Membership classification using Integer Bloom Filter

Abstract

Due to the large quantity of digital information now available, information search engines provide a popular and important Internet service. Issues involved in the improvement of digital content search efficiency include: keyword filtering, inefficient search filtering, and existence search queries. Internet services are currently focusing on improving efficiency and accuracy. Through a pre-processing filter of inefficient search contents, a waste of Internet resources can be avoided and search efficiency can be improved. This study proposes an Integer Bloom Filter (IBF) that combines the concepts of a Bloom Filter (BF) and an artificial neutral network. It is based on the basic structure of the Bloom Filter so that multiple attribute existence algorithms can be developed. The algorithm's characteristics include: error-detected ratio, parallel computing, multiple attribute identification, non-fixed length string sample applications, as well as dynamic sample addition. With the non-fixed length string sample, the research results show that under proper conditions, the error-detected ratio has a very satisfactory performance and an on-line/off-line application field demonstrates its stable and highly efficient performance.