Loading, setting up

library(tidyverse) # analysis and processing
library(DBI) # database interface
library(scales) # plots
library(quanteda) # text analysis

cn <- dbConnect(RSQLite::SQLite(), dbname = here::here("db", "k12-institutions-fb-posts.sqlite"))

ss_collected <- tbl(cn, "posts") %>% 
  filter(year == 2020) %>% 
  collect()

ss_collected

## # A tibble: 3,052,643 x 37
##    page_name user_name facebook_id likes_at_posting created type  likes comments
##    <chr>     <chr>           <dbl>            <int>   <dbl> <chr> <dbl>    <dbl>
##  1 Mountain… mountain…     1.55e14             2282  1.60e9 Link      1        0
##  2 Okaloosa… Okaloosa…     1.73e14            16602  1.60e9 Photo     2        0
##  3 Paul R. … PaulRBro…     2.13e14               NA  1.60e9 Photo    11        0
##  4 Western … WesternC…     1.56e11               NA  1.60e9 Photo    76        6
##  5 English … Englishv…     1.44e15              876  1.60e9 Link      0        0
##  6 Dierks H… DierksOu…     4.82e14               NA  1.60e9 Photo    25        0
##  7 Bartow C… <NA>          4.39e11            16879  1.60e9 Link     16        2
##  8 Muskogee… Muskogee…     4.96e10             6344  1.60e9 Photo   140       19
##  9 Mountain… mountain…     1.55e14             2282  1.60e9 Photo    11        0
## 10 Tompkins… tsmsonli…     3.22e14               NA  1.60e9 Photo     0        0
## # … with 3,052,633 more rows, and 29 more variables: shares <dbl>, love <dbl>,
## #   wow <dbl>, haha <dbl>, sad <dbl>, angry <dbl>, care <dbl>,
## #   video_share_status <chr>, post_views <dbl>, total_views <dbl>,
## #   total_views_for_all_crossposts <dbl>, video_length <chr>, url <chr>,
## #   message <chr>, link <chr>, final_link <chr>, image_text <chr>,
## #   link_text <chr>, description <chr>, sponsor_id <chr>, sponsor_name <chr>,
## #   overperforming_score <dbl>, hour <int>, day <dbl>, year <dbl>, month <dbl>,
## #   day_of_week <dbl>, day_of_month <int>, created_rounded_to_day <dbl>

Links

Overall

ss <- ss_collected %>% 
  mutate(links_list = stringr::str_extract_all(message,
                                        "http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+"))

unnested_urls <- ss %>% 
  unnest(links_list) %>% 
  filter(!is.na(links_list))

# my_long_urls <- longurl::expand_urls(unnested_urls$links_list) # is slow; should only do with short links

my_long_urls_processed <- unnested_urls %>% 
  mutate(orig_url= str_split(links_list, ":=:")) %>% 
  mutate(orig_url_second = map(orig_url, pluck, 2)) %>% 
  rowwise() %>%  
  mutate(url = ifelse(is.null(orig_url_second), orig_url, orig_url_second))

parsed_urls <- urltools::url_parse(my_long_urls_processed$url)

parsed_urls %>% 
  count(domain) %>% 
  arrange(desc(n)) %>% 
  filter(!is.na(domain)) %>%
  slice(1:100) %>% 
  knitr::kable()

