With the rapid development of artificial intelligence technology, large model technology has become the core driving force of a new round of scientific and technological revolution. Its strong generalization ability and multi-modal support features have set off an unprecedented application boom in the industry. The urgent demand for intelligent transformation in all walks of life has given birth to the rapid development of large industry models. The application of large models in vertical industries has gradually become an important force to promote industrial reform. In the past 2024, AI track heat has not decreased, market competition is still fierce, technology and products continue to update iteration. Large models move from Volume Parameters to Volume Applications. The "value orientation" of large model applications has undergone an important shift from pursuing "super applications" to focusing on "super useful". This shift from pursuing comprehensive functions to focusing on practical application effects will enable large models to participate more deeply in public life and work, improve user satisfaction and experience, and become everyone's right-hand man.
The application of TRS-GPT LLM in the enabling industry also presents a vigorous development trend, with technological innovation constantly emerging and increasingly rich application scenarios. It can adapt to different tasks and produce quality answers or content with few or no samples. However, to land in a complex business, a complete set of Agent patterns is needed to drive it, that is, encapsulating business logic and tool interaction components on top of the large model, so that the language model of "passive answer" becomes the agent of "active execution" of the work task. TRS-GPT LLM Integration Platform helps users to accurately customize the intelligent application of large models in vertical fields and specific scenarios by applying the characteristics of easy maintenance, easy connection of capabilities, easy collaboration of business, easy operation of data and easy optimization of models.
TRS-GPT LLM has been successfully implemented in many vertical fields such as media, finance, government affairs, etc., covering a wide range of business scenarios, providing diversified modal capabilities, and bringing rich business experience to enterprise users with excellent model and Agent effects, effectively promoting the digital transformation and high-quality development of industry users.
"TRS-GPT LLM + Finance"
As a representative of data-intensive industries, the financial industry involves extensive and complex financial data and business logic. The application of large model technology is of great significance to improve the automation and intelligence level of financial business, enhance risk management and control ability and decision-making efficiency. At present, Financial TRS-GPT LLM has been widely penetrated in many application scenarios of financial industry, which not only significantly improves business processing efficiency, but also greatly optimizes user experience, injecting strong impetus into innovation and development of financial industry.
Application scenarios
Intelligent risk control：Credit risk early-warning、Industry Risk Analysis、Enterprise risk conduction、Questions and Answers in the Financial Field
Intelligent protection：Smart Consumer Protection Review、Complaint Analysis Early Warning、Financial Customer Service Assistant、Intelligent attribution analysis
Intelligent audit：Compliance FAQ、Audit risk analysis、Compliance Problem Diagnosis、Audit report generation
Marketing extension：Get customers accurately、Marketing lead recommendation、Regional Industry Analysis、Enterprise value assessment
Innovation Research：Ask questions、Information summarized、View Recommended、Research report generation
Typical cases
Case 1——TRS-GPT LLM Energizes a Large-scale Financial Holding Group to Create Accurate Risk Control Early Warning System
A group is one of the few financial control groups in China financial enterprises, with almost all financial licenses such as insurance, banking, securities, funds, trusts, etc. This diversified business layout greatly increases the difficulty of risk management. The risk characteristics and regulatory requirements of different business segments are different. How to effectively coordinate and comprehensively manage risks at the group level is a major challenge, such as complex comprehensive financial business, scattered customer risk data, risk management system to be upgraded, radical expansion and risk control imbalance, frequent credit business violations, etc.
TRS tailor-made intelligent early warning service based on heterogeneous data for nearly 12 financial subsidiaries of a certain group by using digital financial risk control data and large model technology. This service significantly improves the accuracy and efficiency of data processing and production by introducing Financial TRS-GPT LLM. At the same time, combining the integration application of large model and RAG, it not only strengthens the integration of financial risk control knowledge and the updating ability of real-time data, but also effectively breaks through the knowledge limitation of traditional large model in the field of financial risk control, thus greatly improving the quality of generated content. This series of optimization measures directly promoted the accuracy of risk control warning signals to achieve a significant improvement.
