DocumentCode :
35428
Title :
Self-adaptive topic model: A solution to the problem of ???rich topics get richer???
Author :
Ying Fang ; Heyan Huang ; Ping Jian ; Xin Xin ; Chong Feng
Author_Institution :
Sch. of Comput. Sci. & Technol., Beijing Inst. of Technol., Beijing, China
Volume :
11
Issue :
12
fYear :
2014
fDate :
Dec. 2014
Firstpage :
35
Lastpage :
43
Abstract :
The problem of “rich topics get richer” (RTGR) is popular to the topic models, which will bring the wrong topic distribution if the distributing process has not been intervened. In standard LDA (Latent Dirichlet Allocation) model, each word in all the documents has the same statistical ability. In fact, the words have different impact towards different topics. Under the guidance of this thought, we extend ILDA (Infinite LDA) by considering the bias role of words to divide the topics. We propose a self-adaptive topic model to overcome the RTGR problem specifically. The model proposed in this paper is adapted to three questions: (1) the topic number is changeable with the collection of the documents, which is suitable for the dynamic data; (2) the words have discriminating attributes to topic distribution; (3) a self-adaptive method is used to realize the automatic re-sampling. To verify our model, we design a topic evolution analysis system which can realize the following functions: the topic classification in each cycle, the topic correlation in the adjacent cycles and the strength calculation of the sub topics in the order. The experiment both on NIPS corpus and our self-built news collections showed that the system could meet the given demand, the result was feasible.
Keywords :
pattern classification; text analysis; ILDA; NIPS corpus; RTGR problem; adjacent cycles; automatic resampling; infinite LDA; latent Dirichlet allocation; rich topics get richer; self-adaptive topic model; self-built news collections; standard LDA model; strength calculation; topic classification; topic correlation; topic evolution analysis system; topic number; Adaptation models; Big data; Computational modeling; Data models; Integrated circuit modeling; Resource management; Tagging; Dirichlet process; infinite Latent Dirichlet Allocation; topic evolution; topic model;
fLanguage :
English
Journal_Title :
Communications, China
Publisher :
ieee
ISSN :
1673-5447
Type :
jour
DOI :
10.1109/CC.2014.7019838
Filename :
7019838
Link To Document :
بازگشت