Working with Proper Nouns in Search Queries
Queries often contain proper nouns. Idilia’s Named Entity Recognizer (NER) will identify proper nouns in queries giving you more options for how you present your results:
- First proper nouns are rewritten so the original query matches more documents in your index
- Second, Idilia automatically link people in a query to their Twitter accounts and other entities to reference sources like Wikipedia
- Then info-boxes are created right on the results page for instant references
Recognizing and Rewriting Proper Nouns
Let’s say a user is looking for direction to JFK Airport and enters the query:
Idilia’s NER will identify the proper noun intended by “jfk” so it can be expressed it in its several known variations, greatly increasing the chances of matching to the best document in an index. Using Idilia’s Query Rewrite API generates the following rewrites (first 10):
|0.98||JFK Airport directions|
|0.98||JFK International directions|
|0.98||JFK International Airport directions|
|0.90||Kennedy Airport directions|
|0.90||John F. Kennedy Airport directions|
|0.90||John F. Kennedy International directions|
The proper noun originally expressed as “jfk” can now be expressed using Boolean logic as “(JFK or Kennedy or “John F. Kennedy”) and (Airport or International)”.
Integrating Info-Boxes and Twitter accounts with Search Results
The proprietary Idilia Language Graph links millions of proper nouns to reference sources such as Wikipedia, IMDB, MusicBrainz, and Twitter.
For example, in the query:
Idilia’s NER recognizes that “del potro” is the tennis player, Juan Martin del Potro:
As shown above, the Idilia’s Query Rewrite API references the Idilia Language Graph and locate the precise node for Juan Martin del Potro. The node contains links to a Wikipedia page and a Twitter account, indicated by the icons and .
These icons link to info-boxes that can be integrated into a search results page, so instead of just links, you can present rich results: