A couple of weeks ago, the was held at ������Ƶ (20^th—21^st September). This conference focused on exploring what geospatial technologies such as Geographic Information Systems (GIS) have to contribute to humanities research. The main aim was to explore and demonstrate the contributions to knowledge enabled by these technologies, approaches and methods within and beyond the digital humanities. To read a little more about how the conference progressed, take a look at the published on the ������Ƶ History Department website, and find further discussions at on Twitter.

At this conference, we presented two papers related to the Digging Into Early Colonial Mexico project. On the first day, (University of Lisbon) presented ‘Exploring the challenges of Named Entity Recognition in an historical multilingual corpus: Digging into Early Colonial Mexico’. This paper focused on two of our project’s key aims: the creation of the first digital 16th century Spanish-Nahuatl Colonial Gazetteer, as well as a comprehensive Geographic Information System of New Spain. As we have mentioned in a , our corpus presents a key challenge for Natural Language Processing (NLP) – how do we accurately perform Named Entity Recognition tasks in a multilingual corpus, and particularly one with a combination of European and non-European languages? Bruno’s presentation addressed these challenges and explored ways in which they can be resolved, as well as showing some preliminary results of the NLP experiments carried out so far in our project.

On the second and final day of the conference, (������Ƶ) presented ‘Development of an Historical Place-Name Gazetteer for the Viceroyalty of New Spain’. This paper described the main principles behind the development of our gazetteer, the process used for collecting and integrating data from multiple sources, the resulting software for managing and exporting the data (available as open-source at ), and lessons learned from our efforts that could be useful for the development of similar gazetteers. Our gazetteer has adopted the data model of the , which has already considered the association of places to multiple alternative place-names, feature-types, detailed geographical extents, quality and provenance information, and temporal ranges for all the aforementioned elements. For a visual on how these wonderful tables fit together in our project, click .

In addition to these two presentations on our project, our Principal Investigator, , presented the opening keynote of the conference, ‘Subaltern Spatial Thinking: Reflections on the technological integration of non-western and non-cartographic thinking in Humanities research’. An exploration of how GIS has been adopted in Humanities research and the problems this might bring, Paty spoke about how cartography can be seen as a tool of colonial hegemony and power, and how the Humanities need a critical view on the adoption of this or any technology. Advocating for a decolonial approach to technology, this paper considered examples of Mesoamerican and colonial spatial thinking which may be unfamiliar to a modern and western-centric gaze. The talk included a walk through the , and some of the maps of the 16^th century Relaciones Geográficas.

This opening keynote set a valuable tone for conference, highlighting non-western and non-cartographic thinking in Spatial Humanities research, and the importance of recognising alternate methods of representing and analysing space and place in historical research.

A key element of our research on the Relaciones Geográficas is the analysis of the textual information contained within these sixteenth century reports. To do this, we will be drawing on techniques from Computer Science, namely Natural Language Processing (NLP) and Machine Learning. Whilst these areas have been studied a great deal, the vast majority of this research has been conducted using modern languages, and largely in English.

Our corpus is, of course, neither modern nor English. The Relaciones Geográficas that we are currently studying were written in the sixteenth century by Spanish officials and contributed to by indigenous people across Mexico. The mix of Spanish and indigenous languages throughout the Relaciones poses a challenge to these computational methods which have, for the most part, been trained on modern text. ��We are therefore faced with the task of training our own NLP system which takes into account the unique challenges presented by the Relaciones Geográficas.

Corpus Annotation

We have recently established a partnership with , an NLP-tech company that has developed an online text annotation tool with the capacity to train models to annotate large quantities of textual information. Tagtog offers a free version which allows a single user to work with up to 100 documents and to make use of its Machine Learning automatic annotation capabilities. For more information on their free and paid plans, head on over to their .

So far, we have been using tagtog to annotate a few excerpts from our corpus. Annotation essentially means assigning metadata about specific terms or phrases in order to train the machine to recognise key words. For example, in the excerpt below, we have tagged “Yenynguia” as a place-name (you may note that this place is also known as Coyula, which can be recorded through the use of dictionaries – as explained later in this post).

Before annotating anything, it is important to define the types of entity that need identifying within the text. We started with a few key categories (such as place-names, institutions and geographic features), and have since expanded to include around forty categories which reflect the diverse nature of information contained within the Relaciones. Whilst this is a considerable number of categories to be using, tagtog is coping wonderfully so far!

Below is an excerpt from the Relacion de Papaloticpac (Antequera) which gives an indication of the sort of information we have been annotating with tagtog.

