The goal of this project is to implement libraries for accurate handling of place adjectives and demonyms for applications that use Lucene or Lucene-based search systems like Solr and ElasticSearch.
Place Adjectives are adjectival forms of place names, used in sentences to indicate something of that place. For example, Australian for Australia is used in a sentence like "The Australian weather is wonderful."
Demonyms are words used for people of a certain nationality or ethnicity, like Americans, Indians, or Italians.
Sometimes, place adjectives and demonyms are the same - for example, Australian - and sometimes different - for example, Swedish and Swedes.
Read these articles for more information on these concepts:
-
Applications sometimes want to map place adjectives or demonyms in search queries to their related places, or vice versa.
For example, a search query with an place adjective, like "swedish history", should match a title like "A History of Sweden".
Conversely, a search query with a place, like "sweden history", should match a title like "A History of medieval Swedish people".
The former is a stemming operation, while the latter is more of a synonimizing operation.
-
Unfortunately, popular English stemmers don't handle this mapping well. KStemmer is the only one that handles some cases, but not all. Some examples:
word Porter KStemmer English minimal Hunspell Snowball swedish 🚫 swedish ✅ sweden 🚫 swedish 🚫 swedish 🚫 swedish welsh 🚫 welsh 🚫 welsh 🚫 welsh 🚫 welsh 🚫 welsh indian 🚫 indian 🚫 indian 🚫 indian 🚫 indian 🚫 indian american 🚫 american ✅ america 🚫 american 🚫 american 🚫 american british 🚫 british ✅ britain 🚫 british 🚫 british 🚫 british None of them support the reverse operation - mapping place names to their adjectives or demonyms.