With the rapid development of information technology and the growing demand for financial management and financial analyses, bookkeeping is no longer limited to ordinary handwriting or manual recording in software. In order to meet the needs and development of today's era, an intelligent financial system based on OCR and large language model is designed. The system is suit‐ able for mobile phones, with financial management functions as the starting point and simplified management operations as the pur‐ pose, and realizes a system that is mainly based on OCR recognition scanning and recording bookkeeping functions, supplemented by financial analysis functions of big language model, big data consumption classification trends, and multi-version switching of ver‐ sions suitable for people of different age groups. The system not only greatly improves the efficiency of people's bookkeeping, but also promotes the intelligent financial transformation of society.

With the rapid development of information technology, computer network security has become a focus of social attention. This article explores the teaching innovation and practice of Computer Network Security from the perspective of "ideological and po‐ litical guidance, integration of industry and education", aiming to cultivate outstanding talents with high awareness and skills of net‐ work security. The course utilizes Xuexi Tong APP to carry out blended online and offline teaching. Through the reconstruction of course content, improvement of teaching modes, innovation of teaching methods, integration of ideological and political elements, reform of teaching evaluation, and promotion of teaching effectiveness, the educational philosophy of "moral education first, studentcentered, integration of industry and education, and continuous improvement" is adopted to achieve the deep exploration of ideologi‐ cal and political elements in the course while imparting professional knowledge, integrate professional knowledge of network and se‐ curity with production practice, strengthen school enterprise cooperation, continue teaching reform, and build a learning community for teachers and students.

Aiming at the pain points of Java course teaching process, the teaching model of the course is reformed and innovated. The course deeply integrates PBL and OBE to build a teaching model of multiple integration based on Golden Class. O2O teaching platform is used to divide traditional classes into online and offline mixed teaching. According to the standard of "Golden Class", the innovation reform is carried out from the diversification of curriculum resources, teaching methods, assessment mechanism, curricu‐ lum ideology and politics. The curriculum naturally integrates with the curriculum ideology and politics with local characteristics in teaching, which not only pays attention to students' knowledge generation, but also pays more attention to students' practical ability. It has established a diversified examination mechanism, to strengthen the interaction between teachers and students, and cultivate stu‐ dents' computational thinking and innovation ability.

By analyzing the problems in the traditional teaching mode of C Language Programming， this paper discusses the teach‐ ing framework and implementation plan of applying educational robots to C Language Programming teaching， and discusses the transformation of teacher roles．By introducing educational robots， it is possible to stimulate students' interest in learning， mobilize their internal drive for learning， improve their problem-solving ability， strengthen their teamwork awareness and innovative spirit， and effectively improve teaching effectiveness．

In response to the requirements of the Ministry of Education for cultivating "Four New" composite applied talents in re‐ cent years, combining the disciplinary background of university program design courses to cultivate high-quality students with both professional and practical abilities has become a new research topic. This article relies on the advantages of school and enterprise co‐ operation and proposes a practical research on the teaching mode of university program design courses based on the concept of in‐ dustry education integration, and puts it into practice. By adopting a "363" teaching model and an OBE output oriented practical teaching strategy in the teaching of Python Programming, we aim to strengthen students' subject status in information technology empowerment education, in order to better enhance their computational and innovative thinking abilities under the new medical background, and provide reference for future program design teaching.

The channel uplink scheduler scientifically calculates the relative signal strength parameters of the scheduling value, and processes the program area specified by two symbols in a standardized manner, and specifies the sub-parts of the PUSCH area from different angles. The regularized scheduling area allocated by PRB (the area where the uplink scheduling program starts from the PRB allocation) will be allocated to the cell based on the interference and noise power measured in the area, and finally the corre‐ sponding simulation data will be given to verify the operation of the channel uplink scheduling.

Knowledge graph has been innovated in knowledge representation and management, and is widely used in fields such as intelligent search and question answering, but it has relatively few applications in the field of education and teaching. Based on knowledge graph and natural language processing technology, the main components of the course teaching plan are sorted out, the in‐ formation such as course entities, attributes and relationships are extracted from the talent training scheme and the course teaching plan, and the ontology model of the data science and big data technology professional course system is designed, which realizes the construction of the knowledge graph of the data science and big data technology professional course system.

Helmet detection is a computer vision task with important application value, involving safety management in many fields such as construction sites, mines, and electric power. However, helmet detection also faces many challenges, such as large changes in target size and aspect ratio, rapid changes in target velocity, target occlusion, and background interference. In order to solve these problems, this paper proposes a safety helmet detection method based on YOLOv8, which uses the characteristics of high speed and high precision of YOLOv8 combined with the characteristics of safety helmets to achieve effective detection and identification of safety helmets.

In order to improve the quality of talent cultivation and meet the personalized needs of students, it is necessary to reform and innovate the curriculum system under the background of industry and education integration. Take the computer application tech‐ nology major for example, this article analyzes the current status of the curriculum system in talent cultivation in vocational col‐ leges. Based on the personalized needs of students and the needs of society for talents, it proposes the construction ideas and plans of a diversified curriculum system based on“ big major and small direction”. Practice has proven that the construction of a diversified curriculum system greatly enhances students' practical skills and adaptability to professional positions, improves the quality of talent cultivation, meets students' individual needs, and cultivates high-quality applied talents that meet social needs. 

