By Liam Andrew on November 25, 2013
Think of the word “algorithm” and you might picture a data scientist crunching numbers in front of a terminal, analyzing functions and equations that you can’t begin to understand. If they’re building models of weather systems, you might be right (I can’t help you there). But recommendation systems are another story. The algorithms can be complex, but the output tends to be very simple: a list of some news articles, movies, or other content you might like.
Recommendation science is not rocket science, but it has always been the realm of the engineer. There’s no doubt that a good engineer can hone and chisel a recommendation algorithm to near-perfection (whatever that means), but first you have to choose the type of stone. Each recommendation system has certain methods and assumptions baked into it, and determining what kinds of inputs belong can be more of an art than a science.
Recommendation systems are generally divided into two types: collaborative filtering, and content-based filtering. Content-based filtering focuses on the product itself, like a traditional library classification system. Netflix provides one example: after culling their records down to several thousand movies and TV shows, they hire freelance film buffs to tag content with delightfully contrived categories like “Mind-Bending Romantic Foreign Movies” or “Understated Detective TV Shows” (though they also match you to similar users). While these are more fun than generic, automated tags (and Netflix deserves credit for using humans), these categories are still boxes; they place cultural products into certain discourses and implicitly exclude others. Tagging systems are inherently stale and lacking in dynamism, and folksonomies aren’t always feasible and come with their own problems.
One attempt to sidestep classification is via collaborative filtering, which looks at the user, their past behavior, and similar users or social networks for clues into what the user might like. Consider Amazon, who uses this model extensively; given the massive scale of products on offer, many from third parties, it is more manageable to leverage machine-learning algorithms that watch what you buy and browse, rather than attempting to infer the properties of thousands of new products a day. A user-history based approach maximizes efficiency but at the sake of variety, assuming that you want to keep seeing more of the same.
If it’s looking to your social networks, it feeds you what’s already most popular, stifling individual preference and shepherding audiences into identical routes. If you were to sit around a table with “users like you” and start a conversation, would you rather have seen everything they’ve seen, or something a little different? If you’re all on the same pages, how would anyone bring anything new to a discussion? What about all the possible treasures out there that haven’t been discovered yet? Collaborative filtering may work when shopping for consumer products, but it creates a filter bubble for art and culture.
The challenges and limitations behind each of these models is very different, and they point to a need to focus on the objects and users of recommendation systems, rather than the algorithms. As taste plays more of a factor in recommendation, this becomes even more crucial. If you click away from a product on Amazon, maybe it’s because you didn’t like the price, the quality, or the reviews. Regardless, the company assumes something about you. It gets even more complicated with art or music. Art has ever-changing material, cultural, and discursive properties; which ones are most important to a given viewer? Do they like being challenged and broadening their horizons, or prefer staying in their comfort zone within a certain style or mood? If so, is it worth trying to change their mind?
This brings up another variable that goes underserved when the focus is on the algorithm: what is the metric for a “successful” recommendation system, and how does that change from company to institution? Industry tends to give people more of what they want, with the end goal of a click or purchase. Cultural institutions like ours can break convention here, and at HyperStudio we hope to challenge users by making unique connections and new introductions, so long as users are ultimately delighted and informed. Unencumbered by industry demands like growth and scale, we can maintain a very different metric for success, and it could be unique to each project or user.
We also hope to open up further discussion about these systems and their limitations. The extreme secrecy behind companies’ proprietary algorithms and the domain’s traditionally engineering bent make the recommendation system something of a black box. Given the extent to which automatic recommendations affect what we read or hear about, it’s important to understand what can go into them. HyperStudio and other open-source initiatives can play a role in making them more transparent: creation and discussion of recommendation algorithms could lead to insight about the decisions computers are making for us and their assumptions about what we want. Perhaps in our effort to improve them, we’ll discover some ways in which proprietary algorithms are failing us.