At present, this intelligent early warning service has been successfully applied to many key scenarios such as post-loan risk management, post-investment risk management, supplier monitoring, macro research, compliance internal control, etc., providing professional, efficient and comprehensive financial risk control early warning support for the Group's financial subsidiaries.
Case 2——TRS-GPT LLM Energizes a Large State-owned Bank to Improve Industrial Data Fusion and Science and Technology Financial Comprehensive Service Capability
A large state-owned bank hopes to construct an analysis model of advantageous enterprises based on macro-economic and policy data of industry and Yangtze River Delta region, so as to form an accurate portrait of region-industry-enterprise. Focus on strategic emerging and technological frontier industries, look for potential enterprise credit customers with capital demand, high conversion and controllable risks from various dimensions such as industrial customer acquisition, macro insight and marketing risk, so as to form a high-value potential customer pool for development industry. Through enabling banks to build internal and external corporate digitalization capabilities, assist in the construction of digital insight system for industry customers, promote the digital construction of marketing customer acquisition and risk warning, provide decision-making assistance for accurate marketing customer acquisition, and establish financial connection to drive the future development of enterprise and overall social economy.
Based on the four supporting engines of independent research and development of industrial data center, data intelligent label, whole industry knowledge engine and financial large model, TRS realizes intelligent data collection, processing and integration of the whole industry chain, builds public domain industrial data center, and forms rich core scenario application of industrial chain. With its high-quality industrial data and financial risk control data, the company deeply integrates with Financial TRS-GPT LLM, accurately focuses on the science and technology industry chain, scientifically evaluates the value and risk of science and technology enterprises through the application of cutting-edge technology, and effectively expands the ecology of science and technology financial channels. At the same time, actively cultivate science and technology maker group, optimize science and technology financial risk control system, assist a large state-owned bank to realize significant improvement of comprehensive service capability of science and technology finance, and fully reflect the application efficiency of large model.
"TRS-GPT LLM + Media"
In the wave of AI-driven information technology, accelerating the birth of new quality productivity has become a core issue to promote systematic change in mainstream media. TRS actively responds to and assists the development of mainstream media, injecting powerful impetus into the progress of media industry through cutting-edge technology. In response to the unique needs of the media industry, Media TRS-GPT LLM implements precise policies in the entire business process of "planning, editing, reviewing, publishing, and evaluation", which has significantly improved users 'production efficiency, editing and proofreading quality, content creativity, and communication operations. Such efficiency has made positive contributions to the transformation, upgrading and sustainable development of the media industry.
Application scenarios
Content Production and Creative Assistance：Content Selection Planning、Design an interview outline、Proposed headline、Auxiliary graphic creation、Content Smart Summary、Article Style Rewriting、Proofread manuscript errors
News Search and Knowledge Aids：Multi-modal search、Manuscript traceability、Factual content induction、Data summary、Multilingual automatic translation
News dissemination and operation expansion：Communication analysis report generation、Publicity report summary generation、Government Service AI Assistant、Cultural Travel AI Assistant
Typical cases
Case 3——AIGC Energizes People's Daily Media Convergence and Intelligent Transformation
It is necessary to promote the integration and development of traditional media and emerging media, follow the law of news dissemination and the law of development of emerging media, pay attention to integration on one hand and management on the other, so as to ensure that the integration and development advance in the correct direction. This requires systematic changes in mainstream media in terms of content, channels, technology, operations, and mechanisms. In the context of comprehensive media transformation driven by big data and intelligent technology, intelligent logic has penetrated deeply into all aspects of mainstream media news production. How to speed up the process of media convergence and quickly form a new mainstream media with communication power, guidance, influence and credibility is the core goal of media convergence transformation.
