Natural Language Search On Facebook


This was the final project of my Artificial Intelligence class at Olin, taught by Prof. Lynn Stein.  I took the class in spring 2007 as a first-year.


Eric Hwang, Olin College '10


With the ability of computers to store vast quantities of information, accessing that information in an efficient manner is important. For low-level access to database information, the Server Query Language, or SQL, provides a powerful method of access. However, SQL takes time to learn and master, so most information access systems targeted towards people with no specialized knowledge use either keyword searches or form-based searches. Although not as powerful as specialized query languages like SQL, those interfaces are much more intuitive to use. An alternative method that would possess both the power of query languages and ease of use would be a natural language interface. Natural language interfaces allow users to pose questions to databases in a natural language such as English and have their queries answered according to the information in the database.

In this paper, we describe our simple natural language interface to the social networking site Facebook. We choose to use an external knowledge base as the basis of our system to allow for more attention to the problems of natural language processing and query evaluation without getting bogged down in data collection, entry, and maintenance. Additionally, using a sophisticated external knowledgebase such as Facebook allows us to gain easy access to a diverse, information-rich data source, which makes our queries all the more interesting and relevant.


PDF available for viewing and download below.
Mike Hughes,
Jan 13, 2009, 11:38 AM