Regardless, I should hope that people have a different relationship to art than to a product or piece of information, and cultural institutions should treat their audiences differently from companies. It’s important to devise recommendation systems that avoid reducing cultural objects to the level of products, and museum audiences to consumers. At the heart of the collaborative and content-based systems is the notion that more personalization leads to higher quality, and that existing networks, discussions and canons are there to be reinforced. These are meaningful and important signals, but they need not be the only ones. While quality is always important, categories should be fuzzy, as should networks of people; the most important signals are often the nodes that link them, and we hope to surface these new connections. When it comes to art and culture, looking past the current limitations of discovery will be vital for generating new ideas and conversations.
Liam Andrew is a graduate student in Comparative Media Studies and a research assistant in HyperStudio.
By Kurt Fendt on November 20, 2013
We are looking for a part-time Communications Officer to coordinate HyperStudio’s outreach activities, specifically for our NEH-funded Annotation Studio project. Please see the job description and requirements below. Applications will be accepted through MIT’s Human Resources website (link below).
Job Number: 10653
Functional Area: Communications
Department: Comparative Media Studies
Location: Cambridge, MA
Employment Type: Part-Time Temporary
Employment Category: Exempt
Visa Sponsorship Available: No
COMMUNICATIONS OFFICER, Comparative Media Studies/Writing (part-time, 45-50%), to provide outreach and project support to the HyperStudio – Laboratory for Digital Humanities. Will plan and coordinate interactions with educators and developers within and outside of MIT to help them adopt and use HyperStudio’s Annotation Studio web application; work closely with the studio’s team members to develop and implement outreach and adoption strategies for other HyperStudio projects; coordinate the development and assessment of curricular modules for Annotation Studio and organize workshops and additional outreach activities for educators and developers; and maintain public project websites and develop social media strategies for Annotation Studio and other HyperStudio projects.
REQUIRED: strong interpersonal, communication, writing, analytical, and organizational skills; accuracy and an eye for detail; strong computer skills that include experience with web, social media, and cloud-based tools; ability to work both independently with limited supervision and as part of a team; flexibility and resourcefulness; and the ability to multitask. Must enjoy contact with students and teaching staff and be able to use discretion, sensitivity, and tact in all interactions. Humanities and/or education background a plus. Job #10653-P
Schedule to be determined, but will include working 15 to 20 hours/week. The initial appointment is for one year, with the possibility of renewal.
In addition to a CV, please submit a cover letter explaining your qualifications, experience, and reasons for interest in the position; the names of three references; and links to relevant samples of your work. References and links may be included in your cover letter or resume.
and search for job id 10653
By Desi Gonzalez on October 11, 2013
Artist Sherrie Levine is best known for her 1979 series, After Walker Evans, in which she rephotographed Depression-era images by Walker Evans and presented them as her own. With this series, she posed questions about authorship and originality that would remain central to her art production: What is an original work of art? What is a copy?
To me, Levine’s work hints at another interesting idea: How do we learn about art? Levine grew up and attended college in the Midwest, miles away from the mainstream art world. In a 1985 interview the artist revealed that her early experiences with art involved “seeing everything through magazines and books.” To create After Walker Evans, she didn’t take snapshots of the original Evans photographic prints, but instead photographed the images from an exhibition catalog. While museums and galleries allow us to experience art directly, much of what we learn about art is indirect. Most people can recognize da Vinci’s Mona Lisa and Vermeer’s Girl with a Pearl Earring without ever having traveled to Paris or the Netherlands.
Today, in addition to magazines and books, we also learn about works of art through the internet. A spate of recent websites and mobile applications, such as Artsy, ArtStack, and Curiator, allow users to aggregate virtual collections of art. The missions behind such websites often have a democratic impulse, making works of art accessible on the web and transforming users into art collectors. As a museum educator, I am all for democratizing the art experience, but I wonder if we’re asking the right questions. What the internet affords us is a wealth of art images at our easy disposal, and web projects like Artsy allows us to cull through them in one place. But how, then, do we create meaningful experiences from these images? How do we get people to look closely, think critically, and engage in conversations with others about a work of art? Additionally, the majority of the works of art being shared, collected, followed, and Tweeted on these platforms were intended to be experienced in person. While the internet can be a great tool for first forays into learning about art, it doesn’t replace witnessing an artwork firsthand. How can our online art dabblings connect us to onsite art experiences?
