Livestock farming practices and technologies, like many aspects of agriculture and industry, continue to evolve. As technology and attitudes change regarding livestock farming, public response changes as well; this is reflected in the way that people talk and write about the subject. This change and growth is a common topic of both public and technical debate and scrutiny. Databases on the internet collect public articles and documents related to livestock farming dating back to the early 1980’s. The information in these articles can be evaluated using a number of computer science based approaches. These data can help to highlight how significant past events and their impacts were perceived, and possibly predict how future trends within the industry will be described in popular press/media.
What Did We Do?
We gathered popular press articles from an online database, Factiva, with the search terms “livestock and odor,” from the year 2000 to the present. A computer program developed using machine learning processes: (1) cleans and structures the individual articles into text files; and (2) quantifies the importance and frequency of words in individual and groups of articles, by year. The program assigns two measures of importance to each word. Words that frequently occur in many articles per year provide broad overarching ideas and subjects. Words that are deemed important to each individual article provide more nuanced data including companies, people, and equipment discussed in livestock farming. To demonstrate the results, this data is visualized in tables and graphs to show patterns in subjects as they develop and change over time.
What Have We Learned?
This analysis method gives us a quantitative basis for reviewing the change in importance of words over time. All analysis after choosing the subject and search terms is done by a computer program, protecting the outcomes from reader bias. Changes in word importance or frequency can be supported with numerical data and easily visualized from year to year. The different approaches also allow for inferences between long-term subjects and ideas (Table 1), and shorter term players in the industry (Table 2).
This analysis method does not pull out the context that any of the words are used. Manure and waste are two means of describing the same material, with different connotations. Manure and waste appeared at similar frequencies in many, but not all years. Dairy was more prominent in 2011 and 2013, but hogs (or synonyms) appeared in most years. Refinements to the article search protocol could limit the articles to those of opinion (i.e. editorials) or regional perspectives. There are opportunities for this method to inform historical reviews of livestock and the environment, and inform future communication efforts.
There are a number of opportunities to extend this project in the future. One would be to experiment with different search terms and databases to see how outcomes depend on the data source. Another opportunity would be to apply the quantitative method to other applications. The computer program could be applied to any database and so the method has utility to topics other than livestock farming.
Ryan Felton, Undergraduate Research Assistant, University of Minnesota
Erin Cortus, Assistant Professor and Extension Engineer, University of Minnesota
Project support provided by the University of Minnesota UROP program.
The authors are solely responsible for the content of these proceedings. The technical information does not necessarily reflect the official position of the sponsoring agencies or institutions represented by planning committee members, and inclusion and distribution herein does not constitute an endorsement of views expressed by the same. Printed materials included herein are not refereed publications. Citations should appear as follows. EXAMPLE: Authors. 2019. Title of presentation. Waste to Worth. Minneapolis, MN. April 22-26, 2019. URL of this page. Accessed on: today’s date.