Fast Gaussian Process Preference Elicitation via Expectation Propagation and How to apply query expansion to capture and relate concepts and trends in social media
Mark Norrish and Fran Su (ANU)
NICTA SML SEMINARDATE: 2011-01-27
TIME: 11:00:00 - 12:00:00
LOCATION: NICTA - 7 London Circuit
CONTACT: JavaScript must be enabled to display this email address.
ABSTRACT:
Title: Fast Gaussian Process Preference Elicitation via Expectation Propagation
Summer research scholar: Mark Norrish, ANU
Supervisor: Edwin Bonilla, NICTA CRL
----------------------------------------
Title: How to apply query expansion to capture and relate concepts and trends in social media.
Summer research scholar: Frank Su, ANU
Supervisors: Hanna Suominen & Leif Hanlen, NICTA CRL
Abstract: With the rapid growth in the popularity of social media and micro-blogs, health information users are overwhelmed by the amount of information. The focus of this study is on Twitter due to its popularity. However, monitoring tweets and finding relevant concepts from Twitter in person can be time-consuming. This research project aims to make information search faster and more accurate by expanding a query and finding the most related tweets. Our approach is to develop a Twitter search engine and discover the trend in health information from the tweets that are streamed and stored locally. In this 20-minute session, I will explain the methods, challenges and remedy in this study; then, I will discuss the implementation and experiments. The results of query expansion on tweets show that it is stable when the size of documents grows. However, the data set must be large enough to ensure the reliability and confidence that a certain trend of keywords related to a given query is emerging.
BIO:
