| 
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • You already know Dokkio is an AI-powered assistant to organize & manage your digital files & messages. Very soon, Dokkio will support Outlook as well as One Drive. Check it out today!

View
 

Social Network Analysis: Open Questions

Page history last edited by Alan Liu 10 years, 3 months ago

Questions about social network analysis harvested from

discussion in a graduate class on digital humanities

(November 5, 2013)


(English 236, U. California, Santa Barbara.  Instructor: Alan Liu)

 

As part of our class on social network analysis in English 236, we kept a running log of questions that the students and instructor (Alan Liu) asked and left hanging in our discussion.  The instructor edited, filled in the context for, and sometimes elaborated on these questions here.  Most of us are humanities or education scholars from several disciplines who are beginners at many of the subjects included in this course.  These questions remained after our readings and practicums--and, in fact, our questions were often sparked (or stoked higher) by those exercises.  We're aware that there is additional research and discussion on some of these issues, likely at a level beyond that of beginners or occurring in other fields.  Suggestions for reading welcome!

(Thanks to Rebecca Chenoweth (@chenoweth42), a student in the course for acting as transcriber of our questions.  Thanks also to my colleague William Warner, who sat in on the class and added an additional open question) --Alan Liu (@alanyliu)

 


 

  1. For humanities scholars, what can, and cannot, be a "node" on a social network graph?  For example, can the "narrator" or the "reader" be represented as a node in a social network analysis of characters in a novel?  Similarly, what can and cannot be an "edge"?  In the example of a character analysis that includes the narrator and reader, would we need a multidimensional understanding of edges to tie them into the network?
  2. What do we do with the tricksters in humanistic materials that can be considered either nodes or edges?  For example, consider the cats in the Early Modern event recounted by Robert Darnton as "The Great Cat Massacre" in which apprentices slaughtered a village's cats as a symbolic action directed against their masters (and, in the case of la grise, the favorite cat of a master's wife) their mistresses?  Is the cat an entity that should be a node in the graph?  Or is it the medium of a relationship and thus an edge?  Put generally: how do humanists deal with social network analysis methods that insist on strict differentiation between nodes and edges when such differentiation can be subject to representational and discursive effects?
  3. When we watch Franco Moretti ("Network Theory, Plot Analysis") or other humanities scholars use social network analysis to study humanistic materials, we see incisive arguments propounded on the basis of clear contrasts between different social network graphs or states of such graphs.  (In the case of the Moretti essay, for example, we see striking contrasts between Hamlet with and without the character of Hamlet and between a Dickens novel and the 18th-century Chinese novel The Story of the Stone.)  Is the interpretive use of social network analysis always a matter of discrete quantum jumps like this?  Or is there a whole other mode of interpretation based on continuous or incremental transition/change?
  4. Digital humanists often use social network analysis to understand the networked relations within discursive works (e.g., characters in a play, novel, or contemporary TV series).  But should the true subject of such research be the meta-relationship between intra-work networks and their extra-work contexts--e.g., between the networks represented within a Shakespeare play and the networks of real-life actors, playwrights, courtiers, aristocrats, and others that produced the play?  If so, can the humanities recover data about extra-networks at the same level of granularity or explicitness as data about intra-networks?
  5. Social scientists and ethnographers are trained in the methods of observing, surveying, and analyzing that produce primary data for social network analysis--e.g., data about which social actors engage with which others.  In applying social network analysis to their materials (e.g., fictional works, historical events inferred from discursive works, etc.) do humanities scholars need equivalent methods of observation whose protocols are standard and explicit in a manner exceeding the protocols of traditional "close reading"? 
  6. If so (pendant to #5 above), can machine-learning or other computational methods ever be expected to provide the means for such programmatic observation (e.g., discriminating character and relationship data in novels or memoirs)?  Or will humanists always require manual observation of their difficult and messy material in a way that is unrealistic at scale (e.g., researchers and their graduate students marking up novels)?
  7. How can we best discover and represent meaningful absence in a social network?  Consider, for instance, the question of characters in a literary work or historical document who never meet.  Do there exist social-network-analytical methods or metrics that allow us to discriminate between the significant vs. non-significant white spaces on the social graph between people who never meet?
  8. What is the difference between past and present understandings of "network"?  (In class, we put this question into play by comparing the visual paradigms of Victorian wallpaper, 19th-century American quilts, and contemporary visualizations of the Internet--each of which visualizes a variant of node-edge relationships.)  Similarly, what are the differences between Western and Eastern, or Northern and Southern, understandings of networks?
  9. How do the media of interaction (the carrier of an "edge," e.g., an oral speech, gesture, glance, physical contact, letter, email, etc.) influence network structure?  So far, the humanities do not have the equivalent of "media richness theory" in the social sciences to address such issues differentially.  What would media richness theory (related to what N. Katherine Hayles calls "media specific analysis") be in the humanities?
  10. How many kinds of networks (structures, topologies, and growth patterns of networks) are there anyway?  Do we inhabit an open- or closed-system universe of network types?  Also, how are those types distributed over natural, social, and other phenomena?  And a historical question: how have dominant network types changed over time--e.g., from the era of coffeehouses in the Enlightenment to social network media today?
  11. What is the relationship between physical and social networks; and how can that relationship be understood in the terms of recent humanistic explorations of the contact zone between humans and the non-human (e.g., explorations of animal studies, the new materialism, actor-network theory, etc.)?  A more specific form of the question: how do spring-force or force-directed algorithms for visualizing nodes and edges in social network graphs (e.g., ForceAtlas 2 in Gephi) negotiate the leaky membrane between the physical and the social through their fiction that the social universe is a matter of equilibrium between forces of physical attraction and repulsion?
  12. Insofar as digital humanists are humanities scholars, they add their particular perspective to the continuing effort to address the fundamental question: what did, does, and will it mean to be human?  In regard to the problem of the human, is the method of social network analysis ultimately reductive or expansive?

 

  • Additional open question from William Warner (sitting in on the class): "In our discussion in Alan’s DH seminar last week, we at one point near the end of class focused on the problem how to develop a network map of characters in the novel that are linked to one another. This raised the question: how can we visualize the quality of the relationship between characters quantitatively, so a computer program could map it? [The method of counting the] number of times a character speaks (as with Moretti in the pamphlet on Hamlet) is limited for obvious reasons; number of words exchanged might be more valuable; but (as Liz Shayne pointed out—sometimes you can’t tell if a word is spoken or thought). How does one express quantitatively, the amplitude of the meanings articulated, or the meanings withheld in silence? For example, the fraught love triangle at the center of The Portrait of a Lady involving Ralph Touchett, Isabel Archer and Gilbert Osmond, is hardly quantifiable through word count. Much of it unfolds off ‘stage’ or in fraught silences. So this is the general conceptual question that comes to mind: is there not a point where we pass from extractable data (that is usefully {or not usefully} computable), to a domain of affect, connotation, barely graspable cultural codes, and rhetoric, where, for example, the question “how do I love thee, …let me count the ways”, obviously suggests a love so great that the lover (in this case Elizabeth Barrett Browning) can’t count the all the ways she her beloved Robert. "
  •  

     

     

    Comments (0)

    You don't have permission to comment on this page.