These are some of the issues I’ve been thinking about as a new research assistant in HyperStudio. Digital media can provide new ways to discover art, but the best experiences are the ones that put you in direct contact with a work of art. And that’s why we’re starting local. I’m working with the HyperStudio team to develop a digital tool that provides an easy way for Boston-area residents to access information about artworks, exhibitions, and events. But we also want to be more than a mere listings or image-aggregating website, instead asking: what are the as-of-yet-unearthed art threads happening in the Boston area? For example, let’s say you see the print Under the Wave off Kanagawa by Hokusai at the Museum of Fine Arts and are immediately enamored. Our tool might connect you to a workshop on woodblock printing led by a local artist or inform you about a lecture at Harvard about Edo period Japan.
The principles underlying this new project is a concept that runs through many HyperStudio endeavors. For example, Annotation Studio is an open source web tool that aims to enhance the ways a student interacts with a text. A student can add multimedia annotations onto a text, search and link to other content, and ultimately engage in a more active reading of the text. The digital tool doesn’t overshadow the original text, but is instead an avenue to dig deeper into it. Like Annotation Studio, the goal of our new project is to privilege engagement with the art first.
Over the next few months, we will be researching into how digital tools might allow us to create meaningful art experiences through out the Boston area. We have started by looking into what examples are already out there, from event listings to museum collection guides. Ultimately, we want to go from looking at images online to fostering rich, direct experiences with art. For Sherrie Levine, books and magazines were a medium through which she experienced art, far away from art world centers. Here in Boston, we’re lucky; we have many world-class museums and intellectual centers at our disposal. We hope that our tool will be a way to connect the community to these onsite experiences. Like Levine’s books and magazines, this project will be a first step to learning about art—but it won’t be the last.
By Jason Lipshin on July 23, 2013
As a scholar who subscribes to a “big tent” conception of digital humanities, I must admit that I was initially a bit nervous about attending DH 2013 at the University of Nebraska-Lincoln. As the annual international conference of the Alliance of Digital Humanities Organizations, DH 2013 is one of the oldest and most established conferences supporting emerging work in the field of digital humanities, but it also has its roots in old school “humanities computing.” For those who are unfamiliar with the history of the field, humanities computing is the predecessor to digital humanities in its current form and is often associated with work in text mining Shakespeare, using word frequencies to understand an author’s style, or experimenting with TEI to compare different aspects of a text. Although much of this work can be extremely interesting and rigorous in terms of its play with computational affordances, I consider myself much more oriented towards flavors of DH with roots in digital art, design, experimental “multimodal scholarship,” and digital media studies. The distinctions between these camps are, of course, getting fuzzier by the minute, but the history and reputation of the conference implicitly reinforces these boundaries. And if there's one thing I've learned in my short time on the conference circuit, when you’re presenting, it certainly pays to know the composition of your audience.
Upon arriving at the conference, however, I was pleasantly surprised to find scholars from many camps present and conversing in relative harmony. In addition to the humanities computing mainstays, there were also programmers using computer vision to identify dance techniques, literature scholars experimenting with sound art, and even historians using 3-D printing to better understand historical artifacts. Defining what is and what isn’t digital humanities in such an environment is increasingly difficult, but the flip side of this lack of definition is an exciting sense of openness and experimentation. Scholars were tinkering, hacking, failing, and experimenting with the affordances of non-textual forms, while always trying to keep in mind the implications of this work for asking traditional humanities questions.
One of the more interesting talks that I attended was Jentery Sayers’ “Made to Make: Expanding Digital Humanities Through Desktop Fabrication.” As a self-described media archeologist, Jentery is very interested in the materiality of historical artifacts (particularly 19th and 20th century sound technologies) and has begun using 3-D printing in order to better understand this materiality. Leading a “maker lab” on humanities-oriented fabrication at the University of Victoria, Jentery has already begun producing a series of “maker kits for cultural history,” allowing, for instance, students or scholars interested in the history of radio or the telegraph to build their own. Drawing on recent work by Jonathan Sterne, Wendy Chun, Matt Kirschenbaum, and Jeffrey Schnapp, Jentery is interested in “how the hermeneutics of screwing around” with these technologies can fold back into theoretical reflection.
