Italian Cultural Heritage Text Descriptions with Quality Labels

A collection of over 100,000 textual descriptions of cultural items from the Italian national aggregator Cultura Italia. Each description is labeled as HIGH or LOW quality based on adherence to cataloging standards set by the Istituto Centrale per il Catalogo e la Documentazione (ICCD).

Use Cases

• Train a text classification model to predict the HIGH or LOW quality label from the textual description of cultural items.
• Analyze patterns in textual descriptions that correlate with the quality label to identify common features of high-quality metadata.
• Benchmark natural language processing models on the task of acceptability classification for cultural heritage metadata.
• Investigate the linguistic characteristics of Italian-language cultural descriptions flagged as LOW quality.

Strengths

• Over 100,000 textual records provide a substantial sample for analysis.
• Quality labels are defined by adherence to established cataloging guidelines from the ICCD.
• Annotations are expert-generated, suggesting a reliable ground truth.

Limitations

• The specific features (columns) within each textual description are not detailed, limiting feature engineering.
• The dataset is monolingual (Italian), restricting applicability to other languages.
• The criteria for the HIGH/LOW quality label are not fully specified in the provided input.

Provenance

Source

Cultura Italia, the Italian National Cultural aggregator.

Collection Method

Textual descriptions were labeled according to standard cataloguing guidelines from the Istituto Centrale per il Catalogo e la Documentazione (ICCD).

Freshness

Last updated on January 8, 2024.

Geography

Italy