Email communications are a rich source of personally relevant information, which can be leveraged to study cognitive processes. In this project, we designed a Web-based program that analyzed each participant’s year-old sent email archive and applied textual analysis algorithms to identify a set of sentences likely to be memorable from those emails. Our process involved cleaning and fetching the email archive, performing named entity recognition (NER) to extract names from emails, and then developing an iteratively refined scoring algorithm, that weighed email content based on the certain features, like sentiment in emails (indicated by sentiment words and emojis), reflective words, frequency of emailing and conversational indicators (exclamations and question marks). Through this algorithm, we identified memorable sentences in the e-mails and used them as stimuli in a memory task, where participants were shown sentences from their email and asked to recall the name of the email recipient. Our analyses of experimental results showed that the algorithm was successfully able to retrieve memorable sentences from the emails, and recall was predicted by different aspects of the email conversation, such as the age of the email, the frequency of interaction with the recipient and sentiment in the email.
The R files in this repository go over the code use to analyze data from the e-mail communications. Please refer to the original paper published in Behavior Research Methods for more details. For access to the original program, please click here