PulseBench-Tab: A Multilingual Benchmark for Table Extraction with Graph-Based Evaluation
2026-06-09T08:52:19Z•f7cac2d325de9e9af4c98ddc1b7a31e23ca2e29b13910c492fb5ebd3f0f3e18b
agent-frameworkarabicbenchmarkbiomedical-qacjkcompression-transformercyrillicdatasetfairness-reusabilitygemini-flashlong-sequence-modelingmedhopqametadata-enrichmentmsrmultilingualprompt-engineeringpulsebench-tabrecommendation-systemsremote-sensingsemantic-complementaritysemantic-idst-lagtable-extractiontool-dependency-graphtool-retrieval
What happened
This silver document bundles multiple arXiv submissions (09 Jun 2026) across IR, ML, recommender systems, and dataset/benchmark work. Highlights: PulseBench-Tab — a 1,820-table multilingual table-extraction benchmark (9 languages) plus a graph-based T-LAG metric; a bidirectional semantic complementary tool-retrieval method for remote-sensing agents with dynamic tool-dependency graphs; a production-scale semantic-native long-sequence modeling approach for short-form-video recommendation using Semantic IDs and a Global-Aware Compression Transformer; advanced prompt-engineering evaluation of Gem1
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_ir
- Record identifier
- f7cac2d325de9e9af4c98ddc1b7a31e23ca2e29b13910c492fb5ebd3f0f3e18b
- Enrichment time
- 2026-06-09T08:52:19Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.