domain	n
docs.google.com	52064
www.google.com	49849
www.youtube.com	33478
drive.google.com	21261
www.facebook.com	15466
twitter.com	13028
www.smore.com	8718
sites.google.com	8269
accounts.google.com	7389
zoom.us	5491
core-docs.s3.amazonaws.com	5237
us02web.zoom.us	4257
www.surveymonkey.com	4035
mailchi.mp	3892
www.cdc.gov	3877
bit.ly	3279
vimeo.com	3072
www.nhaschools.com	3051
www.signupgenius.com	3024
meet.google.com	2694
www.instagram.com	2665
forms.office.com	2167
content.parent-institute.com	2100
www.applitrack.com	1935
myemail.constantcontact.com	1852
www.eventbrite.com	1786
2020census.gov	1776
www.tulsaschools.org	1724
youtu.be	1516
bookfairs.scholastic.com	1442
www.uticak12.org	1353
4.files.edl.io	1341
ow.ly	1323
tea.texas.gov	1287
flipgrid.com	1241
tinyurl.com	1207
uticak12.org	1202
www.jostens.com	1184
www.careerarc.com	1160
my2020census.gov	1101
forms.gle	1089
us04web.zoom.us	1087
www.hsd2.org	1085
go.boarddocs.com	1044
www.forsyth.k12.ga.us	1023
www.maranausd.org	1010
www.nfhsnetwork.com	999
www.wevideo.com	982
www.mnps.org	876
hhs.texas.gov	865
windsor.weatherstem.com	858
www.colonialschooldistrict.org	847
www.ces95.org	845
www.parentsquare.com	800
spark.adobe.com	799
www.loom.com	782
www.michigan.gov	742
survey.k12insight.com	734
www.garlandisd.net	711
padlet.com	695
www.austinisd.org	684
www.lasd.k12.pa.us	672
teams.microsoft.com	663
striv.tv	657
www.pcsb.org	601
m.facebook.com	597
photos.google.com	595
livestream.com	593
resources.finalsite.net	585
www.shenet.org	585
form.jotform.com	580
my.thoughtexchange.com	568
yourtexasbenefits.com	562
www.bsnteamsports.com	561
www.susd.org	561
www.dallasisd.org	540
nces.ed.gov	524
www2.heart.org	521
wecan.education.wisc.edu	518
www.edjoin.org	517
tpstv.viebit.com	510
s3.amazonaws.com	508
www.ocps.net	503
www.rsdmo.org	500
www.amazon.com	497
play.google.com	489
www.city.waltham.ma.us	487
www.gofundme.com	479
safeyoutube.net	476
www.hy-vee.com	470
achieve.lausd.net	463
apps.apple.com	463
www.dpsk12.org	455
www.donorschoose.org	454
www.gaston.k12.nc.us	449
www.hufsd.edu	445
m.youtube.com	442
www.colorado.gov	442
www.internetessentials.com	432
www.census.gov	425

By year

ss <- ss_collected %>% 
  mutate(links_list = stringr::str_extract_all(message,
                                               "http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+"))

unnested_urls <- ss %>% 
  unnest(links_list) %>% 
  filter(!is.na(links_list))

# my_long_urls <- longurl::expand_urls(unnested_urls$links_list) # is slow; should only do with short links

my_long_urls_processed <- unnested_urls %>% 
  mutate(orig_url= str_split(links_list, ":=:")) %>% 
  mutate(orig_url_second = map(orig_url, pluck, 2)) %>% 
  rowwise() %>%  
  mutate(url = ifelse(is.null(orig_url_second), orig_url, orig_url_second))

parsed_urls <- urltools::url_parse(my_long_urls_processed$url)
parsed_urls$month <- my_long_urls_processed$month
parsed_urls %>% 
  as_tibble()

## # A tibble: 730,902 x 7
##    scheme domain      port  path                   parameter      fragment month
##    <chr>  <chr>       <chr> <chr>                  <chr>          <chr>    <dbl>
##  1 https  www.nfhsne… <NA>  <NA>                   <NA>           <NA>         7
##  2 https  www.rockli… <NA>  <NA>                   <NA>           <NA>         7
##  3 https  www.youtub… <NA>  watch                  v=h3hQ8hK7oj4… <NA>         7
##  4 https  sites.goog… <NA>  caurus.org/caurus-aca… <NA>           <NA>         7
##  5 https  bidcal.hib… <NA>  catalog/220292/yuba-s… <NA>           <NA>         7
##  6 https  www.spring… <NA>  sports/wildcats-perfo… <NA>           <NA>         7
##  7 https  www.indian… <NA>  isaias                 <NA>           <NA>         7
##  8 https  www.youtub… <NA>  watch                  v=WryDzQHEGeY… <NA>         7
##  9 https  www.basset… <NA>  apps/news/article/126… utm_sq=ghb57c… <NA>         7
## 10 https  apps.apple… <NA>  us/app/here-comes-the… <NA>           <NA>         7
## # … with 730,892 more rows

# could probably use table above for this
sum_by_domain <- parsed_urls %>% 
  group_by(month) %>% 
  count(domain) %>% 
  arrange(desc(n)) %>% 
  filter(!is.na(domain),
         n > 0) %>% 
  ungroup() %>% 
  group_by(domain) %>% 
  summarize(sum_n = sum(n)) %>%
  arrange(desc(sum_n)) %>% 
  slice(1:30) # grabbing 30 most common domains

parsed_urls %>% 
  group_by(month) %>% 
  count(domain) %>% 
  arrange(desc(n)) %>% 
  filter(!is.na(domain),
         n > 0) %>% 
  spread(month, n, fill = 0) %>% 
  semi_join(sum_by_domain)

