A Novel Information Retrieval Model Inspired By Quantum Weak Measurement

Recently,one of the most important highlights in information retrieval is how to build an effective user model.Since the quantum mechanics theory was introduced into the filed of information retrieval in 2004,more and more researchers have used quantum theory to model the change of user' information needs in retrieval process.However,there are still some user behaviors that can't be well explained using quantum standard measurement theory.In this paper,we have found many similarities between the process of user relevance judgement and the process of quantum weak measurement through sufficient user experiments.Quantum weak measurement theory is the generalization of quantum standard measurement theory.We try to use quantum weak measurement theory to give a robust explanation to user's cognitive changes in retrieval process from a novel perspective.Further,inspired by the theoretical process of quantum weak measurement,we have also proposed a quasi weak measurement retrieval model.Instead of the common cosine similarity or KL divergence,we use the weak value formula to score the relevance of the documents.Experimental results on Session Search tasks show that our model outperforms classical retrieval models and existing quantum retrieval models.We propose that the theory of quantum weak measurement is more applicable to the analysis and modeling of the user's cognitive behavior in the retrieval process than the quantum standard measurement theory.