My Proposed Project

Project Overview

Iowa church and community cookbooks can reveal a lot about the historic demographics of a community, including showing the influence of different ethnic communities on local foodways, highlighting a shared food culture across Iowa communities, and revealing recipe traditions unique to specific communities. For example, a Pella church cookbook contains a subsection of Indonesian recipes. Closer examination reveals these recipes are not from Indonesian residents in Pella but rather are from church-supported missionaries to Indonesia returning to Pella and bringing food culture back with them (See other examples below). These examples are currently only discoverable by close reading, however, by creating a machine-readable corpus, these and many other examples can be revealed via distant reading and machine learning.

Using Python and the Natural Language Toolkit (and other Natural Language Processing (NLP) Python libraries), this project will examine a text corpus of community cookbooks from the Library’s Iowa Cookbook Collection in order to identify groupings of similar recipes that can reveal a shared food culture among Iowa communities as well as the evolution and adaptation of shared recipes over time and recipes that share ingredients and similar preparation but are called by different or ambiguous names. This analysis will also identify recipes that are unique to particular communities and that may highlight ethnic and cultural histories of those communities that are worthy of further study.

By examining this important Library collection in a new way using Natural Language Processing, this project will build upon existing text analysis and Python programming experience within DSI in order to support future teaching opportunities with an expanded skillset in machine learning. Through this project, DSI will be able to use the resulting corpus and text analysis workflow in bootcamp-style instruction settings and workshop series in collaboration with Research Data Services that move beyond the brief, one-off workshop to support students and faculty in developing deeper knowledge of text analysis from start to finish.

Other areas of possible exploration with the proposed corpus:

  • The emergence of food fads by over-representation of specific ingredients or brief spikes in specific recipes over time
  • Changes in canning/food preservation practices over time
  • The emergence, adoption and adaptation of Italian, Chinese, Mexican, and Indian cuisines in Iowa communities (can be compared to national analysis of same trends in American cuisine via secondary sources)

Project Justification/Benefits

  • Use of machine learning to highlight and examine an important library collection in new ways
  • Supports DSI skills development in text analysis/machine learning for future teaching opportunities
  • Develops a dataset and text analysis workflow for potential future use in DH bootcamps, DSI/RDS workshop series, or graduate fellowships where a deep dive into digitization and corpus development for text analysis from start to finish is desirable
  • Supports DSI collections as data efforts, establishes possible workflow for future collections as data efforts (w/ Preservation)
  • Supports land grant mission to provide greater access to Iowa cultural material
  • Opportunity for partnership with AI/machine learning experts and scholars interested in food history/Iowa history outside of the library
  • Identify opportunities for future development of the Iowa Cookbook Collection from underrepresented Iowa communities