• Journal of Information Processing Systems
• /
• v.16 no.4
• /
• pp.820-831
• /
• 2020

Extractive document summarization aims to select a few sentences while preserving its main information on a given document, but the current extractive methods do not consider the sentence-information repeat problem especially for news document summarization. In view of the importance and redundancy of news text information, in this paper, we propose a neural extractive summarization approach with joint sentence semantic clipping and selection, which can effectively solve the problem of news text summary sentence repetition. Specifically, a hierarchical selective encoding network is constructed for both sentence-level and document-level document representations, and data containing important information is extracted on news text; a sentence extractor strategy is then adopted for joint scoring and redundant information clipping. This way, our model strikes a balance between important information extraction and redundant information filtering. Experimental results on both CNN/Daily Mail dataset and Court Public Opinion News dataset we built are presented to show the effectiveness of our proposed approach in terms of ROUGE metrics, especially for redundant information filtering.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.12
• /
• pp.4275-4291
• /
• 2021

With the unprecedented growth of textual information on the Internet, an efficient automatic summarization system has become an urgent need. Recently, the neural network models based on the encoder-decoder with an attention mechanism have demonstrated powerful capabilities in the sentence summarization task. However, for paragraphs or longer document summarization, these models fail to mine the core information in the input text, which leads to information loss and repetitions. In this paper, we propose an abstractive document summarization method by applying guidance signals of key sentences to the encoder based on the hierarchical encoder-decoder architecture, denoted as KI-HABS. Specifically, we first train an extractor to extract key sentences in the input document by the hierarchical bidirectional GRU. Then, we encode the key sentences to the key information representation in the sentence level. Finally, we adopt key information representation guided selective encoding strategies to filter source information, which establishes a connection between the key sentences and the document. We use the CNN/Daily Mail and Gigaword datasets to evaluate our model. The experimental results demonstrate that our method generates more informative and concise summaries, achieving better performance than the competitive models.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.4
• /
• pp.104-115
• /
• 1993

The motion compensated coding (MCC) technique, which exploits the temporal redundancies in the moving images with the motion estimation technique,is one of the most popular techniques currently used. Recently, a variable block size(VBS) motion estimation scheme has been utilized to improve the performance of the motion compensted coding. This scheme allows large blocks to the used when smaller blocks provide little gain, saving rates for areas containing more complex motion. Hence, a new VBS motion estimation scheme with a hierarchical structure is proposed in this paper, in order to combine the motion vector coding technique efficiently. Topmost level motion vector, which is obtained by the gain/cost motion estimation technique with selective motion prediction method, is always transmitted. Thus, the hierarchical VBS motion estimation scheme can efficiently exploit the redundancies among neighboring motion vectors, providing an efficient motion vector encoding scheme. Also, a restricted search with respect to the topmost level motion vector enables more flexible and efficient motion estimation for the remaining lower level blocks. Computer simulations on the high resolution image sequence show that, the VBS motion estimation scheme provides a performance improvement of 0.6~0.7 dB, in terms of PSNR, compared to the fixed block size motion estimation scheme.