In this context, TRS cooperated with People's Daily to launch AIGC intelligent editorial assistant, which provided an innovative path for intelligent transformation of media. In terms of media resource retrieval and planning, the assistant provides media resource retrieval, topic selection planning, intelligent writing and abstract, as well as efficient and accurate content compliance review services for all editors and reporters of People's Daily. Its multi-mode search function covers intelligent text search, image search, video search, multi-language search, semantic search and other scenes, effectively improving the information collection efficiency of editors and reporters. In addition, the intelligent review and proofreading function can effectively detect errors or violations by accurately identifying text, video, audio, pictures and other contents in multiple dimensions, thus improving content quality and communication security.
After the system was launched, as a representative of the "new quality productivity" of the media industry, it comprehensively improved the quality and efficiency of the two sessions of the People's Daily. More than 200 people from the all-media matrix of People's Daily Chinese client, video client "Vision," People's Daily corporate micro blog, People's Daily Digital Communication, Global Times, National Humanities History and other newspapers used AIGC editorial department for live editing and work creation. Related AI capabilities were invoked for nearly 800 hours.
Case 4——TRS-GPT LLM Energizes China Environmental News "Carbon Engine" to Build a Green Carbon Environmental Protection Future
As a responsible developing country, China is playing an important role as a participant, contributor and leader in global climate governance. The proposal of the "double-carbon" goal is China's solemn commitment to the world, a major strategic decision made by China's highest decision-making level after careful consideration of the overall situation at home and abroad, and a reflection of China's active response to the global climate crisis and efforts to promote the construction of a community of human destiny. As the competent media of the Ministry of Ecological Environment, China Environmental News has great influence in the field of ecological environment protection.
China Environmental News jointly launched the second phase construction of the "Carbon Engine" big data application service publicity platform with TRS in 2023. "Carbon Engine" comprehensive service platform integrates domestic double-carbon related news information, policies and regulations, enterprise and city carbon emission indicators, industry supervision and punishment information, cutting-edge technology, experts and opinions, etc., and provides double-carbon service capabilities such as carbon·school, carbon·information, carbon·technology, carbon·disclosure, carbon·list, carbon·tool, carbon·consultation, carbon· check, etc. Through scientific analysis and data mining, the platform provides information consulting services and solutions for carbon emission management of enterprises and industries, and assists in the implementation of the "double carbon" strategy. At the same time, the platform also integrates enterprise environmental violation punishment data, environmental quality monitoring data and public opinion data, builds environmental public opinion risk early warning platform, and provides risk consultation suggestions and early warning tips for ecological environment departments.
In 2024, the "Carbon Engine" big data application service promotion platform continued to expand, integrating industry and enterprise carbon emissions and product carbon footprint data information, laying a foundation for carbon management application services. In addition, the platform builds intelligent question-and-answer applications based on the TRS-GPT LLM to promote the development of digital intelligence. Based on the massive data and professional knowledge base of TRS-GPT LLM,"Carbon Engine" intelligent Q & A provides users with accurate and comprehensive carbon environmental protection knowledge Q & A service. Users can quickly obtain professional answers on carbon emissions, carbon trading, carbon emission reduction, etc. by inputting relevant questions, so as to deepen their understanding and understanding of the field of carbon environmental protection. The Carbon Engine intelligent Q & A also has a number of core functions, such as carbon emission calculation, carbon footprint tracking, carbon emission reduction scheme recommendation, etc. These functions can provide users with personalized carbon environmental protection services, help users better understand their own carbon emissions, formulate effective carbon emission reduction plans, and track carbon emission reduction effects.
As the first comprehensive carbon service platform in China, the "Carbon Engine" big data application service publicity platform stood out in the 2023-2024 national media convergence technology application innovation case selection. Its core advantages and innovations lie in publishing high-quality information and transmitting and popularizing professional "carbon" knowledge; building a powerful database to collect high-quality "carbon" information; creating eight core functions to provide comprehensive "carbon" services; setting up communication communities to build professional "carbon" friends; and introducing AI technology to build intelligent "carbon" scenarios.