Within the first 800 words or so of this Relación, we already have useful information highlighting numerous pueblos in the area and their location in relation to one another, the relevant “ilustre” and “muy excelente” señores involved in the production of this report, as well as some indication of geographic features in the area: cerros, sierras and quebradas. This is all valuable information which we want to be identifiable for textual analysis.

Dictionaries

As mentioned earlier, in cases where we have alternate names for one place (Yenynguia = Coyula), it is possible to use dictionaries to tell the machine that these entities are one and the same. With the inconsistencies of spelling throughout the Relaciones Geográficas, normalization of entities is essential. You may be able to see that, within the first few lines of the excerpt above, we are given three different spellings of the named pueblo. After ‘Papaloticpac’, we have ‘Papaloticpaque’ as well as ‘Papalotiquipaque’. Of course, this is referring to the same place, but the machine needs to be told this explicitly. In tagtog, this is made possible through the use of dictionaries which enable the normalization of entities. So, in the case of ‘Papaloticpac’, we would include each spelling in the ‘dictionary’ as follows:

(note that the all-caps instance of the word also has to be included for the machine to recognise this as a match)

Our next steps, once we have annotated some more of the corpus, will be to train a model using the annotations we have created. To do this, we will import some ‘raw’, un-annotated text, which the machine will annotate automatically given what it has learned from our manual annotations and dictionaries. Of course, this will not produce 100% accuracy. We will proceed to manually correct any errors, repeating this process until we produce a high level of accuracy. Training a model which can accurately produce automatic annotations will enable far more intuitive interaction with our multilingual corpus of over 3 million words.

Following our last post, , in which I mentioned the problems we face in automatized identification of place-names, I thought it would be worthwhile to take a look at the toponyms we are working with, and why using computational approaches will allow us to further our understanding of the Relaciones Geográficas.

One of our first, and ongoing, challenges with this project is the identification of thousands of place-names across Mesoamerica. The source materials for the gazetteer we are currently compiling include:

Rene Acuña’s

Mercedes de la Garza’s

Alejandra Moreno Toscano’s

Francisco del Paso y Troncoso’s

Peter Gerhard’s and

Our first task was cleaning and converting each of these sources into a computer-readable format, allowing us to extract data more easily. OCR was (sometimes) our friend for this part of the process. We were then able to extract all the place-names listed in the indexes of these works (correcting OCR mistakes along the way), leaving us with a list of almost 14,500 toponyms. Of course, many of these are duplicates or alternate spellings of the same place. We are currently disambiguating these place-names to ensure we are referring to the correct location. (I described this process in our post if you’d like a little more detail about this.)

The wordcloud below was created from the full list of toponyms listed in Rene Acuña’s editions of the Relaciones Geográficas, excluding alternate spellings for the same place. If I had included the alternate spellings, the list would have been over 6,200 names. As it was, I inputted a list of around 4,900 toponyms.

The influence of the Spanish language is clear, though not surprising, with names of saints featuring prominently alongside common descriptors such as Río, Valle and Laguna. However, indigenous toponyms remain prevalent, with frequent mentions of specific locations such as Acámbaro, Tlaxcala and Ixtlahuacan. Yucu, a Mixtec word meaning ‘hill’, appears 33 times, no less frequently than Valle. The occurrence of Yucu in this source material was also exclusively within the region of Antequera (currently Oaxaca), explained by the region being home to the convergence of numerous mountain chains, known as the Complejo Oaxaqueño (Oaxaca Complex).

Disambiguating the thousands of place-names which are mentioned in the Relaciones Geográficas will allow us to effectively interact with the source material using computational methods. Using techniques such as Collocation Analysis in conjunction with our gazetteer will open up opportunities for analysing the text in innovative ways, such as identifying associations between locations, entities, topics etc. For example, it should be possible to search for Tlacotepec and determine whether this place has any relationship to another place, person, or concept. Furthermore, it will be possible to search for the specific Tlacotepec which you may be interested in, and any associated alternate names/spellings for that particular place. As the map below demonstrates, place-names are often repeated across, and within, regions. This is why disambiguating our corpus is so important!

At present, we have a total of 3,650 fully disambiguated place names – meaning that we have definite coordinates assigned to these names. You can see a sample of some of these locations on the tab of our website.

We have a fair few more toponyms which are partially located (i.e. we have identified the region in which they lie), and thousands more awaiting disambiguation. We’re approaching the halfway point…just over the next yucu!