- These tweets were part of Documenting the Now's second Twitter chat on August 11, 2016. The tweets have been curated which means we've attempted to keep conversations and responses to questions grouped together, altering the purely chronological flow of the original stream. We also tried to include replies that did not use the #docnowcommunity hashtag. Please get in touch if you sent a tweet that you would like to have included or removed. info@docnow.io
Join us on Thursday, August 11th, 2016 at 2pm EST for a twitter chat on social media research. #docnowcommunity pic.twitter.com/vSbvKBlyr5Introductions
- Welcome to our Twitter chat about social media research. Please take a moment to introduce yourself #docnowcommunity.
- I’m Bergis Jules, one of the PI’s for the @documentnow project and I’ll be posting the questions today. #docnowcommunity
- I'm Brian Dietz, and I'm on the @documentnow advisory board #docnowcommunity
- Alexandra Dolan-Mescal here, UX designer for @documentnow #docnowcommunity
- hi #docnowcommunity, digital archivist & records manager at a public uni in Ohio. Mostly listening as I compile my #saa16 reimbursement req
- I’m Ian Milligan, historian and #webArchiving researcher. Looking forward to participating in today’s #docnowcommunity
- Hi I'm Ed Summers, i've done some analysis of Ferguson data with @SociologistRay and @BlackFeministMB & work on docnow #docnowcommunity
- Hi, #docnowcommunity. I’m Dorothea Salo, iSchool instructor, interested in digital (incl web, social media) and A/V preservation.
- Hi, I’m Shawn Walker, my research focuses on social media and political participation - especially by social movements. #docnowcommunity
- Hello all! Nicholas Proferes here, I study users' understandings and beliefs about how social media work #docnowcommunity
- Hey #docnowcommunity - I helped organize Documenting #Ferguson http://digital.wustl.edu/ferguson/ & studied what motivates people to contribute to it
- I am Desiree Jones-Smith a member of the @documentnow team. I work to manage the project and am excited be a part of #docnowcommunity.
- I’m Neil Fraistat. I’m the Director of @UMD_MITH. #docnowcommunity
Q1. What types of research do you do with social media: topics, disciplines, methods?
- Q1. What types of research do you do with social media: topics, disciplines, methods? #docnowcommunity
- @documentnow I research designing a method for building collections for stories and events esp. without domain knowledge. #docnowcommunity
- I'm interested in: characteristics of good collections, can we teach a machine to build good collections #docnowcommunity
- A1: quite a bit around collecting, analyzing, + thinking about open-source approaches. Work a lot with @ruebot on all this! #docnowcommunity
- A1. None yet, but based on a small survey we did w researchers, we anticipate future interest http://go.ncsu.edu/smalt #docnowcommunity
- @documentnow I’ve worked with @edsu, @BlackFeministMB & @SociologistRay on twitter data from Ferguson & Baltimore Uprising.#docnowcommunity
- I work with social media methods and the impact of the ephemerality of social media data on research. #docnowcommunity
- A1 Hello #docnowcommunity helping Dutch students collect and make publicly available tweets on their uprising against univ board
- @mjschaap Are you concerned about the University using this tweet collection for negative purposes? #docnowcommunity
- @fromADMwithlove no, but others might be and they are my concern, so it's more ethical #docnowcommunity
- @fromADMwithlove 2nd concern: copyright: is it lawful to harvest tweets from others and publish them on our website? #docnowcommunity
- lawful? twitter's ToS seem to say so (with attribution)...though who really knows. ethical? much harder to suss out https://twitter.com/mjschaap/status/763820213339119616 …
- A1: Started working with social media content on @OurMarathon; particularly interested in archiving / analyzing memes #docnowcommunity
- .@JimMc_Grath interesting, how do you typically identify memes? #docnowcommunity
- @edsu I'm particularly interested in image macros / transformative use of images and text circulating on social media #docnowcommunity
- @edsu but yes, I was recently having a conversation about distinctions between "art" and "memes," so it's hard to define! #docnowcommunity
- @JimMc_Grath @dchud is working on some cool features for DocNow around images. @edsu #docnowcommunity
Q2. What are some of the primary challenges you encounter in your research with social media?
