Corrupted queries in Spanish text retrieval
Juan Otero, Jesús Vilares, Manuel Vilares Ferro · 2008
In this paper, we propose and evaluate two different alternatives to deal with degraded queries on Spanish IR applications. The first one is an n-gram-based strategy which has no dependence on the degree of available linguistic knowledge. On the other hand, we propose two spelling correction techniques, one of which has a strong dependence on a stochastic model that must be previously built from a POS-tagged corpus. In order to study their validity, a testing framework has been formally designed and applied on both approaches.