500 Novels Ramps Up for the Summer

As we’ve begun to scale up our efforts this summer, we have eight full-time research assistants on the 500 Novels project. We’ve been undergoing lots of changes and developments! In order to start off our summer session, here’s a post.

This project, Tagging 500 Novels, seeks to analyze the ways in which physical space and mental constructions of location have worked to organize and represent changing experiences of the city throughout the nineteenth century, a period marked by both a democratization of the novel and rapid urbanization. Specifically, this project will focus on the emotions associated with particular urban locales in London during the nineteenth century, where “the narrative system becomes complicated, unstable: the city turns into a gigantic roulette table, where helpers and antagonists mix in unpredictable combinations” (Franco Moretti, Atlas of the European Novel, 68). Where there once existed one story, centered on the communal experiences of an insular town’s inhabitants, now a different narrative structure must emerge from the crowd - one that can accommodate the emotional intensity associated with shifting allegiances, geographic displacement, and class antagonisms as lower, middle, and upper classes collide in this newfound urban space. By tracking and mapping these fluctuations in characters’ emotional lives, Tagging 500 Novels hopes to search for connections among particular places and arrive at possible explanations for these trends.

The project has changed slightly from its initial conception. The major change we have experienced is a shift away from social networking toward emotions, gender, and class. We previously wanted taggers to perform anaphora resolution: for each pronoun, find its antecedent (he-->Lord Filroy), as well as annotate each instance of dialogue with regard to the speaker and recipient. Anaphora resolution proved too time-consuming, and it introduced hermeneutical problems that could possibly detract from the other tasks we were asking our taggers to perform. For example, discerning the recipient of dialogue requires contextual knowledge only obtained by reading the entire novel. We found that the best tasks were ones which required no prior knowledge of the literature or of English customs during the nineteenth century.

Moving forward, here are some of our preliminary hypotheses and questions:

-Which regions of the city are most represented? Are some locales predominantly referred to or imagined, rather than rendered explicitly in the narrative?

- Does the spatial representation of certain kinds of emotional experience correlate to known class divisions in the city? Can we anticipate the rise of what Franco Moretti has called the “Third London,” a middle-class “wedge” in the city, through changing fictional representations of this region?

- Do characters of different genders or social classes have different spatial distributions, and if so, what would this mean? Does the mobility of women in fiction increase over time due to rising women’s activity in the public sphere?

In order to investigate these wide-reaching research questions, this project has diverged from traditional humanities research in four distinct ways: by making use of distant reading techniques to glean patterns from hundreds of texts; by using crowdsourcing to gather data from distributed human readers that would be impossible to collect by a computer; by emphasizing the role of space as an actor within literary history; and by attempting to trace changes in the emotional experience of fictional characters navigating physical space.

In our future posts, we will unpack each of these divergent aspects in detail. Stay tuned for “What is distant reading?”

Our Year of the Bay Hackathon at the California Historical Society

This is a cross posting by Kerri Young from the Historypin blog.

http://blog.historypin.org/2013/06/12/our-year-of-the-bay-hackathon-at-the-california-historical-society/

Railroad Enthusiast Advisory Meetings Continued

The Mellon Crowdsourcing project held another meeting in a series of encounters with local railroad enthusiasts on Tuesday, April 30.  Robert Bowdidge, from Google (though in a private, enthusiast's capacity) visited CESTA and discussed his own research and interests, as well as the ways he thought the project might be able to develop with partnerships in the railroad enthusiast community.  Robert became interested in railroads through an interest in local history, particularly involving San Jose and the route to Santa Cruz, which highlighted to the whole group how important this dimension will be throughout the rollout of the project.  It also suggests some interesting overlaps with the Year of the Bay initiative--local connections breed intense interests and the accumulation of knowledge.  Robert highlighted the importance for railroad enthusiasts that their materials be maintained in a sustainable manner (if donated or shared).  In attendance: Gabriel, Zeph, Robert, Sophia, Jon V via Skype.

Cesta 2

Stanford Crowdsourcing Meeting, March 14

PART 1

After six months of preliminary work on the Mellon Crowdsourcing project, the principal participants convened for an all-day meeting in the Center for Spatial and Textual Analysis at Stanford University.  Present at the meeting were three members of the WAWWD/Historypin group: Nick Stanhope, Jon Voss, and Kate Ferrier.  Also attending from outside Stanford were advisory board members Justin Quimby and Dave Martens.  From Stanford, Zephyr Frank, Erik Steiner, Ryan Heuser, and Sofia Paliza-Carre all attended the meetings at some point between 10 am and 4 pm.