## # A tibble: 30 x 10
##    domain                    `1`   `2`   `3`   `4`   `5`   `6`   `7`   `8`   `9`
##    <chr>                   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
##  1 2020census.gov             80    76   238   444   226   129   110   145   328
##  2 accounts.google.com       137   163   786  1834  1883   728   442   748   668
##  3 bit.ly                   2727   194    91    35    47    32    29    63    61
##  4 bookfairs.scholastic.c…    60   225   222   314   244    46    14    13   304
##  5 content.parent-institu…   434   297   250   253   246   187   126   152   155
##  6 core-docs.s3.amazonaws…   305   285   775   551   554   364   673   917   813
##  7 docs.google.com          2176  1919  7338  6709  7173  5305  7020  8023  6401
##  8 drive.google.com         1082   898  2942  3470  3485  2089  1842  3014  2439
##  9 forms.office.com           34    82   243   275   329   253   370   322   259
## 10 mailchi.mp                555   415   552   508   471   327   223   391   450
## # … with 20 more rows

Text

my_corpus <- corpus(ss_collected, text_field = "message")

my_tokens <- tokens(my_corpus, remove_symbols = T, remove_numbers = T, remove_punct = T, remove_url = T)

my_dfm <- quanteda::dfm(my_tokens, remove = stopwords('en'))

Frequencies

Overall

textstat_frequency(my_dfm, n = 10, groups = "month")

##        feature frequency rank docfreq group
## 1       school    149568    1   94179     1
## 2     students    109230    2   73992     1
## 3      january     63500    3   50170     1
## 4         high     47742    4   35194     1
## 5          day     39928    5   31695     1
## 6       please     35333    6   29593     1
## 7        today     34809    7   32231     1
## 8         year     32557    8   26056     1
## 9          new     30328    9   24653     1
## 10  basketball     30249   10   23078     1
## 11      school    121725    1   77281     2
## 12    students     92693    2   62703     2
## 13    february     46227    3   37140     2
## 14         day     42859    4   33525     2
## 15        high     40841    5   30432     2
## 16       today     32005    6   29655     2
## 17        week     29075    7   23072     2
## 18      please     26651    8   22420     2
## 19        team     26019    9   18622     2
## 20       grade     24198   10   18306     2
## 21      school    200826    1  105096     3
## 22    students    157069    2   92712     3
## 23       march    100828    3   65255     3
## 24      please     95690    4   69267     3
## 25         can     59730    5   43696     3
## 26 information     56097    6   41682     3
## 27        time     55973    7   41099     3
## 28          de     55772    8    8434     3
## 29     schools     53588    9   36092     3
## 30    families     53367   10   38045     3
## 31      school    148620    1   84403     4
## 32    students    108349    2   69259     4
## 33      please     69050    3   51396     4
## 34       april     57574    4   41228     4
## 35         can     54085    5   40701     4
## 36         day     53120    6   41426     4
## 37    learning     48618    7   31890     4
## 38        week     46363    8   35256     4
## 39        time     42452    9   32862     4
## 40          us     42051   10   34180     4
## 41      school    163099    1   95622     5
## 42    students     96379    2   64098     5
## 43         may     78433    3   52914     5
## 44      please     60296    4   45825     5
## 45         day     59639    5   47287     5
## 46       class     54811    6   45053     5
## 47        high     54272    7   40502     5
## 48       thank     50979    8   44179     5
## 49        week     50621    9   39338     5
## 50        year     50102   10   36563     5
## 51      school    108386    1   60156     6
## 52    students     52003    2   32359     6
## 53        june     42705    3   29954     6
## 54      please     39623    4   31130     6
## 55        year     31849    5   23842     6
## 56        high     30181    6   22256     6
## 57      summer     29429    7   21361     6
## 58       class     25493    8   21196     6
## 59          de     24555    9    4602     6
## 60         can     23835   10   18850     6
## 61      school    114236    1   55727     7
## 62    students     51669    2   26908     7
## 63      please     43653    3   33210     7
## 64        july     38788    4   27747     7
## 65        year     31475    5   23037     7
## 66    learning     25940    6   13706     7
## 67 information     24841    7   19369     7
## 68          de     24436    8    4229     7
## 69         can     23948    9   18454     7
## 70    families     21623   10   15301     7
## 71      school    156879    1   83874     8
## 72    students     93231    2   50934     8
## 73      please     63634    3   47118     8
## 74      august     46586    4   32853     8
## 75        year     44644    5   33375     8
## 76         day     37618    6   27568     8
## 77    learning     36067    7   23004     8
## 78          de     34559    8    6609     8
## 79         can     34317    9   26513     8
## 80 information     33258   10   26042     8
## 81      school    127843    1   72743     9
## 82    students     96937    2   57256     9
## 83      please     51966    3   39899     9
## 84         day     45428    4   33598     9
## 85   september     44324    5   33927     9
## 86          de     33785    6    6957     9
## 87         can     33570    7   26309     9
## 88        week     30901    8   23888     9
## 89    learning     30658    9   21349     9
## 90        high     29324   10   21674     9