With the technical support of TRS, the platform provides information services, consulting services and technical support schemes for carbon emission management of industries and enterprises, promotes the transformation, promotion and application of green and low-carbon technology achievements, promotes the solid implementation of "double-carbon" strategy by industries and enterprises, and further expands the depth and breadth of ecological environment information consulting services of China Environment News.
"TRS-GPT LLM + Think Tank"
The combination of large models and think tanks is the product of the deep integration of contemporary science and technology and decision-making science. In recent years, with the continuous maturity of big data technology and the continuous optimization of deep learning algorithms, the ability of large models in processing massive data, mining deep information, and making complex predictions has been significantly improved. At the same time, think tanks, as an important reference source for decision-making by governments, enterprises and all sectors of society, have a large amount of professional knowledge, rich cases and in-depth industry insight. The introduction of large models into think tanks can not only provide scientific basis for decision-making through data analysis, but also accelerate the transformation and application of think tank research results, promote more accurate and efficient policy formulation, market forecasting closer to reality, and provide strong support for economic and social development.
Application scenarios
Decision Support and Analysis：
Real-time business metrics、Trend prediction、Competitive intelligence analysis、Abstracts and refinements、Research paper assisted writing、Smart Text Translation、Automatic newspaper generation and proofreading
Knowledge management and sharing：Knowledge integration Q＆A、Intelligent knowledge base、Semantic search、Personalized recommendation
Evaluation and Prediction of Policy Effects：Policy Search、Policy interpretation、Policy context、Policy analysis、Policy evaluation
Typical cases
Case 5——TRS-GPT LLM Enabling Biomedical Think Tank Research Innovation Optimization
As a national medical information research and resource guarantee institution, a research institute has deep strength and extensive influence in the fields of medical informatics, health policy and management, etc. It undertakes important tasks such as medical information research, health policy and management research, medical information resource construction and service for the reform and development of national medical and health undertakings, medical science and technology innovation, and is an indispensable important force in the industry. The institute is committed to providing comprehensive and accurate decision support and information services for the reform and development of China's medical and health undertakings and medical science and technology innovation. It plans to focus on the intelligent and professional biomedical literature based on the first phase of the biomedical literature intelligent service platform. Service needs, deepen the application of new technologies such as AI, and develop AI intelligent retrieval, intelligent recommendation, professional knowledge discovery, full-text services, personalized services, fund-funded results remittance services, Key technologies and platform functions such as dynamic management of big data comprehensively improve the intelligent and knowledge-based service capability of biomedical literature information.
As its long-term partner, TRS carries out in-depth function optimization and innovative function expansion on the solid basis of inheriting the first-phase platform based on TRS-GPT LLM, including intelligent semantic retrieval service, supports natural language retrieval and cross-language search in Chinese and English through large model and RAG technology, and realizes rapid and accurate positioning of literature resources; Intelligent recommendation and personalized service based on user portrait. By establishing multi-dimensional user portrait, including scientific research achievements and research interests of users, the platform provides personalized literature retrieval sorting, recommendation of relevant literature and academic social recommendation services. Users can discover cutting-edge problems or potential research topics of interdisciplinary subjects according to their interests and academic background, thus improving academic research efficiency and cooperation opportunities; Professional knowledge discovery service supports structured analysis and extraction of PICOs elements based on clinical real problems based on large models, provides evidence-based medical literature discovery service, automatically identifies high-quality evidence-based medical literature, captures and displays key hot words, realizes efficient aggregation of high-quality clinical research evidence, and assists clinical research.
Case 6——TRS-GPT LLM Energizes a Central Enterprise Think Tank to Improve Research Innovation and Business Decision Efficiency in Energy Field
Under the background of rapid progress of information technology and explosive growth of Internet information, a wholly-owned subsidiary of State Grid Information and Communication Industry Group, as an important think tank in the State Grid system, shoulders the important task of providing intellectual support for business development and strategic decision-making, and faces the urgent demand of efficiently extracting, processing and analyzing valuable information from massive and unstructured information resources. This information is not only about market dynamics and policy orientation, but also involves multiple dimensions such as technology trends and customer needs, which are critical to the company's business decisions and strategic planning. The introduction of large language model technology is precisely to meet this challenge. This technology can make full use of the company's internal knowledge accumulation, and efficiently integrate external policy information data to achieve rapid analysis and intelligent induction of information. Through the powerful ability of large language model, the company can quickly capture key information points in complex and changeable information environment, significantly improve information processing efficiency, and provide more accurate and efficient information support for the company's business development and decision-making.
