A technique of high-speed information processing is realized by determining a method of accessing processing target data so as to allow high-speed access in consideration of a memory architecture. According to the technique, in a method of performing information processing by sequentially referring to element data of the processing target data stored in a main memory according to a predetermined information processing rule such as a recognition dictionary, when generating the information processing rule, a reference order of the element data which improves a cache hit rate is determined based on a rule for storing the element data of the processing target data in the main memory, records of the positions of referred element data, and the cache architecture.