Help

discussion

Preparing Planet Hunters for Kepler's Extended Mission

Started by Tom128
Default_user
about 7 years ago

I thought this may be a good time to begin a dialog concerning improvements here on Talk that may better position Planet Hunters as a "go to" data mining experience for visitors over the next four years. It is time for members to share their thoughts on system upgrades, lessons learned, from using the PH interface and proposals for tools that could enhance light curve analysis and classification. One thought is that Kepler undetected and obvious transit drops are becoming less common. The need for advanced tools to look deeper into noisy or quiet light curves is required. With that said, many veteran members have been using such tools to analyze light curves, but more user friendly programs are needed here at PH for members that do not have the technical background. Members should be able to use tools here at Planet Hunters and not have to move off site, learn code or other system protocol.

As mentioned by members in the past, Kepler identification numbers should be automatically incorporated into the data set. Perhaps APH and SPH numbers are now over come by events and time. I invite our members to post their thoughts and encourage healthy dialog on how to make the experience here at Planet Hunters the best possible.

Default_user
about 7 years ago

More and better tools. Good idea, but it seems anathema to the PlanetHunters man v. machine premise that the human eye can find patterns that computer algorithms might overlook.

Default_user
about 7 years ago

Hmm, well whichever tools are used the most would have to be incorporated and explained to the casual user. Preferably with video tutorials. Or at least guided tutorials and walkthroughs to get users familiar with the tools. Quick lightcurve manipulation would be the ultimate. The entire light curve should be stitched together, with options to detrend, and phase the curve (in more than 1 time dimension). Each star should have quick access to either starview, or something that is not outside of planethunters. VSX is also a common reference for stars ... the option would have to be incorporate all known information about a star from all these different sources, or provide direct links to pages.

A section that lists all stars flagged a certain way (flares, variable, kepler favorite, EB, glitch, etc...). The information is already in a database somewhere, because there is a keyword section on the homepage. Forum threads do a good job of listing stars, but a list with a link that takes a user right to the tools that they want to use on that star would improve both efficiency. I bet it would be very educational as far as users familiarizing themselves with various lightcurves.

The section labeled "boards", that is only visible when my mouse is over it in the upper right corner .. should be the focus of the forum page. I think just a shift in layout would do wonders. Add more forum sections and refocus a users initial attention from most recent comments to a section more favorable to the science with some stickied threads (the featured section doesn't quite do this). And finally, the "trending" section of the forum should use something that has most popular discussion in the last day, week, month, and year. Currently, I think the programming just tallies the number of comments in a thread, and the winner is the highest count. Hence, all threads are 4 months old.

Default_user
about 7 years ago

Two really crucial improvements that essentially require a redesign and revamp of the database but really should be done in order to standardize WHAT something is and WHEN it happens, i.e. names and times

1) Drop the APH/SPH ids and go to Kepler IDs.

2) Use the standard Kepler BJD as found in the MAST FITS files.

The worldwide scientific community understands KICs and KIDs. PH ids are internal only to PlanetHunters. Right now when we come across a mention of a KIC, say in a journal or publication, it is impossible to look up its PH reference. When a discovery is made on an PH light curve, it requires one more step to associate this with a KID. Once upon a time, back in Q1, there may have been some rationale for why PH ids were kept separate. But that time is long past.

A lot of trouble and programming errors can be avoided by just standardizing on BJDs. BJDs are unambiguous, whereas PH times are dependent on the first acquired quarter in the database. Not only that, within each subquarter, timings start from 0 again, adding yet more confusion.

Default_user
about 7 years ago

I'd like to see some of these changes. But I'm wary of seeing big bundles incorporated into the Classify side. It still feels a bit awkward to manipulate on every system I've used it, and I have to wonder if it won't start breaking some browsers with more layers.

For cursory checks I do find having to chase the star to Talk-->Examine Star-->then to MAST extraneous and time consuming. I don't see how the integrity of the vetting statistics would be at risk by switching to Kepler IDs or having them shown in tandem at the 'Would you like to discuss this star' juncture; the APH number is not displayed until after the user has classified and determined transit presence...

No BJD timing is cumbersome because in some cases a feature is noticed afterwards outside Classify, and one has to chase down the PH date back to quarter chunk ID and monthly scale, so as to avoid confusing others. This is an issue for discussions as well when it comes to LC's with multiple spots of interest in multiple quarters.

Every quarter chunk needs to automatically associate and display all links with previous or subsequent quarters and attached reference discussions--failing that at least a highlighted icon or caption to indicate there are attached discussions open in other quarters. That way we can get fresh ideas and perspectives on old discussions and forgotten KIDs. For a while I think tagging SPH/APH at the bottom of posts on particularly interesting stars worked around this but there are getting to be too many chunks for this to be realistic much longer.

I would really like the ability to switch b/t corrected and sap_raw flux, but that's possibly a lot of extra data to host and serve up...

Last on my wish list are some tweaks that would allow users to identify specific portions of an LC, crop/zoom, mark etc and share them on Talk, even if it was limited to small thumbs to minimize hosting. Also what if photobucket goes bankrupt tomorrow this place will have broken links everywhere...!?