Relying on its self-built data center and TRS-GPT LLM technology, TRS tailor-made an Internet information integration platform focusing on energy business for the company. The platform is deeply integrated with expertise in the field of energy and is committed to providing comprehensive and intelligent information support to research teams.
In terms of energy knowledge and industry information, the platform can instantly respond to the query needs of the research team on energy policy, market dynamics, technological innovation and other aspects through the comprehensive question and answer function, and provide accurate and authoritative answers. At the same time, the policy analysis module utilizes the data mining and intelligent analysis capabilities of the large model to deeply analyze the energy policy trend and its impact on the industry, providing a strong basis for the team to formulate research strategies, including policy information interpretation, policy index extraction, policy change analysis, policy comparison analysis, etc.
The manuscript auxiliary writing function aims at the writing needs of professional manuscripts such as energy research reports and policy interpretations, and significantly improves the professionalism and logic of manuscripts through intelligent recommendation of keywords, sentence structure and chart data.
The summary extraction function can quickly extract core views from long reports, facilitating the rapid dissemination and sharing of research results, including summary extraction, report summary results, report knowledge extraction, report viewpoint comparison, multi-report summary and data analysis.
In addition, considering the importance of international exchange and learning in the field of energy, the platform is specially equipped with automatic translation function to support accurate translation of Chinese and English and various professional terms in the field of energy, effectively promoting international academic exchange and cooperation. The text proofreading function uses natural language processing technology to intelligently check the grammar, spelling and other aspects of the manuscript to ensure the accuracy and standardization of the manuscript.
To sum up, this Internet information integration platform built by TRS, with its distinctive features of energy business and comprehensive intelligent functions, is gradually becoming an important support for promoting research and innovation in the energy field and improving business decision-making efficiency, helping enterprises to stand out in fierce market competition and achieve sustainable development.
"TRS-GPT LLM + Government Affairs"
With the vigorous rise of large model technology, the construction of digital government ushered in an unprecedented opportunity for change. Relying on its deep accumulation in the field of government affairs, Government Affairs TRS-GPT LLM integrates a large number of government affairs knowledge resources, and can accurately grasp the professional terms, complex laws and regulations and business processes in government affairs scenarios. This innovation not only significantly improves the efficiency of knowledge management and application within the government, realizes the optimization of workflow and effective control of cost, but also greatly improves the experience of external service, enhances the transparency of government work, provides solid technical support for building a more efficient, convenient and transparent service-oriented and intelligent government, and promotes the construction of digital government to a new height.
Application scenarios
Government consultation：IR、Intelligent question and answer、Policy interpretation and recommendation、Service navigation
Government affairs service：Service、Smart single、Intelligent approval、Aid decision making
Government office：Auxiliary documentation、Document intelligent distribution、Policy Smart Q & A、Smart meeting、Intelligent training
Typical cases
Case 7——TRS-GPT LLM Empowers Digital Government to Lead Efficient Political-Civil Interaction
"What policy support does the government have for college students to start their own businesses?" "How about Guizhou's educational resources?" "How to replace the lost social security card?" How can users intuitively and conveniently obtain the policies and services they really need in the massive policy document data and service information?