- Q2. What are some of the primary challenges you encounter in your research with social media? #docnowcommunity
- @documentnow Restrictions in API (rate limits) #docnowcommunity
- @documentnow One of the greatest challenge involve the ethics and rights of using the social media archives we create. #docnowcommunity
- @documentnow 1.relevance to subject 2.balanced as much as possible (all bias represented).3.context available(conversation) #docnowcommunity
- A2: The ethical dimensions of archiving social media content; balancing close reading and distant reading #docnowcommunity
- A2: the black boxes of social media platforms, how to evaluate the completeness of what they offer is challenging #docnowcommunity
- A2: for me to ones of scoping (what to collect? what hashtags? is that enough?), as well as ethics & preservation Qs. #docnowcommunity
- @ianmilligan1 yes! knowing what to collect and how it was collected is an area that we are focused on in DocNow #docnowcommunity
- A.2 Understanding limits of Twitter’s terms of service is complicated in itself, no less how to share the data with others. #docnowcommunity
- A2:To scrape or not to scrape?sometimes getting stuff is easier by scraping-but scraping isn't stable/general/shareable #docnowcommunity
- .@acnwala that's really interesting -- what about scraping data isn't stable & shareable as compared to data from APIs? #docnowcommunity
- @edsu stability: if DOM changes, code might break if code relies on what changed (DOM usually changes slowly). #docnowcommunity
- @edsu stability: e.g if class attr changes, or there's a new one #docnowcommunity
- @edsu not-shareable:since Twitter doesn't want to be scraped (terms of serv), I am reluctant to share any code that scrapes #docnowcommunity
- @edsu So I only scrape privately #docnowcommunity
- @edsu especially when it's about collecting threaded conversations, scraping is straight forward, api - not #docnowcommunity
- .@acnwala yes, i've had the same issue myself #docnowcommunity
A2: what i call the "tweet titanic problem" - "the tweetanic problem" #docnowcommunity pic.twitter.com/rN4PEi1FDX- .@documentnow for me, a big challenge is understanding what sources users draw on to develop beliefs abt how platforms work #docnowcommunity
- @edsu issues of publicness, transparency, and concept are difficult at this point in time. #docnowcommunity
- .@walkeroh how do notions of publicness impact your work? #docnowcommunity
- @edsu I struggle with the idea that public posts are open for use. What does consenting 20k+ accounts look like? #docnowcommunity
- @edsu while there are guidelines, it’s unclear how to apply these guidelines to social media platforms. #docnowcommunity
- A2: Not having access to tools / data that social scientists have #docnowcommunity
- .@JimMc_Grath what kinds of barriers are there for getting & using those tools? #docnowcommunity
- due to server needs? costs? #docnowcommunity https://twitter.com/JimMc_Grath/status/763803907093127168 …
- @fromADMwithlove for sure! Also access to skills / resources to be able to get lots of social media data #docnowcommunity
- @edsu money, labor, time, digital space: was thinking of the access to, say, Twitter firehose vs. getting data other ways #docnowcommunity
Q3. Do you seek IRB approval (or an equivalent) as part of your research? Why or why not?
