A descriptor vector is added to every bucket in the EH* file. Then the substrings descriptor vector is compared with the keys descriptors vector and the buckets descriptor vector. At last, the keyset, where every key contains the substring, are retrieved.