I know there’s been lots of discussion about ‘junk’ observations, and I don’t intend to re-hash any of that stuff about whether they should be deleted etc.
But on a purely practical level, how do we handle them?
I’m talking about ones that have multiple species in a single photo, such as a photo of a field of flowers, and the user only ever submitted 3 observations five years ago. There’s no possibility of getting more info or better quality pictures, and there’s no way to ID it.
I’m NOT suggesting that it gets deleted, but is there anything I should be putting in the dqa to help filter it out of searches for everyone?
Obviously, I can mark it as reviewed (which still isn’t ideal because I often go back through ‘reviewed’ observations, to annotate etc.), but it’s frustrating that dozens of people have probably already spent time looking at it, leaving helpful comments for the observer, and marking it as reviewed, and many more will.
In cytology there is wonderful concept. It’s called skip cell. It’s cell which is simply too mauled to draw conclusion, in such case cytologist moving to another cell which may be suitable for diagnosis.
In freshwater planarians we getting lot of skip cells. Planarians which are dragged out of water are sometimes just blob. Sometimes I slap at least order over observation in question when I’m sure it is a planarian. Sometimes it’s so bad that item going directly into reviewed.
Items with multiple species are instant reviewed checkbox. Sadly, flow is so big that I really skipping these.
No, the DQA shouldn’t be used simply to make the observation casual grade if it doesn’t otherwise meet the requirements of the DQA.
You can add a broad ID, you can add a specific ID of one subject in the observation if you can, if an observation has reached a CID for something somewhat specific (or is so low quality that only a broad ID is possible), then “good as can be” can be ticked.
For shots of meadows/landscapes without obvious focus I just focus on one plant that I recognize, hoping that this will eventually get it out of need-ID pile if the photo is not to blurry overall. I question the value of all the observation with just a single leaf in them, where we can be even less sure about confident ID, let alone if the overall plant was cultivated or if the location of the photo coresponds to the location of the plant.
“I believe this observation cannot be identified more finely than the current ID. I’m marking “Based on the evidence, can the Community Taxon still be confirmed or improved?” as “No, it’s as good as it can be.” When the community ID is below family level, that makes it “research grade.” This means that other identifiers don’t need to again review the observation. **Please do so** when you’re pretty sure the community ID cannot be further improved!”
Is this possible even if the observation can’t be identified beyond kingdom: plants? Because I am not sure the yes/no is available until you reach a certain level.
I like the idea of leaving a comment, as that can encourage other people to give it a generic ID and mark it as “as good as it can be”.
@asteroidowl From what I’ve figured out I think the yes/no availability is based on number of IDs, rather than whether it’s got beyond Kingdom. I think if it’s ID at Kingdom level by 2 or more, it can be marked as “as good as it can be”, and it goes to ‘Casual’.
Ah. That makes sense. I ID’d a landscape observation as plants, and someone else tried to see if the observer would respond (which I doubt…) and I couldn’t just mark the community taxon thing so no one else would waste time on it.
I’m not trying to use the DQA to ‘fudge’ a result - just wanting to find out what in the DQA it does meet the requirements of. Sounds like the ‘as good as it can be’ works.
I like that. I do take into account when the user was last active and if there are previous unanswered queries to determine if its worth leaving a comment.
I have often wished for two different types of reviewed, one for “I never wish to see this observation again” and one for “I want to/may want to revisit this later”
I have been thinking the same thing, literally yesterday in fact.
Simply ignoring “problem” observations isn’t a good option for me because I’m curating observations in our project in a systematic fashion. I use “reviewed” to indicate observations that are acceptable/usable by our project. In other words, I use “reviewed” as a flag to indicate that an observation is “OK”. I have to review 10’s of thousands of observations every year, so reviewed works well for this purpose, as it can be applied/removed easily from the thumbnail view, and observations can be filtered based on it. Unfortunately, in this work flow “unreviewed” can mean several different things:
haven’t looked at the observation yet
I’m still thinking about the ID
it’s a “bad” observation that I don’t want to work with at all
Having the bad observations included in my unreviewed cue means I have to look at them again each time I go back through my cue to see if there’s any “clean-up” to be done. It would be much better if the bad observations could simply “disappear” from my view forever. I can’t remove the observations from the project, because it’s an old project, and iNat puts them back into the project overnight when observations get re-indexed (or whatever it is that happens overnight).
I’ve considered adding an observation field to the bad observations, but as far as I can tell, one cannot filter (via the URL) on observations “without” a particular field value. I’ve considered creating a project for what I regard as “bad” observations, and excluding the project from my view, but I’m not sure people would appreciate having their observations added to a “garbage observation” project. Of course, I could give the project a non-pejorative name like “extra special observations” …
So yeah, it would be great to have a separate “ignore” flag that works like the “reviewed” flag.
Wouldn’t that be a misuse of the project feature? I don’t know about anyone else, but when I’m looking at an observation, it bugs me to have to scroll through a long list of projects to see if any observation fields have been added. Especially when many of the projects seem to have very similar mandates, and some appear to be personal “vanity” projects.
It appears that on the identify page, you can filter (via the URL) for observations that do not have a particular field (you add “&without_field=[your chosen field]” to the URL).
So pick an innocuous observation field that nobody in your context uses, and add it with a similarly innocuous value to observations that you want to exclude from your view, and then filter out observations that have that field. Garbage observations effectively ignored.
Unfortunately, this won’t work in cases where the observer only allows curators to add fields to their observations.
I don’t see why a misuse. It is an iNat tool for us to use as we see fit. I collapse the list of projects on an obs if there are more than a few. But I seldom have use for obs fields. Seldom.
And if it is the only way for you to avoid seeing an unwanted obs again ?
I use observation fields sometimes, and now that I’ve discovered the universal meta-data tool (which facilitates adding annotations and fields), I’ll probably be using them more.
Note that if you use/value annotations, I’m not sure how you can avoid observations fields, since they’re the only way you can “override” an incorrectly set annotation. For example, I use the life stage annotation, and I’m now starting to use the sex annotation. When those annotations are set incorrectly, I need to use an observation field to enter the correct value (which will replace the incorrect annotation when the data is downloaded).