Putting old data into new system: Web-based catalog card image searching

This paper proposes a new approach to solve the data inputting bottleneck problem for library catalog data, or metadata. The data have been provided by paper cards arranged in wooden boxes. A lot of efforts have been taken to digitize them in order to put these data to be machine-readable. However, despite such efforts, only a small amount of data has been digitized so far because the inputting is done manually. We solve this problem by using the catalog card images digitized by high-speed scanners. This approach has advantages such as: (1) we can deal with the electronic catalog data with remarkably reduced time and cost; (2) it enables the seamless integration of the image-based and keyword-based searches; and (3) it boosts up the process of inputting of the catalog data itself