Guizhou Province People's Government Network relies on the unified knowledge question-and-answer database constructed by the intensive platform, comprehensively upgrades the "Xiaogui" intelligent question-and-answer robot with the help of Government affairs TRS-GPT LLM and introduces Copilot co-driver mode to ensure instant access of high-frequency content. The upgraded "Xiaogui" intelligent question-and-answer robot carefully classifies policies, services and interactive channels with high public attention, and optimizes these information at the top and right prominent positions of the interface, which is convenient for users to consult at any time and enjoy at one touch, greatly improving the convenience of use. At the same time, aiming at different user groups (such as individuals and legal persons) and diversified service demands, the robot provides scene-based service solutions, covering multiple dimensions such as government service, policy consultation and frequently asked questions.
This large-scale model technology innovation enables "Xiaogui" to realize "second return","quasi-return" and "intelligent return" to user consultation problems, which not only significantly improves public satisfaction, but also effectively establishes a good image of efficient government service.
Case highlights
1. Through the construction of intelligent question-and-answer and consulting service applications, standardize service matters, optimize service processes, closely integrate online and offline, and improve the standardization and convenience level of government affairs services.
2. Build intelligent Q & A and consulting service application based on large model technology to provide users with all-weather automatic, accurate, scene-based, active and real-time intelligent interactive services.
3. This project innovatively adopts the three engine coordination mechanisms with "big model engine" as the main body,"vector index engine" as the auxiliary and "intelligent search engine" as the supplement to comprehensively optimize and improve the intelligent question and answer service level, which can accurately understand users 'complex and colloquial questions, support multiple rounds of conversations, provide the public with more accurate, recommended and reference-oriented hierarchical answer services, and further demonstrate the innovative, efficient and convenient image of the government. Close the distance between the government and the masses.
Case 8——AI Political Intelligence Improves User Search Integration Experience
The search function of government website is the main means and way for users to obtain information and services. Limited by the bottleneck of traditional full-text retrieval technology, the search function of government website is still mainly based on keyword search and matching, and there are some problems such as inaccurate search results, delayed information synchronization and ineffective integration of search function with other government service functions, resulting in users needing to search and find many times when handling affairs.
Shenzhen government portal website adopts Government affairs TRS-GPT LLM to launch "AI political inquiry agent" service, which integrates "search" and "question and answer" services through interface fusion, data fusion and function fusion through data intercommunication, technology sharing and user sharing service constructed by big model, vector engine and semantic engine, so as to abandon the information separation between search and question and answer in the past and provide integrated search and question answering service. Many service items provided by government websites are highly interrelated and logical. Users input search terms or questions, and the "AI Inquiries Agent" service will intelligently associate with website information according to relevant topics, cluster and display a complete scene integrating dynamic news, policy document interpretation, service and other information services, and provide integrated, scene-based and navigation-based integrated search services to realize "search-as-a-service". We will continue to optimize the service of "AI Political Intelligence" and lead the construction of Shenzhen digital government to a new height.
Case highlights
1. Intelligent retrieval and policy recommendation
Using natural language processing (NLP) technology, highly intelligent policy document retrieval function can be realized. Users can quickly locate the desired policy through various dimensions such as keyword, theme, publishing organization and time range. Based on large model technology, it provides search and inquiry integrated service, and quickly answers users 'questions about policy terms and implementation details through policy question-and-answer agent. This intelligent question-and-answer service not only improves the user experience, but also makes the government website closer to the user's needs in terms of information services.
2. Deep policy data integration
Through the integration of policy documents from different levels such as the national laws and regulations database, the State Council, Guangdong Province and Shenzhen City, a comprehensive and interrelated policy database has been formed. This deep integration of data enables users to query and analyze data across domains and periods to gain a more comprehensive understanding of policy background and implementation.
3. Real-time update and dynamic monitoring
The system can capture the latest policy information issued by the authoritative channels of the official government website in real time to ensure the timeliness and accuracy of the policy document library. The data is collected and updated every day, and the link availability detection is carried out many times a day. This real-time update mechanism enables users to obtain the latest policy information at any time, improving the timeliness and effectiveness of information services.
4. Personalized intelligent recommendation and policy service
The system can provide personalized recommendations according to user identities (such as Shenzhen enterprises, non-Shenzhen enterprises, individuals, etc.) and user portrait labels. In addition, machine learning algorithms are introduced to intelligently recommend relevant or potentially interesting policy documents based on the user's search history and preferences. This intelligent recommendation method not only improves the user experience, but also makes the government website more intelligent and personalized in terms of information services.
