Incomplete data, such as the interactive information on social platforms and the review contents in Internet movie datasets, widely exist in the real life. However, most existing sentiment classification models are built on the basis of complete data, without considering the impact of incomplete data on classification performance. To address this problem, a stacked denoising neural network model based on BERT (Bidirectional Encoder Representations from Transformers) was proposed for sentiment classification of incomplete data. This model was composed of two components:Stacked Denoising AutoEncoder (SDAE) and BERT. Firstly, the incomplete data processed by word-embedding was fed to the SDAE for denoising training in order to extract deep features to reconstruct the feature representation of the missing words and wrong words. Then, the obtained output was passed into the BERT pre-training model to further improve the feature vector representation of the words by refining. Experimental results on two commonly used sentiment datasets demonstrate that the proposed method has the F1 measure and classification accuracy in incomplete data classification improved by about 6% and 5% respectively, thus verifying the effectiveness of the proposed model.

Grape quality inspection needs the identification of multiple categories of grapes, and there are many scenes such as depth of field changes and multiple strings in the grape images. Grape recognition is ineffective due to the limitations of single pretreatment method. The research objects were 15 kinds of natural scene grape images collected in the greenhouse, and the corresponding image dataset Vitis-15 was established. Aiming at the large intra-class differences and small inter-class of differences grape images, a multi-scale grape image recognition method based on Convolutional Neural Network (CNN) was proposed. Firstly, the data in Vitis-15 dataset were pre-processed by three methods, including the image rotating based data augmentation method, central cropping based multi-scale image method and data fusion method of the above two. Then, transfer learning method and convolution neural network method were adopted to realiize the classification and recognition. The Inception V3 network model pre-trained on ImageNet was selected for transfer learning, and three types of models-AlexNet, ResNet and Inception V3 were selected for convolution neural network. The multi-scale image data fusion classification model MS-EAlexNet was proposed, which was suitable for Vitis-15. Experimental results show that with the same learning rate on the same test dataset, compared with the augmentation and multi-scale image method, the data fusion method improves nearly 1% testing accuracy on MS-EAlexNet model with 99.92% accuracy, meanwhile the proposed method has higher efficiency in classifying small sample datasets.

Most of the existing methods to compute text similarity based on Vector Space Model (VSM) use TF-IDF scores as the weights of feature terms in text, which ignores the word sense relationships among feature terms and lead to inaccurate text similarity. To improve the accuracy of text similarities calculated by methods based on VSM, a new term weight computing method by integrating word sense was proposed in this paper. Firstly, word sense similarities among feature terms were computed based on the Chinese WordNet. And then, the TF-IDF weights were revised according to the word sense similarities for the purpose of reflecting both the frequency and the word sense of feature terms in text. The experimental results on the HIT IR-lab Multi-Document Summarization Corpus show that to use the weights calculated by the proposed method can efficiently improve the differentiation among document clusters.

According to base bias in the three positions of codon and base chemical properties, the YKW graph, a new graphical representation of gene sequences was introduced for recognizing short coding sequences of human genes. Nine effective features of area matrix were extracted in the YKW curves. In the identifying process, the incremental feature selection algorithm was used to add four statistical features to improve the accuracy. Then Principal Component Analysis (PCA) method was adopted to reduce dimensions and Support Vector Machine (SVM) was applied to classify the coding/un-coding sequence in short human genes. Finally, the experimental results show that the proposed method uses fewer features (seven or four) and gets better recognition results than other methods.

The common technology of the network discovery can only fulfil the local management work and can not fulfil the remote topology discovery of networks.A class of rules of the network topology analysis was advanced.Based on it,an algorithm of remote network discovery covering different management domains was presented.This algorithm can resolve the problem of no longer connected graph in the discovery for different management domains and fulfil the demand of accuracy and integrity in the topology discovery.The test effect in CERNET in east China was given and analysed.