Version 0.1
Please note that this a beta version and is subject to further changes
The Card Catalogue application is the result of an experiment with OCR technologies (Object Character Recognition).
Each card from our old Card Catalogue has been scanned and the automated final result is now searchable.
Information Systems at Senate House Library developed a small application which scanned each image held in our old card catalogue and transferred the text to a SQL Server database. The application utitlised Microsoft's 'Office Document Imaging' software library to read the image. The OCR result was then subject to language detection via a small custom made application and Bing services. The OCR text then went back through the OCR application with the correct language scaffolding.
The result you see is not perfect, indeed, it is far from it. However, our intention was to augment the search result rather than replace the 'drawer' search method of our old system.
This application is far from finished; we intend to add the following 'shopping cart' style collection and, more importantly, community contributions. We will also look to changing the technology we use to index the application to Lucene, a more advanced indexer than we are using now.
The latter will further add to the discovery by helping to make our indexed content more accurate.
You may also note that there are a couple of other features on the item display. We have added our own catalogue search, a Microsoft Academic API search and an interesting natural language analysis tool called Alchemy API.
If you have any questions or suggestions regarding our Card Catalogue application please contact us at information.systems@london.ac.uk
Card Catalogue