"TRS-GPT LLM + Public Opinion"
With the rapid growth of Internet data, traditional technical means are increasingly unable to cope with the actual needs of network public opinion discovery, judgment and analysis. In order to assist users to upgrade their services from network public opinion monitoring to network public opinion situational awareness, and to provide intelligent analysis and decision support for users, TRS launched Public Opinion TRS-GPT LLM. The model has been widely applied and implemented in many fields such as network information, government affairs and enterprises.
Application scenarios
Public opinion analysis：Public Opinion Questions、View refinement、Public opinion analysis、Suggestions、Development forecast、Sensitivity analysis
Public opinion report：Public opinion briefing generation、Public opinion special report generation、Statement Report Generation
Typical cases
Case 9——TRS-GPT LLM Energizes Network Information Office to Create a New Paradigm of Intelligent Governance
The large model in the field of network information takes TRS own 1 billion public opinion samples, millions of words of research and judgment suggestions, 100,000 excellent reports and other data as professional training corpus, and at the same time accesses a large number of professional public opinion corpus, broadening and deepening its accumulation in the field of professional knowledge in the industry. Through model incremental training without labeled data (industry knowledge incremental injection), multi-task instruction optimization of large model (industry knowledge update based on model fine-tuning) and LHF (industry value injection), the deep fusion of large model and industry knowledge base is further completed.
Comprehensive data collection, rich knowledge reserve and high-quality public opinion industry data provide stable data support and technical guarantee for the professional and accurate service of large models in the field of network information in practical application. In the construction of intelligent network information projects in many provinces and cities, TRS has combined with the local network information office business, based on the independent controllable products, big data and AIGC large model capability, and from the actual work, constructed a new overall intelligent governance pattern of territorial business.
At the application construction level, TRS focuses on creating two functions: AI intelligent report generation and intelligent question and answer of public opinion events. AI intelligent report generation breaks the bottleneck of traditional system report writing and analysis, covers public opinion report analysis scenarios and analysis angles, and provides users with convenient, professional and satisfactory reporting services; the intelligent question and answer for public opinion events is based on semantic understanding and big data platform. The intelligent interactive question and answer method performs multi-dimensional analysis of public opinion events and provides important supporting basis for public opinion judgment.
At the same time, by gathering public opinion risk data, TRS has built a comprehensive business application system, provided system service capabilities such as public opinion monitoring discovery, collaborative linkage disposal, evaluation statistical analysis, etc., realized multi-terminal display and real-time early warning, and provided strong technical support for local network information offices to respond quickly and make scientific decisions in complex public opinion environment.
In some projects, the cooperation between TRS and Huawei has realized the local deployment of the large model of the network information industry on the domestic server. By adapting the Shengteng 910B chip, the system performance has been significantly improved while ensuring data security and independent controllability. This model not only verifies the feasibility of local deployment of large models, but also indicates its potential for large-scale replication and expansion in the market, opening a chapter of deep collaboration and win-win development in the industry.
At present, Public Opinion TRS-GPT LLM has become the core of TRS public opinion service, building a one-stop service system, covering public opinion acquisition, content analysis, report writing and information push, etc. It is widely used in many industries such as network information, education, finance, public security, etc., showing strong industry adaptability and application value.
"TRS-GPT LLM + Public Safety"
With the widespread penetration of artificial intelligence technology in various fields, the field of public safety has also entered a new stage of intelligence. AI has demonstrated significant advantages in improving management effectiveness, reducing costs, predicting risks, and responding to emergencies. With its deep industry knowledge accumulation and excellent business processing and data understanding reasoning capabilities such as law enforcement intention identification, data analysis and early warning, case logic reasoning, document automatic generation, etc., Public Security TRS-GPT LLM can continuously learn and evolve from the huge data ocean. This evolutionary process accelerated the formation of new service capabilities of "science and technology +","data +" and "intelligence +", which effectively promoted the transformation, upgrading and innovation development of the public security department, and then built an intelligent actual combat application system.