- Q3. Do you seek IRB approval (or an equivalent) as part of your research. Why or why not? #docnowcommunity
- A3. I got IRB approval for the Doc Ferguson study ( http://onlinelibrary.wiley.com/doi/10.1002/pra2.2015.1450520100106/abstract;jsessionid=42814BD61A15BC4EE7A8B301A2441AC6.f01t02 …) b/c we did an in-person usability study #docnowcommunity
- A3: UMD IRB recently told @SociologistRay @BlackFeministMB & I that consent was not required for public tweets #docnowcommunity
- @chrisfreeland we were told we weren't doing research! which was kind of a blessing & a curse :-) #docnowcommunity
- .@documentnow Most of my work involves intervention, so it's an absolute yes. #docnowcommunity
- @moduloone how do you assess the potential harms of an intervention in these spaces? #docnowcommunity
- @walkeroh Very carefully :) Really tho, its not just me assessing. I talk to others to get opinions, IRBs, colleagues, etc. #docnowcommunity
- @moduloone Great answer! These platforms present new harms, etc. So how do we expand our conceptions of harm? #docnowcommunity
- .@walkeroh We have individualistic notion of harm based on the empirical Community harm & harm to agency nd 2be considered. #docnowcommunity
- .@walkeroh @moduloone I'm interested in that question too. it seems harder than usual, but maybe because i haven't done it? #docnowcommunity
- A3: but when publishing tweets of activists it seems that care is needed like @freelon @meredithclark have in their work #docnowcommunity
- @edsu @SociologistRay @BlackFeministMB y'all have probably seen this http://bdes.datasociety.net/council-output/perspectives-on-big-data-ethics-and-society/ … about IRBs and alternatives
- A3. We haven’t, but increasingly thinking that’s going to be the way things need to move. #docnowcommunity
- @ianmilligan1 Why do you belive that you will need to consider IRB aproval? Can you expand your thinking? #docnowcommunity
- @dpjones1983 I guess I’m becoming uneasy with the (at least for us) consent issues inherent in social media research #docnowcommunity
- @dpjones1983 i.e. for us historians, we go through IRB for oral history but not public accessible tweets. #docnowcommunity
- A3: during the occupation of Amsterdam uni there were people who 'did not want to be seen', suppose/hope they didn't tweet? #docnowcommunity
- .@mjschaap did you work with social media data from the occupation? #docnowcommunity
- @edsu indeed, the students estimate we will have to collect ca 6000 tweets from the time of the 5wk occupation #docnowcommunity
Q4. What tools are you currently using as part of your social media research?
- Q4. What tools are you currently using as part of your social media research? #docnowcommunity
- @walkeroh What tools have you been using? #docnowcommunity
- @fraistat Mostly I use tools I’ve developed for Twitter data collection. I use heritrix for web archiving. #docnowcommunity
- @fraistat preservation of metadata is important to my work. Most tools truncate posts post/profile metadata. #docnowcommunity
- @walkeroh Are the tools you’ve developed shareable with others? #docnowcommunity
- @fraistat not yet, but that’s something I’m working on. I’m happy to share privately until they’re more polished. #docnowcommunity
- @walkeroh Excellent. We might follow up with you about this. #docnowcommunity
- A4. and a combination of Dataverse, Zenodo (Invenio), GitHub, & #Islandora for sharing & providing access to our datasets. #docnowcommunity
- A4: Often I have to make my own tools--TSM https://github.com/dfreelon/TSM , fb_scrape https://github.com/dfreelon/fb_scrape_public … & other custom stuff #docnowcommunity
- .@walkeroh how important is it for researchers to build tools themselves, so they understand how they work? #docnowcommunity
- @edsu it’s important to not conflate ability to build tools and understanding the epistemologies/methods under the hood. #docnowcommunity
- @edsu most tools aren’t well documented except by the code, so how do potential users evaluate — especially researchers? #docnowcommunity
- .@walkeroh yes, that's what i'm asking; if researchers reach for their own tools because they will understand their limits #docnowcommunity
- @edsu In light of that, what questions can we ask of the data while respecting its limitations? #docnowcommunity
- @edsu If tool building isn’t the focus of your research then it’s taking time away from it, right? #docnowcommunity
- .@walkeroh seems like step one is being able to talk about the limitations -- which is why I'm a big fan of your work. #docnowcommunity
- A question I've been thinking about is how we can use metadata to support ethics in opendata sets. #docnowcommunity https://twitter.com/ruebot/status/763810235563773956 …
- Like, how do we create provenance to support ethics as part of metadata standards? #docnowcommunity
- @edsu Creating more researchers focused on these limitations would be a good step. but where should the output go? #docnowcommunity
Q5. How important is it for you to publish/share your research data? If yes, how do you do it?
