CyberJustice-TJSP: A Dataset of Brazilian Judicial Cases on Cybercrime

CyberJustice-TJSP is a dataset compiled from publicly available judicial decisions of the São Paulo Court of Justice (TJSP) related to cybercrime. The dataset contains the full text of judicial decisions, along with associated metadata, topic labels obtained through topic modeling, and manual annotations. It is structured in three versions:

Dataset I: 1,874 judicial cases containing the full text of decisions, metadata, and topic labels automatically assigned through Latent Dirichlet Allocation (LDA).

Dataset II: a token-level representation of the same 1,874 cases, resulting in 12,651 rows, where each row corresponds to an identified token linked to its respective case, preserving both the original metadata and the topic label.

Dataset III: a manually annotated subset of 300 cases, with 166 labeled as cybercrime-related and 134 as non-cybercrime, serving as a gold standard for supervised classification.