Weakly Supervised Learning in Text Mining: Utilizing Natural Supervision from Text and Graph Data

PhD Thesis Proposal Defence


Title: "Weakly Supervised Learning in Text Mining: Utilizing Natural 
Supervision from Text and Graph Data"

by

Miss Ziqian ZENG


Abstract:

Currently, supervised learning based methods and techniques such as deep 
learning methods, have achieved great success in the text mining area. When 
researchers develop these giant models, they usually assume the availability of 
massive annotated training data. However, the real-world usefulness of these 
models will be impaired because in the real world, readily available annotated 
data are scarce. Natural supervisions are inexpensive and readily available 
resources. We show how to utilize natural supervision in text and graph data in 
the sentiment analysis, text classification, and the personalized word 
embeddings task. For example, for sentiment analysis, natural supervision can 
be opinion words in the text. For personalized word embeddings learning, 
natural supervision can be corpora from users’ friends via a social graph. To 
utilize supervision signals from text data,  we propose a variational 
weakly-supervised framework for the sentiment analysis task and the text 
classification task. To utilize supervision signals from graph data, we propose 
to utilize social networks as regularization in learning personalized word 
embeddings.


Date:			Thursday, 1 April 2021

Time:                  	2:00pm - 4:00pm

Zoom Meeting:
https://hkust.zoom.us/j/91890133178?pwd=K2poVzROaTF2dHFCelBsUEI3MEZvQT09

Committee Members:	Dr. Yangqiu Song (Supervisor)
  			Prof. Xiaofang Zhou (Chairperson)
 			Prof. Dit-Yan Yeung
 			Prof. Nevin Zhang


**** ALL are Welcome ****