最新刊期
卷 22 ， 期 1 ， 2023
- 摘要：Learning engagement is an important indicator for teachers to study and judge learning activities， which lays a foundation for subsequent teaching diagnosis. With the continuous promotion of the deep integration of AI and teaching， various teaching data media are growing day by day， making multimodal data representation of learning engagement portrait gradually become the technical trend in the teaching field. In view of the lack of complementarities between high and low features of data in previous studies， the research proposes a construction method of multimodal learning engagement feature portrait that integrates text data and image data to improve the representation effect of multi-dimensional learning engagement. Specifically， transformer model is used to extract text semantic features， FasterRCNN model is used to extract visual features， and attention feature fusion is used to obtain the prediction classification results of different dimensions of learning engagement. Finally， visualization technology is used to present the picture of learning engagement. The research shows that the model effectively represents the unbalanced performance of learning groups in all dimensions， helps to carry out teaching diagnosis and collaborative grouping， effectively assists teachers in teaching diagnosis， and improves the teaching efficiency of collaborative grouping activities.关键词：student portrait;teaching diagnosis;cooperative grouping;multi-model data;learning engagement feature
- 158
- |
- 41
- |
- 0
发布时间：2023-02-09 - 摘要：Relevance learning of interdisciplinary knowledge points has always been a key and difficult point in college education. Using knowledge map to push knowledge structure is one of the effective solutions. First， formalize the definition of the knowledge atlas （ECKG） of college education discipline ［S1］； Then， the interdisciplinary knowledge push technology based on the ECKG structure features is proposed； Finally， it analyzes the application of knowledge point retrieval in the discipline education map of colleges and universities， and explains the knowledge point query， knowledge association query and personalized learning resource recommendation technology. The practice shows that the proposed knowledge push tool has a good effect in knowledge point location query， can better complete the personalized intelligent push task of knowledge points， and can provide reference and reference for cross course knowledge push in colleges and universities.关键词：university education;knowledge graph;artificial intelligence-based education;knowledge recommendation
- 88
- |
- 27
- |
- 0
发布时间：2023-02-09 - 摘要：Online teaching is widely adopted， while it suffers from some drawbacks. In online teaching， teachers and students are unable to conduct effective emotional interaction， and students' facial information may be also exposed. Based on the theories of emotion recognition’s neural network， facial blending and 3D animated model binding， propose three solutions to realize the emotional interaction in online teaching， while at the same time protecting the student’s facial privacy. In the first scheme， it maps the emotional analysis results of students to the corresponding emojis by SepConv2D neural network. In the second scheme， styleGAN is employed to generate another facial style to be blended with students’ faces with technic of triangulation and Poisson blending. In the third scheme， using the technics of the face marks detection and Blender 3D animated model binding， it binds the students’ real face with 3D animated models. Based on the technical analysis， propose the different application scenarios for three schemes. The results of experiments and questionnaires demonstrate the emotional interaction feasibility and privacy protection security of the three schemes.关键词：online teaching;emotional interaction;privacy protection;neural network;facial blending;3D animated model binding
- 42
- |
- 12
- |
- 0
发布时间：2023-02-09 - 摘要：With the continuous development of intelligence in higher education schools， how to use massive teaching data to measure educational results and teaching processes through information technology has become a topic of great research challenge. In order to solve this problem， propose a student assessment system based on student profiles. The system uses K-means clustering algorithm to divide student information into six dimensions： mathematics and natural sciences， humanities， professional foundation， professional electives， engineering practice ability， and innovation ability. It also added essays， competitions and other information to paint a profile that comprehensively demonstrates students' attributes. Based on this profile， use SpringBoot， Vue， MySQL and other technologies to design and implement a student assessment system to achieve accurate evaluation of students' comprehensive ability.关键词：K-means algorithm;SpringBoot;student profile;assessment system
- 87
- |
- 14
- |
- 0
发布时间：2023-02-09 - 摘要：With the large-scale popularization of online learning， facing the massive learning resources， the online learning resources recommendation by combining the learners’ cognitive diagnosis analysis， and data mining of learning interests and preferences， has become an important research focus. Based on the bibliometrics analysis via the CiteSpace， the research progress on the personalized online learning resources recommendation is reviewed. The basic research situation is introduced according to the publications amount variation， the key journals， the author's co-occurrence map and affiliations. From the aspects of the co-occurrence map of keywords and co-cited literature cluster map， the hot spots of online learning resources hybrid recommendation are elaborated in detail by combing the traditional algorithms with deep learning， knowledge mapping and cognitive diagnosis. Based on it， the research trends are generalized by supporting the knowledge transfer， fragmented resources integration and dynamic recommendation. Finally， the main directions of future research are pointed out， which could provide important references for promoting the research of online learning resource recommendation.关键词：personalized learning;learning resources recommendation;CiteSpace;hotspots
