Introduction: The goal of this proposed method is to investigate that whether considering query types in query refinement (QR) improve the performance of information retrieval (IR) performance
Inputs:
2009 Million Query track topics (20001-60000)
Output:
The out put would be the proper QR method for each qury type.
Installation:
step 1: Put 2009 Million Query track topics (20001-60000) and query classes files into dataset folder. By running main.py, the first step of our method, which is to preprocess the input queries, will be applied. The result of this step is topics.trecMQ file, which is suitable input for the ReQue(we will explain in the next step).
step 2: Install ReQue. Put topics.trecMQ and prels relevance judgments into ds-->TrecMQ folder. The results of this step would be the evaluation of QR methods.
step 3: We would rank the QR methods in each query type. Finally, we will find the QR methods for each query in a specific query type which improves the IR performance, and all other methods were not able to improve such queries.(This results will bbe published in Experiment phase)