Mining Relevant Text Features for Retrieving Web Information

The quality of extracted features is the key issue to text mining due to the large number of terms, phrases, and noise. Most existing text mining methods are based on term-based approaches which extract terms from a training set for describing relevant information. However, the quality of the extracted terms in text documents may be not high because of lot of noise in text. For many years, some researchers make use of various phrases that have more semantics than single words to improve the relevance, but many experiments do not support the effective use of phrases since they have low frequency of occurrence, and include many redundant and noise phrases. In this paper, we propose a novel pattern discovery approach for text mining. This approach first discovers closed sequential patterns in text documents for identifying the most informative contents of the documents and then utilise the identified contents to extract useful features for text mining. We develop a novel fusion method based on Dempster-Shafer's evidential reasoning which allows to combine the pieces of document to discover the knowledge (features). To evaluate the proposed approach, we adopt the feature extraction method for information filtering (IF). The experimental results conducted on Reuters Corpus Volume 1 and TREC topics confirm that the proposed approach could achieve excellent performance.

It is a big challenge to guarantee the quality of extracted features in text documents to describe user interests or preferences due to large amounts of noise. Over the years, pattern mining-based approaches to RF have attracted great interest to discover knowledge of user interest from text documents. However, the data mining approaches often produce a large set of patterns, which include a lot of noisy patterns, reducing the effective use of pattern mining. In this paper, we present a novel pattern mining approach to RF. This approach mines patterns in both positive and negative feedback and then classifies them into clusters to find user-specific patterns. We also propose a novel pattern deploying method that effectively uses the discovered patterns for improving the performance of searching relevant documents. Experiments are conducted on Reuters Corpus Volume 1 data collection (RCV1) and TREC filtering topics. The results show that the proposed approach achieves promising performance comparing with state-of-the-art term-based methods and pattern-based ones.

Nowadays, the big challenge for community of Information Retrieval domain is to discover the relevance feature in text document which helps to decide whether document is relevant or irrelevant. Most existing text mining methods are based on term-based approaches which extract terms from a training set for describing relevant feature. In termbased approach multiple meaning of same word in different context leads to generate polysemy and synonymy issue. The term based approach also suffers from low level support problem. Even though pattern based text mining approach solve low level support problem but still this approach suffers from large number of noise pattern. In the propose work, a pattern discovery approach for text mining is explored. This approach discovers frequent sequential pattern and closed sequential patterns in text documents for identifying the most information contents of the documents and extract useful features for text mining. It also classifies extracted terms into three categories: positive terms, general terms, and negative terms. In this way, we update extracted features using multiple revising strategies. This technique discovers positive and negative patterns in text documents as higher level features in order to accurately weight low-level terms based on their specificity.

The prediction of time-series data is a challenging and complex issue. For many practical applications, network topology information and text information can play a perfect role in time-series prediction. This article takes stock data as an example by constructing a graph, connecting each stock’s upstream and downstream industries, and obtaining useful text features and topological features to predict the stock time-series. Based on the time-series data features, text features, and the topological features of the stock industry chain of machine learning, we compared our prediction model with other fuzzy time-series prediction methods, which are only based on historical features. The experiment shows that our method is better than the other methods in terms of the performance of multiple stocks in each stock’s time-series prediction. The experimental results show that the stock topology based on the industrial chain effectively improved time-series forecasting accuracy.

As one of the most popular social media platforms, microblogs are ideal places for news propagation. In microblogs, tweets with both text and images are more likely to attract attention than text-only tweets. This advantage is exploited by fake news producers to publish fake news, which has a devasting impact on individuals and society. Thus, multimodal fake news detection has attracted the attention of many researchers. For news with text and image, multimodal fake news detection utilizes both text and image information to determine the authenticity of news. Most of the existing methods for multimodal fake news detection obtain a joint representation by simply concatenating a vector representation of the text and a visual representation of the image, which ignores the dependencies between them. Although there are a small number of approaches that use the attention mechanism to fuse them, they are not fine-grained enough in feature fusion. The reason is that, for a given image, there are multiple visual features and certain correlations between these features. They do not use multiple feature vectors representing different visual features to fuse with textual features, and ignore the correlations, resulting in inadequate fusion of textual features and visual features. In this paper, we propose a novel fine-grained multimodal fusion network (FMFN) to fully fuse textual features and visual features for fake news detection. Scaled dot-product attention is utilized to fuse word embeddings of words in the text and multiple feature vectors representing different features of the image, which not only considers the correlations between different visual features but also better captures the dependencies between textual features and visual features. We conduct extensive experiments on a public Weibo dataset. Our approach achieves competitive results compared with other methods for fusing visual representation and text representation, which demonstrates that the joint representation learned by the FMFN (which fuses multiple visual features and multiple textual features) is better than the joint representation obtained by fusing a visual representation and a text representation in determining fake news.