- 20
- |
- 9
- |
- 0
发布时间：2023-02-09 - 摘要：To understand the development of artificial intelligence in education fields in recent years， VOSviewer and CiteSpace software were used to visualize the relevant literature of artificial intelligence in education fields in CNKI and WoS databases from 2010-2021. By drawing high-frequency keyword relationship network， timezone mapping and timeline mapping， the evolution path， research hotspots and development trend of the domestic and international artificial intelligence in education research fields are introduced. the highly cited and representative literature of different time periods are analyzed in depth to summarize the mainstream research hotspots in the field. The differences， focus and causes of the domestic and international artificial intelligence in education field research are compared， so as to provide reference for promoting the development of domestic artificial intelligence in education research by grasping the development of the field.关键词：artificial intelligence;educational application;bibliometrics;knowledge graph
- 18
- |
- 6
- |
- 0
发布时间：2023-02-09 - 摘要：At present， the traditional collaborative filtering test recommendation only considers the user's test data， ignoring the knowledge points behind the test questions. Therefore， a personalized test question recommendation algorithm （KGeP-CF） based on knowledge atlas and collaborative filtering is proposed. First， build a knowledge map to store knowledge point relationships. Then， use TransE model to learn the vector representation of knowledge point entities， use cosine similarity to calculate the similarity of knowledge points， extract the correlation between knowledge points， and apply it to calculate the similarity of test questions to obtain the similarity of comprehensive test questions. Finally， the user personalized test questions are recommended based on the user's knowledge points. The experiment shows that this method has a good recommendation performance in course test question recommendation. Compared with a variety of recommendation algorithms， KGeP-CF algorithm has better recommendation performance， and has certain reference value for other application scenarios.关键词：knowledge graph;collaborative filtering;TransE model;personalized question recommendation
- 33
- |
- 8
- |
- 0
发布时间：2023-02-09 - 摘要：Under the condition of normalization of epidemic prevention and control， the development of smart education has been vigorously promoted. Propose a "full-stack" smart education system based on the KANO design model and the distributed architecture of Spring Cloud. The system is not only functional but also hierarchical. Different operating logic and system functions are designed for teachers and students. Students can use the system to complete a series of learning activities such as course registration， pre-class material downloading and course viewing. Teachers can use the system to complete the daily management and analysis of the trinity of students and courses and classes. At the same time， the system is based on the KANO design concept and the development method of Spring Cloud's distributed architecture. The "one-stop" education platform can be better expanded into a "full-stack" smart education system，which realizes the development of "full-stack" smart education and the sharing of high-quality teaching resources， and provides new design ideas and development schemes.关键词：full stack;smart education;distributed architecture
- 23
- |
- 5
- |
- 0
发布时间：2023-02-09 - 摘要：COVID-19 has put forward new challenges to university education and teaching， and new requirements for the construction and integration application of university teaching environment， facilities and related platforms， which has enriched the connotation and extension of the construction of university's smart learning environment. Taking Ocean University of China as an example， introduces the digital transformation of higher education in the context of COVID-19， proposes a three-level model for the construction of the university's smart learning environment， and finally analyses the effectiveness of the application of teaching and learning big data empowering education in the smart learning environment， with a view to providing reference for the research， construction and operation of university's smart learning environment， and promoting the further development of smart education in universities.关键词：educational informatization;smart learning environment;teaching mode;teaching big data
- 16
- |
- 2
- |
- 0
