Identifying human morals and values in language is crucial for analysing lots of human- and AI-generated text.
We introduce "MoVa: Towards Generalizable Classification of Human Morals and Values" - to be presented at @emnlpmeeting oral session next Thu
🧵 (1/n)
What is the mathematical structure of many people paying attention to many online items?
Our #TheWebConf2023 paper has answers - "Stability and Efficiency of Personalised Cultural Markets" w Haiqing Zhu and Yun Kuen Cheung arxiv.org/abs/2302.06226 🧵 (1/n)
Replying to @john_c_hawkins
hmm, i don't think so.
@autreche have done a series of nice point process talks/tutorials but that was before talk recording became the norm ...
@shchur_ 's fairly recent IJCAI review paper has a video
youtube.com/watch?v=J7qH7i0E…
On the folktable dataset, we show that different preprocessing decisions not only affect dataset balance, but also affect fairness outcomes.
(5/n)
The smallset software package supports Timeline creation with structured comments in R or python pre-processing scripts.
(3/n)
In a nutshell, Smallset Timelines are compact visual representations with color coded changes and captions for each step. It is intended to be included in publications and data/code releases to help researchers and end-users understand and reproduce preprocessing.
(2/n)
Data preprocessing is messy and under-documented, yet it is full of details that matter.
Ever wonder how best to fill in the preprocessing questions in Datasheets for Datasets, Model Cards, or Data Nutrition Labels? Our #FAccT2022 paper "Smallset Timelines" offers help.
(1/n)
Promotion of right leaning videos tend to be decentralised, with large cascades starting earlier; whereas left-leaning videos tend to involve central actors with tweet cascades starting later. (4/n)
Across three long-running controversial topics - Gun Control, Abortion and BLM over 16 months, we found left-leaning videos *more viewed* overall, more engaging, but *less tweeted* than right-leaning videos. (2/n)
"Conservative advantage" refers to the commonly observed strategic disseminations of right-wing activism. But whose advantage is it when attention is measured across YouTube and Twitter? Our #icwsm2022 paper has an answer, w Jooyoung Lee, @avalanchesiqi, UPitt @PICSOLab (1/n)










