Foods of Abu Dhabi

Sociological Food Histories in a Visual Format

Crafting an investigative, often interpersonal research project can be difficult for shy people. However, the biggest challenge with our Abu Dhabi Foods data set is the freedom of choosing the restaurants. By foregoing a methodical street-by-street calculation, it remains relatively unclear which areas where built up when, which would be my personal area of interest. The Al Raha beach area, for instance, is often discounted by long-term city residents, but this apparently new area seems more popular among expat communities, especially those living out in that part of the city. However, being relatively nearby Zayed University, it’s also possible that Al Raha or areas close to it began building up before one would guess, and without documenting every single restaurant in a given area, there’s no way of knowing when the first restaurant was built there. It’s also hard to gauge the popularity of new restaurants in one area over the other without a comprehensive review of the restaurants in both areas. To further complicate representing the data, either due to personal oversights of data-takers or due to issues with the exportation of the data, some values for data_established appear as 0. For this reason, the date line in the map starts at 0 and I cannot figure out how to change it to a more reasonable date. Further, I’m uncertain whether it would be better to include these places and note that 0 is an unknown or whether I should just eliminate these data points. Because they were numerous, I favoured the former option for one and included these unknown places in the visualization instead of just erasing them.

For the map on a separate webpage however, I decided to edit the .csv dataset, deleting all entries with 0 for date established, uploaded that into Carto and created the map visible here. What I prefer about this 2nd map is that the critical years go by at a reasonable pace. In the first map, going from 0 to 2016, it was necessary to use a fast pace to get through 0-2000-something, and I was unable to determine how to change the speed for the 2000’s. However, one issue I did have with the second map, and it’s quite blatantly apparent in the result, is representing the timeline accurately. The widget at the bottom of the final version has number values that appear to correlate with the “steps” of the timeline, or when it shows one set of data and when it shows the next. I hope to eventually correct this problem and better represent the timeline.

In the future, I would narrow the scope of the research so that specific neighborhoods might be tackled and there will be a higher concentration of data in those areas. This would however likely effect variations in delivery options. While this would also take a lot more interviewing and research, I personally became fascinated by the nationalities of the founders and the chefs. I would find it incredibly interesting to know what restaurants were started by Emiratis and which by foreigners; however, that does raise an issue of massive procrastination because it’s more difficult to find out this information. As seen below, student procrastination was common for data collection even with our relatively simple information required. (It should be noted that I wanted to create these as hexbins for some variety and ease of viewing the data; however, unfortunately, the data would be need to be one numeral, e.g. a year or the like, for this functionality in Carto).

 

However, these maps prove useful to travelers and residents by offering functionality different than that of Zomato or TripAdvisor. For instance, below you can see Emirati restaurants across Abu Dhabi. If you want cheaper food, that information is easily accessible by hovering over the point. This is important for those willing to explore new areas or venture further out for cheaper food and different atmospheres and typically absent from Zomato, the popular app in the Emirates, that focuses on your neighborhood and makes a city-wide search relatively inconvenient.

UAE Map Rectification and Comparison: A Snapshot from Year 1 to Year 45

In honor of GIS Day, a.k.a. Geographic Information Systems Day, and, to some extent, the UAE’s 45th National Day, our Digital Humanities mini-tutorial of the week at the NYUAD Center of Digital Scholarship was on map rectification or geo-referencing using old Soviet maps and current satellite images. The process involved creating shape files of the satellite images, putting them into a program, and matching coordinates or features together. Further explanations are provided by our professor, David Wrisley.

Complicated technical details aside, on a philosophical level, GIS Day was an interesting exercise in transparency and definitions of reality. Often, particularly growing up in America, maps seem incontrovertible in terms of their veracity: bodies of water are defined; borders are established. While these may just be lines drawn on a map when you’re wandering through the woods, the borders seem very clear and delineated when a sign along the highway welcomes you to a new state at a specific point. However, as proven by America’s relentless debates on borders and immigration, as well as Africa’s and the Middle East’s colonization, borders are often far from distinct, incontrovertible separations of states and nations.

