Vol. 2, No. 1, Winter 2012 on Journal of Digital Humanities

Pacing Scholarly Conversations

Sat, 01 Dec 2012 00:00:00 +0000

The advancement of scholarship relies on the timely communication of questions, methods, results, and reflections. The iterative publications Digital Humanities Now and the_Journal of Digital Humanities_ are intended to facilitate this process. DHNow surfaces and distributes the conversations weekly in order to invite participation and feedback. The Journal of Digital Humanities then identifies the conversations that need a stable landing on which to pause and reflect before continuing onward.

Respond to JDH 2.1

Sat, 01 Dec 2012 00:00:00 +0000

The Journal of Digital Humanities “catches the good” by surfacing and highlighting valuable material published on the open web. For this issue we will go one step further, and solicit responses to this edition’s take on topic modeling in the digital humanities.

We offer two ways to provide your response. On this page you can post an open comment or a link to longer piece posted elsewhere. You also can tweet a response using #JDHTopics, which we will aggregate on another page.

Review of MALLET, produced by Andrew Kachites McCallum

Sat, 01 Dec 2012 00:00:00 +0000

MALLET Version: 2.0.7
Requirements: Java

Reviewed: 15 February 2013
Tested on: Mac OS X v. 10.8.2, and Windows 7

The MAchine Learning for LanguagE Toolkit, or MALLET, has been one of the “hottest” tools in digital humanities research. A product of the University of Massachusetts Amherst, written by Andrew McCallum and a team of collaborators, MALLET was originally released in 2002 but has received considerable renewed interest of late.¹ Fruitfully employing Latent Dirichlet Allocation [pdf], or LDA, MALLET can help navigate large bodies of information. It does so by finding clusters of words that frequently appear together, or “topics.” The algorithm imagines that any possible text or document within a corpus is a mixture of different topics; each topic is imagined to be a probability distribution of terms within and across that corpus. This leads to a variety of outputs, including lists of topics and their constituent words, and list of documents and their constituent topics. These results can often be astounding, and have even been (tongue-in-cheek) described as “magic.”

Topic Modeling and Digital Humanities

Sat, 01 Dec 2012 00:00:00 +0000

Introduction

Topic modeling provides a suite of algorithms to discover hidden thematic structure in large collections of texts. The results of topic modeling algorithms can be used to summarize, visualize, explore, and theorize about a corpus.

A topic model takes a collection of texts as input. It discovers a set of “topics” — recurring themes that are discussed in the collection — and the degree to which each document exhibits those topics. Figure 1 illustrates topics found by running a topic model on 1.8 million articles from the New York Times. The model gives us a framework in which to explore and analyze the texts, but we did not need to decide on the topics in advance or painstakingly code each document according to them. The model algorithmically finds a way of representing documents that is useful for navigating and understanding the collection.

Topic Modeling and Figurative Language

Sat, 01 Dec 2012 00:00:00 +0000

… to have them for an instant in her hands both at once,
the story and its undoing…
from “Self Portrait as Hurry and Delay” [Penelope at her loom]

Located at the center of Jorie Graham’s collection The End of Beauty, “Self Portrait as Hurray and Delay” crafts a portrait of the artist, poised at a precarious moment in which thought begins to take shape. Like Penelope, Graham entertains the illusion, if only momentarily, of a choice between bringing a creative impulse into form or allowing it to come undone. A weaver of language, Graham subtly, deftly, but unsuccessfully attempts to delay the inevitable moment in poetic creation in which complexity of thought adopts form through language, and so realized is also reduced. In The End of Beauty, the beginning of the creative act signals an inevitable descent into meaning — language’s ultimate impulse.

Responses from Twitter

Sat, 01 Dec 2012 00:00:00 +0000

We offer two ways to provide your response. On this page you can see the Twitter conversation surrounding this issue, identified by the #JDHTopics hashtag. On another page, you can see comments and links to longer pieces that respond to the content of this issue.

Review of Paper Machines, produced by Chris Johnson-Roberson and Jo Guldi

Sat, 01 Dec 2012 00:00:00 +0000

Paper Machines Version: 0.3.6
Requirements: Zotero, Python 2.7.3, Java.

Reviewed: 25 February 2013
Tested on: Mac OS X v. 10.6.8, and Windows 7
Tested with Zotero for Firefox 3.0 and Zotero Standalone 3.0

Paper Machines is an interactive multi-tool that allows users to perform textual analyses on their Zotero notes, tags, HTML snapshots, or attached pdfs (if OCR layer is present) directly in Zotero. The project provides users with an effective way to get an intellectual grasp of a corpus relatively quickly. This Zotero add-on currently ships with five different tools, which make it possible to determine anything from the topics found in a user’s Zotero library to the geographic distribution of the references present. The project works in both Zotero’s Firefox and standalone versions, which makes it particularly convenient for Zotero users but considerably less appealing for anyone who stores their research material in another program or format. Paper Machines is a promising and visually appealing teaching tool that would be particularly useful for introducing students to topic modeling, but needs some improvements to the code and documentation to be world class.