发布时间：2023-02-09 - 摘要：In order to solve the problems of boring teaching cases，insufficient personalization of teaching，plagiarism of students' online assignments，and large amount of online assignment correction in traditional classroom， design a real-time automatic testing and scoring online assessment system based on the programming codes submitted by students，which can help students solve the problem of unsound knowledge in the online teaching environment in addition to solving the above problems，and help Students quickly improve their programming experience，algorithmic ability，and time-space trade-off decision-making ability. At the same time，students can see other students' problem solving thinking and algorithmic skills on the system，helping them to develop new learning perspectives. The system can clearly show the whole process of assignments from submission，testing，grade correction，to pointing out the problems，which provides an effective verification and analysis method for students to master programming knowledge， and constructs a unique evaluation system in data visualization.关键词：Online Judge;Django;Python3;Docker;distributed system
- 22
- |
- 7
- |
- 0
发布时间：2023-02-09 - 摘要：With the rapid development of science and technology， the traditional classroom has been difficult to meet the changing learning and teaching methods and innovative talent training needs， the necessity and urgency of future classroom construction is more and more prominent. The research based on the development of education， education informatization development stage theory， theory of knowledge ladder integration technology subject teaching knowledge （TPACK） theory and fuzzy mathematics theory as the instruction，On the basis of analyzing the research status of future classroom design and evaluation at home and abroad，designed the future classroom maturity model， building the future classroom design and evaluation index system， The evaluation standard of future classroom maturity is formulated， the development path and calculation method of future classroom maturity are studied， and the effectiveness of models and standards are tested.关键词：future classroom;evaluation criteria;maturity model;fuzzy comprehensive evaluation
- 8
- |
- 5
- |
- 0
发布时间：2023-02-09 - 摘要：Policy tools are an effective means to achieve the expected goals of education policy. The research on intelligent education policy tools not only has high academic value， but also is an important guarantee to promote the development of intelligent education policy.To understand the status of intelligent education policy in China， combing the development stage of intelligent education policy， explore the intelligent education policy tools in China， from the education policy tools， intelligent education development elements two dimensions， using content analysis， statistical analysis， with the help of Nvivo12PLUS analysis software of national level 70 intelligent education policy text quantitative analysis.The study found that：① China has entered from the initial construction stage of intelligent education and promoted the deep integration of intelligent technology and education；② there are structural differences in the use of policy tools； ③ policy tools are widely used in various development elements of intelligent education， but the use is unbalanced； ④ policy tools and intelligent education development factors are not fit.On this basis， the following policy suggestions are proposed： ① government should optimize the combination of policy tools in the future to promote the coordinated use of policy tools； ② increases the supply of policy tools for network security and teaching mode elements； ③ strengthens the fit of policy tools and intelligent education development elements.关键词：intelligent education;policy tools;text analysis;policy advice;educational elements
- 10
- |
- 6
- |
- 0
发布时间：2023-02-09 - 摘要：Traditional classroom teaching analysis based on manual coding has problems such as excessive dependence on experts， low analysis efficiency， and difficulty in large-scale services. With the construction of smart teaching environment and the development of artificial intelligence technology， making the collection and analysis of classroom teaching process possible. Based on the summary of the current situation of classroom teaching analysis， this paper firstly put forward an intelligent classroom teaching analysis framework with three functional modules of “data collection and processing”， “classroom teaching analysis” and “classroom teaching improvement”. Then， designed the function of intelligent classroom teaching analysis system， and constructed the system technical architecture based on cloud edge end collaboration. Finally， the paper described the practical application of the system. The system can realize the intelligent and large-scale classroom teaching analysis， and provide support for classroom teaching evaluation and reflection based on data.关键词：artificial intelligence;classroom teaching analysis;intelligent analysis;teaching evaluation
- 18
- |
- 8
- |
- 0
发布时间：2023-02-09 - 摘要：Aiming at the characteristics of abstract， complex and wide range of operating system courses， combined with modern education information technology， the "trinity" online course resource library is constructed， and a hybrid teaching mode of different teaching scenarios and different intelligent teaching tools is proposed. Before and after class， we use the Chaoxing Fanya platform， use the rain class in class， and digitally record the whole teaching process. By analyzing teaching big data， teachers realize the transformation of teaching from "experience driven" to "data driven"， and carry out targeted teaching according to students' learning process and knowledge points. The practice shows that this teaching mode can improve students' learning interest and classroom participation， teaching quality and effect， promote the development of hybrid teaching of operating system courses， and make positive contributions to improve the quality of system R&D personnel training.关键词：operating system;Chaoxing Fanya;rain classroom;hybrid teaching;data analysis
- 21
- |
- 4
- |
- 0
