As noted in the ASReview documentation:

"Title, Abstract: Each record (i.e., entry in the dataset) should hold metadata on a paper. Mandatory metadata are only title or abstract. If both title and abstract are available, the text is combined and used for training the model. If the column title is empty, the software will search for the next column primary_title and the same holds for abstract and abstract_note."

This means the software requires at least one field—either the title or the abstract—to contain text. Both fields are combined into a single text input for the model. If both fields are empty, the record is skipped as there is no text for the model to process.

Regarding your question about whether the title is specifically mandatory, it likely isn’t required in isolation, but rather some text input is necessary. It’s possible that implementing the requirement to allow "at least one field to be filled" (title or abstract) was simpler with this structure. However, I can’t recall any specific reason why the title itself must always contain text. (ping: @J535D165 )

To confirm this, could you try creating a small example dataset where the title column is entirely empty but the abstract column contains text? This would help demonstrate whether the system can proceed without issues as long as the abstract is filled, and to improve the error message.