With the wide application of Web application development technology, many colleges and universities have set up related courses, and the training of Web application development talents is becoming more and more important. The demands for Web appli‐ cation development talents are collected by obtaining and analyzing job demand data from recruitment websites. According to these, a Web application development course group is constructed, which consists of basic courses, application technology courses and practical courses, and a job-demand-oriented production-education integration system scheme is proposed to provide support and help for the training of Web application development talents in colleges and universities.

Conventional wireless local area network intrusion signal detection nodes are mostly set independently, resulting in low detection efficiency and high false detection rate of intrusion signal detection. Therefore, this article proposes a wireless local area network intrusion signal detection method based on GA-SVM algorithm. This method first uses correlation to extract intrusion sig‐ nal features, improve detection efficiency, set up correlation detection nodes, construct a GA-SVM intrusion signal detection model, and use location separation method to achieve signal detection processing. The test results show that for signal intrusion detection at the selected 300 sampling points, compared to the traditional distributed fiber optic network intrusion signal detection group and the traditional FastICA calculation network intrusion signal detection group, the GA-SVM calculation network intrusion signal detection group has a good detection error rate of less than 20%, indicating that with the assistance of the GA-SVM algorithm, the current de‐ sign has better detection results, stronger pertinence, and practical application value.

In the era of digital transformation, Xinjiang Telecom has built a digital management platform for wireless network engi‐ neering construction, which is used to leave traces of key data in the process of wireless network engineering construction. Since the platform has been used, it has stored a large number of trace data. Playing the value of the data has become an important challenge for Xinjiang Telecom. To address this challenge, Xinjiang Telecom strengthens data-driven full process control, streamlines the pro‐ cess of wireless network planning, design, construction, and post evaluation, and constructs a refined management system for wire‐ less network from planning to post evaluation, guiding the optimization and investment management of wireless network engineer‐ ing construction.

Government extranet is an important infrastructure of e-government in China. In the new era, Guangdong Province pro‐ poses to build a new generation government extranet with "resilience, comprehensive integration and extensive coverage", which puts forward higher requirements for the coverage, stability and security of the network. Therefore, combining SPN technology and Super SIM zero trust security gateway technology, this paper discusses the planning, design and construction of a new generation of government extranet.

Smart campus is the mainstream in the field of education informatization. As its infrastructure, education private network has high requirements for network coverage, quality, time ductility, data security and other performance indicators. Integrating 5G slicing technology and edge computing technology, innovative solutions are proposed to build an educational computing network with 5G+ three-level cloud architecture. This network architecture can optimize the distribution of resources and realize the balanced development of education. Ensure network data security and access performance through dedicated slicing; Gradually replace WIFI with 5G to realize the interconnection of devices and everything; Provide construction conditions for new mobile services and appli‐ cations such as AR/VR teaching; Provide a customized converged network; Provides local access management. It implements classi‐ fied cloud network solutions for different types of schools. For new schools, new infrastructure concepts can be incorporated to re‐ duce repeated investment in wireless network resources. For the information pilot school, it can focus on satisfying the application innovation of computing power and the application demonstration of information education. For non-pilot schools, wireless network coverage can be optimized to reduce network construction costs. For remote schools, the goal is to balance educational resources and access the regional cloud nearby to achieve computing power sharing. Schools of different sizes can achieve an education private net‐ work of high safety, high stability, strong bearing capacity, easy expanding with the smallest construction cost. 

In view of facial expression recognition technology, the advantages of self-repair network (Self-Cure Network, SCN) fa‐ cial expression recognition algorithm model are used to realize the application of SCN on mobile terminal devices, using the trained model to build facial expression recognition system, and develop a wechat small program as the system platform．After the system test, the results show that the system has high expression recognition accuracy, convenient operation, smooth operation and complete functions．The SCN face expression recognition system based on wechat small program has a wide range of application prospects．

This paper uses the CiteSpace tool and literature statistical methods to conduct a visual knowledge graph analysis of the literature on the transformation of scientific and technological achievements in software quality inspection for the past 10 years. The high-frequency keywords "information security"、"software testing" and "network security" are current research hotspots with strong centrality, and the future will shift to "data encryption" which prominence intensity is 2.71. To promote the transformation of scien‐ tific and technological achievements, third-party quality testing institutions should cooperate deeply with research and development institutions, and leverage the data advantages of market testing and practical capabilities of achievement transformation, and strengthen detailed research on different sub characteristics in the field of quality inspection, and conduct more specific and intuitive analysis of each sub feature at the software application, and establish standards for testing technology through big data analysis. At the same time, it can also rely on artificial intelligence for cross penetration of sub features, achieving the sharing and complementar‐ ity of research and development resources. This paper summarizes the research strength, hotspots and cutting-edge trends of soft‐ ware quality inspection, and provides some reference value for promoting the transformation of subsequent achievements.

An improved algorithm is proposed to address the problem that yolov5 detects UAV aerial images with poor monitoring performance for low light background targets. Firstly, data normalization is performed on the target dataset visdrone2019 to improve the detection effect. Then the dynamic convolution kernel with Mish activation function and the C3_DSConv module using distrib‐ uted offset convolution to replace the C3 block are introduced, and the above two convolution structures are fused into the yolov5 network; the BiFormer attention mechanism is embedded to improve the accuracy of small target detection. In summary, the MODByolov5 model is finally obtained, and the experimental results prove that the model's mAP and recall are both improved, and the ac‐ curacy of detecting vehicles in shadows and dark environments is significantly increased, and the FPS is high, which ensures that the model can still be used for rapid detection or real-time monitoring.