|
Title:
|
INCREMENTAL LEARNING OF RELATIONS FROM THE MOST FREQUENT PATTERNS IN CONVERSATIONS ON MICROBLOGGING SERVICES |
|
Author(s):
|
Sijin He, Yike Guo, Moustafa Ghanem |
|
ISBN:
|
978-972-8939-23-6 |
|
Editors:
|
António Palma dos Reis and Ajith P. Abraham |
|
Year:
|
2010 |
|
Edition:
|
Single |
|
Keywords:
|
Twitter, Relation Extraction, Microblogging, Social Media |
|
Type:
|
Full Paper |
|
First Page:
|
37 |
|
Last Page:
|
44 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Microblogging is a popular form of communication in which users can instantly post messages to express their feelings and opinions on the Internet. Those messages often contain valuable structured data that is hidden in regular English sentences. Extracting patterns from microblogging data is useful for different purposes, such as sentiment analysis, opinion mining and relation extraction. In this paper, we explore the power of surface text patterns for relation extraction on the microblogging services. The text patterns we are looking for are the patterns occurring frequently in conversational messages. Our approach is inspired by two bootstrapping-based approaches, DIPRE and Snowball, which automatically learn text patterns and relations. In order to obtain an optimal set of patterns and an accurate set of relations, we define some evaluation criteria for generated patterns and relations based on the properties of microblogging data. The proposed approach is used for extracting relations, such as interpersonal relations, feelings, etc. from messages on microblogging services. Apart from extracting relations, we also propose an incremental learning model that constantly learns relations from users conversations whenever a new message is presented in the conversations. |
|
|
|
|
|
|