发布时间：2023-02-09 - 摘要：For the requirement of emerging engineering education and the characteristics of discrete mathematics course， the blended online-offline teaching based on the data and whole-process feedback is studied. To solve the difficulties the students faced， analyze the teaching designing based on learning theory and data， and realize the whole-process teaching evaluation and feedback. Taking students as center of education， we help students to complete deep learning and self-motivated learning through the discussion， taking part in the designing of exercises， and cooperation of groups. The teaching， involving activities inside-outside textbook and campus， pays attention to the relations of different courses and helps students master the whole logic of computer discipline. The applications show that the teaching practices based on whole-process feedback and data can promote the computational thinking， agree with the output-oriented principle and continuous improvement of professional certification， and help to complete lifetime development of students. The effectivity of the teaching is verified by complementation degree of course goals， changing of CSP certification scores， and prizes in different contests.关键词：discrete mathematics;blended teaching;whole-process feedback;data analysis;self-motivated learning
- 10
- |
- 4
- |
- 0
发布时间：2023-02-09 - 摘要：Augmented reality technology is widely used in augmented reality systems.Whether augmented reality systems help reduce user cognitive load and improve completion efficiency and user experience are still controversial. Therefore， five domestic and international authoritative databases were selected， and a total of 31 sets of data from 25 relevant empirical studies were retrieved and screened for meta-analysis to analyze the moderating effects of different augmented reality devices， system types， and using time. On the whole， augmented reality systems contribute to cognitive load reduction. In terms of device type， cognitive load reduction is better when using non-wearable devices. In terms of system type， both learning systems and informational systems can reduce cognitive load. When using augmented reality information prompting systems， more than 30 minutes will increase cognitive load. In the future design， it is necessary to find the intervention path of AR technology according to user characteristics and task types， so as to improve the system usability. Besides， it’s also necessary to promote the integration and development of multiple disciplines and improve the adaptability of AR system.关键词：human-computer interaction;cognitive load;augmented reality technology;meta-analysis
- 7
- |
- 2
- |
- 0
发布时间：2023-02-09 - 摘要：Quantitative evaluation of students' mastery of knowledge points is an important research topic in the field of intelligence education. The existing cognitive diagnosis methods and collaborative filtering methods ignore the high-order interaction information between test questions， which affects the accuracy of cognitive diagnosis results. In order to solve the above problems， propose a cognitive diagnosis model GCCD based on convolutional neural network. The graph convolutional network is used to construct the student ability vector containing the high-level interaction information of students' test questions as the cognitive diagnosis result， and the interaction process of students' test questions is modeled in a nonlinear way to predict students' performance. Experimental results on students' test data show that the proposed method can effectively reduce errors in students' prediction process and ensure the interpretability of diagnosis results. The accuracy of the method in the student test data set is 93.7%.关键词：GCN;cognitive diagnosis;neural network;high-level interactive information;skill proficiency
- 18
- |
- 7
- |
- 0
发布时间：2023-02-09 - 摘要：Interactive experience of online learning is an important aspect of describing learners' online learning process， and flow experience is an important indicator to evaluate learners' learning experience in online learning environment. By constructing a structural equation model， we conducted a questionnaire survey of online learners， and analyzed the survey results using SPSS and Amos. Practice shows that online learning interaction experience and its dimensions can positively predict emotional engagement， and flow experience plays a partial mediation role in relationship between online learning interaction experience and emotional engagement， student content interaction experience and emotional engagement. Therefore， in the process of online education in the future， learners' emotional engagement in online learning can be promoted by building a benign in-depth interaction mechanism and optimizing resource construction to trigger flow experience.关键词：online learning;interactive experience;flow experience;emotional engagement;structural equation model
- 13
- |
- 7
- |
- 0
发布时间：2023-02-09 - 摘要：At present， the existing evaluation methods of teaching quality in universities are oversimplified， which cannot distinguish the teaching quality of different courses. The comprehensive comments given by experts also have not been well utilized. To effectively use the comprehensive comments， two scoring models based on technical feature and big data feature were proposed， respectively. The results show that the scoring model based on big data feature is more reliable， the accuracy of the model is 87.5%. Meanwhile， the evaluation errors caused by the subjectivity of expert can be reduced， the fuzzy results caused by traditional evaluation methods can also be well avoided，which has important reference value.关键词：teaching quality evaluation;quantification of comprehensive comments;scoring model;technical feature words;feature words of big data
