Machine Learning Apps for Non-programmers
Machine learning is demonstrating increasing usefulness as a tool in varied library applications. However, because machine learning is still an emerging technology in libraries, related commercial apps are of limited availability. This means that the benefits of using this technology are often limited to librarians with coding skill or with IT departments that are not already fully consumed with other priority tasks. This presentation demonstrates how open systems and services, including the Jupyter® notebook tool used within the Kaggle data-science environment, can help overcome this time- and skill-based barrier to entry into machine learning by non-programmer librarians. Specifically, an example is given of using open source topic modeling modules to analyze and visualize publication data, and inform collection development decisions. Other similar applications of general interest to libraries are discussed, along with practical issues of implementation.