Uses the Wayback CDX Server API.
Script expects a urls.json file. It can be retrieved with the following curl command
"https://web.archive.org/cdx/search/cdx?url=*.mit.edu&collapse=urlkey&output=json"
This will give you all unique URLs under the *.mit.edu namespace archived by the Wayback Machine. It's a few hundred megabytes, so the download may take a while.
| Counter({('edu', 'mit'): 518220, | |
| ('edu', 'mit', 'barnowl'): 100414, | |
| ('edu', 'mit', 'agage'): 68239, | |
| ('edu', 'mit', 'blackhistory'): 50227, | |
| ('edu', 'mit', 'scratch', 'api'): 12894, | |
| ('edu', 'mit', 'scratch', 'cdn2'): 12754, | |
| ('edu', 'mit', 'scratch'): 11790, | |
| ('edu', 'mit', 'arts'): 5252, | |
| ('edu', 'mit', 'appinventor', 'community'): 5229, | |
| ('edu', 'mit', 'dspace'): 2942, | |
| ('edu', 'mit', 'news'): 2686, | |
| ('edu', 'mit', 'aeroastro'): 2587, | |
| ('edu', 'mit', 'athena10'): 2435, | |
| ('edu', 'mit', 'archivesspace'): 2385, | |
| ('edu', 'mit', 'alumgroup', 'amita'): 2341, | |
| ('edu', 'mit', 'alum'): 2340, | |
| ('edu', 'mit', 'act'): 2312, | |
| ('edu', 'mit', 'direct'): 2289, | |
| ('edu', 'mit', 'biology'): 2162, | |
| ('edu', 'mit', 'scratch', 'assets'): 2041, | |
| ('edu', 'mit', 'appinventor', 'ai2'): 1914, | |
| ('edu', 'mit', 'appinventor'): 1801, | |
| ('edu', 'mit', 'bcs'): 1748, | |
| ('edu', 'mit', 'ocw'): 1658, | |
| ('edu', 'mit', 'geodata'): 1432, | |
| ('edu', 'mit', 'media', 'conceptnet5'): 1388, | |
| ('edu', 'mit', 'betterworld'): 1336, | |
| ('edu', 'mit', 'media', 'conceptnet'): 1279, | |
| ('edu', 'mit', 'dome'): 1241, | |
| ('edu', 'mit', 'professionalprograms'): 1172, | |
| ('edu', 'mit', 'krbdev'): 1151, | |
| ('edu', 'mit', 'csail', 'people'): 1100, | |
| ('edu', 'mit', 'climate'): 948, | |
| ('edu', 'mit', 'internetpolicy'): 935, | |
| ('edu', 'mit', 'math'): 880, | |
| ('edu', 'mit', 'mitsloan'): 868, | |
| ('edu', 'mit', 'architecture'): 858, | |
| ('edu', 'mit', 'be'): 843, | |
| ('edu', 'mit', 'acdl'): 820, | |
| ('edu', 'mit', 'calendar'): 798, | |
| ('edu', 'mit', 'ai'): 778, | |
| ('edu', 'mit', 'csail', 'groups'): 774, | |
| ('edu', 'mit', 'drl'): 738, | |
| ('edu', 'mit', 'scratch', 'cdn'): 737, | |
| ('edu', 'mit', 'artscal'): 695, | |
| ('edu', 'mit', 'alum', 'encompass'): 679, | |
| ('edu', 'mit', 'jwel'): 642, | |
| ('edu', 'mit', 'history'): 623, | |
| ('edu', 'mit', 'mitpress'): 590, | |
| ('edu', 'mit', 'ailg'): 576, | |
| ('edu', 'mit', 'scheduling'): 576, | |
| ('edu', 'mit', 'calendars'): 570, | |
| ('edu', 'mit', 'mailman'): 531, | |
| ('edu', 'mit', 'csail', 'labelme'): 511, | |
| ('edu', 'mit', 'classics'): 487, | |
| ('edu', 'mit', 'ballroom-media'): 482, | |
| ('edu', 'mit', 'kb'): 479, | |
| ('edu', 'mit', 'csail', 'lis'): 477, | |
| ('edu', 'mit', 'ab-initio'): 407, | |
| ('edu', 'mit', 'cgcs'): 383, | |
| ('edu', 'mit', 'media', 'opera'): 382, | |
| ('edu', 'mit', 'alo'): 379, | |
| ('edu', 'mit', 'alumic'): 373, | |
| ('edu', 'mit', 'anthropology'): 368, | |
| ('edu', 'mit', 'executive'): 361, | |
| ('edu', 'mit', 'openlearning'): 333, | |
| ('edu', 'mit', 'ai', 'ftp'): 330, | |
| ('edu', 'mit', 'cbmm'): 309, | |
| ('edu', 'mit', 'media'): 307, | |
| ('edu', 'mit', 'd-lab'): 293, | |
| ('edu', 'mit', 'agelab'): 286, | |
| ('edu', 'mit', 'media', 'web'): 278, | |
| ('edu', 'mit', 'mrl'): 271, | |
| ('edu', 'mit', 'bioelectronics'): 268, | |
| ('edu', 'mit', 'globalchange'): 264, | |
| ('edu', 'mit', 'alumsso'): 257, | |
| ('edu', 'mit', 'orgchart'): 252, | |
| ('edu', 'mit', 'eapsweb'): 248, | |
| ('edu', 'mit', 'gambit'): 246, | |
| ('edu', 'mit', 'ccc'): 245, | |
| ('edu', 'mit', 'awards'): 234, | |
| ('edu', 'mit', 'orc'): 229, | |
| ('edu', 'mit', 'hsl'): 228, | |
| ('edu', 'mit', 'capitalprojects'): 224, | |
| ('edu', 'mit', 'lawrencesusskind'): 222, | |
| ('edu', 'mit', 'libraries'): 222, | |
| ('edu', 'mit', 'olsenlab'): 222, | |
| ('edu', 'mit', 'csail'): 221, | |
| ('edu', 'mit', 'icampus'): 207, | |
| ('edu', 'mit', 'mitpress', 'thereader'): 206, | |
| ('edu', 'mit', 'energy'): 200, | |
| ('edu', 'mit', 'globalshakespeares'): 193, | |
| ('edu', 'mit', 'aia'): 190, | |
| ('edu', 'mit', 'eecs'): 189, | |
| ('edu', 'mit', '2009', '2022'): 187, | |
| ('edu', 'mit', 'debathena'): 187, | |
| ('edu', 'mit', 'adminconnect'): 186, | |
| ('edu', 'mit', 'admindatahandbook'): 183, | |
| ('edu', 'mit', 'bateslab'): 183, | |
| ('edu', 'mit', 'hr'): 181, | |
| ('edu', 'mit', 'openlearninglibrary'): 180, | |
| ('edu', 'mit', 'registrar'): 178, | |
| ('edu', 'mit', 'billf'): 172, | |
| ('edu', 'mit', 'research'): 172, | |
| ('edu', 'mit', 'acl'): 169, | |
| ('edu', 'mit', 'physics'): 169, | |
| ('edu', 'mit', 'cba', 'gitlab'): 167, | |
| ('edu', 'mit', 'risingstars18-eecs'): 166, | |
| ('edu', 'mit', 'mitmrsec'): 165, | |
| ('edu', 'mit', 'complex'): 162, | |
| ('edu', 'mit', 'mastodon'): 162, | |
| ('edu', 'mit', 'ashdownhouse'): 160, | |
| ('edu', 'mit', 'ovc-dev'): 160, | |
| ('edu', 'mit', 'maslab'): 156, | |
| ('edu', 'mit', 'biotech'): 154, | |
| ('edu', 'mit', 'beta'): 153, | |
| ('edu', 'mit', 'media', 'simonetti'): 153, | |
| ('edu', 'mit', 'engineering'): 152, | |
| ('edu', 'mit', 'alumclub', 'northerncalifornia'): 151, | |
| ('edu', 'mit', 'asa'): 148, | |
| ('edu', 'mit', 'ic2s2'): 148, | |
| ('edu', 'mit', 'impactlabs'): 147, | |
| ('edu', 'mit', '3down'): 145, | |
| ('edu', 'mit', 'alana'): 143, | |
| ('edu', 'mit', 'catalyst'): 143, | |
| ('edu', 'mit', 'economics'): 143, | |
