TY  - CONF
ID  - SisLab2980
UR  - https://www.cicling.org/2018/accepted.html
A1  - Ngo, Thi Lan
A1  - Vu, Van Tu
A1  - Hideaki, Takeda
A1  - Pham, Bao Son
A1  - Phan, Xuan Hieu
Y1  - 2018/03//
N2  - Suggestion analysis of opinion data is classifying a given utterance into one of two classes: suggestion and non-suggestion. In this paper, we introduce a new method, called LLMaxent, to cross-domain suggestion classification. LLMaxent is an approach to lifelong machine learning using maximum entropy (Maxent) method. Based on the main idea of lifelong learning, that is retaining the knowledge learned from past tasks and using it to help future learning, we build a classifier can use labeled data in existed domains for suggestion classification in a new
domain. The experimental results show that the proposed novel model
can improve the performance of cross-domain suggestion classification. This is the first study of lifelong machine learning using Maxent and our method is not only useful for suggestion classification but also for cross-domain text classification in general.
TI  - Lifelong Learning MaxEnt for Suggestion Classification
M2  - Hanoi, Vietnam
AV  - none
T2  - 19th International Conference on Computational Linguistics and Intelligent Text Processing
ER  -