Tang Dynasty costumes are regarded as a brilliant brushstroke in the history and culture of Chinese costumes, and the fate of the whole Tang Dynasty can be analyzed through the evolution of Tang Dynasty costumes. In this paper, we have constructed a dress semiotics system from the social level, psychological level and cultural trait level, through the transfer of the imagery dress structure to the real dress, to express its symbolic meaning, and applied the constructed system to the Tang dress symbols, to interpret the meaning of the Tang dress symbols from the two levels of society and culture. Using CiteSpace information visualization software, combined with the literature of “Tang Dynasty Costume”, the study explores the dynamic evolution law of Tang Dynasty costume culture. The results of the study show that the earliest year for the keyword “Tang Dynasty costumes” is 1985, and the frequency is as high as 68 times. The keywords with the highest degree of centrality are dress and Tang Dynasty culture, both of which are 0.38. A number of new keywords with strong salience emerged in 2013-2023, among which the ones with a sudden increase of intensity greater than 5 are artistic features, clothing styles, clothing colors, clothing shapes, sweater, structural design, cheongsam, and knitted fabrics, and therefore the future hotspot of the Tang Dynasty dress research shifts to these keywords.

In order to be able to study in-depth image recognition technology for the detection of emergencies, this paper firstly adopts the image processing technology image processing, removes the noise in the image, improves the clarity of the image, and reduces the distortion of the image. Secondly, the signal in the image is extracted, and the network transmission algorithm is used to detect the signal in the image and calculate the corresponding transmission energy value. Finally, a standard threshold is set according to the calculation results, and once the transmission energy exceeds the threshold, it is an abnormal event. The analysis of the emergency event detection model based on image recognition technology shows that the image contrast effect is good, around 8.5 points, indicating that the image quality obtained based on image recognition technology is good. For the third emergency detection, the value based on image recognition technology is 93.3%, the detection results are more accurate, the response speed is faster, the fastest can reach about 1.1s, can real-time feedback on the results of the detection of the emergency situation in a timely manner to deal with the emergency situation, to reduce the loss of personnel, and to improve the efficiency of the management of the smart community emergencies of public health events.

The integration of modern information technology and civil litigation promotes the electronic civil litigation, online litigation as a kind of litigation behavior, promoting the development of traditional trial mode. This paper starts from analyzing the relationship and conflict between civil e-litigation and traditional civil litigation, and organizes the relationship between online trial mode and traditional court trial, and the relationship between civil online trial mode and traditional trial mode respectively. Based on the influencing factors of civil trial, the time proof consensus algorithm and data security transmission algorithm are respectively proposed to combine the network nature of online litigation and blockchain storage data information to optimize the online litigation electronic evidence storage. Summarizing the litigation efficiency of online trial and traditional trial under different control variables, from the point of view of the complexity of the case, the litigation efficiency of online trial mode for more complex cases is significantly higher than that of traditional trial mode. For non-complex cases, the efficiency increases but the difference is not significant. The online litigation mode is a part of the civil online trial mode and serves the traditional civil trial mode.

The traditional personnel recruitment is less efficient and difficult to find the talent that meets the job requirements. This paper firstly constructs the personnel management system of colleges and universities, and clarifies the recruitment process and information management program. Secondly, collect the recruitment information in the personnel management system of colleges and universities, and utilize the fuzzy C class mean algorithm to cluster the user portraits of the applicants to get the user characteristics of different positions. Finally, the joint embedded neural network is used to match the user portraits and positions, set the relevant objective function value, minimize the value of the objective function, and complete the matching of positions. In the personnel management system of colleges and universities, the application and exploration of the application of big data technology in intelligent recruitment found that the job matching rate is high, the highest is about 98.1%, the efficiency of recruitment is faster, the job release to the candidate onboarding only takes 25 days, and the system’s shortest response time is 0.5 s. The high efficiency of the talent recruitment of big data technology can provide timely feedback on the effect of interviews, reduce the cost of interviews, reduce the burden of the staff, and promote the intelligent recruitment of talent. Burden and promote the intelligent development of talent recruitment.

In order to accurately assess the financial status of a company and identify potential anomalies, this paper first implements unsupervised classification of financial transaction data based on Support Vector Machines, which automatically classifies the data into normal and abnormal categories. Histograms are introduced in combination with LightGBM to quickly fuse data from multiple sources. The most suitable first layer is selected by different algorithms, and the outputs of these algorithms are combined with industry-wide common abnormal features as inputs for LightGBM’s second layer identification. With this two-layer structure, the model not only takes into account the industry characteristics, but also the common anomaly features. Empirical results show that in the accuracy of smart financial statement generation, the sensitivity of this paper’s model iterates to 99.99% at 41.25% specificity, and the accuracy of this paper’s model is as high as 0.98 when dealing with financial private information, macroeconomic, and market information.In the identification of financial transaction anomalies, the number of anomalous weeks is identified to be 24, 29, 34, and 36, and the fusion of multi-source data effectively identifies the large amount of financial transactions, fluctuating transactions and other suspicious abnormal transactions.