| ('edu', 'mit', 'intramurals'): 142, | |
| ('edu', 'mit', 'agelab-archive'): 141, | |
| ('edu', 'mit', 'abaqus-docs'): 139, | |
| ('edu', 'mit', 'rle'): 139, | |
| ('edu', 'mit', 'milechin'): 138, | |
| ('edu', 'mit', 'environmentalsolutions'): 136, | |
| ('edu', 'mit', 'ki-images'): 135, | |
| ('edu', 'mit', 'ctl'): 133, | |
| ('edu', 'mit', 'impactclimate'): 133, | |
| ('edu', 'mit', 'owens'): 133, | |
| ('edu', 'mit', 'csail', 'underactuated'): 130, | |
| ('edu', 'mit', 'bazantgroup'): 125, | |
| ('edu', 'mit', 'iceo'): 123, | |
| ('edu', 'mit', 'encompassvolunteers'): 121, | |
| ('edu', 'mit', 'idp'): 121, | |
| ('edu', 'mit', 'chemistry'): 120, | |
| ('edu', 'mit', 'medical'): 120, | |
| ('edu', 'mit', 'abd'): 119, | |
| ('edu', 'mit', 'belong'): 119, | |
| ('edu', 'mit', 'ipc'): 119, | |
| ('edu', 'mit', 'applymitsloan'): 118, | |
| ('edu', 'mit', 'cisr'): 118, | |
| ('edu', 'mit', 'autoid'): 117, | |
| ('edu', 'mit', 'saxelab'): 117, | |
| ('edu', 'mit', 'rledevs4'): 114, | |
| ('edu', 'mit', 'berinsky'): 113, | |
| ('edu', 'mit', 'appinventor', 'weblate'): 112, | |
| ('edu', 'mit', 'connection'): 111, | |
| ('edu', 'mit', '6a'): 110, | |
| ('edu', 'mit', 'mafia'): 109, | |
| ('edu', 'mit', 'mitx', 'lms'): 109, | |
| ('edu', 'mit', 'afrotc'): 108, | |
| ('edu', 'mit', 'ase'): 108, | |
| ('edu', 'mit', 'afeldman'): 106, | |
| ('edu', 'mit', 'mitmgmtfaculty'): 105, | |
| ('edu', 'mit', 'radius'): 104, | |
| ('edu', 'mit', 'accreditation'): 103, | |
| ('edu', 'mit', 'barrett'): 103, | |
| ('edu', 'mit', 'education'): 102, | |
| ('edu', 'mit', 'irvine-lab'): 101, | |
| ('edu', 'mit', 'legatum'): 101, | |
| ('edu', 'mit', 'littletonlab'): 101, | |
| ('edu', 'mit', 'asu'): 100, | |
| ('edu', 'mit', 'cpcm'): 100, | |
| ('edu', 'mit', 'covidapps'): 99, | |
| ('edu', 'mit', 'ceepr'): 98, | |
| ('edu', 'mit', 'livingwage'): 98, | |
| ('edu', 'mit', 'access'): 97, | |
| ('edu', 'mit', 'ovc'): 97, | |
| ('edu', 'mit', 'bdc'): 96, | |
| ('edu', 'mit', 'annasun'): 95, | |
| ('edu', 'mit', 'debathena:444'): 95, | |
| ('edu', 'mit', 'ilp'): 95, | |
| ('edu', 'mit', 'alumgroup', 'eesn'): 94, | |
| ('edu', 'mit', 'bearlab'): 94, | |
| ('edu', 'mit', 'affordablewater'): 92, | |
| ('edu', 'mit', 'aipolicy'): 92, | |
| ('edu', 'mit', 'ki-bridge'): 92, | |
| ('edu', 'mit', 'alumclass', '1993'): 91, | |
| ('edu', 'mit', 'ist'): 90, | |
| ('edu', 'mit', 'ai', 'www-swiss'): 89, | |
| ('edu', 'mit', 'cis'): 88, | |
| ('edu', 'mit', 'media', 'dam-prod2'): 88, | |
| ('edu', 'mit', 'ashkanhh'): 87, | |
| ('edu', 'mit', 'bioinspiredoptics'): 86, | |
| ('edu', 'mit', 'inauguration'): 86, | |
| ('edu', 'mit', 'cms'): 85, | |
| ('edu', 'mit', 'giving'): 85, | |
| ('edu', 'mit', 'ai', 'swiss'): 84, | |
| ('edu', 'mit', 'mitpress', 'hdsr'): 80, | |
| ('edu', 'mit', 'asianclub'): 79, | |
| ('edu', 'mit', 'griffingroup'): 78, | |
| ('edu', 'mit', 'lit'): 78, | |
| ('edu', 'mit', 'childcare'): 77, | |
| ('edu', 'mit', 'github'): 77, | |
| ('edu', 'mit', 'alumclub', 'japan'): 75, | |
| ('edu', 'mit', 'biomakers'): 75, | |
| ('edu', 'mit', 'libproxy', 'login'): 75, | |
| ('edu', 'mit', 'mitpressbookstore'): 75, | |
| ('edu', 'mit', 'capd'): 74, | |
| ('edu', 'mit', 'csail', 'pdos'): 74, | |
| ('edu', 'mit', '2n'): 73, | |
| ('edu', 'mit', 'fabric-ideas'): 73, | |
| ('edu', 'mit', 'biomansummit'): 72, | |
| ('edu', 'mit', 'amre'): 71, | |
| ('edu', 'mit', 'climate-dev'): 71, | |
| ('edu', 'mit', 'infinite'): 71, | |
| ('edu', 'mit', 'idhr'): 70, | |
| ('edu', 'mit', 'ems'): 69, | |
| ('edu', 'mit', 'll'): 69, | |
| ('edu', 'mit', 'appinventor', 'code'): 68, | |
| ('edu', 'mit', 'atlas'): 68, | |
| ('edu', 'mit', 'puzzles'): 68, | |
| ('edu', 'mit', 'architecture-archive'): 67, | |
| ('edu', 'mit', 'ksj'): 67, | |
| ('edu', 'mit', 'oge'): 67, | |
| ('edu', 'mit', 'accessibility'): 66, | |
| ('edu', 'mit', 'alumclub', 'newyork'): 65, | |
| ('edu', 'mit', 'makerworkshop'): 65, | |
| ('edu', 'mit', 'media', 'atlas'): 65, | |
| ('edu', 'mit', 'at250'): 64, | |
| ('edu', 'mit', 'gamelab'): 64, | |
| ('edu', 'mit', 'picower'): 64, | |
| ('edu', 'mit', 'acdl-web'): 62, | |
| ('edu', 'mit', 'advancedligo'): 62, | |
| ('edu', 'mit', 'mcgovern'): 62, | |
| ('edu', 'mit', 'asianamericanstudies'): 60, | |
| ('edu', 'mit', 'cba'): 60, | |
| ('edu', 'mit', 'ibk'): 60, | |
| ('edu', 'mit', 'listart'): 60, | |
| ('edu', 'mit', 'media', 'lsm'): 60, | |
| ('edu', 'mit', 'aidm'): 59, | |
| ('edu', 'mit', 'alexpfeiffer'): 59, | |
| ('edu', 'mit', 'codeforgood'): 59, | |
| ('edu', 'mit', 'mindhandheart'): 59, | |
| ('edu', 'mit', 'freightlab'): 58, | |
| ('edu', 'mit', 'convergence'): 57, | |
| ('edu', 'mit', 'gssd'): 57, | |
| ('edu', 'mit', 'apl'): 56, | |
| ('edu', 'mit', 'gradadvisingmentoring'): 56, | |
| ('edu', 'mit', 'corporation'): 54, | |
| ('edu', 'mit', 'jag'): 54, | |
| ('edu', 'mit', 'addir'): 53, | |
| ('edu', 'mit', 'asdogra'): 53, | |
| ('edu', 'mit', 'esp'): 53, | |
| ('edu', 'mit', 'firstyear'): 53, | |
| ('edu', 'mit', 'lemelson'): 53, | |
| ('edu', 'mit', 'akpia'): 52, | |
| ('edu', 'mit', 'doomcom'): 52, | |
| ('edu', 'mit', 'physicsrisingstars'): 52, | |
| ('edu', 'mit', 'reif'): 52, | |
| ('edu', 'mit', 'actionplan'): 51, | |
| ('edu', 'mit', 'appinventor', 'gallery'): 51, | |
| ('edu', 'mit', 'newsoffice'): 51, | |
| ('edu', 'mit', 'ahtt'): 50, | |
| ('edu', 'mit', 'meche'): 50, | |
| ('edu', 'mit', 'ardent'): 49, | |
| ('edu', 'mit', 'humancreativity'): 49, | |
| ('edu', 'mit', 'media', 'robots'): 49, | |
