Volume 7, Issue 2 (6-2015)                   2015, 7(2): 33-40 | Back to browse issues page

XML Print


Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Abdi Ghavidel H, Khosravizadeh P, Rahimi A. Impact of Topic Modeling on Rule-Based Persian Metaphor Classification and its Frequency Estimation. International Journal of Information and Communication Technology Research 2015; 7 (2) :33-40
URL: http://ijict.itrc.ac.ir/article-1-100-en.html
Abstract:   (2913 Views)
The impact of several topic modeling techniques have been well established in many various aspects of Persian language processing. In this paper, we choose to investigate the influence of Latent Dirichlet Allocation technique in the metaphor processing aspect and show this technique helps measure metaphor frequency effectively. In the first step, we apply LDA on Persian or so-called Bijankhan corpus to extract classes containing the words which share the most natural semantic proximity. Then, we develop a rule-based classifier for identifying natural and metaphorical sentences. The underlying assumption is that the classifier allocates a topic for each word in a sentence. If the overall topic of the sentence diverges from the topic of one of the words in the sentence, metaphoricity is detected. We run the classifier on whole the corpus and observed that roughly at least two and at most four sentence in the corpus carries metaphoricity. This classifier with an f-measure of 68.17% in a randomly 100 selected sentences promises that a LDA-based metaphoricty analysis seems efficient for Persian language processing.
Full-Text [PDF 1937 kb]   (1373 Downloads)    
Type of Study: Research | Subject: Information Technology

Add your comments about this article : Your username or Email:
CAPTCHA

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.