Evaluation of IR models
Create cores for BM25 and VSM using the createCore and indexer files
You can git a POST request via postman to delete the data and reindex it using the indexer everytime you modify the core schema: http://{GCP_IP}:8983/solr/VSM/update http://{GCP_IP}:8983/solr/BM25/update Body(JSON)- { "delete": { "query": ":" } } Run the json_trec_eval_lang.py file to generate the output of queries for BM25 and VSM Upload those files to GCP to evaluate the MAP scores. You need to have trec installed in your VM to run the trec_eval query
Query: trec_eval -q -c -M 1000 qrel.txt {output_file_for_BM25/VSM}
Improve MAP scores by modifying parameters (eg: k, b in BM25)
Get the best possible MAP scores