🎲 LudoSym: Gambling Addiction IR Dataset (Spanish)

Overview

This dataset accompanies the paper:

Analyzing Gambling Addictions: A Spanish Corpus for Understanding Pathological Behavior
Manuel Couto-Pintos, Marcos Fernández-Pichel, Mario Ezra Aragón, David E. Losada
Findings of the Association for Computational Linguistics: EMNLP 2025

The dataset is designed to support information retrieval and ranking research focused on the detection and analysis of gambling addiction symptoms from Spanish text. It enables the evaluation and training of retrieval models that associate user-defined symptom queries with relevant textual evidence.

---

Task Description

The dataset follows a standard IR / ranking formulation:

• Queries represent standardized gambling addiction symptoms.
• Documents are short Spanish text passages related to gambling behavior.
• Relevance judgments (qrels) indicate how relevant each document is for a given symptom query.

The task is to rank documents by relevance for each query, enabling:

• neural ranking
• symptom screening
• cross-encoder and bi-encoder training
• IR evaluation (e.g., nDCG, MAP, Recall)

---

Dataset Structure

Each example in the dataset has the following fields:

Field	Type	Description
query_id	int	Unique identifier of the symptom query
doc_id	int	Unique identifier of the document
query	string	Text of the gambling addiction symptom
document	string	Spanish text passage
relevance	float	Relevance score between the query and the document

Citation

If you use this resource, please cite:

@inproceedings{couto-etal-2025,
    title = "Analyzing Gambling Addictions: A Spanish Corpus for Understanding Pathological Behavior",
    author = "Couto-Pintos, Manuel and
              Fernández-Pichel, Marcos and
               Aragón, Mario Ezra and
              Losada, David E.",
    booktitle = "Findings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP)"
}