New tools enable literature retrieval to enter AI mode

Science and Technology Daily, Beijing, May 30th (Reporter He Liang) Searching and reading literature is a basic task in scientific research. According to statistics, researchers spend about 51% of their research time searching and digesting scientific and technological data

Science and Technology Daily, Beijing, May 30th (Reporter He Liang) Searching and reading literature is a basic task in scientific research. According to statistics, researchers spend about 51% of their research time searching and digesting scientific and technological data. Is there a possibility to turn literature into a knowledge base or database, using artificial intelligence methods to reduce the "burden" of researchers searching and reading literature? On May 30th, at the 2023 Zhongguancun Forum "Artificial Intelligence Driven Scientific Research Forum", the ScienceNavigator (hereinafter referred to as the Literature Knowledge Base), a literature knowledge base based on large language models and vector databases, was officially released.

This is a research achievement that allows researchers to conduct literature retrieval, reading, analysis, and management through dialogue and questioning. This achievement was jointly developed by Beijing Institute of Scientific Intelligence, Computer Network Information Center and Mochi Technology.

From the earliest 'search based' retrieval based on eye search and hand flipping, to later 'search based' retrieval based on search engines and the Internet, and now to the leapfrog development of artificial intelligence technology, we have seen for the first time that big language models approach the level of human intelligence in understanding problems and answering questions. "Meng Zhuofei, Vice President of Mocky Technology, said that the release of literature knowledge bases coincides with the development trend of the retrieval model entering the era of dialogue.

The performance advantages of literature knowledge bases can be described by the four words' multiple, fast, good, and economical '. Meng Zhuofei introduced that' multiple 'is reflected in' multimodal, multi model, and multi data '; Fast "refers to" fast query, fast import, and fast iteration "; Good "is reflected in" more real-time data, more reliable citation, and more professional understanding "; The "province" approach significantly reduces the computational cost of data through extreme system optimization and self-developed vector algorithms.

The development direction of literature knowledge base is to incorporate more experimental data into vector databases. At that time, the design principles, experimental methods, experimental conclusions, and corresponding thinking behind the conclusions involved in scientific experiments can all be used as query targets. Meng Zhuofei stated that with the help of large models and vector databases, researchers can propose directional problems. The machine will complete a set of processes such as disassembling problems, questioning, designing experiments, and simulating experiments. It can even reflect, deduce, and iterate problems based on the results, further freeing up the time and energy of researchers to devote themselves to solving key problems and innovative thinking.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])