Pseudoseer: Search Queries Paired with Pseudocode Articles

A dataset created for the Pseudoseer paper evaluation. It pairs search queries with target articles and their associated pseudocode blocks, including LaTeX source and reference context. The data was authored by Levent Toksoz and hosted on Harvard Dataverse, with a metadata update in June 2026.

Use Cases

• Training search ranking models based on query-article pairs.
• Evaluating pseudocode retrieval algorithms based on the provided query-target pairs.
• Analyzing the relationship between LaTeX source and reference context in academic documents.
• Benchmarking NLP models for understanding and generating pseudocode.

Strengths

• Data is explicitly created for a published academic paper evaluation, suggesting a clear research purpose.
• Each record includes multiple structured components: search query, target article, pseudocode blocks, LaTeX source, and reference context.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Harvard Dataverse

Collection Method

Extracted from a pseudocode corpus for the Pseudoseer paper.

Freshness

Last updated 2026-06-12 11:36:51; freshness should be verified.