Application scenarios
Smart Police Affairs：Voice street law enforcement、Alert situation statistics、Case Feature Analysis、Personnel track query、OPersonnel file retrieval、Intelligent document processing、Major case detection
Intelligence＆Command＆Operations：Integrated intelligence analysis、Safety risk assessment、Emergency Plan Recommendations、Command and dispatch、Intelligent dispatching of police resources、Multi-police coordinated operations、Situation awareness、Operational execution aids
Open source intelligence：Global Open Source Intelligence Search、Knowledge Intelligence Questions、Multilingual automatic translation、Target tracking and recommendations、Open source intelligence research and analysis、Open source intelligence report compilation
Typical cases
Case 10——TRS-GPT LLM Energizes Chengdu High-tech Public Security to Improve the New Quality Combat Effectiveness of Public Security
With the rapid development of science and technology, all walks of life are actively exploring the road of digital transformation, and the public security field is no exception. Chengdu High-tech Public Security Bureau actively responded to the call of the times, joined hands with TRS, and launched a brand-new "Smart Cloud Brain-Big Data Service Platform" with cutting-edge scientific and technological innovation, setting a new benchmark for intelligent policing construction.
Based on the ability of naming entity identification and relationship extraction of TRS data platform and large model, a brand-new "data service platform" was constructed, which successfully broke the data barrier and realized the deep integration and sharing of multi-party data such as government affairs, public security, social resources and Internet. This innovative measure not only greatly enriches the data sources, but also enhances the in-depth insight of the public security organs into the data. Through this platform, police can capture potential security risks and social dynamics more quickly and accurately, providing powerful data support for decision-making.
On the basis of data service platform, graph calculation and correlation analysis ability are integrated. By constructing a one-stop label system, the police can accurately depict the professional characteristics, travel characteristics and sensitive group characteristics of personnel in multiple dimensions. For example, through model calculation, the police can quickly identify "suspected delivery logistics personnel","subway travelers","driving travelers" and "suspected sensitive group personnel", which provides great convenience for the daily work of grassroots police. This intelligent labeling system not only improves police effectiveness, but also enables the police to locate and serve the masses more accurately.
With fine person tag screening and flexible combination functions, users can quickly lock in and accurately identify target groups. On this basis, the system further depicts the action track of the target population in the jurisdiction, and shows its activity mode and flow trend through detailed track description. At the same time, using the thermal map analysis technology of aggregation points, the high-frequency areas of crowd aggregation are intuitively revealed, providing decision makers with intuitive insight into hot spots, potential risk points or high-value areas.
Based on Public Security TRS-GPT LLM, by calling police data, combat analysis algorithm and real-time computing capability of data service platform, it can deeply mine and quickly compare massive alarm data, and enable application scenarios such as major incident command and alarm analysis and report. In this process, the system can automatically identify and mark potential unsafe factors of alarm personnel, including but not limited to key information such as past behavior records of alarm personnel and risk events involved. Once the system detects that there are unsafe factors that may pose a threat to the police, the mobile police terminal of the police will receive the warning prompt message sent by the data service platform in real time. The text message content is refined and critical, directly pointing to potential risk points, providing valuable advance information for the police, enabling them to make full preparations before arriving at the scene, thus greatly improving the safety and success rate of the police process. At the same time, the police can carry out all-round analysis of cases and cases in the jurisdiction area through natural dialogue mode with intelligent question-and-answer robot, such as serial case analysis, case statistics, automatic writing of case analysis report, etc.
Relying on the new police operation mode of "professional + mechanism + big data", TRS assisted Chengdu High-tech Public Security Bureau to enhance the new combat effectiveness of public security and open a new era of intelligent policing through the excellent data integration capability of "data service platform", intelligent graph and number fusion technology and efficient police model application.