Ggccg mentioned man v machine and personally I think the pipeline may have us very nearly dead to rights at this point, so anything to make it a little easier and efficient to classify and poke around the light curves would be advantageous. There have got to be more subtle transits with TTV that were missed or tossed, so the timing scheme issue is almost more important than having the KID # right at hand, in my opinion.

Default_user
about 7 years ago

Thanks for the great idea but one thing I'd like to mention is we don't have a lot o

kianjin:

Two really crucial improvements that essentially require a redesign and revamp of the database but really should be done in order to standardize WHAT something is and WHEN it happens, i.e. names and times

1) Drop the APH/SPH ids and go to Kepler IDs.

2) Use the standard Kepler BJD as found in the MAST FITS files.

The worldwide scientific community understands KICs and KIDs. PH ids are internal only to PlanetHunters. Right now when we come across a mention of a KIC, say in a journal or publication, it is impossible to look up its PH reference. When a discovery is made on an PH light curve, it requires one more step to associate this with a KID. Once upon a time, back in Q1, there may have been some rationale for why PH ids were kept separate. But that time is long past.

A lot of trouble and programming errors can be avoided by just standardizing on BJDs. BJDs are unambiguous, whereas PH times are dependent on the first acquired quarter in the database. Not only that, within each subquarter, timings start from 0 again, adding yet more confusion.

Unfortunately it would take a lot of developer time to redo from SPH/APH to kepler id and going to BJD from what we have now (you would be redoing the database structure) all of that and that's not something we have the funding for or the zooniverse has time for (they're building a lot of cool projects that need their attention just like we got all of the developer attention when we launched) - so the APH numbers will be staying, but maybe we can make it easier to find a kepler id. I will ask the development team if there is something we can do about that.

~Meg

Default_user
about 7 years ago

Hi all,

thanks for the ideas. I'm happy to hear all the enthusiasm for November and improving Planet Hunters. I know I am too. Please do share your ideas, but I just want to explain what will actually be realistic for us to do. We're working on ways of speeding up classifications and getting through the data faster on our end and uploaded faster, but the main classification interface will not change. We don't have significant funding at the moment to redesign things in the way we did at launch. I believe there is a planned upgrade to Talk to improve it based on the experience on PH Talk and from the other projects that use Talk. I've already talked to the Zooniverse team about easier ways of linking quarters together and make it easier to transition to the source pages and uploading images is one the zooniverse team's todo list for an improved version of Talk. Switching between raw and reduced flux would take doubling the time to upload data so we probably won't do that, but we are trying to improve uploads and get them faster.

Do keep the ideas coming, if there is a time we do decide to make a version 2 of the classify interace, we will definitely take the ideas posted here under consideration. I just wanted to give some perspective of what the team can do at this time, but if there is something we have time for I will bring them to the development team and see if we can make them happen.

In terms of future tools, keep the ideas coming, but we may not be able to accomodate them all. If there are online tools you use like if having a link to the NASA exoplanet archive on the source pages that we could definitely do.

That being said, the Zooniverse is developing new tools for all the projects to further help promote and encourage science investigations from our volunteers. One new tool being built by the Zooniverse team is Letters which is still in beta and I've been slowly asking for people to contribute to. It's designed to be a supplement for Talk, where you can take the research you're doing or the guide you've started to assemble and have a place to write a long summary and description of it that you can't do with the Talk side discussions or message boards. If you have guides or a tool you developed or want to share your investigation (like the CVs, etc) please do write a letter (it's still in Beta right now, but i've already gotten great feedback on the PH Letters already there from other members of the Zooniverse team), I encourage you to think about writing a Letter. I think Letters wil be a great feature that will add to the Planet Hunters community experience.

Cheers,

~Meg

Default_user
about 7 years ago

mschwamb:

...but maybe we can make it easier to find a kepler id. I will ask the development team if there is something we can do about that.

Thanks for the info, Meg. With respect to KIDs, if it's more of a quick fix option I'd suggest simply placing another link to 'VIEW ON KEPLER ARCHIVE' right beside the header APH # when that is displayed at finale of the Classify process, to save folks from having to return for it via 'Examine Star' from Talk.

I've already talked to the Zooniverse team about easier ways of linking quarters together and make it easier to transition to the source pages and uploading images is one the zooniverse team's todo list for an improved version of Talk.

I'm glad they are working on this!

Switching between raw and reduced flux would take doubling the time to upload data so we probably won't do that, but we are trying to improve uploads and get them faster.

That's kind of what I expected--not a huge issue as I periodically want to go look at sap_raw, but not constantly, so that's not worth all the extra uploading.

As far as the BJD--would it be possible instead to add a pop up style reference/conversion window, accessible from Talk and Examine star, that would show the BJD ranges correlating to each given Quarter chunk we currently have uploaded to PH? Obviously we can still figure this out in a minute or two but it would be nice to have some sort of shortcut!

Default_user
about 7 years ago

A promising, complementary visual tool for looking for changes in light curves is the AKO program that TroyW is working on. Perhaps PH could collaborate with Troy. There had been some discussion in the past about sonification of light curves by Zookeeper. Is that idea still in the works?

Default_user
about 7 years ago

Tom128:

There had been some discussion in the past about sonification of light curves by Zookeeper. Is that idea still in the works?

It's still something we would like to do in the future, but not something we can do right now.

~Meg

Please Log In to make comments.