Information retrieval (IR) is widely used in automatically requirements traceability recovery. Corresponding approaches are built based on textual similarity, that is, the higher the similarity, the higher possibility of artifacts related. A common work of many IR-based techniques is to remove false positive links in the candidate links to achieve higher accuracy. In fact, traceability links can be recovered by different kinds of information, not only the textual information. In our study, we propose to recover more traceability links by exploring both textual features and structural information. Specifically, we use combined IR techniques to process the textual information of the software artifacts, and extract the structural information from the source code, establishing corresponding code relationship graphs. We then incorporate such structural information into the traceability recovery analysis by using graph embedding. The results show that combined IR techniques and using graph embedding technology to process structural information can improve the recovery traceability.

Social network studies are becoming increasingly popular and have been applied to several fields of study such as law enforcement, marketing, spread of disease, as well as in the improvement of organizational performance. One area that is yet to be explored relates to harnessing the power of social networks as recommender systems. The idea that users may provide other users with recommendations that are more relevant than naive approaches is long known. However, the approaches currently implemented are based on the creation of simple relationships such as co-purchase of similar items. Similarly, video websites would like to suggest related videos to users to maximize the time they spend on their sites. Ergo it is crucial that sites like YouTube provide users with recommendations that are relevant to them. Moreover, given the large amount of videos on YouTube, a good recommender system may alleviate users' efforts on finding videos that interest them the most. Existing recommender systems for YouTube are typically based on finding similarities between the videos' textual features (video title or tags annotations) and matching these features to tags in the user's profiles or the videos they are currently watching. This approach is very limited because it restricts the users interests to the current theme of the video being played or the preferences in their profiles. This means that if one is watching a movie about football, it is assumed that the user would like recommendations on other videos about football. In this paper, we attempt to extract information about video relationship using a network formed from reviews left as comments in YouTube videos. We create a network of videos called YouTube Recommender Network (YRN) and use complex network analysis on this network as the basis of a recommender system. Our results show that our list of recommended videos is more diverse than the ones based on textual information. Our YRN provides diversity and captures other important characteristics such as high rating and most-viewed count.

Depression can negatively affect a person's mental health. It can make you unusually sad, worthless, even suicidal. In terms of depression, the depressive tendency is an inevitable experience. The speech of people with depressive tendencies tends to be monotonous and lifeless. In addition, people with depressive tendencies prefer to use negative words to express their inner feelings. Some studies have suggested there are some differences in speech and text information between the depressive tendency and the healthy control group. Consequently, this paper assesses depressive tendency via the Gated Recurrent Unit from speech and text information and makes a comparative analysis of different speech styles, gender, and emotional stimulation. The experimental results indicate that the speech signal and text information are effective in the identification of depression, the speech signal takes a better recognition effect, and the recognition effect of the female is better than that of the male.

In the bases of increasing the volume of text information, the dealing with text information has become incredibly complicated. The text clustering is a suitable technique used in dealing with a tremendous amount of text documents by classifying these set of text documents into clusters. Ultimately, text documents hold sparse, non-uniform distribution and uninformative features are difficult to cluster. The text feature selection is a primary unsupervised learning method that is utilized to choose a new subset of informational text features. In this paper, a new algorithm is proposed based on β-hill climbing technique for text feature selection problem to improve the text clustering (B-FSTC). The results of the proposed method for β-hill climbing and original Hill climbing (i.e., H-FSTC) are examined using the k-mean text clustering and compared with each other. Experiments were conducted on four standard text datasets with varying characteristics. Interestingly, the proposed β-hill climbing algorithm obtains superior results in comparison with the other well-regard techniques by producing a new subset of informational text features. Lastly, the β-hill climbing-based feature selection method supports the k-mean clustering algorithm to achieve more precise clusters.

