Optical-character-recognition (OCR) software development services

Description

The National Library of Norway would like to keep the content of the physical collections for the future through the digitalisation programme. At the same time it would like to form the basis for future digital services on nb.no. The National Library of Norway would like to reproduce the collection in high quality digital format in an authentic way. The content and scope of the collection is so comprehensive that well-defined, stable and cost efficient systems for digitalisation must be used. The National Library of Norway's newspaper collection includes most newspaper editions in Norway through the ages. Large parts of the collection are available as both the original paper newspaper and on silver-based microfilm. As an important element of the digitalisation of the newspaper collection, the National Library of Norway would like to establish a framework agreement with a tenderer who can digitalise considerable amounts of microfilmed newspapers, whilst the National Library of Norway will itself digitalise from the paper newspapers where wanted or required. The National Library of Norway's microfilm collection consists of more than 70 000 microfilm rolls. Most of the collection is 35 mm unperforated, black/white microfilm. The microfilms were produced in the period 1950-2009. The condition and quality will vary somewhat, depending on when and where the films were produced. Tenderers must describe in the tender how they can digitalise microfilm of varying quality. It is particularly important to get clarify any limitations that the National Library of Norway must include in their calculations when large parts of the microfilm collection is to be digitalised. A microfilm roll normally includes between 500 and 1 400 pictures, each one of a single newspaper page. The microfilm format is unperforated 35 mm film, which is normally not longer than 30 m. The contracting authority estimates that the digitalisation service will include approx. 2 000 000 — 4 000 000 pages per annum, depending on the available budget. Digitalisation test: Tenderers who would like to submit a tender must attach examples of digitalised pictures of the quality level covered by the tender. The example pictures are to be digitalised from test microfilms that the National Library of Norway will send out upon request. Requests for test microfilms are to be sent to the contracting authority's contact person. The test digital pictures must reproduce the pictures of the newspaper pages on the microfilm in the best possible way. They must be correctly exposed, and have a good level of focus in order to ensure readability. A random selection of the digitalised pictures will be assessed for quality, and the assessment will be part of the tender evaluation. The digitalised newspaper pages must also be OCR processed. The newspapers include both latin and gothic fonts, they can also Be a combination of these in the same newspaper, depending on when the newspaper was published. Most of the newspapers were published in Norwegian, in one of the 2 official languages, 'New Norwegian' or 'Bokmål'. The oldest newspapers from before 1907 will most likely be in Danish. The OCR processing must have a level of word accuracy of at least 90 % for latin fonts and at least 80 % for gothic fonts. Metadata: There must be relevant metatdata from the OCR processing. As a minimum, these must contain information on: — the OCR technology, — the degree of correctness for each letter, each word and each side, — coordinator for the position of each word on a page. A random selection of digitalised newspaper pages will be assessed for quality, and the assessment of the quality of the OCR processing will be part of the tender evaluation.