Digital Libraries II: National and International Digital Library Initiatives; Institutional Repositories, ROAR, OpenDOAR and Sherpa Romeo; Content Management Systems; AI, Expert Systems, Robotics, SMAC and Cloud Computing; and Ontology, the Semantic Web, Linked Data, Big Data, Data Mining and Data Harvesting

Clifford Lynch (2003) defined a university-based institutional repository as a set of services that a university offers to the members of its community for the management and dissemination of digital materials created by the institution and its members. Repositories collect, preserve and give open access to theses, articles (as preprints or accepted manuscripts, the "green" route to open access), reports, data, learning materials and archives. They are needed to increase the visibility and impact of research, to preserve the institution's intellectual output, to comply with funder and institutional open access mandates, and to support research assessment. Types are institutional repositories; subject or disciplinary repositories, such as arXiv (1991), PubMed Central (2000), RePEc and E-LIS for library and information science; data repositories such as Zenodo (CERN) and Dryad; and national repositories such as Shodhganga. Tools include DSpace (MIT and Hewlett-Packard, 2002), EPrints (University of Southampton, 2000), Fedora, Greenstone and Invenio. In India, ePrints@IISc was among the first institutional repositories, and many IITs, IISERs, universities and CSIR laboratories now run repositories, most on DSpace or EPrints.

A content management system (CMS) lets people create, edit, organise and publish digital content, typically a website, without writing code. Its architecture separates a content repository (a database of content items and metadata), a content management application (authoring, editing, workflow and approval, version control, user roles and permissions) and a content delivery application (templates and themes that present the content); a "headless" CMS delivers content through APIs to any front end. Data integration connects the CMS with other systems: catalogue search boxes and new-arrival feeds from the ILS, repository content through APIs or OAI-PMH, discovery-service widgets, RSS and social media, and single sign-on. Widely used open source CMS software includes WordPress (2003), Drupal (2001) and Joomla (2005); libraries also use hosted platforms for subject guides. Selection weighs functionality, ease of use for non-technical staff, accessibility, security and update record, extensibility through modules and plugins, community and vendor support, hosting and cost. Implementation covers content audit and migration, design, training and launch; evaluation uses web analytics, usability testing and user feedback.

Artificial intelligence (AI) is the capacity of machines to perform tasks that require intelligence, such as understanding language, recognising patterns, learning and reasoning. An expert system captures the knowledge of human experts in a narrow field: it has a knowledge base of facts and rules, an inference engine that reasons with them, a user interface and an explanation facility; early examples were DENDRAL and MYCIN, and library applications have included cataloguing and reference advisers. Current AI in libraries includes chatbots for routine queries, recommendation systems, automatic subject indexing and metadata generation, OCR and handwriting recognition, speech-to-text for accessibility, and generative AI built on large language models, which librarians must also teach users to evaluate, since such tools can fabricate references, reproduce bias and raise copyright and integrity questions. Robotics in libraries covers automated storage and retrieval systems that fetch items from high-density stores, automated materials handling that sorts returned books, shelf-scanning inventory robots, and telepresence or guide robots. SMAC is the convergence of social media, mobile technology, analytics and cloud computing, each reinforcing the others: services reach users on phones and social platforms, generate data that analytics turns into insight, and run on cloud infrastructure.

In computing, an ontology is, in Tom Gruber's phrase (1993), "an explicit specification of a conceptualization": a formal model of the classes in a domain, their properties and relations, the individuals that belong to them, and rules (axioms) that allow reasoning. RDF (Resource Description Framework, a W3C Recommendation from 1999, revised as RDF 1.1 in 2014) expresses data as triples of subject, predicate and object, each named by a URI, which together form a graph. RDFS (RDF Schema) adds a basic vocabulary for describing classes and properties, such as rdfs:Class, rdfs:subClassOf, rdfs:domain and rdfs:range; OWL (the Web Ontology Language, 2004; OWL 2, 2009) adds richer logic; SPARQL (2008) queries RDF. Protégé, the free, open source ontology editor from Stanford University's biomedical informatics research centre, is the tool the syllabus prints as "Potege". The semantic web, set out by Tim Berners-Lee, James Hendler and Ora Lassila in Scientific American (2001), is a web of data that machines can interpret, built in layers from URIs and Unicode through XML, RDF, RDFS and OWL to logic, proof and trust.

• Linked data: Berners-Lee's four principles (2006) are to use URIs as names for things; use HTTP URIs so that people can look them up; provide useful information in standard formats (RDF, SPARQL) when a URI is looked up; and include links to other URIs so more things can be discovered. His five-star scheme (2010) grades open data from one star (available on the web under an open licence) to five (linked to other data). Library examples are the Library of Congress's id.loc.gov (LCSH and other vocabularies as linked data), VIAF (the Virtual International Authority File, hosted by OCLC), Wikidata and BIBFRAME.
• Big data: data sets too large or fast for ordinary tools, described by Doug Laney's three Vs (2001), volume, velocity and variety, often extended with veracity and value; libraries meet it in usage logs, research data and digitised text corpora.
• Data mining: discovering patterns in large data sets as part of knowledge discovery in databases (selection, pre-processing, transformation, mining, interpretation), using classification, clustering, association rules ("users who borrowed this also borrowed that") and text mining.
• Data harvesting: automated collection of data or metadata from many sources, such as OAI-PMH harvesting by aggregators (CORE, BASE, NDLI), web crawling for web archives, and API-based collection; it must respect licences, privacy and robots rules.