An Adaptive Information Retrieval framework for the applications on Tweets and Tropical Moisture Cases

PhD Thesis Proposal Defence


Title: "An Adaptive Information Retrieval framework for the applications
on Tweets and Tropical Moisture Cases"

by

Mr. Xiaotian HAO


Abstract:

We propose an adaptive information retrieval (IR) framework which consists 
of the whole working pipeline. The pipeline includes the procedures of 
problem formulation, background and related work collecting, data driven 
framework development, experimental study and analysis. This framework can 
be adaptively applied in different research domains to solve corresponding 
research problems.

This thesis focus on presenting 2 information retrieval work following the 
proposed framework as application case studies: Towards a Query-less News 
Search Framework and Diagnosis of the Tropical Moisture Exports to the 
Mid-Latitudes the Role of Atmospheric Steering in the Extreme 
Precipitation. In those works, the designed adaptive information retrieval 
framework performed high efficiency and effectively on solving the 
corresponding research problems.

In Tweets related IR work Towards a Query-less News Search Framework, we 
found that given user’s interest in a particular news-related tweet, 
searching for related content may be a tedious process. Formulating an 
effective search query is not a trivial task and may require query 
reformulations. And due to the often small size of smart phone screens, 
instead of typing, users always prefer click-based operations to retrieve 
related content. To address these issues, we introduce a new paradigm for 
news-related Twitter search called Search by Tweet(SbT). In this paradigm, 
a user submits a particular tweet which triggers a search task to retrieve 
further related tweets. We formalize the SbT problem and propose an 
effective and efficient framework implementing such a functionality. At 
the core, we model the public Twitter stream novelly as a dynamic 
graph-of-words, reflecting the importance of both words and word 
correlations. We further develop a streaming clustering algorithm to 
discover dense word clusters in the graph, corresponding to trending 
topics. Given an input tweet, our framework utilizes the graph model and 
word clusters to generate an implicit query.

In Tropical Moisture related IR work Diagnosis of the Tropical Moisture 
Exports to the Mid-Latitudes the Role of Atmospheric Steering in the 
Extreme Precipitatio, three river basins, i.e., the Yangtze river, the 
Mississippi river and the Loire river, were presented as case studies to 
explore the association among atmospheric circulations, moisture exports 
and extreme precipitations in the mid-latitudes. The major moisture source 
regions in the tropics for the three river basins are first identified 
using the Tropical Moisture Exports (TMEs) dataset. We first retrieve the 
space-time characteristics of their respective moisture sources. Then, the 
trajectory curve clustering analysis is applied to the TMEs tracks 
originating from the identified source regions during each basin’s peak 
TMEs activity and flood seasons. Our results show that the moisture tracks 
for each basin can be categorized into 3 or 4 clusters with distinct 
spatial trajectory features. Our further analysis on these clustered 
trajectories reveals that the contributions of moisture release from 
different clusters are associated with their trajectory features and 
travel speeds. In order to understand the role of associated atmospheric 
steering, daily composites of the geopotential heights anomalies data and 
the vertical integral of moisture flux anomalies data from 7 days ahead to 
the extreme precipitation days (top 5%) are retrieved and examined. The 
evolutions of the atmospheric circulation patterns and the moisture fluxes 
are both consistent with the TMEs tracks that contribute more moisture 
releases to the study regions. The findings imply that atmospheric 
steering plays an important role in the moisture transport and release, 
especially for the extreme precipitations.


Date:			Friday, 25 June 2021

Time:                  	2:00pm - 4:00pm

Zoom Meeting: 
https://hkust.zoom.us/j/9892118645?pwd=ZjJLbVpMUHdQOTgyYW45YVVFZ0tQQT09

Committee Members:	Dr. Wilfred Ng (Supervisor)
 			Prof. Dik-Lun Lee (Chairperson)
 			Prof. Qiong Luo
 			Dr. Mengqian Lu (CIVL)


**** ALL are Welcome ****