Icews14: Temporal Knowledge Graph Event Data for Reasoning Tasks

Icews14 is a dataset for temporal knowledge graph reasoning tasks. It was built over the ICEWS and GDELT event databases, which are widely used benchmarks in the field. The dataset was first introduced in the 2022 paper 'TFLEX: Temporal Feature-Logic Embedding Framework for Complex Reasoning over Temporal Knowledge Graph' and was uploaded to Hugging Face by user 'linxy' in May 2024.

Use Cases

• Benchmarking temporal knowledge graph completion models based on event sequences.
• Training models for complex logical reasoning over time-dependent facts.
• Evaluating embedding frameworks on structured event data from sources like ICEWS and GDELT.

Strengths

• Built over established, widely used benchmarks (ICEWS and GDELT) in the temporal knowledge graph community.
• Introduced in a peer-reviewed academic paper (TFLEX) published on arXiv in 2022.
• Dataset metadata was last updated on Hugging Face on 2024-05-05.

Limitations

• Description metadata is limited; actual data quality, column structure, and file formats require manual inspection after download.
• Row count and dataset size are unknown, which may limit suitability assessment for large-scale training.
• Column-level documentation is absent; field semantics must be inferred from the source paper or after download.

Provenance

Source

Built over the ICEWS (Integrated Crisis Early Warning System) and GDELT (Global Database of Events, Language, and Tone) event databases.

Collection Method

Methodology for dataset creation is described in the associated 'TFLEX' research paper.

Time Range

The description suggests the data likely contains temporal event sequences, but the specific time range is not provided.

Freshness

Last updated 2024-05-05 13:39:25; freshness should be verified.

Geography

The source databases (ICEWS, GDELT) typically contain global event data, but the specific geographic coverage of this subset is not stated.