A Dynamic Weighted Method with Support Vector Machines for Chinese Word Segmentation

Evidence-based treatment is the basis of traditional Chinese medicine (TCM), and the accurate differentiation of syndromes is important for treatment in this context. The automatic differentiation of syndromes of unstructured medical records requires two important steps: Chinese word segmentation and text classification. Due to the ambiguity of the Chinese language and the peculiarities of syndrome differentiation, these tasks pose a daunting challenge. We use text classification to model syndrome differentiation for TCM, and use multi-task learning (MTL) and deep learning to accomplish the two challenging tasks of Chinese word segmentation and syndrome differentiation. Two classic deep neural networks—bidirectional long short-term memory (Bi-LSTM) and text-based convolutional neural networks (TextCNN)—are fused into MTL to simultaneously carry out these two tasks. We used our proposed method to conduct a large number of comparative experiments. The experimental comparisons showed that it was superior to other methods on both tasks. Our model yielded values of accuracy, specificity, and sensitivity of 0.93, 0.94, and 0.90, and 0.80, 0.82, and 0.78 on the Chinese word segmentation task and the syndrome differentiation task, respectively. Moreover, statistical analyses showed that the accuracies of the non-joint and joint models were both within the 95% confidence interval, with < 0.05. The experimental comparison showed that our method is superior to prevalent methods on both tasks. The work here can help modernize TCM through intelligent differentiation.

A wide variety of neural-network architectures have been proposed for the task of Chinese word segmentation. Surprisingly, we find that a bidirectional LSTM model, when combined with standard deep learning techniques and best practices, can achieve better accuracy on many of the popular datasets as compared to models based on more complex neuralnetwork architectures. Furthermore, our error analysis shows that out-of-vocabulary words remain challenging for neural-network models, and many of the remaining errors are unlikely to be fixed through architecture changes. Instead, more effort should be made on exploring resources for further improvement.

Chinese automatic word segmentation is the premise of Chinese information processing, which is widely used in Chinese full-text retrieval, Chinese automatic full-text translation, Chinese text-to-speech conversion (TTS) and other fields. A dictionary plays an important role in Chinese word segmentation. The advantages and disadvantages of the word segmentation mechanism directly affect the speed and efficiency of Chinese word segmentation. Therefore, we propose a deep learning method for Chinese word segmentation. First, a separable convolution bidirectional long and short-term memory condition random field word segmentation model with feature points containing dictionary features is constructed. The model parameters are obtained by training on the existing word segmentation corpus. Then, the software engineering field text is used as the small-scale word segmentation training corpus. The word segmentation model of general corpus is fine-tuned. The experimental results show that the transfer learning reduces the iteration times of the domain segmentation model. Meanwhile, compared with other Chinese word segmentation models, the proposed model can reduce the corpus labeling time in training process and realize the cross-domain transfer learning of word segmentation model.

Large alphabet languages such as Chinese present different problems for language modelling compared to small alphabet languages such as English. In this paper, we describe adaptive models of Chinese text based on the Partial Predictive Match (PPM) text compression scheme that learns the language as the text is processed sequentially. We describe several character-based, word-based and part-of-speech (POS) based variants of PPM that achieve significant improvements in compression rate over existing models. Interestingly, results for Chinese text contrast that achieved for English text, with character-based models outperforming the word and POS based models rather than the other way round. We then explore how well these models perform at the task of Chinese word segmentation.

Internet of Things (IoT) is a popular social network in which devices are virtually connected for communicating and sharing information. This is applied greatly in business enterprises and government sectors for delivering the services to their customers, clients and citizens. But, the interaction is successful only based on the trust that each device has on another. Thus trust is very much essential for a social network. As Internet of Things have access over sensitive information, it urges to many threats that lead data management to risk. This issue is addressed by trust management that help to take decision about trustworthiness of requestor and provider before communication and sharing. Several trust-based systems are existing for different domain using Dynamic weight method, Fuzzy classification, Bayes inference and very few Regression analysis for IoT. The proposed algorithm is based on Logistic Regression, which provide strong statistical background to trust prediction. To make our stand strong on regression support to trust, we have compared the performance with equivalent sound Bayes analysis using Beta distribution. The performance is studied in simulated IoT setup with Quality of Service (QoS) and Social parameters for the nodes. The proposed model performs better in terms of various metrics. An IoT connects heterogeneous devices such as tags and sensor devices for sharing of information and avail different application services. The most salient features of IoT system is to design it with scalability, extendibility, compatibility and resiliency against attack. The existing works finds a way to integrate direct and indirect trust to converge quickly and estimate the bias due to attacks in addition to the above features.

