Quality metadata are a vital resource for cultural heritage, as they improve digitisation, research and access to information. Developments in the A.I. field accelerated data gathering, the collected data are beneficial to the enrichment of information by creating accurate labels and minimise biases.
The proper collection and availability of metadata concern all fields; no matter your profession, you are bound to have an archiving system. In cultural heritage sectors, metadata are particularly pertinent as they maintain extensive archives and have recently increased the digitisation of collections. In the long term, A.I. can improve the access and accumulation of information, but to do so needs to learn from the presently available data. These data, however, are scarce and not in standardised formats.
In cultural heritage, some A.I. tools are used to go through images of digital objects and paintings, extract information and create accurate captions. In comparison to identifying objects sourced from photographs (for example: recognising the difference between a horse, donkey and zebra), cultural A.I. does not have enough data to learn. Several initiatives and labs around the globe work to mitigate this issue, two of the main segments are standards for institutions and crowdsourcing data.
Personal capacity: when you view online content of cultural heritage, you can share your story, enhance the label or offer missing information. A great tool for A.I. developers and the public is Wikimedia commons that hosts a vast collection of digital paintings, objects and heritage where users can edit captions, add information or translate in other languages.
Institutional capacity: using metadata standards for archiving and creating descriptions such as the Europeana Data Model (EDM) can increase both A.I. learning abilities and help the users/audience search for information. EDM separates a cultural heritage object from its digital representation to appropriately associate metadata.
Quality assurance: developing quality indicators can help assess the data attributed to a digital object, such as mandatory fields and appropriate tags( e.g. language English=en). Hence, measuring their quality and identifying errors.
Metadata can also help minimise bias; over time, our understanding of concepts and terms shift with societal evolution, antiquated or sometimes inappropriate terminology can be found in museum and archival holdings that reflect the views of their time. Metadata can help identify and minimise biases, facilitate diversity and help reinterpret heritage.
Creating quality metadata in all fields will holistically give access to knowledge and decisively combat misinformation. Cultural heritage benefits considerably, as they feature lengthy contextual information that weaves societal concepts. Unfortunately, A.I. is not magic like us it needs materials to study before it can perform well at its job. We have to help A.I., to help us.
I want to learn more:
METS– (Metadata Encoding and Transmission Standard)
Our other A.I. articles:
PS: “just shower thoughts” a. Sci-fi should draw emphasis on how A.I. collected metadata to draw conclusions and act-e.g. destroy all humans. b. museums in the Matrix universe must be extraordinary in label making.