- 8
- |
- 2
- |
- 0
发布时间：2023-02-09 - 摘要：At present， the vision system of most table tennis robots can only provide the position information of the table tennis without considering the rotation of the table tennis， which makes it difficult to accurately predict the rotating ball. Aiming at this problem， a real-time measurement method of table tennis rotation speed based on high-speed vision is proposed. In order to solve the problem of rotating speed measurement of high-speed rotating table tennis， a new vision system scheme is proposed to meet the requirements of rotating speed measurement of high-speed rotating table tennis. Aiming at the problem that the integrity of trademark is not considered in the process of trademark recognition， a three-dimensional position solution method of trademark center based on contour stitching is proposed， which can reduce the calculation error of trademark center and improve the accuracy of rotation measurement. The rotation of table tennis can be estimated according to the movement of the trademark center in space. Finally， the method is verified by using the self-made rotation verification platform. The experimental results show that the proposed method can realize the accurate speed estimation of high-speed rotating ball， and also has high-precision rotation measurement results under the low-speed rotation of table tennis.关键词：table tennis robot;high speed vision;rotation speed;trademark;contour stitching
- 10
- |
- 5
- |
- 0
发布时间：2023-02-09 - 摘要：In order to improve the accuracy of natural gas load forecasting， according to the periodicity and nonlinearity of natural gas load in different time periods， an optimization model based on the combination of correlation vector machine model （RVM） and generalized regression neural network model （GRNN） is proposed in this paper. RVM is used to preliminarily model the natural gas load data， and GRNN is used to nonlinear model the residual of RVM model. The RVM model， the GRNN model and the first mock exam RVM-GRNN model are used to predict the natural gas load values of the central heating and non heating stages respectively. The combined models are compared with the prediction results of the single model respectively. The first mock exam shows that the first mock exam is more accurate than the single model. In the non heating stage and the central heating stage， the combined models MAE， MSE and MAPE are all less than the single model， which are 0.155 8， 0.047 2， 0.0416 and 0.959 7， 1.660 3， 0.027 9 respectively. In addition to comparing the results of the first mock exam with the traditional model， the results of the combination model predict that the combined model is better than the traditional prediction model. Therefore， RVM-GRNN combined model can capture the change law of natural gas load value， meet the requirements of natural gas load prediction， and provide basis for natural gas transmission and pipe network laying.关键词：RVM;GRNN;MAPE;natural gas load forecasting;combined model
- 34
- |
- 6
- |
- 0
发布时间：2023-02-09 - 摘要：In order to better detect the fall problem of the elderly in the home scene， a fall event detection model based on multi-stream convolutional neural network is proposed. Based on the traditional two-stream network， this model adds a new fusion-stream network to form a three-stream network architecture. The fusion flow takes the spatiotemporal features extracted from the temporal flow and the spatial flow as input， fuses the spatiotemporal features of the corresponding layers through the multimodal fusion module， and fuses all the obtained spatiotemporal fusion features through the fusion module to obtain multi-level spatiotemporal fusion features. . Finally， the outputs of the three tributaries are fused with weighted average scores to obtain the final detection result. The experimental results show that compared with the traditional two-stream network， the method has higher precision， recall and F1 value， reaching 95.8%， 91.3% and 93.5% respectively， which proves the feasibility of the method.关键词：fall detection;multi-stream convolutional neural network;feature fusion;deep learning
- 28
- |
- 4
- |
- 0
发布时间：2023-02-09 - 摘要：Reading Comprehension models have achieved human-like performance on reading comprehension datasets. However， many studies have shown that the models are likely to take advantage of the biases in the datasets and complete the task without understanding the context， which reduces model generalization. To address this problem， propose a reading comprehension debiasing method based on counterfactual inference. Firstly， the model is trained on the original training set. Then， we construct counterfactual inputs based on options and questions to extract biases captured by the model. Finally， the predicted output is adjusted by combining the original output and counterfactual outputs for debiasing. We conduct sufficient experiments on Chinese and English representative reading comprehension datasets C3 and Dream， and the results show that the performance of the proposed method on C3 and Dream is improved by 2.31% and 1.21%， respectively. This method can eliminate biases and improve the ability of the models.关键词：reading comprehension;counterfactual inference;bias distillation;bias removal
- 14
- |
- 3
- |
- 0
