Loading...

Automatic Detection and Disambiguation of Metaphorical and Metonymic Concepts

Abdi Ghavidel, Hadi | 2013

463 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: English
  3. Document No: 44258 (31)
  4. University: Sharif University of Technology
  5. Department: Languages and Linguistics Center
  6. Advisor(s): Bahrani, Mohammad; Khosravizadeh, Parvaneh; Vazirnejad, Bahram
  7. Abstract:
  8. Disambiguating concepts in language is one of the most challenging tasks in the field of natural language processing and computational linguistics. A sort of these ambiguities can be found in metaphorical and metonymic concepts. In the present study, we try to introduce the structure of the detecting and disambiguating system of metaphor and metonymy. Detection step is based on semantic valency of the verbs occurring within the subject and object positions and the disambiguation step on the key topical concepts. The data of the detection step involves semantic valency of the seven metaphorically and metonymically frequent verbs. First of all, the data were collected by extracting the frequent verbs from metaphor and metonymy data-bank called PerMett and then their semantic valencies were specified in the annotated Bijan-Khan Corpus in a supervised mode. Therefore, in the case of observing any anomalous semantic valencies occurring in subjects and objects positions, the system represents it as a metaphorical one. In the disambiguation part, the system is based on the key words occurred in 66 topics included in Bijan-Khan corpus. At first, all of the words related to each topic were categorized through calculating tf-idf and finally 100 most significant ones were chosen for the disambiguation phase. The topically incongruent word in accordance with the general topic of the sentences or the topic of the other words was used as general representation of metaphor in a sentence. The part of disambiguating the metonymic concepts functions precisely according to the part to whole or container and contained relation. This system which is known as Estejaz and identifies metaphorical concepts with the precision of 85% and interprets them with the precision of 43%, promises the extraction the semantic valency of all the verbs in Persian and also solving the meaning ambiguity problem in this language
  9. Keywords:
  10. Metaphor ; Detection ; Semantic Disambiguation ; Metonymy Conceps ; Disambiguation ; Semantic Valency

 Digital Object List

 Bookmark

No TOC