Scholars have an effective way of informing colleagues of lengthy papers that they find not worth their time to read.
They tag such documents “tl;dr” which means too long, didn’t read. The Allen Institute for AI in Seattle has taken both opinions to heart and the current week revealed a system that offers a summary of lengthy computer-science reports to cut the time it takes to review such literature. A research tool known as a semantic scholar by Artificial Intelligence is used for scientific research. It got a summarization feature which surveys massive numbers of a scientific paper and reduces them to one-sentence summaries. There are nearly 10M computer-science papers in Semantic Scholar’s database. According to Dan Weld, who supervises the database commented that papers from other disciplines will gradually be added. The system offers a great use to researchers who have to rely on numerous scanning titles and lengthy abstracts. Dan Weld mentioned people like the system.
Semantic Scholar achieved the largest compression rate as compared to all summarizing tools. As for a Scientific paper averaging are 5,000 words and for Semantic Scholar summaries are around 21 words, which averages to summaries 1/238th the report size. The closest Semantic Scholar competitor reduces documents to 1/36th of the report size.
So many varieties of Natural Language Processing programs have been developed over the years to summarize documents. They use one of two approaches, the extractive approach that focuses on selecting representative text and using it verbatim in the summary. For example, Paper Digest that developed in 2018, appears to extract sentences rather than rewriting decisions in its own words. The second approach is abstractive that uses natural language generation algorithms to generate summaries with the original wording. This approach becomes a favorite among programmers as improvements in Artificial Intelligence natural language generation is going on.
According to Jevin West, an information scientist at the University of Washington in Seattle who tested the new program mentioned that he predicts that this kind of tool will become a standard feature of scholarly search in the coming future.
