Loading...

Persian Abstractive Summarization using Graph-based Abstract Meaning Representation

Haddadan, Shohreh | 2017

585 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 49467 (31)
  4. University: Sharif University of Technology
  5. Department: Languages and Linguistics Center
  6. Advisor(s): Bahrani, Mohammad
  7. Abstract:
  8. This study attempts to introduce a novel approach to abstractive summarization in Persian. According to the methodology the first step is to represent input text sentences into an abstract meaning representation structure. This representation is syntax free thus, it helps the summarization system to represent sentences more semantic based and free of the sentence syntactic structure. In order to select suitable content for the summary output semantic and structural features are extracted from the representation. Data used in this research consists of approximatelty 200 senctences summarized in 30 sentences of a famous story book named: ”The little prince”. An SVM is trained on 80% of train-data to assign weights to the extracted features. The precision percent achieved for classifying nodes and edges of into two classes of present-absent in summary graph is 66 and 89.82 percent recprectively. The next step is to choose an optimal subgraph from the aggregated graph structre of the input text. A linear function uses sums of features times their weights to assign a score to each possible subgraph. The efficiency of this method is estimated by means of Precision and Recall of the concepts in the summary subgraph.A precision of 70.04 And recall of 75.23 is gained for the output from a 10-fold cross-validation evaluation with compression rate of 50%
  9. Keywords:
  10. Feature Extraction ; Abstractive Summarization ; Abstract Meaning Representation ; Optimal Subgraph ; Weighting Features

 Digital Object List

 Bookmark

No TOC