Yale University

04/25/2018
Speaker(s):
Title:
Audience: Restricted (Call)

Dear CIRA Affiliate,

As part of our effort to keep CIRA researchers and community affiliates abreast of the newest methods, we are conducting a review of advanced methods in text mining and natural language processing on Wednesday April 25th at 1pm in the CIRA conference room 202, 135 College Street, Suite 200. While the IRM core has promoted the use of text mining as a complimentary tool to traditional qualitative analysis, new, easily accessible methods in natural language processing. On January 20th a new package in R “ Tidy Text” was published. This excellent, easy to use new software allows a number of text mining functions and natural language processing. I used in the class I teach in African Studies and students who had never used R were able to do this fairly sophisticated text mining and topic on USAID administrators’ publications over the study period (1962-2008). FYI, LDA = Latent Dirichlet Allocations. In natural language processing, Latent Dirichlet Allocations is a generative statistical model that allows sets of observations to be explained by unobserved groups that explain why some parts of the data are similar.

I would like to invite all CIRA affiliates to the next IRM core meeting to review this innovative methodology to explore whether it would be appropriate for CIRA researchers and community organizations. I believe that application of these methods would make CIRA affiliated research and evaluations of implementation science community efforts more distinctive and advanced.

Trace Kershaw, IRM Core director has noted that “This type of analysis actually has a lot of potential uses for community members. For example, it could be used to analyze feedback forms, field notes, meeting notes, clinical notes, responses on social media---particularly when there is a lot of information that would be hard or labor intensive to do by hand using traditional qualitative methods. It is also useful when different people are involved and you want to try to understand common themes when people use different language or notations. So let’s say you have 5 peer navigators who take field notes, but all use different short hand or terms to describe similar clinical factors or risks---this type of analysis can use its computer brain to figure out that these different terms all mean the same thing and would group them together---making it easier for the community agency to see patterns in the behaviors and risks of their clients and modify services accordingly.”

Please find below a link to this important new software:
https://www.tidytextmining.com

Light refreshments will be provided. Contact elizabeth.cappello@yale.edu if you plan on joining in person or by Zoom video conference.


Location: CIRA, Suite 200, Room 202