Web People Search (or how name ambiguity may compromise your online reputation)
Le 19 mai 2011 - 14h LINA - Bât 11 - Salle 3
Searching online information about people is a common activity of Internet
users, with user needs ranging from finding a phone number to acquiring a
complete profile of the online reputation of a person.
But typing a person’s name as a query has a crucial problem : name
ambiguity. Conventional search engines simply return a ranked list of
results, where different pages usually refer to different people sharing
the same name. Ideally, one would like to see results grouped into sets
that refer to the same individual, together with automatically extracted
profiles for each of these sets.
This task, known as Web People Search, has connections with several
problems in Artificial Intelligence, such as Word Sense
Disambiguation,Text Clustering and Information Extraction, Web Mining,
etc. We will review the advances made so far, and identify the remaining
In addition, the Web People Search problem will serve, in this
presentation, as a practical example of how difficult it is to properly
define a problem in Artificial Intelligence so as to enable
measurable,sound scientific progress on it.