发布时间：2023-02-09 - 摘要：Aiming at the problem that it is difficult for convolutional neural networks （CNN） to obtain the global features of the text， and bidirectional long short term memory （Bi-LSTM） does not consider the impact of local features on the text classification results， a Chinese text classification method based on the hybrid model of CNN and Bi-LSTM is proposed. In this method， words are taken as the basic unit of text representation， and two channels are constructed in the feature extraction layer. CNN and Bi-LSTM are used to extract the local and global features of the text， and then the local and global features are fused to generate a feature vector containing rich semantics， which is sent to the softmax classifier for classification. Experimental results show that this method can effectively improve the accuracy of Chinese text classification.关键词：convolutional neural network;bidirectional long short-term memory network;deep learning;text classification;feature vector
- 33
- |
- 9
- |
- 0
发布时间：2023-02-09 - 摘要：Most of the current research on task assignment focuses on roles of tasks and workers， and they usually assume that tasks are homogeneous， ignoring the influence of task requirement and pick-up location on task assignment. In order to solve the problem that the travel cost and execution time are too high due to the lack of workers near the pick-up location designated by tasks， propose a time-delayed matching greedy algorithm. On the one hand， the waiting time of tasks is used to obtain more high-quality workers to improve the allocation effect. And on the other hand， matching triples are quickly filtered by the setting of adaptive threshold and constraints to reduce the traversal size and speed up the allocation efficiency. Finally， the performance of the proposed algorithm under different parameters is compared and analyzed under the simulated dataset and the real dataset through experiments， which proves the feasibility and effectiveness of the proposed method.关键词：spatial crowdsourcing;online task assignment;travel cost;greedy algorithm
- 5
- |
- 3
- |
- 0
发布时间：2023-02-09 - 摘要：In order to meet the testing requirements of enterprise Web applications with frequent iterations， a Web automated testing method based on SVM is proposed. Based on selenium framework， this method builds a three-tier architecture of data acquisition layer， machine learning layer and use case execution layer， and takes the Web application of FPGA public cloud platform as the research object to automatically test its front-end page function. The experimental results show that the system can identify the test cases corresponding to the Web elements on the front-end page， and can select the appropriate test cases for the web elements of the new page， and the prediction accuracy is 96.71%. The proposed method has certain reference and reference significance for the existing web automated testing methods.关键词：Web automated testing;SVM;Selenium framework;FPGA public cloud platform
- 11
- |
- 2
- |
- 0
发布时间：2023-02-09 - 摘要：For the erosion failure of gas elbow， DPM discrete model， Forder particle wall bounce model and RNG k-ε turbulence model were used for simulation calculation by FLUENT simulation software. The law between impact point， impact Angle and erosion area of solid particles was studied. The results show that among the influencing factors， such as particle content， particle size， elbow angle and gas velocity， the gas velocity and particle content have significant effects on erosion rate， and the elbow angle and gas velocity have significant effects on erosion rate in the area with serious erosion.关键词：pipe elbow;multiphase flow;numerical simulation;erosion wear
- 7
- |
- 5
- |
- 0
发布时间：2023-02-09 - 摘要：Simulated annealing algorithm is widely used in combinatorial optimization because of its simple structure， but there are few research results on the analysis of algorithm calculation time. Therefore， for the calculation time analysis problems of simulated annealing algorithms， a stochastic process model of simulated annealing algorithm is proposed， which can describe the calculation time of the algorithm. Based on the stochastic process model， the average gain model of simulated annealing algorithm is given， which can estimate the upper bound of the expected first hitting time of simulated annealing algorithm. In order to verify the feasibility of the proposed theoretical method， the expected first hitting time of the simulated annealing algorithm for the linear function is solved， and the closed expression of the time upper bound of the linear function is obtained for different mutation operators. Theoretical analysis and experimental results show that the average gain model can describe the upper bound of the expected first hitting time of simulated annealing algorithm，which provides a new idea for the calculation time analysis of simulated annealing algorithm.关键词：simulated annealing algorithm;average gain;first hitting time;evolutionary algorithms;calculation time
- 19
- |
- 9
- |
- 0
发布时间：2023-02-09 -
Research on Standardization of Traffic Data Elements for TOCCTaking Changsha City as an Example AI导读
摘要：Transportation Operations Coordination Center （TOCC） is a city's transportation intelligence brain， which brings together massive data in the transportation field. However， because there is no unified data element standard among various departments in Changsha to guide and regulate the construction of TOCC， traffic data is faced with the problems of irregular data， difficult data fusion， and low utilization rate. Based on this， combined with the actual situation of Changsha traffic data， this paper firstly explains the attributes of data elements， in which the traffic-related business fields are divided into four categories in the coding attributes of data elements； secondly， the construction data for Changsha traffic data is proposed. Based on the network structure of data elements， the depth-first algorithm is used for business processes and the breadth-first algorithm is used for user views to extract data elements completely and not heavily； finally， according to the business focus and center of the business field Elements build a concise table of data element frameworks. The research on traffic data element standards carried out in this paper can ensure the quality of traffic data interaction， promote data integration， improve data utilization， and better serve TOCC.关键词：TOCC;data element;standardization;smart transportation- 11
