We also collected some of the most frequently asked questions here: How can we get users involved in the knowledge graph creation? There are low-threshold systems such as simple content editors or card sorting tools that serve as entry points to knowledge graphs. These can be included as elements of the knowledge graph lifecycle and enterprise architecture to make sure the roll out of knowledge graphs doesn’t have any delays or difficulties. (See The Knowledge Graph Cookbook page 219) What are the low-cost approaches that one can use to demonstrate the power of KGs for organisations that are unfamiliar with them? Our advice is to start small and grow based on concrete use cases and examples to show the value the knowledge graph can bring to your organization early. You don't need to start with the perfect data model (ontology) for all your data or a complete taxonomy or controlled vocabulary. (See The Knowledge Graph Cookbook page 139). Are there publicly-viewable good examples of knowledge graphs in use that we can show others as examples? Typically, users benefit from applications that use a knowledge graph through custom interfaces, but they do not see or touch the KG directly. One of those applications is the GBA Thesaurus. Their datasets are coded with thesaurus terms, while the thesaurus is linked to INSPIRE terminology at the same time. Another interesting example is the Australian Health Thesaurus (AHT), which serves as the backbone of Healthdirect, Australia's largest citizen health portal. It uses a knowledge graph to curate and organize hundreds of different data sources. In addition, there are several reusable knowledge graphs which are publicly available, such as Wikidata, KBpedia and DBpedia, or upper ontologies like the Basic Formal Ontology (BFO) or Schema.org. These can be extremely useful in accelerating the process of developing an enterprise knowledge graph. (See The Knowledge Graph Cookbook page 107) What are some of the direct examples of Return on Investment (ROI) that can be used to market the value of Knowledge Graphs to companies that are not familiar with the concept? As with most quality-oriented initiatives, success is difficult to measure. Similar to cooking, tastes are different and influenced by cultural conditions. One of the main criteria could be to reduce the time needed for experts/knowledge workers to obtain information. Since one of the main application scenarios for implementing knowledge graphs is improved search and retrieval, one way to calculate the ROI of such an initiative could be to calculate the reduction in time spent searching for information and people. (See The Knowledge Graph Cookbook page 87). Manually establishing ontologies is a high barrier for companies to start using knowledge graphs. I think we really need to generate ontologies automatically with the help of machine learning. Do you agree? No. Considering that even big data technology giants have to employ thousands of people who have to manually classify their content, one can easily deduce how impossible it will be—at least in the near future—to rely on any AI without the human-in-the-loop (HITL). (See The Knowledge Graph Cookbook page 206). |