Novel Automatic Query Building Algorithm Using Similarity Thesaurus

One of the most effective factors on the natural language researches is the data set which plays a significant role in designing, improving and evaluation the information retrieval systems and other applications for natural language processing. Unfortunately, building a proper data set consume time, labor and effort, in particular the query extraction from the data set documents. In this study, a novel algorithm for query extraction from any collection of documents was suggested, the algorithm elaborate the similarity thesaurus for query extraction, which leads to the ability of using the algorithm on any language, to evaluate the suggested algorithm a data set that consist of 242 Arabic documents and 60 queries was used, 48 queries was extracted 20 of them appeared in manual data set and all of them was relevant with more than one document in the used collection.