Arabic text summarization via Knapsack balancing of effective retention
Alaidine Ben Ayed, Ismaïl Biskri, Jean-Guy Meunier · Procedia Computer Science · 2021
This paper presents a new extractive Arabic text summarization approach based on the Knapsack balancing of effective retention. Effective retention refers to maximizing retention while prioritizing salient themes. First, text segments are mapped to salient topics, and an effective retention score is computed for every sentence. Next, the summarization task is formulated as a combinatory optimization problem. The final output is generated through the maximization of effective retention. Aka selected sentences to be part of the summary are the best ones to encode salient ideas while covering most central themes. Experimental results show that the proposed approach outperforms three state-of-the-art Arabic summarization protocols.