BackgroundNasopharyngeal carcinoma (NPC), an endemic malignancy in Southeast Asia and southern China, necessitates precise delineation of the gross tumor volume (GTV) for radiotherapy to optimize patient prognosis. Positron emission tomography (PET) imaging offers valuable metabolic insights to guide radiotherapy planning but is hampered by manual GTV segmentation, which is labor-intensive and prone to inter- and intra-observer variability. Conventional segmentation methods are inherently limited, and comparative evaluations of deep learning (DL) frameworks for PET-based primary NPC lesion segmentation remain scarce. Thus, this study aimed to optimize the delineation of primary NPC radiotherapy target volumes by leveraging PET-image-based DL segmentation technology, with the goal of enhancing both the precision and efficiency of lesion delineation.MethodsResearchers retrospectively collected PET imaging data from 212 NPC patients at the Affiliated Hospital of Guangdong Medical University. The patients were randomly divided into training (170 patients) and testing (42 patients) datasets. A radiation oncologist and a nuclear medicine physician collaboratively delineated the cancer lesion boundaries through consensus. After data preprocessing, three DL models (Res-Unet, Nn-Unet, and Nn-Former) were used to automatically segment the lesions. Training was based on the training dataset. During evaluation, the models’ segmentation performance was comprehensively assessed using Dice similarity coefficient (DSC) and Hausdorff distance (HD). A visual analysis of the results was also conducted to intuitively understand the models’ segmentation capabilities.ResultsIn the training set, Nn-Unet achieved the highest DSC of 0.869, whereas in the testing set, its DSC was 0.833. The DSC for Nn-Former hovered around 0.8 in both the training and testing sets. In contrast, Res-Unet demonstrated the lowest DSC values among the three models, particularly in the testing set (DSC =0.794). Statistical analysis revealed that the DSC of Nn-Unet was significantly higher than that of both Res-Unet and Nn-Former in the training set (P<0.01 for both comparisons). Regarding the 95% HD and average surface distance (ASD) values, Nn-Unet outperformed the other two models in both the training and testing sets. However, Res-Unet exhibited a much higher HD value in the testing set compared to the training set. The Loss curves for all models gradually decreased as training progressed, indicating that the models were learning relevant features of NPC lesions. The final Loss value for Res-Unet was approximately 0.7, whereas Nn-Unet and Nn-Former had final Loss values below 0.8.ConclusionsThe PET-image-based automatic segmentation model for the primary tumor of NPC established in this study demonstrates the clinical potential of Nn-Unet in primary lesion of NPC segmentation tasks.

Document-level sentiment classification is dedicated to predicting the sentiment polarity of document-level reviews posted by users about products and services. Many methods use neural networks have achieved very successful results on sentiment classification tasks. These methods usually focus on mining useful information from the text of the review documents. However, they ignore the importance of users’ review habits. The reviews posted by the same user when commenting on different products contain similar review habits, and reviews that contain highly similar review habits often have similar sentiment ratings. In this paper, we propose a novel sentiment classification algorithm that utilizes user’s review habits to enhance hierarchical neural networks, namely as HUSN. Firstly, we divide the reviews in the training set according to the users. All the reviews of each user are aggregated together and called the historical reviews of this user. Secondly, the target review in the test set and its multiple historical reviews in the training set are sent to the Long Short-Term Memory based hierarchical neural network to obtain the corresponding review document representations containing the user’s review habits. Finally, we calculate the similarities between the target review document representation and multiple historical review document representations. The higher the similarity, the closer the review habits of different reviews from the same user, and the closer the corresponding sentiment ratings. Experimental results show that the similarities between the review habits of different reviews from the same user can further improve the performance of document-level sentiment classification. The HUSN algorithm performs better than all baseline methods on three publicly available document-level review datasets.

