CORPUS

The UNL EOLSS Interface processes metadata from the EOLSS as of July 2006. Our corpus included:

• Titles (29,411 entries) and their corresponding URLs
• Keywords (56,148 entries) and their corresponding definitions (if available)
• Authors (4,871 entries) and their corresponding biographical sketches
• Institutions (1,870 entries)
• Cities (1,023 entries)
• Countries (101 entries)

The corpus was automatically downloaded from the EOLSS website and semi-automatically represented as a relational database comprising 13 tables:

• TOC = Table of Contents (Titles) and their corresponding URLs
• Words = Keywords
• Authors = Authors and their corresponding biographical sketches
• Institutions
• Cities
• Countries
• Definition: keywords and their corresponding definitions
• AI: Authors x Institutions
• CN: Countries x Nations
• IC: Institutions x Countries
• TA: Titles x Authors
• TT: Titles x Titles
• WT: Words x Titles

Due to many inconsistencies in the EOLSS, the process of populating the UNL EOLSS Database was not a trivial task. There were found many divergent representations for the same concepts: University of Tokyo x Tokyo University; Bagrov Alexei Mikhailovich x Bagrov Alexey Mikhailovich; Berman Cilingir Kayis x Berman C. Kayis; and so on. In many cases, authors were associated to specific departments; in others, to the whole university.

The resulting database can be downloaded from here (Microsof Access 2003 database).