Early opinion monitoring relied on keyword queries chained together with AND, OR and NOT logic, and usually needed heavy manual work to re-process the results. Intelligent monitoring instead uses natural language processing (NLP) to recognise content along multiple dimensions, which improves data accuracy.
In the era of universal internet access, opinion is generated around the clock. China now has more than one billion netizens, the internet has become a major channel for personal expression, and public discussion has largely moved online.
Daily opinion data volume has grown from the millions to the hundreds of millions in under ten years, according to Liu Pengfei, deputy editor-in-chief of People’s Online. That scale is far beyond human reading capacity, so AI processing and analysis have become unavoidable.
From search to algorithms
Compared with traditional systems, an AI opinion system automates analysis of Chinese multimedia content and big-data intelligence mining. Two shifts stand out: monitoring moves from retrieval to algorithms, and computation moves from simple counting to deep learning.
AI systems can perform entity recognition, semantic disambiguation, knowledge-graph construction, topic classification, automatic summarisation and sentiment analysis on text, plus brand, face, object and text recognition on images.
Accuracy improves because AI systems use classification models to filter noise and sentiment analysis to surface sensitive information, track the state of discussion in real time and assess where it is heading. On that basis the system can summarise how an event develops, how its features are distributed and how risky it is, and produce trend forecasts.
AI monitoring also handles un-themed information. Monitoring a defined department, company or event is relatively easy; monitoring a broad topic such as campus safety nationwide is hard to do with keyword lists, because no list can cover every school. By setting regional, media-weight, heat, industry, sentiment and sensitivity attributes, an AI system can locate and classify events nationwide, then extract, cluster and analyse the information to provide intelligent monitoring, early warning and assessment.
Wider use in governance and business
Market regulation departments increasingly use opinion data to focus on the industries under the most pressure, plan ahead for recurring problems, and combine precise identification with precise enforcement – improving efficiency while lowering their own reputational risk.
Enterprises treat opinion data as an important input for big-data risk control and smart marketing. After the misuse of personal credit data was curbed, many financial firms now use opinion data for risk identification, estimation and evaluation.
An AI system can turn known risks into experience knowledge, and use knowledge graphs to generalise one case into a category, enabling more accurate prediction of future risks – including risks that have not appeared yet but could. Its reasoning capability can surface potential risks that have never materialised before.
Media intelligence takes shape
The China Smart Media Development Report (2020-2021), released by Communication University of China and partners, identifies three emerging models of media intelligence: smart broadcasting, smart newspapers and the intelligent transformation of commercial platforms.
AI is now embedded across broadcasting workflows – video restoration, intelligent editing, smart distribution, cloud supervision, multi-dimensional review and emergency information services. Newspaper groups have progressed from single-point AI assistance to full editorial-chain support and platform-level products. Commercial platforms, driven by the market, apply AI across the whole chain from aggregation to publishing, with AI plus video a new battleground under the shift to all-video content.
Want to talk through what you just read?
We work end to end, from product design to launch. Tell us about your scenario and we can start there.
Shenzhen XZS Technology Co., Ltd. · Tel 0755-82494862 · 13760424861