The digital library is a searchable store of text, audio and visual materials, and more. However, as the amount of data that must be stored and made searchable increases, efficient management and retrieval can be a major headache for the digital library. Research in the International Journal of Information and Communication Technology could lead to a new solution to the problem by integrating image processing, big data analytics, and deep learning techniques.
Xiaoyan Wang and Meimei Jia of Nanyang Normal University, Nanyang, Henan, China, like others, recognise the complexities of enhanced multimedia search and retrieval and have turned to the modern tool of deep learning to help. The team has developed a cross-media semantic search framework. This finds and uses the correlations between different types of media to assist in search and retrieval. Their deep learning algorithm can analyse and organize multimedia resources to improve search accuracy and system performance. Indeed, the use of this cross-modal correlation analysis and hierarchical knowledge inference in refining search results gave the team an almost 12 percent boost in search performance when compared with conventional approaches.
The approach will be useful for generic digital libraries and could be extended to personal devices such as smartphones, enabling greater access to useful information silos for the lay public. In more specialist applications, the same approach might be used in medical and scientific information systems. This could allow complex medical imagery, such as MRI scans and diagnostic test data and chemical databases, to be more readily searchable. In enterprise knowledge management systems, with such a system, companies could better handle their ever-accumulating data and information. Moreover, in any field where vast amounts of multimedia data accumulate every day and need efficient search and retrieval methods to make the most of the information they hold, an improved system will benefit users.
Wang, X. and Jia, M. (2024) ‘Development of a unified digital library system: integration of image processing, big data, and deep learning’, Int. J. Information and Communication Technology, Vol. 24, No. 3, pp.378–391.