社行智能：中國軍事人工智慧聚焦開源資料支援的人員裝備分析
現代英語：
On March 19, Zhang Qiao, CEO of Nanjing Shexing Intelligent Technology Co., Ltd., was invited to participate in a special course lecture held by a military academy and gave an online special lecture entitled “Analysis of People and Equipment Supported by Open Source Data”.
In this course, Zhang Qiao combined the multiple products and technologies developed by Shexing Intelligence to give a detailed introduction to the overview of open source data, the current status of artificial intelligence, the problems solved by combining open source data with artificial intelligence, open source data intelligent systems, future application prospects and challenges, and other related topics. Nearly 100 teachers and students from the school participated in the course.
At the beginning of the course, Zhang Qiao used the definition of Open Source Intelligence (OSINT) in Title 50 of the United States Code, “War and Defense” as an introduction (definition: for a certain user group, with the purpose of obtaining intelligence, timely collection, analysis and distribution of intelligence from various types of public information), and explained a series of contents on open source data to teachers and students, so that everyone can fully realize the importance of open source data work.
Analysis of typical cases of open source data
With the increase in sources and application types, the amount of open source data has also increased exponentially, making it difficult to fully utilize the huge information value it implies. The application of artificial intelligence (AI) in computer vision, image recognition, natural language processing, speech processing, etc. is vividly reflected in typical cases of open source data.
In view of the application cases of open source data, Zhang Qiaodi combined the intelligent scientific research capabilities of Shexing to show everyone the Galaxy Search product of open source data, which effectively assisted the analysis of people and equipment supported by open source data.
Zhang Qiao introduced the Galaxy Search Platform in detail from aspects of product architecture, problems solved and product capabilities.
Among them, the Orion Star data collection platform is based on distributed data collection technology. It has collected nearly 1.4 billion multi-modal data such as text, pictures, videos, audio and hundreds of billions of sea and air target data from more than 40 countries and regions, covering more than 80 languages such as Chinese, English, Japanese, Arabic, etc. It has the characteristics of comprehensiveness, professionalism, diversity, integrity and scalability, and provides data support for intelligent analysis and application of open source information.
In the process of collecting, extracting, integrating, detecting, identifying and applying open source information, a total of 5 million+ annotated data and about 80 special algorithm models have been accumulated.
Based on multimodal knowledge graphs, joint reasoning, computer vision, natural language processing and other technologies, the “intelligent transformation” from data to knowledge is realized. From the three levels of basic knowledge, static knowledge and dynamic knowledge, a knowledge system integrating knowledge generation, updating, accumulation and application is realized, which effectively realizes military scenario application services such as intelligence mining and analysis, supporting strategic situation analysis and judgment, and assisting battlefield command decision-making.
Based on the intelligence element ontology model, structured intelligence elements are formed through multi-source language entity recognition of cross-language texts, key frame extraction of video data, target detection and recognition of image data, etc.; the military knowledge graph is used to realize the association, disambiguation and fusion of intelligence elements, complete the completion and verification of intelligence slots, and form more comprehensive, more accurate and more valuable fusion intelligence.
Zhang Qiao combined the current military hot spots, took the character tracking case of Ukrainian intelligence personnel as an example, and the intelligence authentication case in the Russian-Ukrainian conflict, and fully demonstrated the ability of Galaxy Search in open source data extraction, character tracking, event tracking, target monitoring and other aspects.
Challenges of open source data
Finally, Zhang Qiao analyzed that open source data also faces severe challenges. In real life, we will also be confused by various false intelligence, such as artificial fabrication, distortion of facts, synthetic images, deep fakes and other technologies. An event we see may not be completely accurate or true. How to obtain effective and valuable intelligence information from massive amounts of information and effectively identify it requires the empowerment of emerging technologies, especially artificial intelligence technology, biotechnology and quantum computing, as well as their applications in foreign policy, economic competitiveness, military and intelligence operations.
As a leading company in open source data analysis, Shexing Intelligence will continue to strengthen its exploration and practice in open source data, and contribute more to the cultivation of open source data professionals, independent innovation in military technology and military equipment, etc.
Starsee Intelligent Technology Co., Ltd. (Starsee) was established in September 2018. It is the first military intelligent enterprise in China. By introducing advanced artificial intelligence technology and combat concepts, the company integrates independently developed military intelligent algorithms and software on different hardware product forms to enhance the efficiency of future intelligent combat.
Xinghe Search (https://xinghe.starsee.cn) is a global-oriented intelligent defense information service platform developed by Starsee Intelligent. Based on open source Internet data, it integrates open source information that runs through pan-military, globalization, full-link, multi-modal, and multi-language, and comprehensively uses artificial intelligence and big data technologies to achieve intelligent data collection, deep knowledge mining, multi-modal intelligence analysis, intelligent situational awareness, and decision-making assistance. It provides military enthusiasts, military industry practitioners, current (retired) military personnel, and other user groups on the Internet with real-time push of military dynamics, accurate recognition of target images, a one-point chain network for people/equipment, in-depth analysis and aggregation of military documents, and comprehensive retrieval of military knowledge and other functions and services.