| ('edu', 'mit', 'applymisti'): 48, | |
| ('edu', 'mit', 'emergency-vent'): 47, | |
| ('edu', 'mit', 'portela'): 47, | |
| ('edu', 'mit', 'anime'): 46, | |
| ('edu', 'mit', 'atlas-config', 'get'): 46, | |
| ('edu', 'mit', 'iso'): 46, | |
| ('edu', 'mit', 'ai', 'swissnet'): 45, | |
| ('edu', 'mit', 'duspconnect'): 45, | |
| ('edu', 'mit', 'infinitemile'): 45, | |
| ('edu', 'mit', 'policies'): 45, | |
| ('edu', 'mit', 'africanfutures'): 44, | |
| ('edu', 'mit', 'exec'): 44, | |
| ('edu', 'mit', 'mites'): 44, | |
| ('edu', 'mit', 'appinventor', 'appathon'): 43, | |
| ('edu', 'mit', 'atlas-auth'): 43, | |
| ('edu', 'mit', 'docubase'): 43, | |
| ('edu', 'mit', 'grandhack'): 42, | |
| ('edu', 'mit', 'misti'): 42, | |
| ('edu', 'mit', 'mitmuseum'): 42, | |
| ('edu', 'mit', 'aristide'): 41, | |
| ('edu', 'mit', 'hs-dev'): 41, | |
| ('edu', 'mit', 'mlkscholars'): 41, | |
| ('edu', 'mit', 'noxdotorg'): 41, | |
| ('edu', 'mit', 'opendoclab'): 41, | |
| ('edu', 'mit', 'aihardware'): 40, | |
| ('edu', 'mit', 'fnl'): 40, | |
| ('edu', 'mit', 'projectalpha'): 40, | |
| ('edu', 'mit', 'aiclub'): 39, | |
| ('edu', 'mit', 'ballroom'): 39, | |
| ('edu', 'mit', 'media', 'tangible'): 38, | |
| ('edu', 'mit', 'mit150'): 38, | |
| ('edu', 'mit', 'arcturus'): 37, | |
| ('edu', 'mit', 'chancellor'): 37, | |
| ('edu', 'mit', 'commencement'): 37, | |
| ('edu', 'mit', 'ide'): 37, | |
| ('edu', 'mit', 'professional'): 37, | |
| ('edu', 'mit', 'asianamerican'): 36, | |
| ('edu', 'mit', 'cmsw'): 36, | |
| ('edu', 'mit', 'officer179'): 36, | |
| ('edu', 'mit', 'bestudents'): 35, | |
| ('edu', 'mit', 'cmrae'): 35, | |
| ('edu', 'mit', 'globalmedia'): 35, | |
| ('edu', 'mit', 'ikim'): 35, | |
| ('edu', 'mit', 'media', 'alumni'): 35, | |
| ('edu', 'mit', 'media', 'pantheon'): 35, | |
| ('edu', 'mit', 'mitpress', 'jods'): 35, | |
| ('edu', 'mit', 'biophysics'): 33, | |
| ('edu', 'mit', 'ki'): 33, | |
| ('edu', 'mit', 'mitnano'): 33, | |
| ('edu', 'mit', '24hourchallenge'): 32, | |
| ('edu', 'mit', 'allegro'): 32, | |
| ('edu', 'mit', 'catalog'): 32, | |
| ('edu', 'mit', 'comms'): 32, | |
| ('edu', 'mit', 'dmse'): 32, | |
| ('edu', 'mit', 'furstlab'): 32, | |
| ('edu', 'mit', 'innovation'): 32, | |
| ('edu', 'mit', 'lcau'): 32, | |
| ('edu', 'mit', 'mitp-content-server'): 32, | |
| ('edu', 'mit', 'sciwrite'): 32, | |
| ('edu', 'mit', 'ashdown'): 31, | |
| ('edu', 'mit', 'begradhandbook'): 31, | |
| ('edu', 'mit', 'covid'): 31, | |
| ('edu', 'mit', 'csail', 'theory'): 31, | |
| ('edu', 'mit', 'idss'): 31, | |
| ('edu', 'mit', 'mobility'): 31, | |
| ('edu', 'mit', 'ai-collective'): 30, | |
| ('edu', 'mit', 'alumgroup', 'mitsaaa'): 30, | |
| ('edu', 'mit', 'beyondmic'): 30, | |
| ('edu', 'mit', 'gsc'): 30, | |
| ('edu', 'mit', 'infinitehistory'): 30, | |
| ('edu', 'mit', 'beach'): 29, | |
| ('edu', 'mit', 'biscuit'): 29, | |
| ('edu', 'mit', 'climateprimer'): 29, | |
| ('edu', 'mit', 'csail', 'dig'): 29, | |
| ('edu', 'mit', 'csail', 'up'): 29, | |
| ('edu', 'mit', 'ai', 'prep'): 28, | |
| ('edu', 'mit', 'anderson-lab'): 28, | |
| ('edu', 'mit', 'appinventor', 'explore'): 28, | |
| ('edu', 'mit', 'biomimetics'): 28, | |
| ('edu', 'mit', 'mirrors'): 28, | |
| ('edu', 'mit', 'mitcet'): 28, | |
| ('edu', 'mit', 'computing'): 27, | |
| ('edu', 'mit', 'csail', 'action-sense'): 27, | |
| ('edu', 'mit', 'lids', 'zelda'): 27, | |
| ('edu', 'mit', 'next'): 27, | |
| ('edu', 'mit', 'scm'): 27, | |
| ('edu', 'mit', 'alumgroup', 'lamit'): 26, | |
| ('edu', 'mit', 'atlas-apps', 'cdn'): 26, | |
| ('edu', 'mit', 'cas'): 26, | |
| ('edu', 'mit', 'cics'): 26, | |
| ('edu', 'mit', 'jobconnector'): 26, | |
| ('edu', 'mit', 'mmi'): 26, | |
| ('edu', 'mit', 'infectiousdiseases'): 25, | |
| ('edu', 'mit', 'lancet'): 25, | |
| ('edu', 'mit', 'lns'): 25, | |
| ('edu', 'mit', 'ai6034'): 24, | |
| ('edu', 'mit', 'askcos'): 24, | |
| ('edu', 'mit', 'atlas-apps', 'tim-tickets'): 24, | |
| ('edu', 'mit', 'be-it'): 24, | |
| ('edu', 'mit', 'design'): 24, | |
| ('edu', 'mit', 'dusp'): 24, | |
| ('edu', 'mit', 'lae'): 24, | |
| ('edu', 'mit', 'migration', 'vidamigrante'): 24, | |
| ('edu', 'mit', 'ome'): 24, | |
| ('edu', 'mit', 'sasilab'): 24, | |
| ('edu', 'mit', 'alumclub', 'utah'): 23, | |
| ('edu', 'mit', 'archfac'): 23, | |
| ('edu', 'mit', 'bgsa'): 23, | |
| ('edu', 'mit', 'cdo', 'cdn'): 23, | |
| ('edu', 'mit', 'commitments'): 23, | |
| ('edu', 'mit', 'jclinic'): 23, | |
| ('edu', 'mit', 'mitpress', 'openaccesseks'): 23, | |
| ('edu', 'mit', 'qcc'): 23, | |
| ('edu', 'mit', 'alumclass', '2014'): 22, | |
| ('edu', 'mit', 'autismresearch'): 22, | |
| ('edu', 'mit', 'belcherlab'): 22, | |
| ('edu', 'mit', 'csail', 'jmlr'): 22, | |
| ('edu', 'mit', 'dltoolkit'): 22, | |
| ('edu', 'mit', 'alumclub', 'boston'): 21, | |
| ('edu', 'mit', 'aryanm'): 21, | |
| ('edu', 'mit', 'azizan'): 21, | |
| ('edu', 'mit', 'eccl'): 21, | |
| ('edu', 'mit', 'electionlab'): 21, | |
| ('edu', 'mit', 'library'): 21, | |
| ('edu', 'mit', 'oeit-sites'): 21, | |
| ('edu', 'mit', 'psfc'): 21, | |
| ('edu', 'mit', 'csail', 'sls'): 20, | |
| ('edu', 'mit', 'gas-turbine-lab'): 20, | |
| ('edu', 'mit', 'now'): 20, | |
| ('edu', 'mit', 'president'): 20, | |
| ('edu', 'mit', '6191'): 19, | |
| ('edu', 'mit', 'aerobatics'): 19, | |
| ('edu', 'mit', 'berwick'): 19, | |
| ('edu', 'mit', 'bioinstrumentation'): 19, | |
| ('edu', 'mit', 'cba', 'phm'): 19, | |
| ('edu', 'mit', 'facts'): 19, | |
| ('edu', 'mit', 'media', 'citysymphonies'): 19, | |
| ('edu', 'mit', 'media', 'lcl-discuss'): 19, | |