Ryan, Zephyr, and Nick

Our agenda was packed with discussions of work done to date, of next steps, and of more general concerns regarding data management, storage and accessibility.  Nick and Jon kicked things off with a discussion of the Mapcha tool developed by our collaborators at Stamen design, as well as with an overview of how things were progressing on the Year of the Bay project.  The whole group took a look at the King's College London crowd-sourcing paper that Jon V had passed around prior to the meeting.  Therein, we agreed on the desirability of using some of the concepts and terminology presented by the researchers at King's.  Zephyr spoke briefly about the status of the Stanford research and emphasized the importance of:
1) experimental design, both in inputs and tasks and interface (that is, we want to leverage the digital platform to get data on how users deal with different inputs/tasks as well as how different interface designs (games, etc.) influence their activity
2) basic data collection, including better information on current usage stats relating to the Year of the Bay, to be rolled out and extended to the other two projects

Justin chimed in here and throughout the meeting with detailed advice on how to think about metrics.  What do we want to measure and how do we share our measurements with the crowd?  He emphasized, in particular, the importance of relative rankings versus absolute standings.  That is, people care about how they stack up against their friends more than how they are doing against Ms power user.  A detailed summary of his ideas on these matters was promised (and has since been delivered, thanks Justin!)

Zephyr digs into the Maptcha interface prototype by Stamen design on the big screen.

Discussion turned to the 500 novels project, with Ryan taking the lead.  A shift towards geography in the project, namely the identification of places in London (and eventually other cities) and the extraction of other information about those places (moods, class, gender of participants) was proposed.  The Historypin team responded with enthusiasm to the new emphasis on maps.

To be continued...

As an initial pilot project, Ryan suggested using the city of London, where roughly 50,000 mentions of places occur in the corpus of 19th-century novels.  An initial test of what can be done by the crowd will be devised using Mechanical Turk.  A more fleshed out system will follow, mutatis mutandis, on Historypin.

The afternoon session focused on the Living with the Railroads project.  Zephyr showed some examples of photographs of RR right-of-ways with rich visual information relating to native flora versus invasive species.  Richard White, lead academic PI on the LWRR project has indicated a strong interest in identifying plants along the RRs.  Maria Santos, CESTA/Lane Center postdoc, has provided some leads with respect to communities of interest/experts who might be called upon for this kind of work.

Dave Martens spoke about his experiences trying to set up a common space for railroad buffs to share information and archive their valuable collections.  The discussion that followed included questions about where and how materials would be archived, what kind of pipeline might connect Historypin to the website that Dave is developing, and so forth.  These questions of archiving dominated the last hour of our meeting.  Would Stanford step up and provide permanent storage?  How would this relate to other cultural organizations and their content, which might appear in the Historypin interface for YotB and LWRR?

Steiner related his recent conversation with Stanford Libraries in this connection, suggesting that for some datasets, there is a new system up and running that may serve our needs.

Year of the Bay Cultural Heritage meeting

California Historical Society hosted a Year of the Bay meeting on March 11, 2013, which we advertised as

An informal meeting of cultural heritage partners, enthusiasts, developers, and organizers to review the current status of the Year of the Bay project, and the upcoming phases which we're moving into. We'll hear updates from the Historypin team from San Francisco and London, as well as Project Director Jon Christensen.

We had a great turnout, especially given the late notice, with representatives joining us from California Historical Society, SFMTA Photo Archive, San José Public Library, Oakland Museum of California, San Francisco History Room/San Francisco Public Library, and Shaping San Francisco.

Participants shared their plans for current and upcoming exhibits and identified opportunities for collaboration throughout 2013. We talked at length about Phases 2 and 3, and looked at the current status of the Maptcha project by the Stamen Design team, as well as reviewed possibilities for phase 3 content and outreach.

We followed the wider meeting with more specific planning and strategy about incorporating multiple methods of crowdsourcing into the upcoming interactive exhibit, Curating the Bay, at California Historical Society.  The exhibit, curated by Christensen, will feature the Year of the Bay project as one component of how we can all explore and contribute to the evolving environmental and cultural history of the bay.

Lessons from Paul Hagon and Trove

Unfortunately I just missed Paul Hagon's presentation at ALIAOnline in Brisbane several weeks ago, but I did have a chance to catch up with him at the conference and hear more about his latest work.  His presentation looked at some of the lessons learned from the data behind Trove's usage stats and is very instructive to our research as well.

Information online trove crowdsourcing behaviour from Paul Hagon

AHRC Crowdsourcing Survey

At the end of the year, Mark Hedges and Stuart Dunn of Kings College London released their Crowdsourcing Scoping Study, which seeks to "establish a credible definition for, and the current state of the art of, crowd-sourcing in the humanities."  The study was funded by the Arts & Humanities Research Council, and has already played an important part in our work for this project.  Our hope is to leverage many of the learnings documented here and begin putting them into practice in the field.

You can download the paper, or learn more about their project.