# Knowledge Graph Question Answering on WebQSP (Macro-F1 and Hallucination Reporting) Source: [Knowledge Graph Question Answering on WebQSP (Macro-F1 and Hallucination Reporting)]() Collection index: [llms.txt]() Data updated: 2026-04-14T10:41:30.985Z Evaluation of model performance on the WebQSP dataset for Knowledge Graph Question Answering, specifically reporting Macro-F1, Hit rate, and Hallucination metrics. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Knowledge Graph Question Answering | | Dataset (including any stated split) | WebQSP | | Metric | Hit Rate | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 21 of 21 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Hit Rate](): Higher is better; unit: PERCENT. - [Macro-F1](): Higher is better; unit: PERCENT. - [Hallucination Rate](): Lower is better; unit: PERCENT. ## Recorded results | Method | Hit Rate | Result date | Report paper | | --- | --- | --- | --- | | ParallaxRAG + GPT-4o (200) | 0.9353 | Not recorded | 68fa7c461e209131cea8d5a4 | | ParallaxRAG + Qwen3-30B (500) | 0.9326 | Not recorded | 68fa7c461e209131cea8d5a4 | | REL-RAG + GPT-4o-mini | 0.925 | Not recorded | 68fa7c461e209131cea8d5a4 | | ParallaxRAG + Qwen3-30B (200) | 0.9248 | Not recorded | 68fa7c461e209131cea8d5a4 | | ParallaxRAG + Qwen3-30B | 0.916 | Not recorded | 68fa7c461e209131cea8d5a4 | | SubgraphRAG + GPT-4o | 0.898 | Not recorded | 68fa7c461e209131cea8d5a4 | | ParallaxRAG + Llama3.1-8B (↔) | 0.8951 | Not recorded | 68fa7c461e209131cea8d5a4 | | RoG + GraphRAG-FI | 0.8925 | Not recorded | 68fa7c461e209131cea8d5a4 | | ParallaxRAG + Llama3.1-8B | 0.8685 | Not recorded | 68fa7c461e209131cea8d5a4 | | RoG-Joint | 0.8667 | Not recorded | 68fa7c461e209131cea8d5a4 | | GNN-RAG | 0.857 | Not recorded | 68fa7c461e209131cea8d5a4 | | SubgraphRAG + Llama3.1-8B (↔) | 0.8342 | Not recorded | 68fa7c461e209131cea8d5a4 | | ToG (GPT-4) | 0.826 | Not recorded | 68fa7c461e209131cea8d5a4 | | EtD | 0.825 | Not recorded | 68fa7c461e209131cea8d5a4 | | RoG-Sep | 0.8219 | Not recorded | 68fa7c461e209131cea8d5a4 | | Retrieve-Rewrite-Answer | 0.7936 | Not recorded | 68fa7c461e209131cea8d5a4 | | StructGPT | 0.7469 | Not recorded | 68fa7c461e209131cea8d5a4 | | G-Retriever | 0.7346 | Not recorded | 68fa7c461e209131cea8d5a4 | | KD-CoT | 0.686 | Not recorded | 68fa7c461e209131cea8d5a4 | ### ParallaxRAG + GPT-4o (200) Hit Rate: 0.9353. Result ID: 69de196047f34ce924c6430d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"GPT-4o","Number of retrieved triples":"200"}. This record does not assert independent reproduction. ### ParallaxRAG + Qwen3-30B (500) Hit Rate: 0.9326. Result ID: 69de196047f34ce924c6430c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"Qwen3-30B","Number of retrieved triples":"500"}. This record does not assert independent reproduction. ### REL-RAG + GPT-4o-mini Hit Rate: 0.925. Result ID: 69de196047f34ce924c64308. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"GPT-4o-mini"}. This record does not assert independent reproduction. ### ParallaxRAG + Qwen3-30B (200) Hit Rate: 0.9248. Result ID: 69de196047f34ce924c6430b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"Qwen3-30B","Number of retrieved triples":"200"}. This record does not assert independent reproduction. ### ParallaxRAG + Qwen3-30B Hit Rate: 0.916. Result ID: 69de196047f34ce924c6430a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"Qwen3-30B","Number of retrieved triples":"100"}. This record does not assert independent reproduction. ### SubgraphRAG + GPT-4o Hit Rate: 0.898. Result ID: 69de196047f34ce924c64307. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"GPT-4o"}. This record does not assert independent reproduction. ### ParallaxRAG + Llama3.1-8B (↔) Hit Rate: 0.8951. Result ID: 69de196047f34ce924c6430f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"Llama3.1-8B","Cross-dataset generalization":"true"}. This record does not assert independent reproduction. ### RoG + GraphRAG-FI Hit Rate: 0.8925. Result ID: 69de196047f34ce924c64303. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### ParallaxRAG + Llama3.1-8B Hit Rate: 0.8685. Result ID: 69de196047f34ce924c64309. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"Llama3.1-8B","Number of retrieved triples":"100"}. This record does not assert independent reproduction. ### RoG-Joint Hit Rate: 0.8667. Result ID: 69de196047f34ce924c64301. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GNN-RAG Hit Rate: 0.857. Result ID: 69de196047f34ce924c64306. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SubgraphRAG + Llama3.1-8B (↔) Hit Rate: 0.8342. Result ID: 69de196047f34ce924c6430e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"Llama3.1-8B","Cross-dataset generalization":"true"}. This record does not assert independent reproduction. ### ToG (GPT-4) Hit Rate: 0.826. Result ID: 69de196047f34ce924c642fe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Generator":"GPT-4"}. This record does not assert independent reproduction. ### EtD Hit Rate: 0.825. Result ID: 69de196047f34ce924c64305. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### RoG-Sep Hit Rate: 0.8219. Result ID: 69de196047f34ce924c64302. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Retrieve-Rewrite-Answer Hit Rate: 0.7936. Result ID: 69de196047f34ce924c64300. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### StructGPT Hit Rate: 0.7469. Result ID: 69de196047f34ce924c642ff. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### G-Retriever Hit Rate: 0.7346. Result ID: 69de196047f34ce924c64304. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### KD-CoT Hit Rate: 0.686. Result ID: 69de196047f34ce924c642fd. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.