search

ropensci / ozunconf18

repository for the rOpenSci ozunconference 2018
31 stars 7 forks source link

Untweetable species Twitterbot thing 🐜 🐌 #13

Open hollylkirk opened 5 years ago

hollylkirk commented 5 years ago

Some of my colleagues just published a paper showing that 57% of Australian threatened species have been mentioned fewer than 20 times on Twitter. We have named the species that have never been tweeted about the #untweetables. :frog: :bug: :snake: :beetle:

I propose a project to do the following:

NB. I have 0 experience with doing Twitter stuff in R, but I am very enthusiastic! :smiley_cat:

adam-gruer commented 5 years ago

Noice

Maybe the tweetbot could also tweet pics and info about the #untweetables. Give them a boost. the earlier dataset could filter out tweetbot generated tweets.

hollylkirk commented 5 years ago

Yes! I love this! Although, online pics are not available/hard to find for many of the #untweetables... But I guess we could pre-generate that content in advance?

adam-gruer commented 5 years ago

I don't know if its appropriate, I certainly don't know how doable it is but it could cause buzz interest- raise awareness in a geeky fun way. There is a twitter user who posts "photos" of his 6 year son's drawings as if these creations existed in the real world - he uses some machine learning to generate the pictures. Maybe a machine learning, natural language processing algorithm to generate fantastical images of the untweetables based on textual descriptions of them or just their latin names? https://www.boredpanda.com/kid-drawings-things-i-have-drawn-dom/?utm_source=google&utm_medium=organic&utm_campaign=organic

hollylkirk commented 5 years ago

Wow! That would be so much fun!! Some of these species have really strange common names too (e.g. antbed parrot moth and robust white-eye) which maybe we could use to generate the image, and then compare it to the real thing...?

peggynewman commented 5 years ago

Yep @adam-gruer, this exists, check it out: Microsoft researchers build a bot that draws what you tell it to ! Also @hollylkirk, you can use ALA via ALA4R to find:

adam-gruer commented 5 years ago

un@ables? I'll see myself out

hollylkirk commented 5 years ago

These are all excellent ideas folks! @peggynewman I didn't know about ALA4R, that's cool!

Let's see if it gets picked and we can discuss what to prioritise. I will bring along a spreadsheet with the original data for reference on the day.

RPanczak commented 5 years ago

Perhaps we could also generate a hex sticker for each animal! ;-)

hollylkirk commented 5 years ago

Oh my! 😍

RPanczak commented 5 years ago

@hollylkirk authors of the study that you pointed to mention

496 species from multiple taxa

Do you think there is a chance to get a list of them?

We have 5.6 milion of geolocated 🌐 tweets for Oz 🇦🇺 available (ie since mid May this year) - I could have a quick look 👓 if anything got mentions there? Not surprisingly & replicating Table 1 - what was not tweeted then is not mentioned now.

Btw, could #Rstats ninjas help to advise how to efficiently search for occurrence of string in a column of dataframe?

hollylkirk commented 5 years ago

Wow!

@RPanczak I do have access to the full list of threatened species used in the original paper, but I think it would be great to narrow in on the subset that have never been mentioned, or only been mentioned <5 times.

I will attach the appropriate species list to my next comment... 🔢

peggynewman commented 5 years ago

Potentially useful active twitter accounts for testing: @TSR_Hub (Threatened Species Recovery Hub) and @TSCommissioner (Threatened Species Commissioner)

smwindecker commented 5 years ago

cool @hollylkirk! I wonder if instead of a static untweetables list, the list could update as tweets change their rankings?

hollylkirk commented 5 years ago

Follow developments on the Untweetables issue over at https://github.com/ropenscilabs/UntweetdAus All work by @peggynewman @sadian @dataroni8