DeepSeek Introduces Patent for Reducing Network Resource Usage in Data Acquisition
3 April 2025 路 Uncategorized 路
Source: 路 https://technews.tw/2025/04/02/deepseek-new-patent/
According to Chinese media reports, the 'Method and System for Broad Data Retrieval' patent applied by Hangzhou Depth Seek Artificial Intelligence Basic Technology Research Co., Ltd. (a related company of China's AI startup DeepSeek) was published on April 1st according to information from the State Intellectual Property Office鈥檚 Patent Publication Announcement website in China.
The abstract highlights this invention, which aims at discovering as many web links as possible while minimizing traffic impact on websites; analyzing downloaded content and inferring quality for undownloaded links through prioritized downloading allocation methods. This reduces low-quality webpage downloads and redundant ones, thereby enhancing data quality and download efficiency while reducing network resource consumption during the acquisition process.
The patent also mentions that numerous large language models (LLMs) are trained to be applied in natural language processing fields aimed at researching theories and methodologies for effective communication between humans and computers using natural languages. Existing data retrieval techniques face several issues, such as incomplete link collection from complex websites; excessive downloading leading to website crashes; lack of content quality analysis before download causing redundant or low-quality downloads affecting efficiency.
Furthermore, DeepSeek was first used in international earthquake rescue efforts recently during the strong quake relief operation in Myanmar where a Chinese team-developed Sino-Burmese-English translation system played an important role. After the earthquake occurred, China's Embassy to Burma stated that they utilized this emergency developed tri-language translation system based on DeepSeek for their rescue work.
(Translated with permission from MoneyDJ News; Image source: Unsplash)
The abstract highlights this invention, which aims at discovering as many web links as possible while minimizing traffic impact on websites; analyzing downloaded content and inferring quality for undownloaded links through prioritized downloading allocation methods. This reduces low-quality webpage downloads and redundant ones, thereby enhancing data quality and download efficiency while reducing network resource consumption during the acquisition process.
The patent also mentions that numerous large language models (LLMs) are trained to be applied in natural language processing fields aimed at researching theories and methodologies for effective communication between humans and computers using natural languages. Existing data retrieval techniques face several issues, such as incomplete link collection from complex websites; excessive downloading leading to website crashes; lack of content quality analysis before download causing redundant or low-quality downloads affecting efficiency.
Furthermore, DeepSeek was first used in international earthquake rescue efforts recently during the strong quake relief operation in Myanmar where a Chinese team-developed Sino-Burmese-English translation system played an important role. After the earthquake occurred, China's Embassy to Burma stated that they utilized this emergency developed tri-language translation system based on DeepSeek for their rescue work.
(Translated with permission from MoneyDJ News; Image source: Unsplash)