- |
- 4
- |
- 0
发布时间：2023-02-09 - 摘要：The collaborative filtering recommendation algorithm based on matrix factorization can mine user preferences and hidden features， but the algorithm only uses user-item scoring matrix， without considering semantic similarity between entities. In order to solve the above problems， a recommendation algorithm based on user preference and semantic similarity is proposed. Firstly， the semantic data of entities and relationships are embedded into the low-dimensional semantic space by using the knowledge graph representation learning method， and the semantic similarity between entities is calculated. Then， the semantic similarity of entities is fused in the objective function of matrix factorization model to make up the deficiency of the recommendation algorithm. Finally， the algorithm is tested in public movie data set. The recommendation performance of the proposed algorithm is better than that of the comparison algorithm. The accuracy rate， recall rate and F1 value of the recommendation evaluation index are improved by 9.25%， 5.54% and 8.7% on average. The experimental results show that this algorithm makes up for the shortcomings of the recommendation algorithm and improves the recommendation effect of the recommendation algorithm.关键词：recommendation algorithm;semantic similarity;expression learning;knowledge graph
- 16
- |
- 8
- |
- 0
发布时间：2023-02-09 - 摘要：The data analysis for anti-money laundering is one of the vital technologies for financial security. With the rapid development of the Internet and big data， it is urgent to form a full-dimensional view of corporate customers in the financial field， but traditional anti-money laundering methods lack in effective data analysis capabilities of mining the valuable customer information effectively to identify and fight against financial risks. One of the effective tools to address this problem is to build a financial knowledge graph platform， to enable better business department fusion in financial industries. To this end， describe the motivation and framework of developing the anti-money laundering software based on knowledge graph. Then， the key technologies of anti-money laundering software are presented in detail. Further， the proposed software is validated with real cases. Applications show the higher identification rate of this software， which enables big data based active early-warning and prevention software updating for anti-money laundering.关键词：knowledge graph;anti-money laundering software;artificial intelligence;data association
- 15
- |
- 4
- |
- 0
发布时间：2023-02-09 - 摘要：In order to improve the fusion effect of multispectral and panchromatic images， a new method based on anisotropic gradient domain guided filtering （AnisGGF） and parameter adaptive pulse coupled neural network was proposed. First， the MS image is converted from RGB to YUV color space， and NSST multi-scale decomposition is performed on the luminance component Y and PAN images； secondly， for the low frequency subband， the PAN low frequency subband is used as the guide map， and AnisGGF is used to analyze the low frequency of the Y component. The sub-band completes the injection of spatial information； for the high-frequency sub-band， a PAPCNN fusion strategy is introduced in which the link strength is determined by the variance； finally， the inverse NSST and YUV transforms are performed to obtain the fusion image. Comparing experiments with 4 methods are carried out on the images of GeoEye， GF-2 and GF-6 satellites. Visually， the proposed method has stronger spatial detail preservation and spectral information fidelity ability. At the same time， the 6 quantitative evaluation indicators show that the average performance of the method is improved by 68.24%， 25.8%， 22.05% and 12.28% compared with SE， NSCT-PCNN， ISCM and GF， respectively.This method can provide a new solution for the field of image fusion， and can significantly improve the effect of image fusion.关键词：image fusion;non-subsampled shearlet transform;guided filtering;pulse coupled neural network;anisotropic
- 18
- |
- 6
- |
- 0
发布时间：2023-02-09 - 摘要：Aiming at the error caused by skip connections and the overburden of the decoder in the process of fisheye image correction by generative adversarial network， an algorithm combining generative adversarial network and fisheye correction parameter prediction network is proposed. The algorithm adds a fisheye correction parameter prediction module to the generative network of the model， takes the output of the encoder of the generative network as the input of the prediction network， and then embeds the predicted parameters into the decoder of the generative network， and uses the predicted parameters to guide the decoding process of the fisheye image to reduce the error loss and the burden of the decoder caused by the skip connection， and improve the quality of the model-corrected fisheye image. Experiments show that on the fisheye image dataset， The peak signal to noise ratio （PSNR） of the corrected image and background image can reach 28.039， which is 9.5% higher than that of the advanced method， and the structural similarity （SSIM） can reach 0.875， which is 7.5% higher than that of the advanced method.关键词：generative adversarial network;fisheye image;image correction;full convolution neural network
