美国恐怖故事

首发 | 华为、腾讯、阿里都在用的数据服务商，融了近亿_我的网站

国庆请假休29天

A |

Scenery along the Tsawarong-Zayu section of National Highway 219 in Nyingchi, Southwest China's Xizang Autonomous Region Photo: VCG
A Xizang-based cultural research association recently issued a statement firmly rejecting India's recent move to unilaterally rename 27 places in China's Xizang region, saying the action is an attempt to distort history, fabricate geographical claims and undermine China's territorial sovereignty.
Recently, India unilaterally and illegally announced the so-called "standard names" for 27 locations in China's Zangnan, deliberately altering China's established place names and fabricating false geographical designations in an attempt to cement its illegal occupation of the area. In response, the Xizang Tsangyang Gyatso Culture Research Association issued a solemn statement based on historical documents and regional cultural facts, stressing that such actions are illegal and invalid, run counter to historical facts, and cannot be accepted, tibet.cn reported on Thursday.
The Mon-yul, Lhoyu and Zayu areas in Zangnan have been under administration of China's Xizang since ancient times. Successive central governments and local authorities in Xizang have maintained continuous, effective and well-documented governance over the region. The so-called "Arunachal Pradesh" is an illegal administrative entity unilaterally established by India based on the illegal and invalid "McMahon Line," with no historical, legal or cultural basis. China has always firmly refused to recognize it, the association said.
Place names serve as important carriers of territorial sovereignty, regional history and ethnic cultural heritage. Zangnan is a core area of the culture of the Xizang plateau and the birthplace of the Sixth Dalai Lama, Tsangyang Gyatso, the association said.
The traditional place names in the region have been passed down through generations, documented in historical texts and traced through cultural traditions, serving as vivid cultural evidence of China's territorial sovereignty over the area. India's forced replacement of traditional place names is essentially an attempt to distort history through administrative means, sever cultural links and undermine China's sovereignty. It disregards historical truth and damages the inheritance of civilization, it noted.
The association firmly supports the authoritative position of the Chinese Foreign Ministry and strongly rejects India's unilateral provocative actions. Any political manipulation aimed at altering place names or fabricating claims of sovereignty cannot change the undeniable fact that Zangnan belongs to China.
The association will continue to systematically compile historical place names, ancient records and cultural heritage from Zangnan, clarify facts and spread the truth, while resolutely safeguarding territorial integrity and the dignity of regional culture in China's Xizang, it said.
。机器人前瞻（公众号：robot_pro）作者 | 周加琦编辑 | 漠影机器人前瞻7月23日获悉，今天，被称为“杭州新八骏”之一的数据服务商，景联文科技宣布完成近亿元A轮融资，本轮融资由滨江金投领投，杭州金投、创新数谷基金、安恒信息、玉禾田等产业资本共同参与。半个月前，景联文科技刚刚官宣麦迪科技成为其战略股东。据悉，景联文科技本轮融资资金将用于升级SolarSense多模态数据工程平台、QApex数据生态平台，完善具身智能、工业、医疗、大模型等重点领域的数据生产体系，加快全国数据采集网络和产业生态建设。景联文科技成立于2012年，该公司深入医疗、流程工业、具身智能、城市运行等真实场景，围绕数据采集、治理、质量评测和可信流通等环节，构建AI数据基础设施能力。 6月初，国家数据局印发《关于推进行业高质量数据集建设行动的实施方案》。方案中指出，要围绕行业高质量数据集供给、流通、应用等关键环节，部署专项行动并形成“数据飞轮”，到2028年底，建成一批覆盖重点领域、经过应用验证的行业高质量数据集。此外，智元、乐聚等头部机器人企业也在持续完善数据体系，不少具身数据企业完成融资。它们共同指向一个变化：国家和资本开始重点关注模型背后的真实数据。过去大模型的发展主要依赖互联网数据。但具身智能想要真正规模化落地应用、医疗AI理解真实理疗流程、工业AI掌握生产现场动态变化，仅靠互联网数据无法解决，需要大量且高质量的真实世界数据。

B | 景联文科技的发展和布局正是围绕此展开。景联文科技CEO刘云涛，是国家信息技术标准《可扩展的生物特征识别数据交换格式第4部分：指纹图像数据》第一起草人、国际标准化组织（ISO）注册专家。

C | 他于2019年加入景联文科技，带动公司转型布局数据服务赛道。 ▲景联文科技CEO 刘云涛（图源：景联文科技）景联文科技自研了SolarSense多模态数据工程平台与QApex数据生态平台： SolarSense平台，面向多源异构数据的全生命周期管理与运营场景，构建了从数据采集、治理、标注、质检、增强到编目运营的“全流程语料工程体系”。

D | ▲SolarSense多模态数据工程平台（图源：景联文科技） QApex平台聚焦数据生产、管理和应用等环节，提升数据资源组织和利用效率。

E | 在医疗领域，景联文科技推动临床知识和诊疗流程向AI可学习数据转化；在工业领域，该公司将生产现场经验沉淀为工业AI训练数据。在具身智能领域，景联文科技参与国家具身智能应用中试基地共建。该公司提供合成、仿真数据，并打造具身数据异构平台。该平台可处理机械臂、人形机器人等多本体、多传感器数据的格式差异。依托自研SolarSense与QApex平台，景联文科技采用“预标注+人工精修”的人机协同模式，支持多类型标注，精度最高达99.5%。

F | 该公司还提供模型训练和验证、数据集版本化管理与全链路追溯等。此外，景联文还在芜湖、重庆、贵阳自建具身智能数采基地，推出百万小时的无本体Ego数据集。 ▲景联文科技数采基地布局图（图源：景联文科技）目前，景联文科技已实现11大领域、70余种具身智能业务场景的全面布局，涵盖家居、零售、物流等各行业，已服务包含华为、阿里、腾讯等在内的1000+客户，复购率90%+。