To quantitatively research about ethical level of network businesses, we propose the evaluation model of network business's professional ethics based on dynamic weighting method. The model combined BP neural network and particle swarm optimization algorithm (PSO). It can help to get a simulation and prediction network business's professional ethics level by using the practical investigation data. The experiment results show that the algorithm has the advantages of fast convergence speed, high precision and it is an effective assessment of occupation moral score model.

PurposeThis study identifies a methodology to explore the issues of enterprise innovation ecosystem health (EIEH).Design/methodology/approachFirst, this study constructs the indicator system of EIEH based on the research objective; second, the dynamic vertical projection method (DVPM) and entropy weight method are proposed to analyze the status and influencing factors of EIEH; finally, the future development of EIEH is analyzed using GM (1,1).FindingsIn terms of methodology, the DVPM can effectively analyze EIEH, which can not only analyze the development status and potential of EIEH every year but also analyze the comprehensive state of EIEH for many years. In terms of practice, the value and grade of EIEH in China have been gradually increasing from 2016 to 2020, but the overall development is unbalanced, and five key factors affecting EIEH have been identified. The EIEH in China is predicted to steadily grow from 2021 to 2025.Originality/valueThe analytical method employed in this study can effectively analyze EIEH, which provides a new analytical perspective for the evaluation of EIEH and enriches the research content of the enterprise innovation ecosystem (EIE). By analyzing the results, we can gain a comprehensive understanding of the state of different EIEs, enabling each EIE to design tailored remedial measures to enhance EIEH and achieve sustainable development.

Chinese word segmentation is a very important problem for Chinese information processing. Chinese word segmentation results are the basis for computers to understand natural language. However, unlike most Western languages, Chinese words do not have fixed symbols like white space as word segmentation marks. Moreover, Chinese has a very complex grammar, and the word segmentation criteria are varied with the contexts. Therefore, Chinese word segmentation is a very difficult task. Many existing works have proposed many algorithms to solve this problem. However, to our best knowledge, none of them could outperform all the other methods. In this paper, we develop a novel algorithm based on semantics and contexts. We propose a semantic-based word similarity measure using the concept hierarchy in knowledge graphs, and use this measure to prune the different results which are generated by several state-of-the-art Chinese word segmentation methods. The idea is to respectively compute the concept similarity of these words to other words in the text, and choose the word with the highest concept similarity score. To evaluate the effectiveness of the proposed approach, we conduct a series of experiment on two real datasets. The results show that our method outperforms all the state-of-the-art algorithms by filtering out wrong results and retaining correct ones.

Particle Swarm Optimization (PSO), a meta-heuristic algorithm inspired by swarm intelligence, is widely applied to various optimization problems due to its simplicity, ease of implementation, and fast convergence. However, PSO frequently converges prematurely to local optima when addressing single-objective numerical optimization problems due to its inherent rapid convergence. To address this issue, we propose a hybrid differential evolution (DE) particle swarm optimization algorithm based on dynamic strategies (MDE-DPSO). In our proposed algorithm, we first introduce a novel dynamic inertia weight method along with adaptive acceleration coefficients to dynamically adjust the particles’ search range. Secondly, we propose a dynamic velocity update strategy that integrates the center nearest particle and a perturbation term. Finally, the mutation crossover operator of DE is applied to PSO, selecting the appropriate mutation strategy based on particle improvement, which generates a mutant vector. This vector is then combined with the current particle’s best position through crossover, aiding particles in escaping local optima. To validate the efficacy of MDE-DPSO, we evaluated it on the CEC2013, CEC2014, CEC2017, and CEC2022 benchmark suites, comparing its performance against fifteen algorithms. The experimental results indicate that our proposed algorithm demonstrates significant competitiveness.

In this paper, we represent a dynamic context-dependent weighting method for vector space model. A meaning is relatively decided by a context dynamically. A vector space model, including latent semantic indexing (LSI), etc. relatively measures correlations of each target thing that represents in each vector. However, the vectors of each target thing in almost method of the vector space models are static. It is important to weight each element of each vector by a context. Recently, it is necessary to understand a certain thing by not reading one data but summarizing massive data. Therefore, the vectors in the vector space model create from data set corresponding to represent a certain thing. That is, we should create vectors for the vector space model dynamically corresponding to a context and data distribution. The features of our method are a dynamic calculation of each element of vectors in a vector space model corresponding to a context. Our method reduces a vector dimension corresponding to context by context-depending weighting. Therefore, We can measure correlation with low calculation cost corresponding to context because of dimension deduction.

