The business
NBD Biblion is a non-profit company founded in 1970. They aim to provide support to libraries in their core business: central buying, binding, cataloguing, and getting books ready for the public library. NBD Biblion has developed a system that allows us to select, purchase, bind and distribute the collection while performance continuously gets better, smarter and faster. NBD Biblion cares for, and takes care of libraries.
Their challenge
Libraries need to build up and maintain a collection that matches the needs of the visitors and the institutional mission. Among thousands of titles per year, they have to select the ones that will be a valuable addition to the collection. After building up a collection it remains a challenge to respond well to the changing needs of their visitors in order to make as many people enthusiastic about books and reading as possible. Therefore, NBD Biblion needs to match the expectations of library visitors with a relevant collection in libraries. Libraries are unique and serve a unique audience. As a result, both an eye for individual libraries as efficiently serving libraries as collective are important to NBD Biblion.
Our solution
Together with NBD Biblion a plan is made to optimise their ordering advise for libraries. This helps to advise libraries on book titles that have just appeared based on their unique audience and expectations about the collection. While developing the ordering advise engine we worked closely together with experts in order to understand the current process. Together with experts we tuned the ML model to understand the context based on the right assumptions.
Our solution consists of multiple steps:
- First a score is produced for each individual book based on the quality and expected popularity.
- Those books are then matched to unique libraries and their corresponding profiles. The engines solve the puzzle to create a unique order advise for each organisation.
- A feedback loop ensures a constant improvement of the ordering advices.
In this way we were able to develop an automated yet personal ordering advice engine that matches the visitor’s behaviour, the collection of a library and the institutional mission in order to provide better service to libraries.
Looking forward
NBD Biblion its increased knowledge of library needs will also enable them to optimise internal purchasing and production planning processes.
Technology
To realise a scalable and future-proof solution the engines are built in Python and are containerised deployed using Docker, the data pipelines are set up using Apache Airflow, all running on AWS.