- Q5. How important is it for you to publish/share your research data? If yes, how do you do it? #docnowcommunity
- A5: I put the Ferguson dataset we worked with up on Internet Archive, but only as a dataset of Tweets IDd (per ToS) #docnowcommunity
- A5: please do for you're far ahead in the US, bit pioneering here so I need all the inspiration i can get from you!!! #docnowcommunity
- A5. Very important to share research data! Our team has hosted through Dataverse, i.e. http://dataverse.scholarsportal.info/dvn/dv/wahr/faces/study/StudyPage.xhtml?globalId=hdl:10864/11311 …. #docnowcommunity
- A5: very important for scholarly work for reproducibility. Sharing tweet IDs for large dataset #docnowcommunity
- .@acnwala Big issue for consent! Many users don't know abt the dozens of metadata fields that may be pulled up #docnowcommunity
- A5: I've published Twitter IDs in the past per its TOS. Ppl still ask me for the full datasets! #docnowcommunity
- A5 ^^ I guess that answers that question too. #docnowcommunity #elxn42 https://digital.library.yorku.ca/yul-642801/elxn42-crawl … #panamapapers https://digital.library.yorku.ca/yul-669273/panamapapers-crawl-may-1-7-2016 …
- .@dfreelon thanks for doing that! it has been very useful to the @documentnow project #docnowcommunity
- A5: But Twitter datasets degrade quickly. Our BLM dataset suffered a 10% attrition rate only a year later. What to do? #docnowcommunity
- .@dfreelon yes, I still get some too for our Ferguson dataset; do any of them end up hydrating the IDs? #docnowcommunity
- @dfreelon same with the Egyptian revolution: https://arxiv.org/abs/1209.3026 (@hanysalaheldeen ) #docnowcommunity
- @acnwala I know, I did some of the earliest published work on Arab Spring tweets. @hanysalaheldeen
- @dfreelon Do you think it would be useful for those folks to have a simple tool to recreate the tweets from the IDs? #docnowcommunity
- @documentnow that would be nice, but 1. still need code to handle large data volumes and 2. attrition still a big problem
- A5: Sharing data is very important. We share it within our lab. Share IDs for those outside our lab. Not ideal at all! #docnowcommunity
- .@walkeroh do you keep any record of the data's provenance when you are doing that sharing? #docnowcommunity
- A5 Challenge is finding the best way to model & share these datasets, along w/the best documentation & metadata around them #docnowcommunity
- A5 @scholarsportal gets us a hdl, but replicating the dataset w/sameAs in Zenodo gets us a DOI, which is nice for analytics #docnowcommunity
- A5. Example: http://hdl.handle.net/10864/11311 https://zenodo.org/record/55889#.V6zKjdFBuCg … More descriptive metadata in Dataverse. But DOI in Zenodo. #docnowcommunity
- .@moduloone @ruebot agreed, it seems super important #docnowcommunity reminded of @kwelle's https://www.researchgate.net/publication/303363035_A_manifesto_for_data_sharing_in_social_media_research … #docnowcommunity
- A5. All that said, another challenge is getting folks to actually cite your data when they use it. #docnowcommunity
- #docnowcommunity is now trending in USA, ranking 49
Q6. What social media platforms do you focus on for your research?
- Q6. What social media platforms do you focus on for your research? #docnowcommunity
- A6: Just Twitter, ‘tho am curious to maybe try using some Reddit data at some point .. #docnowcommunity
- A6: Mostly Twitter and FB, but always looking to expand. Have a forthcoming paper using federal comment data, fun stuff #docnowcommunity
Q7. Do you see any opportunities for improved social media collection and analysis tools?
