A text analysis approach to analyse nostalgia in social media

Nostalgia, some people see it as a warm fuzzy feeling or a state of mind but it is a complex psychological construct that can be triggered in various ways such as revisiting your childhood neighborhood or looking at gifts reminding you of loved ones. Like many feelings, it’s double-sided! If you become too drawn to it, it may stop you from looking forward and experiencing new things in the world. This is nostalgia overload or the curse of nostalgia! In the digital era, it is easier than ever to reunite with old friends. You can take a virtual tour of a place from the old day! With over 1.8 billion active monthly users, Facebook is the biggest social media network on the Internet.  It also serves as a unique platform for nostalgic reunions with family and friends. It is full of “in the moment” expressions, reflecting nostalgic memories.

In a recent study, Davalos et al. (2015) used WordStat text mining software to analyze nostalgia on Facebook. The authors investigated how people expressed nostalgia in nearly 400,000 Facebook posts and extracted themes of nostalgic conversations.  Using the cluster analysis tool in WordStat, they found the word clusters, hidden themes in the data, as well as the associated words. The Jaccard’s index, also known as intersection over union, was used to assess the similarity. Davalos and his colleagues applied further measures available in WordStat to post process the clusters such as eliminating clusters with only one or two members. Their final solution contained nine clusters. Based on their content analysis, there exists a significant evidence of nostalgic expression in Facebook. The main topics people harkened back to were: family, life stories, historical events such as presidential elections or man on the moon, spirituality, appreciation of life, romanticism and fun.

WordStat, Provalis Research text mining and content analysis tool, is a powerful quantitative content miner which can detect hidden patterns in the text data based on different measures such as word co-occurrences and automatic coding. It can automatically detect and analyze the most important words and phrases and is perfect fit for most text data corpus such as interviews, electronic files, public speeches, etc. In addition to the automatic categorization, WordStat can also apply existing dictionaries to further fine tune the output.