• 3958 university libraries in 43 countries and 3 languages ​​were evaluated.
  • The possibility of manually extracting the data was ruled out 100%, as this generates bias and subjectivity that eliminates the neutrality of the ranking.
  • A semantic analysis of the keywords was made, taking into account linguistic aspects of the different Latin American countries.
  • A structured language was built with the keywords that identified the presence of each variable evaluated in each of the 390 libraries evaluated in 2020.
  • Metadata were identified in the source code of the web pages of each library, which contained the keywords of the structured language.
  • A “robot” was built for the collection of unstructured data, which reads each web portal of the university libraries, extracts the keywords that identify each variable of each dimension, verifies the existence of the variable and assigns a compliance score, 1 if the library meets the variable and 0 if the library does not meet the variable.
  • For the construction of the data extraction robot, the Small Data methodology was used.
  • The MOREQ methodology for document management was taken into account as a methodological guide.
  • With the results of the semantic analysis, the knowledge base is built to tag the keywords that would be searched for in each of the 3,958 libraries evaluated in 2021.
  • The robot was modeled and created in January 2021 and refined 11 times, to the highest quality of data collection.
  • The web robot was executed 3 times, during the months of June, July and August 2021, to verify the quality of the information, resulting in a reliability of 99.7% in the data collected from the 3958 libraries.