This paper examines the differences and convergence of regional real estate markets based on panel statistics of 28 provinces, autonomous regions and municipalities directly under the central government in China from 2010 to 2023. Relevant variables such as urban construction land area, population and economic growth are set and the data are processed. The data show that the degree of industrial convergence and circulation costs have a positive spatial correlation and an upward trend from 2015 to 2021. From the viewpoint of regional real estate market divergence, the proportion of the real estate industry in GDP has remained above 5% since 2015, and this proportion is larger in the eastern region, for example, it was 8.74% in Beijing in 2015, but it has slightly decreased in some provinces and cities. The proportion in central and western provinces and cities has been rising faster year by year. The extreme deviation and standard deviation coefficient of the eastern region are relatively large, with the extreme deviation of the eastern region being 4.35% and the standard deviation coefficient being 1.45529 in 2021, indicating that the internal development is not balanced. From the analysis of convergence, the rate of convergence in the absolute convergence test is 3.66%, and the rate of convergence in the conditional convergence test is 2.89%, with a half-life of about 23.8 years. It indicates that the regional real estate market differences are shrinking, showing a trend of convergence, but the convergence process is relatively slow, which provides an important basis for an in-depth understanding of the characteristics of the regional real estate market.

With the increasing complexity of the financial market, corporate financial fraud events occur frequently, posing a serious challenge to investors and market regulators. Aiming at the limitations of traditional financial fraud recognition methods, this paper constructs a financial fraud recognition model MCN based on the topological data analysis method. The model consists of two parts: the Mapper algorithm and one-dimensional convolutional neural network (1DCNN), which combines the global topology extracted by the Mapper algorithm with the local features of the IDCNN to realize the effective identification of financial fraud samples. In order to evaluate the recognition performance of the model, this paper controls the topological feature extraction method unchanged and the classifier unchanged respectively, and compares the performance of the MCN model with other financial fraud recognition models. The results show that the Acc and F1-score of the MCN-based financial fraud recognition model in this paper are 98.69% and 97.64%, respectively, which are better than other models in both perspectives, proving the superiority of the financial fraud recognition model based on topological data analysis constructed in this paper, and thus providing powerful technical support for the regulation of the financial market and the risk management of enterprises.

In this paper, the image parameters are preprocessed by the gray scale histogram statistical image parameters, which reflect the gray scale distribution information of the plant images, using the zero-mean normalization formula. According to different lighting conditions, the plant image is segmented, and the texture feature information in the plant image is extracted by using the improved grayscale covariance matrix. The hyperspectral linear mixing model is constructed, and the MVSA algorithm meta-decomposes the mixing model to solve the solution optimization problem. Using the natural gravity embossing method, produce plant embossed flowers and analyze the features and spectral curves of different parts of the embossed flowers to evaluate the comprehensive use of the embossing method proposed in this paper. The ROI images of 1200 embossed pattern petals were calculated to obtain the sample spectral matrix of embossed petals, in which the reflectance of the central petal was the highest among the three parts at a wavelength of 450 nm, with a reflectance of 0.46487, and then decreased, and then gradually increased to one place after the wavelength was equal to 694, with a reflectance of 0.8. The reflectance of the Shaanxi Weixiang (Weixia), the single side-embossed Yuanbaosi (Yuanbao maple), the hammered elm (fruits), and the pachypodium (Green) obtained a full score of 35 in the comprehensive evaluation after drying, which is a perfect embossed plant material, and all the plant materials embossed using the method proposed in this paper averaged above 30, and the comprehensive effect of plant embossing was good.

Since the financial crisis, the economies of all countries have been affected by the recession triggered by global events, and the uncertainty brought by the changes in economic policies has also become a risky shock, and the uncertainty of economic policies has been climbing worldwide. This paper firstly briefly analyzes the mechanism of economic policy and financial market, in order to comprehensively study the changes of market economic liquidity, this paper starts from the return of the market economy, and adopts the symbolic time series analysis method to analyze the prediction of the financial market by taking the stock market as an example. Then construct the regression model, and then study the impact of economic policy uncertainty on market liquidity. The regression coefficient of economic policy uncertainty is 0.064, which is significant at 1% level. Secondly, when GDP growth rate and inflation level are added as control variables, the regression coefficient of economic policy uncertainty obtained is 0.108, which is still significant at 1% level, implying that a rise in economic policy uncertainty brings about a decline in market liquidity. This study provides an effective analytical tool for the impact of economic policies on market liquidity. It also provides a basis for the government to improve market liquidity and enhance market vitality.

The study adopts a detection followed by tracking paradigm. In the detection stage, the BiFormer dynamic sparse attention module is embedded in the YOLOv8 network model, while the original nearest neighbor interpolation upsampling is improved by replacing it with the lightweight upsampling operator CARAFE. In the target tracking stage, a multi-vehicle steering trajectory tracking algorithm based on particle filtering is proposed, and the particle filtering algorithm is improved by combining the target motion direction weighted resampling algorithm. The two improved algorithms are combined for multi-vehicle detection and tracking in tunnel scenarios, and the average tracking accuracy can reach 97.3%. Compared with the traditional YOLOv8 combined with particle filtering algorithm for tracking, the method in this paper is more advantageous.