In regards to the Soviet maps specifically, the choice of map accuracy is illuminating and peculiar. As outlined in the article, Hyper Detailed Soviet Maps of Washington, Soviet precision was remarkable as pertained to government buildings and vital infrastructure, but haphazard when charting residential areas. Before this exercise, I had never assumed that maps would be “filled in” with such carelessness. I imagined maps either as fully researched entities or transparent charts that would say, circle an area lacking detail and simply note “residential area,” instead of providing approximated or fictionalized houses. In our own project, you could see distinct differences in the water borders. Any number of things could have caused this difference: changes in the tide, long-term environmental changes, inaccuracies in the Soviet map, etc. Thus, unfortunately, further information, from the site itself or other maps, is required to determine the actual root of the discrepancy.

gis-mapping

Consequently, the project was simultaneously encouraging and discouraging. The subjective decisions taken in the map-making process allows greater analysis in terms of the mapmakers’ motivation. However, this step also necessitates prior knowledge of the area due to the opaque nature of maps. In the Soviet Union, where access to maps of Washington, DC. was likely limited, there’s little possibility for the readers of the map to know of its discrepancies with reality. Map scaling and “bucketing” of certain information only exacerbates these problems of precision as it draws away from the specific data and into a simpler representation. Of course, whether the specific data is necessary will vary based on the map’s goals and subjects; however, while university teaches these ideas, it’d be nice to see more dissemination of these ideas of GIS and subjectivity in high schools. Such a change in the general curriculum would ensure greater awareness and critical thinking skills among the general populace, of whom some cannot afford higher education, and likely foster greater interest in the oft-overlooked field of geography in American classrooms.

Networking with Nodegoat

Our recent project networking with Nodegoat evidenced how quickly  webs of information can become complicated. For instance, the network of films, their directors, and relevant actors, writers, etc. quickly became a bit muddled in our Egyptian cinema project. The names of films and people overlap; the exact connections between that one large mass on the left and its branches blur. Although the outline colours can be charged, there is no clear-cut way to choose one colour for one type and another colour for another type, facilitating easy visualization. These have to be established through “conditions” within the design of the data, instead of, for instance, in the visualization panel.
Nodegoat-cinema-network

Meanwhile, devoid of spatial data, the geographic representation of the project is currently meaningless, as seen here:

nodegoat-cinema-map

Conversely, other projects are not conducive to social representations, but work fairly well with geographical representations. For instance, this image presents a preliminary work of Sudanese authors and their “migration,” from birthplace to university to current residence (that will hopefully include lines representing their migration patterns in the future).

sudan-prelim-authorsAlthough this map does not yet express the distinction between universities, birthplaces and cities of current residence, the points are revelatory simply by representing where the current Sudanese literary “elites” have lived at some point over the courses of their lifetimes. However, it’s likely that even this map won’t present the whole picture when complete, principally because I have left out childhood statistics. Due to the relative difficulty of ascertaining where all these authors went to primary school or in what specific city they lived, I have decided not to include this data at all. Similar phenomena would have occurred even in the overly complicated network of Egyptian cinema shown in the first image. A young actor or director may have played a minor role in one of the included films before becoming a star and taking on a role large enough to constitute Person 1, 2 or 3. Therefore, there would be a link between that person and a film that wouldn’t be documented in the network.

Ultimately, Nodegoat tries and overall succeeds at simplifying the process of creating networks and relationships. However, the very nature of these networks often makes them inherently complex, with countless executive decisions to condense and present the information in an easily digestible way.

Brief Overview of Paris’ 2005 Race Riots

In October and November of 2005, riots broke out in Paris as the result of grievances of those living in Paris’ banlieues, roughly similar to American slums or ghettos. Most of the banlieues’ residents are immigrants, or the children of immigrants, and never felt fully accepted by French society. As unemployment aggravated these discontents and violence affected one banlieue in particular, Clichy-sous-Bois, the citizens rose up in protest.