| ('edu', 'mit', 'media', 'paradiso'): 19, | |
| ('edu', 'mit', 'ocw2'): 19, | |
| ('edu', 'mit', 'plaza'): 19, | |
| ('edu', 'mit', 'sandbox'): 19, | |
| ('edu', 'mit', 'accelergy'): 18, | |
| ('edu', 'mit', 'actionlearning'): 18, | |
| ('edu', 'mit', 'appinventor', 'iot'): 18, | |
| ('edu', 'mit', 'bearing-witness'): 18, | |
| ('edu', 'mit', 'ferrari'): 18, | |
| ('edu', 'mit', 'lcs', 'graphics'): 18, | |
| ('edu', 'mit', 'libproxy', 'com'): 18, | |
| ('edu', 'mit', 'media', 'heyplix'): 18, | |
| ('edu', 'mit', 'alum-dev'): 17, | |
| ('edu', 'mit', 'amam2015'): 17, | |
| ('edu', 'mit', 'beingmaterial'): 17, | |
| ('edu', 'mit', 'canvas'): 17, | |
| ('edu', 'mit', 'cee'): 17, | |
| ('edu', 'mit', 'chemepro2'): 17, | |
| ('edu', 'mit', 'future'): 17, | |
| ('edu', 'mit', 'garg-lab'): 17, | |
| ('edu', 'mit', 'hz'): 17, | |
| ('edu', 'mit', 'libproxy', 'bmo-bmiresearch-com'): 17, | |
| ('edu', 'mit', 'media', 'admissions'): 17, | |
| ('edu', 'mit', 'ono'): 17, | |
| ('edu', 'mit', 'aimlabs'): 16, | |
| ('edu', 'mit', 'aphi'): 16, | |
| ('edu', 'mit', 'appinventor', 'hourofai'): 16, | |
| ('edu', 'mit', 'backup-w'): 16, | |
| ('edu', 'mit', 'csail', 'cap'): 16, | |
| ('edu', 'mit', 'drake'): 16, | |
| ('edu', 'mit', 'exchange', 'owa'): 16, | |
| ('edu', 'mit', 'introcomp'): 16, | |
| ('edu', 'mit', 'ir'): 16, | |
| ('edu', 'mit', 'lcs', 'haystack'): 16, | |
| ('edu', 'mit', 'lcs', 'pag'): 16, | |
| ('edu', 'mit', 'lids', 'lidsmag'): 16, | |
| ('edu', 'mit', 'lookit'): 16, | |
| ('edu', 'mit', 'media', 'characters'): 16, | |
| ('edu', 'mit', 'mitei'): 16, | |
| ('edu', 'mit', 'nanousers'): 16, | |
| ('edu', 'mit', 'polisci'): 16, | |
| ('edu', 'mit', 'scale'): 16, | |
| ('edu', 'mit', 'admitted'): 15, | |
| ('edu', 'mit', 'alum-r1'): 15, | |
| ('edu', 'mit', 'alumclub', 'centralohio'): 15, | |
| ('edu', 'mit', 'alumsso-dev'): 15, | |
| ('edu', 'mit', 'downloads'): 15, | |
| ('edu', 'mit', 'elx'): 15, | |
| ('edu', 'mit', 'equitytransit'): 15, | |
| ('edu', 'mit', 'iceoreport'): 15, | |
| ('edu', 'mit', 'lcs', 'theory'): 15, | |
| ('edu', 'mit', 'learn-xpro'): 15, | |
| ('edu', 'mit', 'me'): 15, | |
| ('edu', 'mit', 'neet'): 15, | |
| ('edu', 'mit', 'alumclub', 'sloannewyork'): 14, | |
| ('edu', 'mit', 'alumgroup', 'mitaaa'): 14, | |
| ('edu', 'mit', 'apsdfd2015'): 14, | |
| ('edu', 'mit', 'bc'): 14, | |
| ('edu', 'mit', 'biorefs'): 14, | |
| ('edu', 'mit', 'csail', 'courses'): 14, | |
| ('edu', 'mit', 'csail', 'db'): 14, | |
| ('edu', 'mit', 'csail', 'gdp'): 14, | |
| ('edu', 'mit', 'due'): 14, | |
| ('edu', 'mit', 'ideabank'): 14, | |
| ('edu', 'mit', 'interactive'): 14, | |
| ('edu', 'mit', 'kircher'): 14, | |
| ('edu', 'mit', 'libproxy', 'onlinelibrary-wiley-com'): 14, | |
| ('edu', 'mit', 'media', 'labcast'): 14, | |
| ('edu', 'mit', 'media', 'reality'): 14, | |
| ('edu', 'mit', 'odl'): 14, | |
| ('edu', 'mit', 'openlearning-dev'): 14, | |
| ('edu', 'mit', 'persci'): 14, | |
| ('edu', 'mit', '6004'): 13, | |
| ('edu', 'mit', 'aaclab'): 13, | |
| ('edu', 'mit', 'amphibious'): 13, | |
| ('edu', 'mit', 'appinventor', 'appserver'): 13, | |
| ('edu', 'mit', 'assassin'): 13, | |
| ('edu', 'mit', 'cba', 'fab'): 13, | |
| ('edu', 'mit', 'cba', 'pages', 'jpellet'): 13, | |
| ('edu', 'mit', 'cognet'): 13, | |
| ('edu', 'mit', 'crc'): 13, | |
| ('edu', 'mit', 'csail', 'graphite'): 13, | |
| ('edu', 'mit', 'edgerton'): 13, | |
| ('edu', 'mit', 'global'): 13, | |
| ('edu', 'mit', 'haiti'): 13, | |
| ('edu', 'mit', 'mcdermottlab'): 13, | |
| ('edu', 'mit', 'media', 'detectfakes'): 13, | |
| ('edu', 'mit', 'media', 'login'): 13, | |
| ('edu', 'mit', 'adt'): 12, | |
| ('edu', 'mit', 'aera'): 12, | |
| ('edu', 'mit', 'alumclub', 'washingtondc'): 12, | |
| ('edu', 'mit', 'alumgroup', 'bglata'): 12, | |
| ('edu', 'mit', 'bad-ideas'): 12, | |
| ('edu', 'mit', 'beyondthelab'): 12, | |
| ('edu', 'mit', 'bigdawg'): 12, | |
| ('edu', 'mit', 'boning'): 12, | |
| ('edu', 'mit', 'codeit'): 12, | |
| ('edu', 'mit', 'csail', 'chemie'): 12, | |
| ('edu', 'mit', 'haystack'): 12, | |
| ('edu', 'mit', 'laublab'): 12, | |
| ('edu', 'mit', 'le-ministry'): 12, | |
| ('edu', 'mit', 'libproxy', 'www-sciencedirect-com'): 12, | |
| ('edu', 'mit', 'lids'): 12, | |
| ('edu', 'mit', 'media', 'ai4comm'): 12, | |
| ('edu', 'mit', 'mitp-content-server:18180'): 12, | |
| ('edu', 'mit', 'pkgcenter'): 12, | |
| ('edu', 'mit', 'accounts'): 11, | |
| ('edu', 'mit', 'apt'): 11, | |
| ('edu', 'mit', 'ati'): 11, | |
| ('edu', 'mit', 'gcfp'): 11, | |
| ('edu', 'mit', 'idm'): 11, | |
| ('edu', 'mit', 'idss-celebration'): 11, | |
| ('edu', 'mit', 'learning-modules'): 11, | |
| ('edu', 'mit', 'media', 'family'): 11, | |
| ('edu', 'mit', 'media', 'mf'): 11, | |
| ('edu', 'mit', 'mitcommlab'): 11, | |
| ('edu', 'mit', 'mitoc'): 11, | |
| ('edu', 'mit', 'acoustics'): 10, | |
| ('edu', 'mit', 'ajw-group'): 10, | |
| ('edu', 'mit', 'alumapi'): 10, | |
| ('edu', 'mit', 'alumclass', 'mit10'): 10, | |
| ('edu', 'mit', 'alumgroup', 'aacre'): 10, | |
| ('edu', 'mit', 'alumgroup', 'alumnxpride'): 10, | |
| ('edu', 'mit', 'ashford'): 10, | |
| ('edu', 'mit', 'cheme'): 10, | |
| ('edu', 'mit', 'civic'): 10, | |
| ('edu', 'mit', 'credentials'): 10, | |
| ('edu', 'mit', 'libanswers'): 10, | |
| ('edu', 'mit', 'nextres'): 10, | |
| ('edu', 'mit', 'accessmichigan'): 9, | |
| ('edu', 'mit', 'adphi'): 9, | |
| ('edu', 'mit', 'alumclub', 'singapore'): 9, | |
| ('edu', 'mit', 'alumclub', 'venezuela'): 9, | |
| ('edu', 'mit', 'apps'): 9, | |
