Abstract
The aim of information retrieval (IR) process is to respond effectively to user queries by retrieving information that are better meets their expectations. A gap could exist between user information needs and his/her defined context as the level of the user's expertise in the search domain is directly influencing the query richness. The information retrieval system (IRS) must be intelligent enough to identify information needs and respond effectively to meet the expected needs, regardless the level of user's expertise level. This process is difficult and it remains a major and open challenge in the domain of IR. IR models integrate many sources to achieve an effective retrieval system. Semantic IR is an environment in which semantic techniques were applied to sort the documents according to their degree of relevance to the query. The present work proposes a hybrid model to rank documents. The proposed model is based on a query likelihood language model and the semantic similarity between concepts to assess the relevance between query-document pairs. Concepts were extracted by projection on WordNet ontology then word sense disambiguation was conducted. A semantic index was built to validate the proposed model. The conducted empirical experiments show that the proposed model is outperformed the compared benchmarks in the measured IR metrics.