Here’s a brief timeline:

  • 27 October: Death of two juveniles in Clichy-sous-Bois due to police
  • 30 October: Tear gas grenade thrown into a mosque in Clichy-sous-Bois
  • Riots…
  • 8 November: Declared state of emergency
  • 17 November: Violence abates, but still apprehensive

In my corpus, I attempted to identify the different perceptions of the events as expressed in different newspapers following different events. My search began with results for the word “émeutes,” or riots, in Libération (a left-wing paper), Le Monde (a center, left-leaning paper), and Le Figaro (a right-wing paper) over November 7th, 8th and 9th in 2005. I choose these particular dates to highlight perceptions on a day when tensions were likely highest (November 7th), the day action was taken (November 8th) and the following day, when concerns may have eased. The results, as seen above, were less revealing than I hoped. Many neutral words consistently appeared in all three papers, like police, France and banlieues. As a consequence, the word cloud shows the elements contributing to tensions but not the nuances of the publics’ perceptions of the events.

Further, the inability to add metadata in VoyantTools created a number of difficulties. For instance, I couldn’t sort by date, paper and switch between the two. Instead, I had to group three articles from each paper on each date into one “text” and use that for comparisons.

Additionally, outside of metadata, I experienced a few other problems with the corpus, namely:

  • I did not take the time to distinguish between full-time journalists, free-lancers for the papers, and opinion papers or letters to the editor
  • Certain words could be used sarcastically or in quotation without referencing the theme in the manner it appears on paper
  • I randomly selected papers from among the many published on each date
  • There were search limitations (I relied on the one word, riots or émeutes, and date ranges to find relevant articles but a particularly pro-protest article may not use the word riot or have it as a tag)

This final graph reveals a bit more information. For instance, the sympathetic term, “jeunes,” was used significantly more often in Libération, the left-wing paper, than in other papers, like the right-wing paper, Le Figaro. However, the majority of the terms were too neutral to gain any useful information and one word, “feu,” is difficult to analyse and understand as “feu” alone means fire but “couvre-feu” means curfew. In the context of the riots and the state-of-emergency, both fire and curfew were relevant and therefore, without further programming, the word “feu” alone cannot reveal anything useful. Ultimately, I could have heavily improved the project with the use of metadata and increased programming skills.

Themes: Not Just for Web Design

In the Digital Humanities, three particularly unexpected themes surfaced throughout my studies: (1) collaboration, (2) pre-1800’s emphasis, and (3) the interaction between the modern and the pre-modern through technology. First, in terms of collaboration, the numerous tools–cited in the “Digital Tools” chapter of The Digital Humanities: A Primer for Students and Scholars by Eileen Gardiner and Ronald G. Musto–require collaboration on some levels, at the very least as some scholars create the tools for others to utilize. Further, tools for blogging and collaboration themselves are included in the chapter. Collaboration is also seen in projects like Linguistic Landscapes of Beirut to facilitate the collection of more data. However, unlike Linguistic Landscapes of Beirut, many digital projects interestingly focus on what I, as a scholar focused on the 20th century, just generally consider “old.” A good example of such projects would be 18th Century Connect, which also exhibits the aforementioned theme of collaboration as anyone with an internet connection could participate in correcting the OCR for the digitization of these old texts. Before studying OCR, I had presumed that old texts would be overly difficult to access for the digital world or that people of the digital age wouldn’t be interested in these older texts and therefore there would not be an overlap. It’s quite interesting to me that there is, and evidences the breadth of study available to digital humanists. Finally, a recent workshop on NYU Abu Dhabi campus about preserving cultural heritage highlighted the theme of modern and pre-modern interaction. In the workshop, a team went out onto Saadiyat Island, one of the islands of Abu Dhabi city, and took photos of an archaeological site. Those photos were then transformed into a 3D image. Additionally, the team took an oral history from an Emirati man to discover the sites around Saadiyat pre-development. These data were laid over a Google Map to build a historical map that is clearly contrasted with the modern state of the area.

In my efforts to make examples of these themes, I tried to be exclusionary; yet failed to do so with collaboration particularly. This failure to extract collaboration from other examples showcases its centrality in the digital humanities, while also sharply differentiating the digital humanities from traditional humanities. Meanwhile, this collaboration can contribute to projects dealing with pre-19th century histories or even facilitate comparisons of pre-modern and modern times.