| ('edu', 'mit', 'cite'): 9, | |
| ('edu', 'mit', 'csail', 'dcai'): 9, | |
| ('edu', 'mit', 'csail', 'haystack'): 9, | |
| ('edu', 'mit', 'geoweb'): 9, | |
| ('edu', 'mit', 'libproxy', 'search-alexanderstreet-com'): 9, | |
| ('edu', 'mit', 'media', 'edev'): 9, | |
| ('edu', 'mit', 'mphotonics'): 9, | |
| ('edu', 'mit', 'alumclass', '1954'): 8, | |
| ('edu', 'mit', 'alumclass', '1991'): 8, | |
| ('edu', 'mit', 'alumgroup', 'dlc'): 8, | |
| ('edu', 'mit', 'apply'): 8, | |
| ('edu', 'mit', 'astrogazers'): 8, | |
| ('edu', 'mit', 'asymptones'): 8, | |
| ('edu', 'mit', 'bahai'): 8, | |
| ('edu', 'mit', 'bitcoin'): 8, | |
| ('edu', 'mit', 'blossoms'): 8, | |
| ('edu', 'mit', 'blueprintlabs'): 8, | |
| ('edu', 'mit', 'bootcamps'): 8, | |
| ('edu', 'mit', 'csail', 'labelme2'): 8, | |
| ('edu', 'mit', 'languages'): 8, | |
| ('edu', 'mit', 'lcs'): 8, | |
| ('edu', 'mit', 'lcs', 'beowulf'): 8, | |
| ('edu', 'mit', 'libguides'): 8, | |
| ('edu', 'mit', 'mimic'): 8, | |
| ('edu', 'mit', 'pgp:11371'): 8, | |
| ('edu', 'mit', '2-70'): 7, | |
| ('edu', 'mit', 'adminappsts'): 7, | |
| ('edu', 'mit', 'aiti'): 7, | |
| ('edu', 'mit', 'alumclub', 'newmexico'): 7, | |
| ('edu', 'mit', 'alumgroup', 'ees'): 7, | |
| ('edu', 'mit', 'archivesspace-staff'): 7, | |
| ('edu', 'mit', 'aro-muri'): 7, | |
| ('edu', 'mit', 'bearlab-s1'): 7, | |
| ('edu', 'mit', 'computing-dev'): 7, | |
| ('edu', 'mit', 'cre'): 7, | |
| ('edu', 'mit', 'csail', 'lists'): 7, | |
| ('edu', 'mit', 'csail', 'missing'): 7, | |
| ('edu', 'mit', 'csail', 'nil'): 7, | |
| ('edu', 'mit', 'csail', 'places'): 7, | |
| ('edu', 'mit', 'csail', 'projects'): 7, | |
| ('edu', 'mit', 'cshub'): 7, | |
| ('edu', 'mit', 'darwinproject'): 7, | |
| ('edu', 'mit', 'dci'): 7, | |
| ('edu', 'mit', 'gradapply'): 7, | |
| ('edu', 'mit', 'hrweb'): 7, | |
| ('edu', 'mit', 'jarilloherrero'): 7, | |
| ('edu', 'mit', 'libproxy'): 7, | |
| ('edu', 'mit', 'media', 'acg'): 7, | |
| ('edu', 'mit', 'media', 'matter'): 7, | |
| ('edu', 'mit', 'media', 'resenv'): 7, | |
| ('edu', 'mit', 'mit2016'): 7, | |
| ('edu', 'mit', 'open'): 7, | |
| ('edu', 'mit', 'pgp'): 7, | |
| ('edu', 'mit', 'raise', 'playground'): 7, | |
| ('edu', 'mit', 'scratch', 'beta'): 7, | |
| ('edu', 'mit', '6031'): 6, | |
| ('edu', 'mit', 'abracadabra'): 6, | |
| ('edu', 'mit', 'actionresearch'): 6, | |
| ('edu', 'mit', 'advancedstudy'): 6, | |
| ('edu', 'mit', 'ai', 'll2'): 6, | |
| ('edu', 'mit', 'alecyen'): 6, | |
| ('edu', 'mit', 'alfalfa'): 6, | |
| ('edu', 'mit', 'alumclub', 'delawarevalley'): 6, | |
| ('edu', 'mit', 'anagha'): 6, | |
| ('edu', 'mit', 'arwhite'): 6, | |
| ('edu', 'mit', 'azhang'): 6, | |
| ('edu', 'mit', 'bigdataanalytics'): 6, | |
| ('edu', 'mit', 'bio-mimetics'): 6, | |
| ('edu', 'mit', 'broad'): 6, | |
| ('edu', 'mit', 'cbcl'): 6, | |
| ('edu', 'mit', 'cehs'): 6, | |
| ('edu', 'mit', 'collabgov'): 6, | |
| ('edu', 'mit', 'csail', 'calendar'): 6, | |
| ('edu', 'mit', 'designx'): 6, | |
| ('edu', 'mit', 'essigmann'): 6, | |
| ('edu', 'mit', 'media', 'dbn'): 6, | |
| ('edu', 'mit', 'media', 'www-prod'): 6, | |
| ('edu', 'mit', 'neuroqwerty'): 6, | |
| ('edu', 'mit', 'orbit-kb'): 6, | |
| ('edu', 'mit', 'pleasure'): 6, | |
| ('edu', 'mit', 'risingstars21-eecs'): 6, | |
| ('edu', 'mit', 'adrianld'): 5, | |
| ('edu', 'mit', 'ai', 'll3'): 5, | |
| ('edu', 'mit', 'aicures22'): 5, | |
| ('edu', 'mit', 'alumclass', '1957'): 5, | |
| ('edu', 'mit', 'alumclass', '1966'): 5, | |
| ('edu', 'mit', 'alumclub', 'southerncalifornia'): 5, | |
| ('edu', 'mit', 'alumniadvisors'): 5, | |
| ('edu', 'mit', 'americanwatershutoffs'): 5, | |
| ('edu', 'mit', 'annualgiving'): 5, | |
| ('edu', 'mit', 'appinventor', 'login'): 5, | |
| ('edu', 'mit', 'asphodel'): 5, | |
| ('edu', 'mit', 'aspire'): 5, | |
| ('edu', 'mit', 'baker'): 5, | |
| ('edu', 'mit', 'barber70'): 5, | |
| ('edu', 'mit', 'belowthebelt'): 5, | |
| ('edu', 'mit', 'chemepro3'): 5, | |
| ('edu', 'mit', 'climategrandchallenges'): 5, | |
| ('edu', 'mit', 'colab'): 5, | |
| ('edu', 'mit', 'diversity'): 5, | |
| ('edu', 'mit', 'ehs'): 5, | |
| ('edu', 'mit', 'girlswhobuild'): 5, | |
| ('edu', 'mit', 'hacks'): 5, | |
| ('edu', 'mit', 'hjkgrp'): 5, | |
| ('edu', 'mit', 'law'): 5, | |
| ('edu', 'mit', 'libraries', 'cdn'): 5, | |
| ('edu', 'mit', 'll', 'llevents'): 5, | |
| ('edu', 'mit', 'lns', 'cosmicwatch'): 5, | |
| ('edu', 'mit', 'media', 'eradio'): 5, | |
| ('edu', 'mit', 'mitpress2'): 5, | |
| ('edu', 'mit', 'pallas'): 5, | |
| ('edu', 'mit', 'ace'): 4, | |
| ('edu', 'mit', 'africa'): 4, | |
| ('edu', 'mit', 'aieducation'): 4, | |
| ('edu', 'mit', 'almlab'): 4, | |
| ('edu', 'mit', 'anjchang'): 4, | |
| ('edu', 'mit', 'appinventor', 'beta'): 4, | |
| ('edu', 'mit', 'army'): 4, | |
| ('edu', 'mit', 'atlas-test'): 4, | |
| ('edu', 'mit', 'atmoschem'): 4, | |
| ('edu', 'mit', 'axo-sisters'): 4, | |
| ('edu', 'mit', 'bandicoot'): 4, | |
| ('edu', 'mit', 'be-ehs'): 4, | |
| ('edu', 'mit', 'cdo'): 4, | |
| ('edu', 'mit', 'csail', 'cb'): 4, | |
| ('edu', 'mit', 'csail', 'pag'): 4, | |
| ('edu', 'mit', 'csail', 'publications'): 4, | |
| ('edu', 'mit', 'csail', 'regina'): 4, | |
| ('edu', 'mit', 'daper'): 4, | |
| ('edu', 'mit', 'doylegroup'): 4, | |
| ('edu', 'mit', 'ebusiness'): 4, | |
| ('edu', 'mit', 'eecseduportal'): 4, | |
| ('edu', 'mit', 'facultygovernance'): 4, | |
| ('edu', 'mit', 'homaesf'): 4, | |
| ('edu', 'mit', 'hst'): 4, | |
| ('edu', 'mit', 'latino'): 4, | |
| ('edu', 'mit', 'lids', 'alpha-leonis'): 4, | |
