Geographical queries reformulation using a parallel association rules generator to build spatial taxonomies

Geographical queries need a special process of reformulation by information retrieval systems (IRS) due to their specificities and hierarchical structure. This fact is ignored most web search engines. In this paper, we propose an automatic approach for building spatial taxonomy, that models’ the notion adjacency will be used in part geographical query. exploits documents are top retrieved list when submitting entity, which composed relation noun city. Then, transactional database constructed, considering each document extracted as transaction contains nouns cities sharing country submitted query’s The algorithm frequent pattern growth (FP-growth) applied his parallel version (parallel FP-growth: PFP) order generate association rules, form country’s taxonomy Big Data context. Experiments has been conducted on Spark results show query using constructed based our proposed improves precision effectiveness IRS.