaped them into enlightenment-era administrative methods. Breton nobles, urban magistrates, engineers, and villagers reshaped their political and professional identities at eighteenth-century highway construction sites. Road construction introduced a new kind of political interaction between country and city. My study of provincial highway construction brings to light long overlooked connections between early modern people, their local built environment, and their political sensibilities. "Public" once meant royal, but the road construction experience was key in broadening the word's definition to include civic engagement and provincial administration. Use of the corvée to build highways introduced rural communities to the idea of provincial public administration. The experience of highway construction offered the Breton Estates an opportunity to rise in political and administrative prominence. With the corvée, the provincial population participated in the transfer of technological information among experts, estate owners and managers, and local practitioners. While social and economic reform in Old Regime France is usually discussed in terms of orders or classes, a person's perception of his responsibilities to village, province, and kingdom was spatial. The highway -- part of the royal domain -- was culturally transformed by the act of construction into an active, communal space. The experiences of workers in this space, combined with debates about corvée reforms, laid the groundwork for thinking about roads and other parts of the public domain as democratic spaces.