The permutation flow shop scheduling problem (PFSP) is one of the hot issues in current research, and its production methods are widely used in steel, medicine, semiconductor, and other industries. Due to the characteristics of permutation flow (optimize the production process through the principle of symmetry to achieve efficient allocation and balance of resources), its task processes only need to be sorted on the first machine, and the subsequent machines are completely symmetrical with the first machine. This paper proposes an enhanced particle swarm optimization algorithm (EPSO) for the PFSP. Firstly, in order to enhance the diversity of the algorithm, a new dynamic inertia weight method was introduced to dynamically adjust the search range of particles. Secondly, a new speed update strategy was proposed, which makes full use of the information of high-quality solutions and further improves the convergence speed of the algorithm. Subsequently, an interference strategy based on individual mutations was designed, which improved the universality of the model’s global search. Finally, to verify the effectiveness of the EPSO algorithm, six benchmark functions were tested, and the results proved the superiority of the EPSO algorithm. In addition, the average relative error of the improved algorithm is at least 21.6% higher than that of the unimproved algorithm when solving large-scale PFSPs.

In order to improve the ability of the modular multi‐level converter to quickly clear and isolate faults in the multi‐terminal flexible DC grid, this paper analyzes the characteristics of current faults in the multi‐terminal flexible direct current grid, introduces the comprehensive distance of current fault waveform, and proposes a line protection method based on the similarity of current fault waveform. This protection method firstly uses Euclid Dynamic Time Warping (DTW) distance and entropy weight method to process electrical quantity signals, calculates the comprehensive distance of current fault waveform of two converter stations, and builds the protection criterion based on the similarity of current fault waveform. Then, the fault identification is realized by using the calculated comprehensive distance, and the fault pole identification is realized by using the ratio of comprehensive distance between positive and negative poles. Finally, PSCAD was used to build a simulation model to output the fault data under different fault conditions, and MATLAB was used to process the fault data to verify the protection algorithm. The simulation results show that the proposed method has good resistance to high resistance, and its rapidity and adaptability are further improved. Compared with other similarity methods, this method combines Euclidean DTW distance and entropy method to process the signal, realizes the fault pole identification, and can still operate correctly under the interference of 500 Ω transition resistance and 20 dB noise ratio. The research results will provide a theoretical basis for the safe and reliable operation of flexible DC power grid.

Chinese word segmentation is the basis of the Chinese natural language processing (NLP). With the development of the deep learning, various neural network models are applied to the Chinese word segmentation. However, current neural network models have the characteristics of artificial feature extraction, nonstandard word-weight, inability to effectively use long-distance information and long training time of models in Chinese word segmentation. To solve a series of problems, this article presents a CNN-Bidirectional GRU-CRF neural network model (CNN Bidirectional GRU CRF Network, CBiGCN), which breaks through the limit of conventional method window, truly realizes end-to-end processing and applies to the neural network model by the five-Tag set method, bias-variable-weight greedy strategy and supplements by Goldstein-Armijo guidelines. Besides, this model, with simple structure, is easy to be operated. And it can automatically learn features, reduces large amounts of tasks on specific knowledge in the form of handcrafted features and data pre-processing, makes use of context information effectively. The authors set an experiment with two data corpuses for Chinese word segmentation to evaluate their system. The experiment verified their new model can obtain better Chinese word segmentation results and greatly reduce training time.

In the absence of inter-word spaces, Chinese readers rely on other available information for word segmentation. An earlier study demonstrated that the valence of words influences word segmentation (Huang et al., Psychonomic Bulletin & Review, 31 (4), 1548-1557, 2024). The current study further investigated the influence of arousal, another key dimension of emotion, on Chinese word segmentation. We first re-analyzed the segmentation results from Huang et al.'s study and found that arousal had an independent effect on Chinese word segmentation. In the experimental study, we manipulated the arousal levels of words while keeping valence at a neutral level. The results provide evidence that the arousal of words can affect Chinese word segmentation, with higher-arousal words being more likely to be segmented than low-arousal words. Moreover, our findings are also essential for understanding the impact of arousal on word processing and suggest that it impacts the early stage of activating a word's representation.

Chinese word segmentation has received extensive attention in recent years. The word segmentation method based on character-based tagging improves the performance of word segmentation greatly. This method transforms the word segmentation problem into a sequence labeling problem, which has become the main word segmentation method. In order to further study the word segmentation performance of this method, we use the maximum entropy sequence labeling model in this paper. We used two different word position sets and three feature templates to compare the experimental results. We have done further research on the unknown words and segmentation ambiguity in the word segmentation results. First we combined N-Gram with cohesion and degree of freedom to solve the problem of unknown words. Then the maximum entropy model is used to train the new participle to eliminate the ambiguity. The closed test was conducted on the Bakeoff 2005 corpus of the international Chinese word segmentation evaluation. Experiments show that the six-tag position combined with the corresponding feature templates can achieve better word segmentation performance. After adding unknown words and disambiguation processing, the word segmentation performance of some data sets can be further improved to optimal results of Bakeoff 2005.