Another fascinating talk that I attended was Michael Aeschbach’s “Dyadic Pulsations as a Signature of Sustainability in Correspondence Networks.” Although Michael’s talk fit much more easily into traditional DH paradigms than Jentery’s, I thought it was equally innovative for the way that it proposed a new direction for research in data visualization. At the core of Michael’s talk was the argument that data viz research has traditionally privileged spatial approaches like topological analysis and graph theory over the temporal. Temporal approaches like measuring response time in a communication exchange between two participants, Michael argues, can be extremely instructive in measuring the “sustainability” or “health” of a communication network. Although he took the long standing Usenet discussion group as his primary case study, he also mentioned that this approach could be applied to non-online phenomena – for instance, mapping the temporality of letter correspondences in Stanford’s “Republic of Letters” project. Many historians in the audience, including Brown’s DH librarian Jean Bauer, seemed to be extremely excited about this possibility.
I delivered my own talk, “Visualizing Centuries: Data Visualization and the Comedie Francaise Registers Project,” on Friday morning to a fairly packed crowd. As an overview of HyperStudio’s use of data visualization in order to understand a large archive of theatrical ticket sales, the talk seemed to strike a chord across many different disciplines, with positive feedback coming from theater historians, computer scientists, and statisticians alike. In particular, many questions following the talk seemed to address the unique dynamism of our data visualization tools and the ways that they allow the user to dynamically combine multiple parameters to help facilitate what we have termed an “exploratory research process.” One scholar was particularly struck by my notion of “combinatorial research” and how this process of combinatorial play might facilitate intellectual exploration, while many others were interested in the ways that our tools could become open source and generalizable enough to be applied to other kinds of data. Hearing such positive feedback was incredibly gratifying, especially after having worked on the project for almost a year.
In all, I felt that attending DH 2013 was an extremely valuable experience, both in terms of promoting HyperStudio’s own work and learning about emerging work in the field. Although I’m still very new to the conference circuit, I’ve already had many wonderfully stimulating discussions, met many like-minded grad students and faculty, and received many encouraging words about my work. Certainly, all of the fault-lines and fissures, boundary work and growing pains of a field in transition are ever present. But so is the nascent sense of an ever expanding scholarly community.
By Kurt Fendt on February 28, 2013
HyperStudio recently launched a new Fellows Program! The HSF gathers a range of postdocs and visiting scholars in the humanities, librarians, technologists, artists, curators, and other members of the MIT community to generate questions and energy around the Digital Humanities. This spring, our Fellows meet bi-weekly and attend DH-related events to brainstorm possibilities for potential programming, incubate projects, and discuss our varied but shared scholarly, pedagogical, and publishing practices in the digital age. The HSF program aims to cultivate community and collaboration at the edges of disciplines and fields, finding overlaps, and investigating productive tensions that suggest opportunities for new creative and critical engagements through digital realms. By bringing together staff and non-permanent faculty at MIT, the HyperStudio Fellows program hopes to grow a hub of DH activity not only within HyperStudio and between existing MIT departments, but also as our Fellows interconnect with the Boston DH consortium, and beyond.
Our inaugural HyperStudio Fellows for Spring 2013 include:
Gretchen E. Henderson
Elizabeth Anne Watkins
If you are interested in participating, or have suggestions of good candidates, please keep watch for an application for the HyperStudio Fellows program for the 2013-14 academic year (to be posted later this spring). We welcome suggestions about our evolving mission, as we seek ways to address existing needs at MIT, to reach out and tap members of the community with relevant experiences and interests, and to encourage cross-disciplinary collaborations.
Questions? Ideas? Interest? Please contact Gretchen Henderson at hendersong [at] mit [dot] edu.