By month - weighted

my_dfm %>% 
  dfm_tfidf() %>% 
  textstat_frequency(n = 10, groups = "month", force = TRUE) %>% 
  as_tibble() %>% 
  select(feature, rank, group) %>% 
  group_split("group")

## <list_of<
##   tbl_df<
##     feature: character
##     rank   : integer
##     group  : character
##     "group": character
##   >
## >[1]>
## [[1]]
## # A tibble: 90 x 4
##    feature     rank group `"group"`
##    <chr>      <int> <chr> <chr>    
##  1 january        1 1     group    
##  2 school         2 1     group    
##  3 students       3 1     group    
##  4 basketball     4 1     group    
##  5 high           5 1     group    
##  6 boys           6 1     group    
##  7 girls          7 1     group    
##  8 day            8 1     group    
##  9 grade          9 1     group    
## 10 jan           10 1     group    
## # … with 80 more rows

By month - not weighted

my_dfm %>% 
  textstat_frequency(n = 10, groups = "month", force = TRUE) %>% 
  as_tibble() %>% 
  select(feature, rank, group) %>% 
  group_split("group")

## <list_of<
##   tbl_df<
##     feature: character
##     rank   : integer
##     group  : character
##     "group": character
##   >
## >[1]>
## [[1]]
## # A tibble: 90 x 4
##    feature     rank group `"group"`
##    <chr>      <int> <chr> <chr>    
##  1 school         1 1     group    
##  2 students       2 1     group    
##  3 january        3 1     group    
##  4 high           4 1     group    
##  5 day            5 1     group    
##  6 please         6 1     group    
##  7 today          7 1     group    
##  8 year           8 1     group    
##  9 new            9 1     group    
## 10 basketball    10 1     group    
## # … with 80 more rows

Key words in context

From report

kwic(my_tokens, "learning") %>% slice(1:10)

##                                                                            
##     [text1, 5]                           The Parent Letter and | Learning |
##   [text11, 42]                    snapshot of our daily remote | learning |
##   [text16, 34]                    year on August with Distance | Learning |
##  [text16, 116]                   a different and more complete | learning |
##  [text16, 151]            families We are relaunching Distance | Learning |
##  [text16, 195]               new state requirements for online | learning |
##  [text16, 205]                         be made to our Distance | Learning |
##  [text16, 225]   classmates and participate in socio-emotional | learning |
##  [text16, 228] in socio-emotional learning-Teacher facilitated | learning |
##  [text16, 233]      learning including live instruction guided | learning |
##                                              
##  Options Survey is now available             
##  schedule#TigerPride                         
##  due to the mandates provided                
##  experience for our students I               
##  with significant changes and upgrades       
##  the following changes will be               
##  program-Daily opportunities for students    
##  -Teacher facilitated learning including live
##  including live instruction guided learning  
##  and small group instruction throughout

kwic(my_tokens, "technology") %>% slice(1:10)

##                                                                      
##    [text352, 2]                                                 Our |
##   [text402, 17]                              will need to check out |
##    [text414, 4]                         STUDENT INFORMATION LUNCHES |
##   [text414, 20]                  this year's student lunch schedule |
##   [text448, 51]                       Middle School student to have |
##  [text639, 202]                    these tools please reference the |
##  [text665, 471]             Classroom 5-12 incorporating new Google |
##    [text793, 2]                                              TUSD's |
##  [text815, 164]                       means of securing funding for |
##  [text1059, 73] additional professional development associated with |
##                                                                   
##  technology | team is the BEST in                                 
##  technology | from the school                                     
##  TECHNOLOGY | ETC Click link to access                            
##  technology | and more                                            
##  technology | it will deliver a new                               
##  Technology | Resources page on the MPHS                          
##  technology | Provide visibility to teacher summer                
##  Technology | Department making preparations for the              
##  technology | professional development intervention programming On
##  technology | integration The change in the

kwic(my_tokens, "student") %>% slice(1:10)

