In one of the first papers using ChatGPT-4 for text analysis, I find that it beats both crowdworkers and experts in a difficult tweet annotation task
This work has convinced me that LLMs will completely transform how we use text in the social sciences 🧵
arxiv.org/abs/2304.06588
Communalytic is not just for social media data anymore. If you have textual data, you can now upload them to run sentiment, toxicity and topic analysis.
DYK that you can ( re)analyze your existing social media dataset with @Communalytic?
Here's how to format and import a CSV data file to Communalytic for further analysis. Note:
EDU CSV data file limit is 30k records, PRO can support up to 10M records. communalytic.org/docs/tutori…
We're finalizing documentation for our new @Communalytic 3D Semantic Similarity Map & will make a formal announcement in early Sept. But if you like to give it a try, the ability to create embeddings semantic maps are now available in both the EDU and PRO versions. #SoftLaunch
@Communalytic EDU is meant for teaching & learning. Its feature set is more limited but it's free to use and is an ideal platform to help students learn about social media data analytics. communalytic.org/
@Communalytic PRO is designed for researchers, journalists and data analysts and is ideal for large-scale research projects. It provides users with the resources and infrastructure necessary for conducting independent research in the public interest. communalytic.org/
How did parties campaign for the #EU2024 Parliament elections? Together with @wiebkedrews, @andreas_neumeie and @RiedlJasmin from @projekt_sparta, we published the EUDigiParty data set today including social media handles, website links etc. of 401 European parties. Link follows!
Are you studying online toxicity or online moderation?
Our @Communalytic research tool has a built-in toxicity analysis module that can automatically pinpoint various types of toxic interactions in a dataset. communalytic.org/docs/tutori…#ICWSM
One of the unique features of the built-in toxicity analysis module in @Comunalytic is that it can automatically pinpoint toxic & anti-social interactions in your dataset & visually highlight those interactions in a sign network. communalytic.org/tutorials/t…#ICWSM#ICWSM24
A signed network is a network w/edges that contains additional info such as toxicity or sentiment polarity scores as weights. This feature can be used to identify & visually highlight interactions of interest (e.g., anti-social interactions).
If you have any questions about the GESIS web data collection for the 2024 EU election, you can contact the team via e-mail: css.datacollection(at)gesis(dot)org
CrowdTangle will be shutting down on August 14, 2024. If you've access to CT, you can use @Communalytic to collect & analyze public FB & IG posts that share the same URL (ex., a URL to a single NYT story or any domain name). #coordinatedlinksharingcommunalytic.org/docs/crowdt…