Automatic knowledge classification based on keyword co-occurrrence network

Hope (Huang Tian-Yuan)

Introduction

Short for automatic knowledge classification, akc is an R package used to carry out keyword classification based on network science (mainly community detection techniques), using bibliometric data. However, these provided functions are general, and could be extended to solve other tasks in text mining as well. Main functions are listed as below:

Features

Generally provides a tidy framework of data manipulation supported by dplyr, akc was written in data.table when necessary to guarantee the performance for big data analysis. Meanwhile, akc also utilizes the state-of-the-art text mining functions provided by stringr,tidytext,textstem and network analysis functions provided by igraph,tidygraph and ggraph. Pipe %>% has been exported from magrittr and could be used directly in akc.

Logo of akc package.

Logo of akc package.

Example

Load package and inspect data

# load pakcage
library(akc)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

# inspect the built-in data
bibli_data_table
#> # A tibble: 1,448 × 4
#>       id title                                                   keyword abstr…¹
#>    <int> <chr>                                                   <chr>   <chr>  
#>  1     1 Keeping the doors open in an age of austerity? Qualita… Auster… "Engli…
#>  2     2 Comparison of Slovenian and Korean library laws         Compar… "This …
#>  3     3 Analysis of the factors affecting volunteering, satisf… Contin… "This …
#>  4     4 Redefining Library and Information Science education a… Curric… "The p…
#>  5     5 Can in-house use data of print collections shed new li… Check-… "Libra…
#>  6     6 Practices of community representatives in exploiting i… Commun… "The p…
#>  7     7 Exploring Becoming, Doing, and Relating within the inf… Librar… "Profe…
#>  8     8 Predictors of burnout in public library employees       Emotio… "Work …
#>  9     9 The Roma and documentary film: Considerations for coll… Academ… "This …
#> 10    10 Mediation effect of knowledge management on the relati… Job pe… "This …
#> # … with 1,438 more rows, and abbreviated variable name ¹​abstract
#> # ℹ Use `print(n = ...)` to see more rows

The data set contains bibliometric data on topic of “academic library”,it is a data.frame of 4 columns(with docuent ID,article title,keyword and abstract), more information could be found via ?bibli_data_data.If the user want to carry out tasks by simply copying the example codes,make sure to arrange the data in the same format as biblio_data_table and set the same names for the corresponding columns.

Keyword cleaning

The entire cleaning processes include: 1.Split the text with separators; 2.Reomve the contents in the parentheses (including the parentheses); 3.Remove whitespaces from start and end of string and reduces repeated whitespaces inside a string; 4.Remove all the null character string and pure number sequences; 5.Convert all letters to lower case; 6.Lemmatization (not in default setting because it is not recommended unless you need a relatively rough result. For better merging, use keyword_merge displayed below).

bibli_data_table %>% 
  keyword_clean() -> clean_data

clean_data
#> # A tibble: 5,378 × 2
#>       id keyword                          
#>    <int> <chr>                            
#>  1     1 austerity                        
#>  2     1 community capacity               
#>  3     1 library professional             
#>  4     1 public libraries                 
#>  5     1 public service delivery          
#>  6     1 volunteer relationship management
#>  7     1 volunteering                     
#>  8     2 comparative librarianship        
#>  9     2 korea                            
#> 10     2 library legislation              
#> # … with 5,368 more rows
#> # ℹ Use `print(n = ...)` to see more rows

Keyword merging

Merge keywords that have common stem or lemma, and return the majority form of the word.

clean_data %>% 
  keyword_merge() -> merged_data

merged_data
#> # A tibble: 5,372 × 2
#>       id keyword                      
#>    <int> <chr>                        
#>  1  1163 10.7202/1063788ar            
#>  2   619 18th century                 
#>  3  1154 1password                    
#>  4    81 1science                     
#>  5   361 second-career librarianship  
#>  6   662 second life                  
#>  7  1424 2016 us presidential election
#>  8    42 21st-century skills          
#>  9  1114 21st century skills          
#> 10  1051 24-hour opening              
#> # … with 5,362 more rows
#> # ℹ Use `print(n = ...)` to see more rows

Keyword grouping

Create a tbl_graph(a class provided by tidygraph package) from the tidy table with document ID and keyword. Each entry(row) should contain only one keyword in the tidy format.

merged_data %>% 
  keyword_group() -> grouped_data

grouped_data
#> # A tbl_graph: 207 nodes and 1332 edges
#> #
#> # An undirected simple graph with 1 component
#> #
#> # Node Data: 207 × 3 (active)
#>   name                  freq group
#>   <chr>                <int> <int>
#> 1 information literacy    58     2
#> 2 academic libraries     145     2
#> 3 archives                12     1
#> 4 open access             32     3
#> 5 bibliometrics           31     3
#> 6 higher education        16     1
#> # … with 201 more rows
#> #
#> # Edge Data: 1,332 × 3
#>    from    to     n
#>   <int> <int> <int>
#> 1     1    97    14
#> 2     1     2    12
#> 3     2    14     8
#> # … with 1,329 more rows

Output the table of results

The output table would show the top 10 keywords (by occurrence) and their frequency. Keywords are separated by “;”.

grouped_data %>% 
  keyword_table(top = 10)
#> # A tibble: 5 × 2
#>   Group `Keywords (TOP 10)`                                                     
#>   <int> <chr>                                                                   
#> 1     1 public libraries (74); libraries (65); digital libraries (31); library …
#> 2     2 academic libraries (145); information literacy (58); librarians (25); l…
#> 3     3 open access (32); bibliometrics (31); library and information science (…
#> 4     4 university libraries (39); collection management (13); leadership (12);…
#> 5     5 social media (23); spain (9); sustainability (6); disinformation (5); f…

Visualize the results

Keyword co-occurrence network in different groups. Colors are used to specify the groups, the size of nodes is proportional to the keyword frequency, while the alpha of edges is proportional to the co-occurrence relationship between keywords.

grouped_data %>% 
  keyword_vis()

Keyword extraction from abstract

To extract keywords from the abstract using the keywords as a dictionary. More pre-processing filter should be implemented afterward, such as cleaning, keyword merging and filtering by term frequency or tf-idf. It is suggested to keep the size down before using keyword_group.

bibli_data_table %>%
  keyword_clean(id = "id",keyword = "keyword") %>%
  pull(keyword) %>%
  make_dict -> my_dict

bibli_data_table %>%
  keyword_extract(id = "id",text = "abstract",dict = my_dict) %>%
  keyword_merge(keyword = "keyword")
#> # A tibble: 27,130 × 2
#>       id keyword                      
#>    <int> <chr>                        
#>  1   619 18th century                 
#>  2  1223 18th century                 
#>  3  1154 1password                    
#>  4    81 1science                     
#>  5   983 1science                     
#>  6    15 2016 us presidential election
#>  7   662 3d environment               
#>  8   910 fourth museum assembly       
#>  9   624 55th library week            
#> 10   747 aasl standards               
#> # … with 27,120 more rows
#> # ℹ Use `print(n = ...)` to see more rows

END