Text: Gretchen Henderson; photo: Kurt Fendt
By Jason Lipshin on December 5, 2012
As is often the case with DH conferences, this year’s Chicago Colloquium on Digital Humanities and Computer Science was something of a grab bag. Featuring a hodge podge of panels on everything from network analysis and data visualization, to locative media and alternate reality games, the conference certainly lived up to the ethos of inclusiveness often characterized as DH’s “big tent.” And yet, despite the diversity of research questions being addressed, I also think that the conference had a great sense of coherence, with a lot of scholars operating in the tradition of humanities computing. Coming out of disciplines such as classics, English, and history, many of the scholars at DHCS were interested in mining large corpuses of data, visualizing such data, and analyzing the results to ask new kinds of research questions.
Although I found many of the projects interesting, two presentations stood out as particularly relevant to the kind of work being done at Hyperstudio. On the first morning of the conference, Diane Cline from the University of Cincinnati presented a fascinating project working firmly within the humanities computing vein. Drawing on her dual background as a classicist and an intelligence analyst for the NSA, she attempted to apply the insights of network analysis and graph theory to study what she called “the social network of Alexander the Great.” Using the NodeXL program created by Ben Schneiderman (of “Direct Manipulation” fame) from the University of Maryland, Cline color-coded each edge within her network to indicate the type of relationship, whether it be family, officer, courtier, enemy, ally, or peer. Affording a kind of macro scale view which is alien to the classicist’s typical tool set, Cline argued that she was able to see “the bridges, brokers, and hubs” which were central to Alexander’s relationships, contributing a new understanding of her topic that would have been unavailable to her without the tool.
This emphasis on network analysis was also echoed in Hoyt Long’s fascinating presentation on modern Japanese poetry. While many DH projects have thought about citation as a way to quantitatively measure networks of influence, Long’s presentation instead focused on less explicit metrics of influence like poetic style and content. Although these elements are less quantifiable, and thus must be coded by an expert researcher rather than automated by a machine, visualizing these influences helped Long to arrive at new insights about the relationships between different schools of aesthetic thought within Japan, as well as to discern which nodes and clusters were most central. As was the case with so many of the panel discussions that day, he ended his presentation with an advocation for visualization tools as part of the humanistic scholarly process, rather than as a means to simply generate “evidence” as part of a research product.
Overall, I believe DHCS was a successful conference, in that it was able to balance the impulse towards creative (and often, chaotic) interdisciplinary collaboration with a sense of common goals and directions for the future of this particular niche within the digital humanities. Importantly, it also proved to be a great space for meeting and exchanging ideas with researchers who are working on similar issues – I, for one, met many faculty and graduate students located on the east coast, and even some in the greater Boston area, who I believe have many overlaps with Hyperstudio’s research goals. Since Hyperstudio is constantly looking to strengthen its connections with other DH researchers outside of MIT, DHCS proved to be a wonderful space for both getting our ideas out there and for learning from others producing innovative projects in the field.
*All photos are from Peter Leonard.
By Katie Edgerton on November 15, 2012
On Monday, November 19, HyperStudio’s Research Assistants Jia Zhang and Jason Lipshin will be presenting a collaboratively authored paper about HyperStudio’s ongoing Comédie-Française Registers Project (CFRP) at the Chicago Colloquium on Digital Humanities and Computer Science. Joint authors are: MIT Professor Jeffrey Ravel, HyperStudio’s Executive Director Kurt Fendt, Jia Zhang and Jason Lipshin.
The paper entitled “Visualizing centuries: Deep insights into cultural production before the French Revolution,” centers on the challenges of visualizing historical data. For several years, the HyperStudio team has been working with Professor Ravel in MIT’s History Department and the Comédie-Française in Paris to digitize paper registers containing more than a century of box office sales. This wealth of information – covering over 113 seasons – is a vital resource for theater and literature scholars as well as humanists interested in the political, social, and cultural history of France and the Western world.
The Comédie-Française Registers Project addresses two main audiences. The first is French theatre historians, to whom the project will provide greater disciplinary knowledge by giving them digital access to rare archival materials and offering innovative research tools that allow them to detect patterns in cultural production and consumption in the 17th and 18th centuries.
The second audience is digital humanists. Through the Comédie-Française Registers Project, HyperStudio is developing innovative methodologies that enable macro and micro-level analysis. Many of these methodologies turn on the development of tools for data visualization that can enlighten both the research process as well as the outcome.
