Users rely increasingly on online reviews, forums and blogs to exchange
information, practical tips, and stories. However, such user-authored
content is in a free text format, usually with very scant structured
metadata information, and is therefore difficult for computers to
understand, analyze, and aggregate. Users often face the daunting task
of accessing and reading a large quantity of text to discover
potentially useful information. This work addresses the need to
automatically leverage useful information from a large quantity of
user-authored text to improve search and to provide personalized
recommendations.
In my research, I first identify topical and sentiment information from
free-form text, and use these to make accurate text-based
recommendations and to make fine-grained predictions of user sentiments
towards product features. Secondly, I learn similarities between online
users via multiple indicators like similar information needs, user
profiles, or topics of interest to predict future social interactions
and to improve ranking in keyword search. I then propose an efficient
search over user authored content at multiple levels of granularity to
enable finding the precise information without having to read a large
quantity of text.