Screening and Notification of News on Social Media (網路社群媒體的新聞篩選與通報)

In this service, we have built a fully automated news gathering and reporting module that enables the system to automate the following functions:

  1. Automated news gathering (authorized source data collection or automated online data collection)
  2. Automated keyword collection (including logic analysis)
  3. Automated similar news verification (self-built lexical vector analysis)
  4. Automated short URL conversion (self-built shortened URL system or serialized third-party system)
  5. Automated semantic analysis (self-built semantic vector analysis system or serialized third-party system)
  6. Automated web page screenshot
  7. Automated special field data extraction (such as main photos)
  8. Automated news summary extraction (such as a specific number of words or a specific segment)
  9. Automated data format conversion (in accordance with specific format or typesetting requirements)
  10. Automated scheduling and news notification (social media or traditional email)

In such a system design, all actions are targeted at unattended and minimal parameter settings, allowing the system to operate continuously with minimal human intervention. Since the launch of the service in August 2017, this system has been upgraded from a single server design to multi-server collaborative processing. Currently, the number of processed data is about 1 million articles per month. Since each news enters the system until the notification process is completed, the processing speed is continuously controlled within 3 minutes, and the average processing speed is maintained within 1 minute. The total amount of monthly news or public opinion of important news and forums in Taiwan falls on the number of millions of articles. If the emergency notification time is 15 minutes after the information is published, it will be determined that all news and public opinion in Taiwan can be processed efficiently while the system can continue to expand in parallel.
在這樣的系統設計中,所有的動作都以實現無人管理與最少參數設定為目標,使系統能在最小的人為介入下持續運行。自 2017年8月上線服務以來,此系統已從單伺服器設計升級為多伺服器協同處理,目前每月處理資料的數量約在 100萬篇文章。自每則新聞進入系統至通報處理完成,我們仍持續將處理速度控制於 3分鐘以內,及平均處理速度維持在 1分鐘以內完成。由於台灣主流新聞與論壇的每月新聞或輿論的資料總量落在百萬篇文章的數量規模,倘若以資料揭露後 15分鐘為黃金通報時間來判斷,以及本系統依舊能持續平行擴展的情況下,應足以確定能負荷台灣地區所有的新聞及輿論資料處理。

We specialize in the invention of application methods, optimization of algorithm and system process development, so we look forward to working with copyright owners to achieve more efficient information transfer. Finally, welcome to contact us, if you have a service application requirement for social media.

 Patent in Taiwan: No. I627545