##                                                               
##  [text16, 295]     -New web-based tools to support | student |
##  [text16, 417]       along with information on our | student |
##   [text37, 69]              the s off your child's | student |
##   [text37, 91]         password So if your child's | student |
##   [text46, 11]               Comes the Bus if your | student |
##   [text51, 29]    access Class Choice through your | student |
##    [text57, 6] A message regarding powerschool and | student |
##   [text57, 28]  virtual platform The finalizing of | student |
##   [text58, 16]                  the life of a YISD | student |
##  [text77, 327]     been fully developed-During the | Student |
##                                               
##  learning including adaptive standards aligned
##  breakfast and lunch program We               
##  number and put a lowercase                   
##  number was s1234567 you would                
##  rides the bus and you                        
##  PowerSchool account and click on             
##  schedules Because of COVID-19 some           
##  schedules had to be put                      
##  during online learning If you                
##  and Family Orientation Days August

kwic(my_tokens, "education") %>% slice(1:10)

##                                                                          
##     [text3, 35]                  Carolina She was a special | education |
##    [text15, 74]              support our students and their | education |
##   [text16, 280]       achieve grade level standards-Special | Education |
##   [text77, 287]      that incorporates an in-person outdoor | education |
##    [text87, 35]               School District and San Diego | Education |
##   [text87, 157] mathematics science social studies physical | education |
##   [text118, 22]                       of the LBUSD Board of | Education |
##   [text193, 34]                   coach and teacher He left | education |
##  [text203, 278]                   be provided with the best | education |
##   [text235, 13]             and the Minnesota Department of | Education |
##                                              
##  teacher in Lenoir county and                
##  #HCSDStrong                                 
##  and related services and support            
##  component The district will roll            
##  Association announced an agreement yesterday
##  and the arts Students will                  
##  the information presented at the            
##  to serve as a firefighter                   
##  that they deserve We are                    
##  have released information as it

kwic(my_tokens, "teaching") %>% slice(1:10)

##                                                           
##    [text3, 27]     spent the last three years | teaching |
##   [text24, 12]   Year Round Schools have been | teaching |
##  [text164, 15] questions about the restart of | teaching |
##  [text271, 17]       the newest member of our | teaching |
##  [text505, 85] teachers represent the best of | teaching |
##  [text637, 18]    Family Coach Fuller will be | teaching |
##  [text662, 18]  Jaguar Family Jessica will be | teaching |
##  [text662, 39]   High School and received her | teaching |
##  [text691, 49]         are so glad she picked | teaching |
##   [text716, 9]           link to view the AEA | Teaching |
##                                 
##  in North Carolina She was      
##  remotely and what to expect    
##  and learning in the fall       
##  team We are glad to            
##  in the SRVUSD and we           
##  P.E this year and has          
##  First Grade Language Arts and  
##  degree from Texas State Jessica
##                                 
##  and Learning Plan for 2020-2021

kwic(my_tokens, "online") %>% slice(1:10)

##                                                                     
##  [text16, 194]             and new state requirements for | online |
##   [text23, 43]                    sold August 7th and 8th | online |
##   [text51, 81]           with teachers at Timpview Option | Online |
##   [text51, 89]               with eSchool We are offering | online |
##  [text51, 101]               your schedule You can choose | online |
##  [text51, 134]                       days or Odd days All | online |
##   [text58, 18]                   of a YISD student during | online |
##   [text58, 35]              to face learning or continued | online |
##   [text81, 32]              platform for the fall whether | online |
##   [text87, 42] announced an agreement yesterday regarding | online |
##                                     
##  learning the following changes will
##  at bidcal.com If you haven't       
##  at home with eSchool We            
##  eSchool classes in every period    
##  classes by taking them The         
##  All students will engage in        
##  learning If you haven't let        
##  learning after the mandatory three 
##  or face to face GO                 
##  learning for the 2020-2021 school

kwic(my_tokens, "teacher") %>% slice(1:10)

##                                                                   
##    [text3, 36]             She was a special education | teacher |
##     [text8, 6] Matthew Catlett Muskogee Public Schools | Teacher |
##  [text16, 218]   relationships and interact with their | teacher |
##   [text50, 17]               Warren ISD family The new | teacher |
##  [text51, 147]      Advisory class with their assigned | teacher |
##  [text51, 155]         from 00am-9 45am Their assigned | teacher |
##    [text85, 8]            posting 7-12 grade PE Health | Teacher |
##    [text98, 4]                        Virtual Meet The | Teacher |
##   [text101, 4]                        Virtual Meet The | Teacher |
##  [text166, 68]             may also email your child's | teacher |
##                                            
##  in Lenoir county and recently             
##  of the year is all                        
##  and classmates and participate in         
##  luncheon with principals and mentors      
##  every Friday from 00am-9 45am             
##  information is forthcoming via PowerSchool
##  North Miami Community Schools has         
##  Slide Show and Videos K-5                 
##  Slide Show and Videos MS                  
##  and they will forward your

