2.3.ii.2.a. CARTO
Within CARTO researchers can use their free service of mapping geotagged tweets that were tweeted within the last 30 days. As CARTO maps geotagged tweets, it should only have access to the approximately 5% of tweets that actually geotagged. However, the company claims that through “geo- enrichment enhancements” CARTO increases its Twitter search results to 15% to 20%. This is done by finding location identifying keywords and then giving that Tweet a geotag according to that location. For instance, if someone mentioned the “Detroit” in the body of their tweet, a geotag will be assigned to that tweet that gives it a random location within Detroit’s boundaries. This is a huge limitation when one is using Twitter data geospatially because it disables the researcher from getting accurate distribution information about where people are tweeting. It also means that the researcher will have to comb through a large amount of geo-enriched data in the dataset to see whether each tweet is actually talking about a place where it has been geotagged.
Twitter data allows for the data to potentially be compiled in a geodatabase. However, the images cannot be included with ESRI’s ArcGIS. It is possible to do this on CARTO and make an interactive geodatabase, which could be a possible option if the researcher was importing their own Twitter data, because the reliability of CARTO Twitter data is questionable.
If one were to use CARTO for Twitter in spite of this limitation, there are three current options available for using their services to get tweets from their several Twitter APIs and map the data with their online mapping platform. The default and free option is to use a Search API which allows the researcher to query geolocated Twitter data from the last 30 days and connect this dataset to their GIS map. There is also a student option for the Search API which allows students to query and map up to 10,000 tweets per month. The second option, which has additional costs, is called “streaming.” This streams and maps current tweets coming in for a charge. The most valuable option both in research value as well as monetary value is the the Historical API option. This option returns tweets back to 2006 when Twitter first began up until 30 days in the past.83
The researcher would need to get a paid CARTO Enterprise plan which starts at $8,000, although they offer 20% discount for the academic community. In addition to that cost would be $3,000 for the consultancy, and $15 per day just for the search depending on how many days one would like the search
42 for. 10,000 tweets are free, but beyond that the cost of tweets would be $1 per 1000 tweets.84
When operating a query for CARTO’s Search API service there are four categories, which then are used to search for tweets using keywords and hashtags. Within each category, one can search multiple hashtags. But once searched, it is impossible to separate the hashtags for further filtering, within the mapping service or dataset. Only the categories themselves can be filtered. The hashtags or keywords are actually searched for individually, however, which might be useful if the terms within a specific category are themed and each one is specific enough to be lumped together. There is also maximum number of characters that can be entered and that limits the searching possibilities.
In order to explore if this way of using this tool could be viable for cross tabulation of Twitter data, each category was themed according to LGBTQ place indicators or LGBTQ community identifying hashtags. Category 1 was for LGBTQ Specific Identity terms, Category 2 was reserved for LGBT Place Specific hashtags taken from NYC LGBT Historic Sites Project top 100 and priority was given to the better-known sites. Category 3 was reserved for Specific People and Movements attributed to NYC LGBT Historic Sites Project’s Sites to try to see if people were talking about any people or movements possibly attributed to LGBTQ sites. Category 4 was for Social Outreach/ Civil Rights tweets to identify again where the LGBTQ community is most active within the city.
Query 1:
Category 1: (LGBTQ Identity Terms)
#lesbian, #lgbtyouth, #lgbtnyc, #queeryouth, #gay, #lgbtqi, #lgbtqia, #ftm, #pansexual, #transwoman, #bisexual, #transsexual, #queer, #bi, #transman, #intersex, #pan, #transgender, #genderqueer, #mtf, #ftm, #drag, #nonbinary, #lgbt, #lgbtq, #boi, #femme, #lipsticklesbian, #butch
Category 2: (LGBT Place Specific - taken from NYC LGBT Historic Sites Project top 100)
#lgbtcenternyc, #pfaff, #herstoryarchives, #portofino, #riispark, #nycitycenter, #wowcafe, #liberationhouse, #moma, #orchardbeach, #judsonmemorialchurch, #juliusbar, #christopherstreetpier, #bathesdafountain, #harlemymca, #iplgy, #greenwoodcemetery, #caffecino, #aliceaustenhouse, #littleredschoolhouse, #liberationhouse, #stonewall, #cubbyhole, #apollotheater, #belascotheater, #bumbumbar, #theramble, #hamiltongrange, #lgbtcommunitycenter
43 Category 3: (Specific People and Movements attributed to NYC LGBT Historic Sites Project’s Sites) #langstonhughes, #arnoldscaasi, #gmhc, #aaroncopeland, #bayardrustin, #andywarhol #aliceausten, #corduroyclub, #gatliberationfront, #dob, #daughtorsofbilitis, #mattachine, #mattachinesociety, #actup, #audreylorde
Category 4: Social Outreach/ Civil Rights Tweets
#gayrights, #notafraid, #itgetsbetter, #wedeservebetter, #trevorproject, #bornthisway, #pride, #gaypride, #transpride, #lgbtpride, #nycpride, #bornperfect, #loveislove, #transrights, #transisbeautiful, #samesex, #protecttranslives, #gayparents
Query 2:
This mapping technique proved to be unreliable because it produced too few tweets than were possible. This was known because when tweets were queried one at a time for each category, there were about the same number of tweets in the same area for these four separate hashtags as for all of the above hashtags. The four separately quarried tweets were tweets that were used in one already
Query 3:
Another query was done to try to simply identify places through hashtags that were place specific and specific to New York City LGBTQ existing hashtags.
Category 1: #safespace Category 2: #NYCLGBTSites Category 3: #LGBTheritage Category 4: #LGBThistoryNYC
2.3.ii.2.b. MAPD
MAPD allows researchers to search hashtags and keywords such as, “LGBT” or #stonewall over the last few months and zoom in for detail and categorizes the tweets by language. It is also free and easy to use.
44 Through MAPD, one has to filter manually, which takes more time. However, this tool offers a free demo, allows for as many searches as desired, allows for all tweets to be queried within a period of three months, and allows researchers to see the mapped data as well as the individual hashtags. There is no “geoenriching” and is thus more reliable than CARTO data. This is a powerful, fast and easy to use tool that also does not have any fees, so this study used this method of collecting Twitter data.
The twitter data that was collected through the Twitter geodatabase MAPD using the keyword results that were previously researched and imported through making a shape file through rasterized screenshots of combined MAPD results.
The actual tweets are available for research as well, which gives the research an insight for filtering these sites and for locating the site the tweet is referring to. It also allows the research to have strong qualitative information in the form of photographs and personal values statements.