2.txt

Do people feel better when on vacation?  We study the social media sentiment of families when on vacation and during normal periods.  Using a snowball sampling method, we solicited participants active on various social media platforms (N = 1,337) to voluntarily disclose when and where they went on vacation. We then crawled our participants’ social media posts to determine whether specific time periods and destinations for vacations were correlated with higher levels of enjoyment.  Initial analysis showed that such features do yield significant prediction performance improvement: for example, vacations taken close-by tended to have markedly higher satisfaction. Our model, Vacay-ok, accounts for such factors in novel gating network architecture, improving joint vacation period and sentiment detection by over 10% F1.

Our analysis validates patterns of interest: vacations with immediate family exhibit a bimodal distribution, and that ones without in-laws have a significantly higher satisfaction and detection rate. 

For robustness in identification, we collect and study the pattern and text of social media posts by parents, children, extended family members and friends of vacation-goers.
To ensure both the accessibility and the reproducibility of our experiments, we have gathered the raw social media data, culled with permission from the original posters, and released it as a zip archive, available without moderation, in the footnote.