This paper firstly researches English text emotion expression and information communication, classifies English text emotion expression and information communication according to the human emotion-value relationship, and summarizes the characteristics of English emotion expression and information communication. Secondly, using artificial intelligence technology, it is proposed to construct an analysis model for English text emotion and information communication using the BiLSTM neural network. To deal with the characteristics of English text quickly and efficiently, it is necessary to encode the emotional information of English text, and based on encoding, the BiLSTM neural network is applied to extract the emotional features of English text and solve the problem of the loss of emotional features through the loss function. Then, the crawler tool is used to obtain the dataset from the Chinese English module under the MOOC of Chinese universities, and the evaluation indexes are set according to the model’s performance, followed by the experimental analysis of the English text emotion expression and information conveyance. The results show that compared with the original CNN, LSTM, and T-LSTM, the BiLSTM-based neural network performs better in the task of text emotion expression and information conveyance, with the accuracy rate staying above 0.925, and the effect on the English dataset is a bit better than that on the Chinese dataset. This study aims to enhance English teaching and communication between Chinese and foreign cultures.

This focus issue of Physiological Measurement follows the 38th Annual International Computing in Cardiology (CinC) Conference, hosted in Hangzhou, China in September 2011 by Zhejiang University. Each year, the NIH-sponsored PhysioNet resource (http://physionet.org/) runs an open competition lasting several months, aimed at encouraging the development of solutions to an unsolved or poorly solved problem in biomedicine, in most cases making use of relevant clinical and experimental data provided freely by PhysioNet. Participants in these annual challenges discuss their diverse approaches to the Challenge problems during dedicated scientific sessions at CinC. The topics of these PhysioNet/CinC Challenges range from physiologic signal processing and analysis to forecasting and modelling clinically important events and processes.

Web apps are now widely used in many different business domains and are vital tools for billions of people in their everyday lives. Unfortunately, a lot of these programs are harmful, which poses a serious risk to Internet users as they may spread spam, install malware, and steal private data. Because of the intricacy of extracting representative features, the vast amount of data, the dynamic nature of dangerous patterns, the stealthiness of assaults, and the limits of conventional classifiers, detecting malicious websites by online content analysis is useless. Static Uniform Resource Locators (URL) characteristics may often provide you instant information about a website without requiring you to load its content.However, complicated feature extraction, large data volumes, changing attack patterns, and the limits of conventional classifiers sometimes make it difficult for current systems to identify malicious web apps using online content analysis. It is inadequate to rely just on lexical URL properties, which might result in incorrect classifications. In order to improve the efficacy of dangerous website identification, this research suggests a multimodal representation strategy that combines textual and image-based information. While picture features are useful for identifying more generic harmful patterns, textual features help the deep learning model comprehend and convey specific semantic information pertaining to attack patterns. By doing this, it may be possible to identify patterns in picture format that are obscured in text format. To extract the hidden features from both textual and image-represented information, two Convolutional Neural Network (CNN) models were built. Both models' output layers were merged and fed into a classifier that uses artificial neural networks to make decisions. The usefulness of the suggested model in comparison to other models is shown by the results. While the false positive rate decreased by 1.5%, the overall performance as measured by the Matthews Correlation Coefficient (MCC) increased by 4.3%.

In order to further mine the deep semantic information of the microbial text of public health emergencies, this paper proposes a multichannel microbial sentiment analysis model MCMF-A. Firstly, we use word2vec and fastText to generate word vectors in the feature vector embedding layer and fuse them with lexical and location feature vectors; secondly, we build a multichannel layer based on CNN and BiLSTM to extract local and global features of the microbial text; then we build an attention mechanism layer to extract the important semantic features of the microbial text; thirdly, we merge the multichannel output in the fusion layer and use soft; finally, the results are merged in the fusion layer, and a surtax function is used in the output layer for sentiment classification. The results show that the F1 value of the MCMF-A sentiment analysis model reaches 90.21%, which is 9.71% and 9.14% higher than the benchmark CNN and BiLSTM models, respectively. The constructed dataset is small in size, and the multimodal information such as images and speech has not been considered.