Run description: Axiomatic Relevance feedback using ClueWeb12 FULL collection. The expansion working set used was the commercial search engine snippets crawled this year. The ranking model is BM25 with b=0.5 (instead of F2Log)
Run description: Axiomatic Relevance feedback using ClueWeb12 FULL collection. The expansion working set used was the wiki dump at June 2018. The ranking model is BM25 (instead of F2Log). The Beta in Axiomatic Reranking was set to 0.5
Run description: Axiomatic Relevance feedback using a subset ClueWeb12 collection by first running the Dirichlet Language Model and sample the first 10K documents from each query results. The expansion working set used was the subset collection itself. The ranking model is Dirichlet Language Model (instead of F2Log). The Beta in Axiomatic Reranking was set to 1.7
Run description: Axiomatic Relevance feedback using a subset ClueWeb12 collection by first running the Dirichlet Language Model (using Anserini not Indri) and sample the first 10K documents from each query results. The expansion working set used was the subset collection itself. The ranking model is Dirichlet Language Model (instead of F2Log). The Beta in Axiomatic Reranking was set to 1.7
Run description: Axiomatic Relevance feedback using a subset ClueWeb12 collection by first running the Dirichlet Language Model (using Anserini not Indri) and sample the first 10K documents from each query results. The expansion working set used was the wikipedia dump from June 2018. The ranking model is BM25 (instead of F2Log). The Beta in Axiomatic Reranking was set to 0.5