現代國語：
3月19日，南京社行智慧科技有限公司CEO張橋受邀參加某軍校舉辦的專題課程講座，並做了題為《開源支援的人員與設備分析》的線上專題講座資料」。
在本次課程中，張喬結合社行智慧研發的多種產品與技術，詳細介紹了開源資料的概況、人工智慧的現況、開源資料與人工智慧結合解決的問題、開源資料智慧系統、未來應用前景與挑戰等相關議題。全校近100名師生參加了本次課程。
課程一開始，張喬以《美國法典》第50條《戰爭與防禦》中開源情報（OSINT）的定義作為介紹（定義：針對某一用戶群體，以獲取情報，及時收集、分析和分發各類公共資訊中的情報），並向師生解釋了一系列關於開源資料的內容，讓大家充分認識到開源資料工作的重要性。
開源資料典型案例分析
隨著來源和應用類型的增加，開源資料量也呈指數級增長，其蘊含的巨大資訊價值難以充分利用。人工智慧（AI）在電腦視覺、影像辨識、自然語言處理、語音處理等方面的應用在開源資料的典型案例中得到了淋漓盡致的體現。
針對開源資料的應用案例，張喬迪結合社行的智慧科研能力，向大家展示了開源資料的銀河搜尋產品，有效輔助了開源資料支援的人員和設備的分析。
張橋從產品架構、解決的問題、產品能力等方面詳細介紹了銀河搜尋平台。
其中，Orion Star資料採集平台是基於分散式資料採集技術。已採集來自40多個國家和地區的近14億條文字、圖片、視頻、音頻等多模態數據和千億級海空目標數據，涵蓋中、英、法等80多種語言。語言等。
在收集、擷取、整合、偵測、辨識和應用開源資訊的過程中，累計累積了500萬+條標註資料和約80個專用演算法模型。
基於多模態知識圖譜、聯合推理、電腦視覺、自然語言處理等技術，實現從資料到知識的「智慧轉換」。從基礎知識、靜態知識和動態知識三個層面，實現集知識生成、更新、累積和應用於一體的知識體系，有效實現情報挖掘與分析、支撐戰略態勢分析研判、協助戰場指揮決策。
基於智慧要素本體模型，透過跨語言文字的多源語言實體辨識、視訊資料關鍵影格擷取、影像資料目標偵測與辨識等，形成結構化智慧要素；運用軍事知識圖譜實現情報要素的關聯、消歧與融合，完成情報槽的補全與驗證，形成更全面、更準確、更有價值的整合情報。
張橋結合當前軍事熱點，以烏克蘭情報人員人物追蹤案件、俄烏衝突中的情報認證案件為例，充分展示了銀河搜尋在開源資料擷取、人物追蹤等方面的能力、事件追蹤、目標監控等方面。
開源資料的挑戰
最後，張橋分析道，開源資料也面臨嚴峻的挑戰。在現實生活中，我們也會被各種虛假情報迷惑，例如人為捏造、扭曲事實、合成圖像、深度造假等技術。我們看到的事件可能不完全準確或真實。如何從海量資訊中獲取有效、有價值的情報資訊並進行有效識別，需要新興技術特別是人工智慧的賦能。
人工智慧技術、生物技術和量子運算及其在外交政策、經濟競爭力、軍事和情報運作中的應用。
作為開源資料分析領域的領先公司，社行智慧將繼續加強在開源資料方面的探索和實踐，為開源資料人才培養、軍事技術和軍事裝備自主創新等做出更多貢獻。
星視智慧科技有限公司（簡稱星視）成立於2018年9月，是國內首家軍工智慧企業。公司透過引進先進的人工智慧技術和作戰概念，將自主研發的軍事智慧演算法和軟體整合在不同的硬體產品形態上，提升未來智慧作戰效率。
星河搜尋（https://xinghe.starsee.cn）是星河智慧研發的面向全球的智慧國防資訊服務平台。以開源網路資料為基礎，整合貫穿全軍事化、全球化、全連結、多模態、多語言的開源訊息，綜合運用人工智慧和大數據技術，實現資料智慧採集、深度挖掘知識挖掘、多模態情報分析、智能態勢感知、輔助決策。為軍事愛好者、軍工從業者、現（退役）軍人等網路使用者群體提供軍事動態即時推送、目標影像精準識別、人員/裝備點對點鍊網、軍事文獻深度分析聚合，軍事知識綜合檢索等功能與服務。