| ('edu', 'mit', 'magnav'): 4, | |
| ('edu', 'mit', 'media', 'llk'): 4, | |
| ('edu', 'mit', 'media', 'smg'): 4, | |
| ('edu', 'mit', 'mitxonline'): 4, | |
| ('edu', 'mit', 'mta'): 4, | |
| ('edu', 'mit', 'sap'): 4, | |
| ('edu', 'mit', 'scienceimpact'): 4, | |
| ('edu', 'mit', 'ai', 'www-cs101'): 3, | |
| ('edu', 'mit', 'aipolicyforum'): 3, | |
| ('edu', 'mit', 'appinventor', 'hack2020'): 3, | |
| ('edu', 'mit', 'appinventor', 'punya'): 3, | |
| ('edu', 'mit', 'appinventor', 'space'): 3, | |
| ('edu', 'mit', 'appinventor', 'summit'): 3, | |
| ('edu', 'mit', 'arab'): 3, | |
| ('edu', 'mit', 'archthesis'): 3, | |
| ('edu', 'mit', 'army-rotc'): 3, | |
| ('edu', 'mit', 'aschoar'): 3, | |
| ('edu', 'mit', 'atlas-dev'): 3, | |
| ('edu', 'mit', 'banda'): 3, | |
| ('edu', 'mit', 'bellagio'): 3, | |
| ('edu', 'mit', 'cfp'): 3, | |
| ('edu', 'mit', 'cocreationstudio'): 3, | |
| ('edu', 'mit', 'counter'): 3, | |
| ('edu', 'mit', 'cranes'): 3, | |
| ('edu', 'mit', 'csail', 'csg'): 3, | |
| ('edu', 'mit', 'csail', 'css'): 3, | |
| ('edu', 'mit', 'csail', 'swiss'): 3, | |
| ('edu', 'mit', 'e-vent'): 3, | |
| ('edu', 'mit', 'electron'): 3, | |
| ('edu', 'mit', 'emvogil-3'): 3, | |
| ('edu', 'mit', 'entrepreneurship'): 3, | |
| ('edu', 'mit', 'imbrium'): 3, | |
| ('edu', 'mit', 'imes'): 3, | |
| ('edu', 'mit', 'lbourouiba'): 3, | |
| ('edu', 'mit', 'lcs', 'cag'): 3, | |
| ('edu', 'mit', 'lcs', 'sdg'): 3, | |
| ('edu', 'mit', 'lcs', 'sls'): 3, | |
| ('edu', 'mit', 'libproxy', 'com', 'dieselnet'): 3, | |
| ('edu', 'mit', 'll', 'careers'): 3, | |
| ('edu', 'mit', 'mcgeelab'): 3, | |
| ('edu', 'mit', 'media', 'affect'): 3, | |
| ('edu', 'mit', 'media', 'dkroy'): 3, | |
| ('edu', 'mit', 'media', 'fluid'): 3, | |
| ('edu', 'mit', 'media', 'mentalhealth'): 3, | |
| ('edu', 'mit', 'media', 'socialmachines'): 3, | |
| ('edu', 'mit', 'mitpress', 'covid-19'): 3, | |
| ('edu', 'mit', 'mitpressjournals'): 3, | |
| ('edu', 'mit', 'mtl'): 3, | |
| ('edu', 'mit', 'nanomedicine'): 3, | |
| ('edu', 'mit', 'opa'): 3, | |
| ('edu', 'mit', 'openspace'): 3, | |
| ('edu', 'mit', 'pact'): 3, | |
| ('edu', 'mit', 'pog'): 3, | |
| ('edu', 'mit', '18thpresident'): 2, | |
| ('edu', 'mit', '2009'): 2, | |
| ('edu', 'mit', 'acisweb'): 2, | |
| ('edu', 'mit', 'act2020'): 2, | |
| ('edu', 'mit', 'admissions'): 2, | |
| ('edu', 'mit', 'aeroastro-ad'): 2, | |
| ('edu', 'mit', 'affiliated'): 2, | |
| ('edu', 'mit', 'ai', 'biobricks'): 2, | |
| ('edu', 'mit', 'aikido'): 2, | |
| ('edu', 'mit', 'alumclass', '1964'): 2, | |
| ('edu', 'mit', 'ams'): 2, | |
| ('edu', 'mit', 'andrescampero'): 2, | |
| ('edu', 'mit', 'anonsvn'): 2, | |
| ('edu', 'mit', 'arclab'): 2, | |
| ('edu', 'mit', 'argonaute'): 2, | |
| ('edu', 'mit', 'asians'): 2, | |
| ('edu', 'mit', 'asl'): 2, | |
| ('edu', 'mit', 'ats'): 2, | |
| ('edu', 'mit', 'audit'): 2, | |
| ('edu', 'mit', 'audivsurveys'): 2, | |
| ('edu', 'mit', 'b-b'): 2, | |
| ('edu', 'mit', 'biogels'): 2, | |
| ('edu', 'mit', 'c2c-ctf-2022'): 2, | |
| ('edu', 'mit', 'cams'): 2, | |
| ('edu', 'mit', 'cci'): 2, | |
| ('edu', 'mit', 'ccs'): 2, | |
| ('edu', 'mit', 'colabradio'): 2, | |
| ('edu', 'mit', 'csail', 'alg'): 2, | |
| ('edu', 'mit', 'csail', 'cag'): 2, | |
| ('edu', 'mit', 'csail', 'cgh'): 2, | |
| ('edu', 'mit', 'csail', 'nms'): 2, | |
| ('edu', 'mit', 'csail', 'physadept'): 2, | |
| ('edu', 'mit', 'csail', 'pmg'): 2, | |
| ('edu', 'mit', 'csail', 'rvsn'): 2, | |
| ('edu', 'mit', 'csail', 'vis'): 2, | |
| ('edu', 'mit', 'dbertsim'): 2, | |
| ('edu', 'mit', 'dc'): 2, | |
| ('edu', 'mit', 'disruptiveplanets'): 2, | |
| ('edu', 'mit', 'diswww'): 2, | |
| ('edu', 'mit', 'educationgroup'): 2, | |
| ('edu', 'mit', 'gradadmissions'): 2, | |
| ('edu', 'mit', 'grades'): 2, | |
| ('edu', 'mit', 'hardjono'): 2, | |
| ('edu', 'mit', 'ideonomy'): 2, | |
| ('edu', 'mit', 'info-libraries'): 2, | |
| ('edu', 'mit', 'infoprotect'): 2, | |
| ('edu', 'mit', 'julia'): 2, | |
| ('edu', 'mit', 'jwafs'): 2, | |
| ('edu', 'mit', 'lcs', 'nms'): 2, | |
| ('edu', 'mit', 'lcs', 'sds'): 2, | |
| ('edu', 'mit', 'lcs', 'supertech'): 2, | |
| ('edu', 'mit', 'lgo'): 2, | |
| ('edu', 'mit', 'lib'): 2, | |
| ('edu', 'mit', 'lids', 'people'): 2, | |
| ('edu', 'mit', 'lidssmart2017'): 2, | |
| ('edu', 'mit', 'media', 'act'): 2, | |
| ('edu', 'mit', 'media', 'dam-prod'): 2, | |
| ('edu', 'mit', 'media', 'el'): 2, | |
| ('edu', 'mit', 'media', 'hlt'): 2, | |
| ('edu', 'mit', 'media', 'mitcityfarm'): 2, | |
| ('edu', 'mit', 'media', 'neri'): 2, | |
| ('edu', 'mit', 'media', 'scratched'): 2, | |
| ('edu', 'mit', 'mimo'): 2, | |
| ('edu', 'mit', 'mitili'): 2, | |
| ('edu', 'mit', 'mitpress', 'mitpressonpubpub'): 2, | |
| ('edu', 'mit', 'mitpress', 'wip'): 2, | |
| ('edu', 'mit', 'mitpress-request'): 2, | |
| ('edu', 'mit', 'mitsfs'): 2, | |
| ('edu', 'mit', 'moralmachine'): 2, | |
| ('edu', 'mit', 'msl'): 2, | |
| ('edu', 'mit', 'mtlsites'): 2, | |
| ('edu', 'mit', 'negotiation'): 2, | |
| ('edu', 'mit', 'newdigs'): 2, | |
| ('edu', 'mit', 'nrl'): 2, | |
| ('edu', 'mit', 'oastats'): 2, | |
| ('edu', 'mit', 'odl', 'greetings'): 2, | |
| ('edu', 'mit', 'parts'): 2, | |
| ('edu', 'mit', 'police'): 2, | |
| ('edu', 'mit', 'proquet'): 2, | |
| ('edu', 'mit', 'qis'): 2, | |
| ('edu', 'mit', 'queerspace'): 2, | |
| ('edu', 'mit', '200b'): 1, | |
| ('edu', 'mit', '3091'): 1, | |
| ('edu', 'mit', 'aacii'): 1, | |
| ('edu', 'mit', 'aai'): 1, | |
| ('edu', 'mit', 'aamp'): 1, | |
| ('edu', 'mit', 'abatelab-dev'): 1, | |
| ('edu', 'mit', 'abel'): 1, | |