- 12
- |
- 2
- |
- 0
发布时间：2023-02-09 - 摘要：In the process of industrial wastewater treatment， the detection of sludge sedimentation ratio plays a vital role in the effect of wastewater treatment the traditional sludge settling ratio test is mainly manual， which is a large and uncontrollable workload and can cause unaccountable errors in the results. Therefore， propose a detection method of sludge sedimentation ratio based on improved LeNet-5 neural network. Before training， the color threshold of the target is determined. During training， a lightweight feature reuse network model and a regular classifier model are proposed to eliminate the edge effect of Label-dropout， Finally， the output results are used to build the whole system with the help of the client/server （C/S structure） mode. The experimental results show that the accuracy of the improved neural network to the test set is as high as 96%， which is far higher than the traditional neural network and artificial methods. Moreover， the improved neural network is more suitable for the classification and recognition of small sample data sets， greatly improving the accuracy and efficiency.关键词：sludge sedimentation;LeNet-5;sewage treatment;C/S structure;Label-dropout;regular classifier
- 13
- |
- 3
- |
- 0
发布时间：2023-02-09 - 摘要：Aiming at the slow development of small target detection， poor detection effect and poor smoking prevention and control effect in public places， a cigarette target detection model based on YOLOv5 algorithm combined with SENet attention mechanism is proposed. Firstly， the model adds the Squeeze-and-Excitation Networks （SENet） attention mechanism to the backbone network. SENet adopts a new feature re-calibration strategy to automatically obtain the weight of each feature channel through self-learning. The weight value of the important feature channel of the big cigarette target. Secondly， CIoU is used as the location box regression loss function， and the center point distance and aspect ratio are normalized to quickly match the real box and the predicted box. Comparing YOLOv5 and the improved model in the self-built data set， the results show that the accuracy of this model reaches 0.85， the recall rate reaches 0.75， the F1 reaches 0.797， and the mAP reaches 0.81. Compared with other mainstream models in the field of object detection， the detection performance has been improved.关键词：cigarette object detection;YOLOv5;SEnet attention mechanism;CIoU
- 128
- |
- 47
- |
- 0
发布时间：2023-02-09 - 摘要：Unsupervised learning text clustering is an important branch of natural language processing， which is widely used in practice. In order to make this technology play a leading role in text clustering technology， firstly， the text clustering process， clustering evaluation indicators and data sets are described in detail， then the text clustering algorithms are classified， explained and compared， and finally the text clustering technology is summarized and prospected. By summarizing the current text clustering technology and integrating the latest research results after the deep learning method， so that to provide reference for the in-depth study of this field.关键词：unsupervised learning;text clustering;deep learning;clustering algorithm
- 29
- |
- 3
- |
- 0
发布时间：2023-02-09 - 摘要：With the update iteration of algorithm technology and the wide application of big data technology， the problem of algorithm bias has gradually attracted the attention of academia and society. In order to explore the research progress and development status of this issue， the relevant English literature in the core database of Web of Science is taken as the research object by using the bibliometric method. Through clustering analysis， the basic knowledge and research frontiers in this field are sorted out to integrate the current research hot issues. At the same time， according to the existing literature analysis results， the overall idea and framework of the current foreign research on the subject of algorithm bias are constructed， and on this basis， the existing bias in the algorithm is classified to analyze the sources of various categories of bias and feasible correction methods， with a view to providing reference for researchers engaged in algorithm bias.关键词：algorithmic bias;CiteSpace;knowledge base;research frontier;source of bias;correction of bias
- 15
- |
- 2
- |
- 0
发布时间：2023-02-09 - 摘要：There are a large number of wireless devices with different protocols in the 2.4GHz band， and the coding mode， transmission rate and power of different protocols are not uniform， cross-technology interference（CTI） is easy to occur among devices. Therefore， in the complex network environment， identifying the interference and improving the anti-interference ability of the equipment become the research hotspot of wireless network. Firstly， the characteristics of different technologies in the frequency band are summarized， and the causes of CTI are analyzed. Secondly， the methods of interference identification and avoidance are analyzed. Finally， summarize the current problems of interference in wireless sensor networks，so as to provide a reference for wireless sensor network interference problems..关键词：wireless sensor network;cross-technology interference;interference identification;interference avoidance;interference coexistence
- 15
- |
- 4
- |
- 0
发布时间：2023-02-09
- Technical support is provided by Beijing Founder electronics co., LTD 鄂ICP备12016542号-70
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0