Identifying Lichens on iNaturalist

Identifying Lichens on iNaturalist

  1. Do not identify without the required identification characteristics.
    Ask the user for additional missing characters before making an ID.
    Identify only to the rank warranted by the features available.

  2. Submit a comment without an ID if you are uncertain of an identification.
    Explain what features are precluding the ID.

  3. Justify Disagreements.
    Provide reasons if you disagree with an existing ID or suggestion. This allows others to asses the validity of the reasons. Without reasons it is impossible to evaluate disagreements.

  4. Also justify an Agreement.
    Leave a comment on why you are agreeing.

< added: >

  1. !! Never use data for research that you have not checked the IDs for yourself, or had a specialist check for you.
  2. !! If you are specialist, annotate your IDs liberally - they will be useful for training future lichenologists.
  3. !! When vetting data for research, be sure to annotate issues on the source data: do not just discard or disregard data when you are suspicious as to its quality.

< and: >

  1. Look out or unusual or rare species or complexes on submitted observations.
    If useful, suggest to the observer to duplicate (or triplicate) the observation to ID those species.

< P.S. >

  • !! Learn how to use the Identification Curation tool on iNaturalist. Use it to manage your workflow.
    Two-minute tutorial here: learn it:: https://vimeo.com/246153496

from:
R. Troy McMullin and Jessica L. Allen (2022) An assessment of data accuracy and best practice recommendations for observations of lichens and other taxonomically difficult taxa on iNaturalist. Botany 100: 491–497 https://dx.doi.org/10.1139/cjb-2021-0160

由使用者 tonyrebelo tonyrebelo2023年09月26日 11:57 所貼文

評論

Limit reliance on machine generated identifications.

The iNaturalist AI CM (artificial intelligence, computer model) uses identifications for training. If specialists have not checked and fixed the identifications on iNaturalist, then the model will deliver inaccurate identifications (just so you know whom to really blame if the IDs suggested by iNaturalist are wrong!).
Note that the model is only trained on taxa with over 100 Research Grade observations - so it does not know about the rarer species and will identify them as the most similar species it has been trained with.

Beware of the AI CM suggesting too fine identifications. Always check and justify your identifications.

發佈由 tonyrebelo 5 個月 前

新增評論

登入註冊 添加評論