Other terms

kwic(my_tokens, "covid") %>% slice(1:10)

##                                                                    
##   [text77, 169]           and or regulations the terrible | COVID |
##  [text317, 364]         as possible during the continuing | COVID |
##   [text499, 51]               4th All practices will have | COVID |
##   [text509, 63]                    Roma High Class of ICU | COVID |
##   [text782, 20]              school year Head to scoe.org | covid |
##   [text929, 28]               Based on State and National | COVID |
##   [text957, 20]              school year Head to scoe.org | covid |
##   [text1203, 3]                                    Due to | COVID |
##  [text1209, 59]               has worked closely with our | COVID |
##  [text1292, 15] Instructional Operational Plan During the | COVID |
##                                     
##  numbers for Grant County I         
##  crisis                             
##  screenings and temp checks prior   
##  Unit Internal Medicine Physician at
##  to access all of SCOE's            
##  trends it is anticipated that      
##  to access all of SCOE's            
##  circumstances the Prom was canceled
##  liaison in order to establish      
##  Pandemic Plan For a Strong

kwic(my_tokens, "school") %>% slice(1:10)

##                                                                   
##    [text3, 52]                    to pursue a degree in | school |
##   [text10, 21]               about how to safely reopen | school |
##    [text11, 9]                     for our first day of | school |
##   [text15, 13]                  today a big donation of | school |
##   [text16, 28]                 we would be starting our | school |
##   [text16, 67]               the state has not provided | school |
##  [text16, 241] group instruction throughout the regular | school |
##  [text16, 351]     elementary and secondary middle high | school |
##  [text16, 394]               to you from your student's | school |
##    [text29, 5]        Congratulations Eden Prairie High | School |
##                                      
##  psychology Welcome aboard ma'am     
##  this Fall Parents can share         
##  on August For the safety            
##  supplies from the United Way        
##  year on August with Distance        
##  districts with the criteria form    
##  day-High quality subject specific   
##  students NOTE The links were        
##  Information on loan of laptops      
##  #CLASSOF2020 Check out our Instagram

kwic(my_tokens, "website") %>% slice(1:10)

##                                                            
##     [text1, 14]   now available on the district | website |
##     [text34, 8]           created a page on our | website |
##    [text85, 18]    has open positions Visit our | website |
##    [text89, 27]          at pm Please visit our | website |
##   [text109, 12]    live on the Martinsville ISD | website |
##   [text118, 41]     are available online on our | website |
##  [text162, 180]       be viewed on our district | website |
##    [text200, 6] Information is available on our | website |
##  [text203, 321]           year If not visit our | website |
##   [text207, 36]         this link or our school | website |
##                                             
##                                             
##  with Hurricane Isaias information including
##  employment section to see a                
##  to view the Zoom meeting                   
##  Please follow the link here                
##                                             
##  An important reminder that due             
##  regarding what you can expect              
##  coastal prep.cfacademy.school and Apply Now
##  to view the broadcast on

kwic(my_tokens, "resource") %>% slice(1:10)

##                                                                         
##   [text575, 58]                       The LCSS website is a | resource |
##  [text582, 611]             Please note students who attend | Resource |
##  [text718, 139]                     School Safety Plan as a | resource |
##   [text1252, 5]                             Here is a great | resource |
##   [text1618, 5]                        Check out this great | resource |
##  [text1948, 20] website's homepage and Coronavirus Response | resource |
##  [text1953, 25]               now available in the COVID-19 | resource |
##   [text2256, 5]                             Here is a great | resource |
##  [text2256, 23]                   Pediatrics This is a good | resource |
##   [text2288, 4]                           The Alisal Family | Resource |
##                                   
##  for parents guardians staff and  
##  Classrooms have supply lists that
##  when it is released or           
##  one of our parents shared        
##  for families and members of      
##  page                             
##  section of the District website  
##  from passed on by our            
##  for parents in relation to       
##  Center Newsletter has a plethora

kwic(my_tokens, "link") %>% slice(1:10)

