We develop a new, generalized language modeling approach for information retrieval by adopting the probabilistic framework of quantum theory to account for both single and compound terms at once without having to extend the term space artificially as in previous studies. Moreover, we propose a natural solution to the weight-normalization problem, which arises from the fact that current retrieval models mix scores from matching compound terms and from matching single terms. Our model is the first practical application of quantum probability to show significant improvements over a robust bag-of-words baseline and achieves better performance on a stronger non bag-of-words baseline.

**Cette présentation est une "pratique" pour une communication à être présentée à SIGIR 2013**