Polis is opinion clustering, TTTC is topic clustering
from Talk to the City Try
Polis is Opinion Clustering, Talk to the City is Topic Clustering
TTTC is not "clustering of opinions" but "clustering of topics"
I thought this might be the case when I saw Talk to the City clustering, but was captured by the concrete facts.
BERTopic, TF-IDF, and the actual implementation uses CountVectorizer.
What this means is that "ban X" and "do not hinder X" fall into the same cluster.
https://gyazo.com/1dce5751a19f8b8a3aa63f7b610c478f
https://gyazo.com/a89d24e0979cb84b551be5de6ec47b80
The Copyright Act, Article 30-4. restricts copyright with respect to (Use not for the purpose of enjoying the ideas or sentiments expressed in the work).
The former says this provision is bad because it undermines the welfare of creators.
The latter is saying that if right holders such as the media interpret the proviso of this provision ["However, this shall not apply in cases where the interests of the copyright holder would be unreasonably impaired in light of the type and use of the work and the manner of such use"] in a broad and strong manner, it would prevent "large-scale linguistic data collection for AI development" (and is therefore "bad"). (and therefore bad).
So he's saying "large scale linguistic data collection for AI development" is Good.
These two are plotted close together
There is no mechanism in place to analyze this type of good/bad sentiment (sentiment).
---
This page is auto-translated from /nishio/Polisは意見のクラスタリング、TTTCはトピックのクラスタリング using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.