##                                                          
##     [text7, 7] GRADUATION Here's the live stream | link |
##    [text37, 4]                  Please click the | link |
##    [text63, 6]            AISD parents here is a | link |
##   [text64, 30]           You will be receiving a | link |
##   [text73, 11]            AISD parents here is a | link |
##   [text81, 11]      families please click on the | link |
##  [text109, 16]     ISD website Please follow the | link |
##  [text186, 38]                at 8PM or use this | link |
##  [text190, 23]               QR Code or use this | link |
##  [text195, 38]                at 8PM or use this | link |
##                                    
##  if you cannot be there            
##  below to view a video             
##  to our Reopening Plan and         
##  to the meetings at pm             
##  to our Reopening Plan and         
##  to reach communications about next
##  here to view them Also            
##  #Classof2020YouDidIt              
##                                    
##  #Classof2020YouDidIt

kwic(my_tokens, "laptop") %>% slice(1:10)

##                                                                  
##   [text203, 163]        the teacher's proximity to the | laptop |
##    [text956, 10]               to pick up your student | laptop |
##    [text956, 28]                  work and to get your | laptop |
##   [text1006, 11] providing school supplies including a | laptop |
##   [text1485, 50]              does not have a district | laptop |
##   [text1531, 48]                 gr 2-5 will receive a | laptop |
##    [text1648, 9]             you who cannot access the | laptop |
##   [text1872, 40]          including but not limited to | laptop |
##  [text1872, 157]          including but not limited to | Laptop |
##  [text2085, 126]              don't already have one a | laptop |
##                                            
##  and other background noise-External       
##  Available until pm tonight-7              
##  or call 615-810-8370 to schedule          
##  to all of our students                    
##  or access to the internet                 
##  Please bring your student ID              
##  troubleshooting document from our previous
##  computers tablets and internet access     
##  Desktop computer tablet Wiﬁ Hotspot       
##  sleeve or case that fits

kwic(my_tokens, "internet") %>% slice(1:10)

##                                                                      
##   [text16, 401]                   on loan of laptops and | Internet |
##   [text308, 44]         option IMPORTANTE Las clases por | internet |
##   [text308, 76]                  Clases en persona y por | internet |
##  [text423, 122]     learning programs and providing free | Internet |
##   [text448, 66]               learning and access to the | internet |
##   [text561, 10] Sparklight has extended their discounted | internet |
##   [text561, 22]      make available its Mbps residential | internet |
##    [text698, 6]               Alabama to give million in | internet |
##   [text805, 54]            or remote instruction via the | internet |
##  [text805, 103]            or remote instruction via the | internet |
##                                         
##  hot spots will also be                 
##  comienzan para todos los alumnos       
##  disponibles para el ciclo escolar      
##  access to serving millions of          
##  This initiative will directly correlate
##  package The company will continue      
##  plan for per month for                 
##  vouchers for low-income students       
##  with daily live and recorded           
##  with daily live and recorded

kwic(my_tokens, "access") %>% slice(1:10)

##                                                                          
##    [text51, 24]                        needed You will need to | access |
##    [text53, 41]                   School Video for students to | access |
##  [text203, 110]                            will not be able to | access |
##  [text203, 131]                        and our google suite to | access |
##   [text230, 19] orientation chromebook distribution and Aeries | access |
##  [text331, 325]                             of the web page to | access |
##    [text414, 9]                   TECHNOLOGY ETC Click link to | access |
##  [text423, 123]           programs and providing free Internet | access |
##   [text448, 63]                 innovative way of learning and | access |
##   [text476, 88]                        using this link You can | access |
##                                       
##  Class Choice through your student    
##  on Monday August 3rd                 
##  the live lessons At the              
##  the learning platform Our middle     
##  for schedules                        
##  all information and updates Notices  
##  file with important information about
##  to serving millions of free          
##  to the internet This initiative      
##  breakfast and lunch menus using

kwic(my_tokens, "packet") %>% slice(1:10)

##                                                                   
##    [text89, 45]               of the meeting agenda and | packet |
##  [text1170, 24]       an AEC program their registration | packet |
##  [text1170, 33]               Check your email for this | packet |
##  [text1773, 19]         didn't pick up your culmination | packet |
##  [text1974, 20] school year Each student's registration | packet |
##  [text2174, 44]                 to pick up a substitute | packet |
##   [text2906, 5]                  BACK TO SCHOOL STUDENT | PACKET |
##  [text2906, 21]           your student's back to school | packet |
##  [text2906, 41]           your student's back to school | packet |
##  [text2906, 87]         High School Facebook pages This | packet |
##                              
##  This is an open meeting     
##  must be reutrned Check your 
##  complete return it to the   
##  you can come on Tuesdays    
##  tells us to expect that     
##  If you have any questions   
##  Click on the link to        
##  This year we are asking     
##  online You'll need to submit
##  is required for each student