Digital in Abu Dhabi?

After exploring OCR a bit more and ruminating on potential projects for the Arab world, two projects arose as particularly intriguing ideas. The first would mirror the digital project Linguistic Landscapes of Beirut but in the Emirates. As multinational as the Emirates is, linguistic data, drawing from either ambience recordings of conversations (which may be illegal) or available texts (in the form of signs, newspapers, etc.), could cast a light on the nature of the Emirate’s multinationalism. For instance, the much older nations of France and the United States debate whether their nations have become assimilationist with all immigrants conforming to the national ideal, “melting pots” where immigrants assimilate to some extent and influence their new society, or “salad bowls” with distinct cultures. The persistence of native languages in daily conversation or leisure reading (newspapers) would reflect the identities of the foreigners in the Emirates. Such a project may be particularly appealing because it can be crowd-sourced, both in terms of the raw data–photos–and the analysis–determination of the language in the photo. Depending on the results, or possibly even independent of the results, the largest problem would likely be legal, as the Emirates and other Gulf nations don’t tend to enjoy other peoples analyzing their identity or the identities of people within their countries (no offense to them for this decision, of course). Another challenge would be the language for publicity and the interface to provide data. Indeed, the utilization of English or Arabic alone for the interface may prevent those living mostly in their native language and culture from discovering the project and participating, thereby biasing the results.

An alternative idea would be to create a corpus of unknown or hard-to-access Arabic literature. Indeed, from foreign nations, the US for example, foreign texts are not often readily available to begin with and Arabic novels are particularly difficult to find. Consequently, running a project to digitize Arabic texts, as is being done for early English literature through TypeWright, could facilitate the acquisition, analysis and ultimately translation of Arabic texts for other scholars and eventually, inshallah, a non-Arabic speaking public.

OCR beyond Pleco

Working with OCR for digitizing texts on Thursday, I experienced the applications of OCR for the first time outside of the typical Chinese learner’s experience with Pleco. Compared to recognition for dictionary purposes, the challenges in document digitization are similar, but more pronounced due to the extent of the project. For example, the lack of spaces between characters–although more than one character may form a word–is oft cited as a language-specific challenge for OCR in Chinese. However, in Pleco, these issues are rarely prohibitive of successful use because the user can manually edit the combination of characters until they make sense. Conversely, for document digitization, the OCR guesses the combination for you, prepares the information as a document, and then you must correct any errors post-facto. Additionally, the ease of OCR to identify Chinese characters in Pleco contrasts sharply with the OCR recognition of Arabic in Abbyy FineReader. This distinction testifies to the utility and progress of machine learning. However, with the struggles in recognizing handwriting, one must wonder if the ancient, more stylized Chinese texts have similar problems to Arabic. As OCR can read not even English handwriting yet, it will be interesting to see which language acquires this capability first. Indeed, if/once accomplished, the ability for OCR to recognize handwriting could facilitate building a corpus of author’s notes, about their literature or their personal lives, and presenting this added contextual information alongside the novels already available.

Digital Projects Reflection

As the field of digital humanities continues to grow, the field becomes harder and harder to describe, as a concept initially vague becomes increasingly broad. In a review of digital projects, our class covered mapping projects, including language mapping and animations of ancient sites laid out over a timeline, as well as digitization of pamphlets from a former French colony. Even this variety of projects doesn’t begin to represent the field of digital humanities, in which one can also include heritage gaming museums. And, compared to traditional “ivory tower” academia, the benefits are also endless, as digital humanities provides ample opportunity for collaboration, public peer review, transparency and increased dissemination. Despite all these factors, since beginning to study digital humanities just a few weeks ago, the largest takeaway for me has been the accessibility of the digital. Coming from a humanities background, the digital realm has always seemed to be something populated by math geniuses, an idea enforced by the math course requirements for Computer Science majors in university. I’m encouraged now to see that not only are numerous free sources and plug-ins available to streamline the process, but also ample resources are provided by universities like mine to the student body at large to gain higher digital literacy.