Twitter Search Personalization Research

This repository contains assets that were used on research about personalization on Twitter Search.

Our article Is There Personalization in Twitter Search? A Study on polarized opinions about the Brazilian Welfare Reform on WebSci2020 was accepted. https://dl.acm.org/doi/abs/10.1145/3394231.3397917

Structure:

• datasets/daytrends: dataset of daily Brazilian Twitter trending topics from 2019-02-01 to 2019-08-30
• datasets/training-data: dataset of tweets for hashtags: #EuApoioANovaPrevidencia and #LutePelaSuaAposentadoria. From each, we extracted the top 100 profiles for #EuApoioANovaPrevidencia (reforma_top_100_profiles_a_favor.json) and #LutePelaSuaAposentadoria (reforma_top_100_profiles_contra.json)
• datasets/twitter-search-results: dataset of the captures search results. Method for capturing were described on coming article and we use our tool (multiquerier-tool) based on Puppetter to capture these data.
  • _dataset_1__en_pt-merged.json contains the raw captured data
  • exported_dataset_metrics.csv contains the processed results we calculated metrics.
• multiquerier-tool: tool that we developed instantiating Twitter agents that run queries and capture the results.

Contacts: jonatas.santos@uniriotec.br