Term co-occurence

not working with the entire dataset

my_fcm <- fcm(my_dfm)
feat <- names(topfeatures(my_fcm, 50))
my_fcm_select <- fcm_select(my_fcm, pattern = feat, selection = "keep")

size <- log(colSums(dfm_select(my_dfm, feat, selection = "keep")))
textplot_network(my_fcm_select, min_freq = 0.8, vertex_size = size / max(size) * 3,
                 edge_alpha = .3,
                 edge_size = 2)

Using regex

text <- ss %>% 
  select(message, year) %>% 
  collect() %>%  
  mutate(learning = str_detect(message, "(?i)learning"),
         technology = str_detect(message, "(?i)technology"),
         student = str_detect(message, "(?i)student"),
         education = str_detect(message, "(?i)education"),
         teaching = str_detect(message, "(?i)teaching"),
         online = str_detect(message, "(?i)online"),
         teacher = str_detect(message, "(?i)teacher"),
         covid = str_detect(message, "(?i)covid")) %>% 
  filter(!is.na(message))

text_to_plot <- text %>% 
  select(year, learning:covid) %>% 
  gather(key, val, -year) %>% 
  group_by(key, year) %>% 
  summarize(mean_val = mean(val, na.rm = TRUE))

text_to_plot %>% 
  filter(year >= 2010) %>% 
  ggplot(aes(x = year, y = mean_val, color = key, group = key)) +
  geom_point() +
  geom_line() +
  scale_color_brewer(type = "qual") +
  hrbrthemes::theme_ipsum() + 
  scale_x_continuous(breaks = 2010:2020) +
  ylab("proportion")

text <- tbl(cn, "posts") %>% 
  filter(year == 2020) %>% 
  select(message, month) %>% 
  collect() %>%  
  mutate(learning = str_detect(message, "(?i)learning"),
         technology = str_detect(message, "(?i)technology"),
         student = str_detect(message, "(?i)student"),
         education = str_detect(message, "(?i)education"),
         remote  = str_detect(message, "(?i)remote"),
         virtual = str_detect(message, "(?i)virtual"),
         remote = str_detect(message, "(?i)remote"),
         teacher = str_detect(message, "(?i)teacher"),
         meal = str_detect(message, "(?i)meal"),
         covid = str_detect(message, "(?i)covid")) %>% 
  filter(!is.na(message))

text_to_plot <- text %>% 
  select(month, learning:covid) %>% 
  gather(key, val, -month) %>% 
  group_by(key, month) %>% 
  summarize(mean_val = mean(val, na.rm = TRUE))

text_to_plot %>% 
  ggplot(aes(x = month, y = mean_val, color = key, group = key)) +
  geom_point() +
  geom_line() +
  scale_color_brewer(type = "qual") +
  hrbrthemes::theme_ipsum() + 
  scale_x_continuous(breaks = 1:9) +
  ylab("proportion")

Only for ST math, a digital tool (of the kind that may be used more during the pandemic); what other tools can we examine? some ideas here: https://www.educationnext.org/google-search-data-reveals-about-learning-pandemic/

text <- tbl(cn, "posts") %>% 
  filter(year == 2020) %>% 
  select(message, month) %>% 
  collect() %>%  
  filter(!is.na(message)) %>% 
  mutate(st_math= str_detect(message, "(?i)st math"))

text_to_plot <- text %>% 
  select(month, st_math) %>% 
  gather(key, val, -month) %>% 
  group_by(key, month) %>% 
  summarize(mean_val = sum(val, na.rm = TRUE))

text_to_plot %>% 
  ggplot(aes(x = month, y = mean_val, color = key, group = key)) +
  geom_point() +
  geom_line() +
  scale_color_brewer(type = "qual") +
  hrbrthemes::theme_ipsum() + 
  scale_x_continuous(breaks = 1:9) +
  ylab("proportion")

Photos

Not run just yet, working on API

ss_1000 <- tbl(cn, "posts") %>% 
  filter(year == 2020) %>% 
  select(facebook_id) %>% 
  collect() %>% 
  sample_n(1000)

files_to_download <- ss_collected %>% filter(type == "Photo") %>% pull(link) %>% head(5)

map2(files_to_download,
     str_c("downloaded-photos/", 1:5, ".jpeg"),
           download.file)

Content analysis of FB posts

2020-10-21

Loading, setting up

Links

Overall

By year

Text

Frequencies

Key words in context

Term co-occurence

Using regex

Photos