| ('edu', 'mit', 'abhishek'): 1, | |
| ('edu', 'mit', 'absence'): 1, | |
| ('edu', 'mit', 'abugoot'): 1, | |
| ('edu', 'mit', 'acavallo'): 1, | |
| ('edu', 'mit', 'accel'): 1, | |
| ('edu', 'mit', 'account'): 1, | |
| ('edu', 'mit', 'account-registration-dev'): 1, | |
| ('edu', 'mit', 'adainspired'): 1, | |
| ('edu', 'mit', 'adalca'): 1, | |
| ('edu', 'mit', 'adminappsts-dev'): 1, | |
| ('edu', 'mit', 'admitone'): 1, | |
| ('edu', 'mit', 'adp'): 1, | |
| ('edu', 'mit', 'adviseme'): 1, | |
| ('edu', 'mit', 'aeroastro-dev'): 1, | |
| ('edu', 'mit', 'aeroastro-ehs'): 1, | |
| ('edu', 'mit', 'aeroastro-refs'): 1, | |
| ('edu', 'mit', 'aeroprop'): 1, | |
| ('edu', 'mit', 'agelab-stream'): 1, | |
| ('edu', 'mit', 'agi'): 1, | |
| ('edu', 'mit', 'agiledrones'): 1, | |
| ('edu', 'mit', 'agimpacts'): 1, | |
| ('edu', 'mit', 'agingbrain'): 1, | |
| ('edu', 'mit', 'ai', 'gnu', 'z'): 1, | |
| ('edu', 'mit', 'ai', 'll1'): 1, | |
| ('edu', 'mit', 'ai-education-research-portal'): 1, | |
| ('edu', 'mit', 'aidinoff'): 1, | |
| ('edu', 'mit', 'airlinedataproject'): 1, | |
| ('edu', 'mit', 'ajhart'): 1, | |
| ('edu', 'mit', 'akira'): 1, | |
| ('edu', 'mit', 'alertextended'): 1, | |
| ('edu', 'mit', 'alfagroup'): 1, | |
| ('edu', 'mit', 'allanore'): 1, | |
| ('edu', 'mit', 'althost'): 1, | |
| ('edu', 'mit', 'alumclass', '1960'): 1, | |
| ('edu', 'mit', 'alumclub', 'minnesota'): 1, | |
| ('edu', 'mit', 'alumclub', 'rtp'): 1, | |
| ('edu', 'mit', 'alumclub', 'semichigan'): 1, | |
| ('edu', 'mit', 'alumclub', 'vermont'): 1, | |
| ('edu', 'mit', 'alumgroup', 'crew'): 1, | |
| ('edu', 'mit', 'alumgroup', 'mitarcha'): 1, | |
| ('edu', 'mit', 'alumlist'): 1, | |
| ('edu', 'mit', 'amadae'): 1, | |
| ('edu', 'mit', 'amnesty'): 1, | |
| ('edu', 'mit', 'analyticsfair'): 1, | |
| ('edu', 'mit', 'analyzeralloy'): 1, | |
| ('edu', 'mit', 'andersk'): 1, | |
| ('edu', 'mit', 'andreacampbell'): 1, | |
| ('edu', 'mit', 'annotationapi'): 1, | |
| ('edu', 'mit', 'annotationstudio'): 1, | |
| ('edu', 'mit', 'annualfund'): 1, | |
| ('edu', 'mit', 'ans'): 1, | |
| ('edu', 'mit', 'antarcticaworldpassport'): 1, | |
| ('edu', 'mit', 'antoniou'): 1, | |
| ('edu', 'mit', 'aolearn'): 1, | |
| ('edu', 'mit', 'api'): 1, | |
| ('edu', 'mit', 'appinventor', 'alexa'): 1, | |
| ('edu', 'mit', 'appinventor', 'alexaskills'): 1, | |
| ('edu', 'mit', 'appinventor', 'ar'): 1, | |
| ('edu', 'mit', 'appinventor', 'c1'): 1, | |
| ('edu', 'mit', 'appinventor', 'conv-ai'): 1, | |
| ('edu', 'mit', 'appinventor', 'coolthink'): 1, | |
| ('edu', 'mit', 'appinventor', 'datasci'): 1, | |
| ('edu', 'mit', 'appinventor', 'extension-test'): 1, | |
| ('edu', 'mit', 'appinventor', 'files'): 1, | |
| ('edu', 'mit', 'appinventor', 'iosbuildservertest'): 1, | |
| ('edu', 'mit', 'appinventor', 'rendezvous'): 1, | |
| ('edu', 'mit', 'archivematica'): 1, | |
| ('edu', 'mit', 'archrefs'): 1, | |
| ('edu', 'mit', 'arcsubmit-stage'): 1, | |
| ('edu', 'mit', 'aresnick'): 1, | |
| ('edu', 'mit', 'armail'): 1, | |
| ('edu', 'mit', 'arnabs'): 1, | |
| ('edu', 'mit', 'asartori'): 1, | |
| ('edu', 'mit', 'ask'): 1, | |
| ('edu', 'mit', 'asp'): 1, | |
| ('edu', 'mit', 'aspace-staff-dev'): 1, | |
| ('edu', 'mit', 'atlas-apps', 'commencement-checkin'): 1, | |
| ('edu', 'mit', 'atlas-apps', 'kiosk'): 1, | |
| ('edu', 'mit', 'atlas-apps', 'pass'): 1, | |
| ('edu', 'mit', 'atlas-apps', 'spaces'): 1, | |
| ('edu', 'mit', 'atlas-apps-dev'): 1, | |
| ('edu', 'mit', 'atlas-apps-dev', 'commencement-checkin'): 1, | |
| ('edu', 'mit', 'atlas-apps-dev', 'jh-okta-dev'): 1, | |
| ('edu', 'mit', 'atlas-apps-test', 'commencement-checkin'): 1, | |
| ('edu', 'mit', 'atlas-auth-dev'): 1, | |
| ('edu', 'mit', 'atlas-center'): 1, | |
| ('edu', 'mit', 'atlas-service-center'): 1, | |
| ('edu', 'mit', 'atlascenter'): 1, | |
| ('edu', 'mit', 'atlasservicecenter'): 1, | |
| ('edu', 'mit', 'atopensigma'): 1, | |
| ('edu', 'mit', 'attendance'): 1, | |
| ('edu', 'mit', 'atweb'): 1, | |
| ('edu', 'mit', 'auvlab'): 1, | |
| ('edu', 'mit', 'azubair'): 1, | |
| ('edu', 'mit', 'b2g-lab'): 1, | |
| ('edu', 'mit', 'babi-mitra'): 1, | |
| ('edu', 'mit', 'bablab'): 1, | |
| ('edu', 'mit', 'bakerlab'): 1, | |
| ('edu', 'mit', 'balchange'): 1, | |
| ('edu', 'mit', 'baphomet2'): 1, | |
| ('edu', 'mit', 'bathroom'): 1, | |
| ('edu', 'mit', 'bazki'): 1, | |
| ('edu', 'mit', 'bcs-dev'): 1, | |
| ('edu', 'mit', 'bcs-dev-www-2'): 1, | |
| ('edu', 'mit', 'bcs-dev-www-3'): 1, | |
| ('edu', 'mit', 'bcs-wp-spare-1'): 1, | |
| ('edu', 'mit', 'bcs-wp-spare-2'): 1, | |
| ('edu', 'mit', 'bcs-wp-spare-3'): 1, | |
| ('edu', 'mit', 'bcs-wp-spare-4'): 1, | |
| ('edu', 'mit', 'bcsrefs'): 1, | |
| ('edu', 'mit', 'be-prod-web-1'): 1, | |
| ('edu', 'mit', 'be-prod-web-2'): 1, | |
| ('edu', 'mit', 'be-prod-web-3'): 1, | |
| ('edu', 'mit', 'be-prod-web-4'): 1, | |
| ('edu', 'mit', 'be-retreat'): 1, | |
| ('edu', 'mit', 'be-scheduling'): 1, | |
| ('edu', 'mit', 'beach-dev'): 1, | |
| ('edu', 'mit', 'bec5'): 1, | |
| ('edu', 'mit', 'becurious'): 1, | |
| ('edu', 'mit', 'beekeepers'): 1, | |
| ('edu', 'mit', 'belcher10'): 1, | |
| ('edu', 'mit', 'benweiss'): 1, | |
| ('edu', 'mit', 'betterscience'): 1, | |
| ('edu', 'mit', 'beughandbook'): 1, | |
| ('edu', 'mit', 'bevfusion'): 1, | |
| ('edu', 'mit', 'beyondthejob'): 1, | |
| ('edu', 'mit', 'bhangra'): 1, | |
| ('edu', 'mit', 'bif'): 1, | |
| ('edu', 'mit', 'bioanalytics'): 1, | |
| ('edu', 'mit', 'bioengcareerexpo'): 1, | |
| ('edu', 'mit', 'biogeochemistry'): 1, | |