These new kinds of data visualizations will enable scholars to address questions beyond the scope of traditional humanistic research methods. For example, what do sales records and repertory trends tell us about changing patterns of author, play, and genre popularity in the years leading up to the French revolution? Integrated research tools such as faceted browsers allow for complex and fine-grained filtering of all data and enable their analysis in novel combinations.
The Comédie-Française Registers Project team will tackle these issues on Monday’s visualization Panel, beginning at 11:15 am in Ida Noyes Hall at the University of Chicago.
For more information, please visit the Comédie-Française Registers Project website.
By Ayse Gursoy on July 19, 2012
I sat in on several sessions today, the first day of conferences. In the morning I checked out the short paper session, and later I went to two long paper sessions. All of the talks were interesting, quick paced (only ten minutes each for the short paper sessions, something to watch out for tomorrow!), and made me think of digital humanities in, if not a new, a richer light. The twitter stream (hashtag #dh2012) was lively throughout the day and offered many a chance to respond to the talks (only a few minutes for questions, unfortunately—though questions were many). Some of the key issues according to the twitter stream: the constant issue of digitized versus born-digital projects, the openness of data and access, and the insufferable hardness of the chairs in the lecture halls (no, really).
There were a few talks that resonated with my background and interests, and I”ll probably get to each of them in some way. But I”d like to focus one one of the long papers, and the questions of design, purpose, and audience that it made clear. The talk in question is Claire Ross, presenting “Engaging the Museum Space: Mobilising Visitor Engagement with Digital Content Creation”. While at first glance this might seem like an interesting topic that isn”t directly related to academic research in Digital Humanities, the central question was actually an extremely important one for any DH researcher: how do you get audiences engaged with digital content? Ross”s case study, the use of QRator at UCL”s Grant Museum top online casino of Zoology, offered one approach to this question. If you have a discussion forum tied to the object on display, accessible with a smartphone or tablet, then give your visitors a tablet and let them talk! We could go on for days about the actual mechanics of this approach or the validity of competing approaches such as short URLs and established social media sites. What I”d like to suggest, however, is that Ross” talk asks us to reframe the question.
It isn”t about engaging with digital content.
It”s about engaging through digital content.
We need to design systems that unobtrusively let people do what they are already good at doing, that step back and make conversations happen. I know this is hard. I know it might mean changing the definition of “conversation”, or expanding it a teensy bit. But I do believe this will impact digital engagement in a positive way.
I”d like to offer the example of the games Demon”s Souls (From Software, 2009) and Dark Souls (2011). These games feature a unique and much-discussed multiplayer engagement in an essentially single-player game: the two games are always-online (unusual for a single-player game), allowing other players to intrude upon your game. The game has been written about in many places and in better ways than I could right now. But there is one particular engagement that I think is highly relevant here: players can leave messages tied to in-game locations, that appear to any player who accesses that location. Yes, it takes an active will to leave a message, but once that message is placed, any player, not just a player with the equivalent of a smartphone, can read it. The systems that we design for digital engagement with physical spaces need to, dare I say it, intrude upon the corporeal experience of the visitor. And doing this is hard. I don”t pretend to have a solution, but it”s something that”s been on my mind.
By Kurt Fendt on May 7, 2012
Text by Elyse Graham and Jia Zhang.
By Kurt Fendt on April 9, 2012
We are happy to announce that HyperStudio has received a Level II Start-Up grant from the Office of Digital Humanities (ODH) at the National Endowment for the Humanities (NEH) for “Annotation Studio – Multimedia Text Annotation for Students”.
Annotation Studio is an open-source web-based application that actively engages students in interpreting literary texts and other humanities documents. Initial features will include:
1) Easy-to-use annotation tools that facilitates linking and comparing primary texts with multi-media source, variation, and adaptation documents;
2) Sharable collections of multimedia materials prepared by faculty and student users;
3) Multiple filtering and display mechanisms for texts, written annotations, and multimedia annotations;
4) Collaboration functionality;
5) Multimedia composition tools.
While strengthening new media literacies, Annotation Studio will help students develop traditional humanistic skills including close reading, textual analysis, persuasive writing, and critical thinking.