How does Elements check for duplicate items in AI assisted data curation?
Elements' AI assisted data entry feature helps users to create structured metadata records quickly from free text input. To mitigate the risk of users creating duplicate items, Elements uses a number of strategies to look for matching items within Elements and matching records in external data sources. Found matches are offered to the user for claiming. Different data categories use different matching strategies, as appropriate.
N.B. This matching process is used from Elements 7.3 upwards. In Elements 7.2, AI assisted data entry uses a simpler matching process (see below).
Publications
Check whether the input text matches any input text previously used by any user to create or claim a publication using AI Assist.
Use publications identifiers found in the input text to search for matching publication records within Elements and in external data sources. The following identifier types are used for this:
DOI
Dimensions publication ID
arXiv ID
PMCID
Scopus document ID (a.k.a. EID)
Using the extracted metadata, search for a matching publication within Elements using the standard algorithm for matching publication records.
Grants
Check whether the input text matches any input text previously used by any user to create or claim a grant using AI Assist.
If the Dimensions Grants data source is enabled and a funder name and grant reference number are extracted from the input text, search for a matching grant within Dimensions.
Using the extracted metadata, search for a matching grant within Elements using the standard algorithm for matching grant records.
Professional activities
Check whether the input text matches any input text previously used by the current user to create a professional activity using AI Assist.
Teaching activities
Check whether the input text matches any input text previously used by the current user to create a teaching activity using AI Assist.
Elements 7.2 matching process
The initial release of AI assisted data entry in Elements 7.2 uses a simplified matching process.
For all supported data categories, Elements will check whether the input text matches any input text previously used by the user to create or claim an item (of the same category) using AI assisted data entry.
Additionally for publications, if the input text you contains a DOI, Elements will use it to check for matching publications within Elements and in CrossRef.