The Digital Humanities Contribution to Topic Modeling

Sat, 01 Dec 2012 00:00:00 +0000

Topic modeling could stand in as a synecdoche of digital humanities. It is distant reading in the most pure sense: focused on corpora and not individual texts, treating the works themselves as unceremonious “buckets of words,” and providing seductive but obscure results in the forms of easily interpreted (and manipulated) “topics.” In its most commonly used tool, it runs in the command line. To achieve its results, it leverages occult statistical methods like “dirichlet priors” and “bayesian models.” Were a critic of digital humanities to dream up the worst stereotype of the field, he or she would likely create something very much like this, and then name a popular implementation of it after a hammer.

Topic Model Data for Topic Modeling and Figurative Language

Sat, 01 Dec 2012 00:00:00 +0000

The topic model discussed in “Topic Modeling and Figurative Language” was created with MALLET. Drawing from 4,500 English-language poems from the “Revising Ekphrasis” corpus, the model was generated using the following parameters:

mallet train-topics --input poems-seq.mallet --num-threads 2 --num-topics 60 --optimize-interval 10 --output-model poems08072012test1.model --output-doc-topics poems08072012_test1.txt --output-topic-keys poems08072012-test1keys.txt

The following table contains the number of the topic (0-59); hyper-parameter estimation; and top 20 key words most likely to be found in each topic.

Topic Modeling: A Basic Introduction

Sat, 01 Dec 2012 00:00:00 +0000

The purpose of this post is to help explain some of the basic concepts of topic modeling, introduce some topic modeling tools, and point out some other posts on topic modeling. The intended audience is historians, but it will hopefully prove useful to the general reader.

What is Topic Modeling?

Topic modeling is a form of text mining, a way of identifying patterns in a corpus. You take your corpus and run it through a tool which groups words across the corpus into ‘topics’. Miriam Posner has described topic modeling as “a method for finding and tracing clusters of words (called “topics” in shorthand) in large bodies of texts.”

The Details: Training and Validating Big Models on Big Data

Sat, 01 Dec 2012 00:00:00 +0000

In this video, David Mimno discusses some of the different choices one can make in training models and what their implications are for efficiency, scalability, and topic quality, using the MALLET topic modeling package. This presentation was recorded on November 3, 2012 at the Maryland Institute for Technology as part of the Topic Modeling Workshop, sponsored by the National Endowment of the Humanities and MITH, at the University of Maryland. Slides are available here.

What Can Topic Models of PMLA Teach Us About the History of Literary Scholarship?

Sat, 01 Dec 2012 00:00:00 +0000

Of all our literary-historical narratives it is the history of criticism itself that seems most wedded to a stodgy history-of-ideas approach — narrating change through a succession of stars or contending schools. While scholars like John Guillory and Gerald Graff have produced subtler models of disciplinary history, we could still do more to complicate the narratives that organize our discipline’s understanding of itself.

Figure 1: A browsable network based on Underwood’s model of PMLA. Click through, then mouse over or click on individual topics.

Words Alone: Dismantling Topic Models in the Humanities

Sat, 01 Dec 2012 00:00:00 +0000

As this issue shows, there is no shortage of interest among humanists in using topic modeling. An entire genre of introductory posts has emerged encouraging humanists to try LDA.¹ So many scholars in humanities departments are turning to the tool in their research that it is sometimes described as part of the digital humanities in itself. Last fall, the NEH sponsored a workshop at Maryland which expressed concern that “the most promising work in topic modeling is being done not by humanists exploring literary or historical corpora but instead by scholars working in natural language processing and information retrieval.” There is not, it seems safe to say, another machine-learning algorithm in the world anyone would expect humanists to lead the progress of. Newcomers to the field could be forgiven for thinking that digital humanists need to topic model to prove their mettle; analog humanists could be forgiven for assuming that the computational interests of literary scholars and historians are particularly focused on the sorts of questions that topic models answer.

Code Appendix for “Words Alone: Dismantling Topic Models in the Humanities”

Sat, 01 Dec 2012 00:00:00 +0000

Topic Modeling Ships

Begin by getting the data in order. (This data is available on request.)

# Oceans2
rm(list = ls())
require(ggplot2)
require(plyr)
require(lubridate)
source("ICOADS parsing.R")
source("../Map Functions.R")

This step pulls in the Maury data and splits it. This is not fully documented as part of this article, since the purpose is to show general geodata parsing.

data = loadInData("~/shipping/ICOADS/maury.txt")
data = splitDataByVoyage(data)

Next, save the R data.frame into a directory that MALLET will be able to interpret as a text, and run a topic model on it. The rounding gives the resolution at which the topic will be run.