- Q7. Do you see any opportunities for improved social media collection and analysis tools? #docnowcommunity
- A7: A code-free frontend for Twarc would be awesome for smaller projects. #docnowcommunity
- .@dfreelon still workin' on it when I can, thanks for the idea way back when https://github.com/DocNow/hydrator/blob/master/README.md … #docnowcommunity
Q8. Are you interested in using tools that can extract images and videos from tweets?
- Q8. Are you interested in using tools that can extract images and videos from tweets? #docnowcommunity
- .@walkeroh y it's difficult because as @moduloone has written about recently the APIs are not fixed http://firstmonday.org/ojs/index.php/fm/article/view/6793 … #docnowcommunity
- Q8: Yes, images and links are an integral part of many social media posts though most analysis treats them as only text #docnowcommunity
- A8: Something that pulled the most RTed/faved/commented images and videos would be very useful #docnowcommunity
- @dfreelon We have some good stuff to share with the advisory board during the upcoming meeting. #docnowcommunity
- A8: I wrote some code to do this for our BLM report but it's not well-documented... #docnowcommunity
Additional related conversation, that wasn't in response to a specific question.
- paper w/@kevindriscoll comparing Twitter Streaming API and Firehose. Also discuss black boxes http://ijoc.org/index.php/ijoc/article/view/2171 … #docnowcommunity
- Paper: A systematic analysis of twitter research by @moduloone and @michaelzimmer http://www.emeraldinsight.com/doi/abs/10.1108/AJIM-09-2013-0083?journalCode=ajim … #docnowcommunity
- @documentnow My diss on users' beliefs about info flow on Twitter. (Condensed ver. hopefully soon) http://dc.uwm.edu/etd/909/ #docnowcommunity
- .@kwelle published a nice Zotero group with lots of papers on social media research as data https://www.zotero.org/groups/social_media_as_research_data … #docnowcommunity
- @documentnow is there an existing zotero group or other place to curate papers and other resources? #docnowcommunity
- @walkeroh Not sure but there is a lot of sharing in our slack channel. About 150 members so far. #docnowcommunity
- .@acnwala Here's an awful question: If what Twitter is is constantly changing, is reproducability actually possible? #docnowcommunity
- .@moduloone very difficult to impossible:Twitter's tools seem to favor the now, archives may be crucial for reproducibility #docnowcommunity
- @moduloone @acnwala this is an issue for research in general, right? Social media data put a very special spin on it. #docnowcommunity
- @walkeroh @moduloone absolutely, research papers - esoteric, code links broken, datasets - god sent #docnowcommunity
- @moduloone @acnwala if we archive the data and are willing to share it eventually, we can create a dataset and make it stable
- .@diuhtez What if 10% of tweets later deleted by users? You can ignore, put this pits reproducability vs respecting users. #docnowcommunity
- .@diuhtez Good point! Does seems to be about the scale of time we need reproducibility in. #docnowcommunity
- This has been a really great conversation. Thanks for joining us. Storify coming soon. Feel free to continue the chat here #docnowcommunity
- @documentnow thanks so much for hosting this #docnowcommunity chat - great stuff!
- Thanks to the #docnowcommunity for an awesome talk today!
- Adding links to any website - You can also embed a link to any website, like the official site for a company or event, a Wikipedia page to give background on a subject, or anything else that might give your readers more information. Click the Google source to search for the right site. If you know the direct URL of something you want to embed, use the Embed URL source (the icon looks like a link) and enter it there.
- Notify -
Because your stories are social, you can also let the people who are quoted know that they are now part of your story. This is a great way to help your story spread further, as people who are quoted are likely to also share it with their friends. After your story is published, you will be prompted to use the Notify feature. Give it a try - we think you'll love the reaction you get! - Feedback, questions? - Got questions, problems or thoughts about Storify? Please tell us! Send us a tweet to @storify, post to our Facebook page or email support@storify.com.
- Enjoy, and thanks for using Storify!
