< Back
Posted: 08 Jan 2022 03:00

“Extractive Summarization” January 2022 — summary from Arxiv and Crossref

Brevi Assistant
Brevi Assistant

Business performance assistant

“Extractive Summarization” January 2022 — summary from Arxiv and Crossref main image

The content below is machine-generated by Brevi Technologies’ NLG model, and the source content was collected from open-source databases/integrate APIs.


Arxiv - summary generated by Brevi Assistant


Text summarization techniques have attracted attention regularly. Nonetheless, the majority of the current text summarization methods based upon deep learning require large datasets, which is hard to achieve in sensible applications. Based on the routed graph algorithm, we change the standard technique of determining the sentence ranking at one time to multi-round calculation, and the summary sentences are dynamically enhanced after each round of calculation to better suit the characteristics of the text.

The transformer multi-head self-attention mechanism has been thoroughly examined just recently. On the other hand, they suggest new focus enhancement techniques to make transformers a lot more precise, reliable and interpretable. After locating 3 meaningful interest patterns in the popular BERTSum model, experiments indicate that when we infuse such patterns, both the original and the smaller sized model show improvements in efficiency and perhaps interpretability.

Document summarization supplies a tool for faster recognizing the collection of text records and has several real-life applications. Nonetheless, the existing extractive summarization models pay little attention to and use the central subject info to aid the generation of summaries, which causes models not ensure the generated summary under the main topic. Particularly, a neural topic model is included in the neural-based sentence-level representation learning to effectively take into consideration the central topic details for recording the crucial content in the original document.


Source texts:



Crossref - summary generated by Brevi Assistant


Text summarization has a vital duty in natural language processing. Research on text summarization in Indonesian Language is still unusual and not examined comprehensively. The function of this research study is to supply an automatic text summarization experience utilizing the extractive technique within an application executed on Android. The code execution using Flutter Software Development Kit on client-side and Python with Flask on server-side is intended to have far better assimilation and maintenance procedure. Presently, nearly all enterprises are oriented into building text information in wealth savoring the benefits of big data idea However, the fact is that it's not practically feasible to experience all this data/documents for decision making due to the time restraint. By embracing these summarizing strategies, the accuracy in data access of summarized content via search queries can be boosted contrasted to performing search over the broad series of initial textual content.

This paper provides a query-based extractive text summarization approach by utilizing sense-oriented semantic relatedness action. It aids in drawing out question importance sentences while determining the sense-oriented sentence semantic relatedness score in between the query and input text sentence.


This can serve as an example of how to use Brevi Assistant and integrated APIs to analyze text content.


Source texts:


logo

The Brevi assistant is a novel way to summarize, assemble, and consolidate multiple text documents/contents.

Partners:

© All rights reserved 2022 made by Brevi Technologies