| ('edu', 'mit', 'biomakerspace'): 1, | |
| ('edu', 'mit', 'bioman'): 1, | |
| ('edu', 'mit', 'bolashak'): 1, | |
| ('edu', 'mit', 'borderline'): 1, | |
| ('edu', 'mit', 'brassrat2024'): 1, | |
| ('edu', 'mit', 'cba', 'ng'): 1, | |
| ('edu', 'mit', 'cctr'): 1, | |
| ('edu', 'mit', 'chemvideos'): 1, | |
| ('edu', 'mit', 'climbing'): 1, | |
| ('edu', 'mit', 'collaborative'): 1, | |
| ('edu', 'mit', 'compton'): 1, | |
| ('edu', 'mit', 'connect'): 1, | |
| ('edu', 'mit', 'cooperation'): 1, | |
| ('edu', 'mit', 'corporateinnovation'): 1, | |
| ('edu', 'mit', 'corporaterelations'): 1, | |
| ('edu', 'mit', 'courseroad'): 1, | |
| ('edu', 'mit', 'covid19challenge'): 1, | |
| ('edu', 'mit', 'csail', '8300', '6'): 1, | |
| ('edu', 'mit', 'csail', 'cis'): 1, | |
| ('edu', 'mit', 'csail', 'ei'): 1, | |
| ('edu', 'mit', 'csail', 'f3d'): 1, | |
| ('edu', 'mit', 'csail', 'gandissect'): 1, | |
| ('edu', 'mit', 'csail', 'ganocracy'): 1, | |
| ('edu', 'mit', 'csail', 'iros2022minirobots'): 1, | |
| ('edu', 'mit', 'csail', 'moocdb'): 1, | |
| ('edu', 'mit', 'csail', 'propnet'): 1, | |
| ('edu', 'mit', 'csail', 'sceneparsing'): 1, | |
| ('edu', 'mit', 'csail', 'sdg'): 1, | |
| ('edu', 'mit', 'csail', 'start'): 1, | |
| ('edu', 'mit', 'csail', 'supertech'): 1, | |
| ('edu', 'mit', 'csail', 'toc'): 1, | |
| ('edu', 'mit', 'cua'): 1, | |
| ('edu', 'mit', 'deiactionplan'): 1, | |
| ('edu', 'mit', 'designforamerica'): 1, | |
| ('edu', 'mit', 'devavrat'): 1, | |
| ('edu', 'mit', 'digital'): 1, | |
| ('edu', 'mit', 'drennan'): 1, | |
| ('edu', 'mit', 'ecg'): 1, | |
| ('edu', 'mit', 'eltahir'): 1, | |
| ('edu', 'mit', 'esi'): 1, | |
| ('edu', 'mit', 'esp-mail'): 1, | |
| ('edu', 'mit', 'events'): 1, | |
| ('edu', 'mit', 'executive-education'): 1, | |
| ('edu', 'mit', 'feynman'): 1, | |
| ('edu', 'mit', 'flavell'): 1, | |
| ('edu', 'mit', 'flightgoggles'): 1, | |
| ('edu', 'mit', 'flippingfailure'): 1, | |
| ('edu', 'mit', 'fullsteam'): 1, | |
| ('edu', 'mit', 'futurebook'): 1, | |
| ('edu', 'mit', 'gallowaylab'): 1, | |
| ('edu', 'mit', 'gecd'): 1, | |
| ('edu', 'mit', 'genes'): 1, | |
| ('edu', 'mit', 'gitstream'): 1, | |
| ('edu', 'mit', 'groups'): 1, | |
| ('edu', 'mit', 'hecps1web01:44300'): 1, | |
| ('edu', 'mit', 'hkinnovationnode'): 1, | |
| ('edu', 'mit', 'hrst'): 1, | |
| ('edu', 'mit', 'ideation'): 1, | |
| ('edu', 'mit', 'ihouse'): 1, | |
| ('edu', 'mit', 'integrity'): 1, | |
| ('edu', 'mit', 'iquise'): 1, | |
| ('edu', 'mit', 'isn'): 1, | |
| ('edu', 'mit', 'its'): 1, | |
| ('edu', 'mit', 'jdj'): 1, | |
| ('edu', 'mit', 'k12maker'): 1, | |
| ('edu', 'mit', 'kendallsquare'): 1, | |
| ('edu', 'mit', 'lbrc'): 1, | |
| ('edu', 'mit', 'lcs', 'khavrinen'): 1, | |
| ('edu', 'mit', 'lcs', 'larch-www:8001'): 1, | |
| ('edu', 'mit', 'lcs', 'nil'): 1, | |
| ('edu', 'mit', 'lea'): 1, | |
| ('edu', 'mit', 'lebeau'): 1, | |
| ('edu', 'mit', 'lfe'): 1, | |
| ('edu', 'mit', 'lgsa'): 1, | |
| ('edu', 'mit', 'libproxy', 'com', 'wiley', 'onlinelibrary'): 1, | |
| ('edu', 'mit', 'lids', 'mitter'): 1, | |
| ('edu', 'mit', 'linc2016'): 1, | |
| ('edu', 'mit', 'lingphil'): 1, | |
| ('edu', 'mit', 'linguistics'): 1, | |
| ('edu', 'mit', 'lns', 'blast'): 1, | |
| ('edu', 'mit', 'lns', 'ctp'): 1, | |
| ('edu', 'mit', 'lns', 'ctpweb'): 1, | |
| ('edu', 'mit', 'madry'): 1, | |
| ('edu', 'mit', 'meddevdesign'): 1, | |
| ('edu', 'mit', 'media', 'blog'): 1, | |
| ('edu', 'mit', 'media', 'courses'): 1, | |
| ('edu', 'mit', 'media', 'duoskin'): 1, | |
| ('edu', 'mit', 'media', 'ic'): 1, | |
| ('edu', 'mit', 'media', 'lcl'): 1, | |
| ('edu', 'mit', 'media', 'learn'): 1, | |
| ('edu', 'mit', 'media', 'learning'): 1, | |
| ('edu', 'mit', 'media', 'open'): 1, | |
| ('edu', 'mit', 'media', 'robotic'): 1, | |
| ('edu', 'mit', 'meet'): 1, | |
| ('edu', 'mit', 'micromasters'): 1, | |
| ('edu', 'mit', 'mitaly'): 1, | |
| ('edu', 'mit', 'mitcmse'): 1, | |
| ('edu', 'mit', 'mitcre'): 1, | |
| ('edu', 'mit', 'mitgsl'): 1, | |
| ('edu', 'mit', 'mitpress', 'cfrp'): 1, | |
| ('edu', 'mit', 'mitpress', 'rapidreviewscovid19'): 1, | |
| ('edu', 'mit', 'mitsloanedtech'): 1, | |
| ('edu', 'mit', 'mobilityinitiative'): 1, | |
| ('edu', 'mit', 'mollylab-1'): 1, | |
| ('edu', 'mit', 'mvl'): 1, | |
| ('edu', 'mit', 'mysloan'): 1, | |
| ('edu', 'mit', 'nasa'): 1, | |
| ('edu', 'mit', 'nb'): 1, | |
| ('edu', 'mit', 'nightmare'): 1, | |
| ('edu', 'mit', 'oceans'): 1, | |
| ('edu', 'mit', 'officesdirectory'): 1, | |
| ('edu', 'mit', 'ogcr'): 1, | |
| ('edu', 'mit', 'playful'): 1, | |
| ('edu', 'mit', 'poggio-lab'): 1, | |
| ('edu', 'mit', 'prathergroup'): 1, | |
| ('edu', 'mit', 'price'): 1, | |
| ('edu', 'mit', 'process'): 1, | |
| ('edu', 'mit', 'psfc', 'www-internal'): 1, | |
| ('edu', 'mit', 'psfc', 'www-new'): 1, | |
| ('edu', 'mit', 'puddle'): 1, | |
| ('edu', 'mit', 'rambax'): 1, | |
| ('edu', 'mit', 'resil'): 1, | |
| ('edu', 'mit', 'saveseniorhouse'): 1, | |
| ('edu', 'mit', 'scattering'): 1, | |
| ('edu', 'mit', 'science'): 1, | |
| ('edu', 'mit', 'scopeweb'): 1}) |
| import json | |
| from collections import Counter | |
| from pprint import pprint | |
| with open("urls.json") as f: | |
| index = json.load(f)[1:] | |
| counter = Counter( | |
| [ | |
| tuple(page[0].split(")", 1)[0].split(",")) | |
| for page in index | |
| ] | |
| ) | |
| pprint(counter) | |
| # for page in index: | |
| # urlkey = page[0] | |
| # unparsed_domain, path = urlkey.split(")", 1) | |
| # domain = tuple(unparsed_domain.split(",")) | |
| # break | |