Automated classification of web pages in hierarchical browsing

BackgroundTo report the diversity of MRI features of brainstem encephalitis (BE) induced by Enterovirus 71. This is supported by implementation and testing of our new classification scheme in order to improve the diagnostic level on this specific disease.MethodsNeuroimaging of 91 pediatric patients who got EV71 related BE were hospitalized between March, 2010 to October, 2012, were analyzed retrospectively. All patients underwent pre- and post-contrast MRI scan. Thereafter, 31 patients were randomly called back for follow-up MRI study during December 2013 to August 2014. The MRI signal patterns of BE primary lesion were analyzed and classified according to MR signal alteration at various disease stages. Findings in fatal and non-fatal cases were compared, and according to the MRI scan time point during the course of this disease, the patients’ conditions were classified as 1) acute stage, 2) convalescence stage, 3) post mortem stage, and 4) long term follow-up study.Results103 patients were identified. 11 patients did not undergo MRI, as they died within 48 hours. One patient died on 14th day without MR imaging. 2 patients had postmortem MRI. Medical records and imaging were reviewed in the 91 patients, aged 4 months to 12 years, and two cadavers who have had MRI scan. At acute stage: the most frequent pattern (40 patients) was foci of prolonged T1 and T2 signal, with (15) or without (25) contrast enhancement. We observed a novel pattern in 4 patients having foci of low signal intensity on T2WI, with contrast enhancement. Another pattern in 10 patients having foci of contrast enhancement without abnormalities in T1WI or T2WI weighted images. Based on 2 cases, the entire medulla and pons had prolonged T1 and T2 signal, and 2 of our postmortem cases demonstrated the same pattern. At convalescence stage, the pattern observed in 4 patients was foci of prolonged T1 and T2 signal without contrast enhancement. Follow-up MR study of 31 cases showed normal in 26 cases, and demonstrated foci of prolonged T1 and T2 signal with hyper-intensity on FLAIR in 3 cases, or of prolonged T1 and T2 signal with hypo-intensity on FLAIR in 2 cases. Most importantly, MR findings of each case were thoroughly investigated and classified according to phases and MRI signal alteration.ConclusionsThis study has provided enhanced and useful information for the MRI features of BE induced by EV71, apart from common practice established by previous reports. In addition, a classification scheme that summarizes all types of features based on the MRI signal at the four different stages of the disease would be helpful to improve the diagnostic level.

Googling for Health Information

Previous work shows that a web page can be partitioned into multiple segments or blocks, and often the importance of those blocks in a page is not equivalent. Also, it has been proven that differentiating noisy or unimportant blocks from pages can facilitate web mining, search and accessibility. However, no uniform approach and model has been presented to measure the importance of different segments in web pages. Through a user study, we found that people do have a consistent view about the importance of blocks in web pages. In this paper, we investigate how to find a model to automatically assign importance values to blocks in a web page. We define the block importance estimation as a learning problem. First, we use a vision-based page segmentation algorithm to partition a web page into semantic blocks with a hierarchical structure. Then spatial features (such as position and size) and content features (such as the number of images and links) are extracted to construct a feature vector for each block. Based on these features, learning algorithms are used to train a model to assign importance to different segments in the web page. In our experiments, the best model can achieve the performance with Micro-F1 79% and Micro-Accuracy 85.9%, which is quite close to a person's view.

In structural pattern recognition, a common problem is the difficulty of constructing classification models or rules from a set of examples. This paper illustrates a method for generating class descriptions from a training set by using inductive logic programming, and in particular an inductive learning algorithm based on the FOIL system. The goal of the method is to find descriptions which are general, i.e. are successfully applicable to recognize objects different from the ones in the training set, while preserving their discrimination ability. The application of the method to a difficult real-world problem (handprinted character recognition) is presented, proposing a hierarchical description and classification scheme in order to reduce the complexity of the task to a level which can be profitably handled by this kind of learning methodologies.

Although there have been a lot of studies on mobile business, it is hard to find ones in which an integrated perspective of mobile business based on both customers’ and companies’ needs is suggested. This research develops a mobile business model classification scheme and applies it to analysis of current status in mobile business. In the first case study on 65 firms in Korea, mobile solutions are categorized based on the classification scheme in order to analyze an overall market environment of mobile business. In the second case study on mobile service trend analysis, a set of evaluation indices including four general indices, ten effectiveness indices and 13 measurement indices were developed to analyze mobile service environment in the perspective of B2C (business to customer). For the second case study, 200 mobile heavy users were questioned on the mobile service‐related issues.

In this paper we proposed a robust classification scheme capable of classifying audio signal as speech, music, silence and environmental sound. The classification scheme is divided into the three sections, in the first section we classify the audio clip as speech, non-speech or silence, the classifier that we have used here is a novel K-nearest neighbour classifier. The second section comprises of labelling the nonspeech signal as music or environmental sound, this section is based on rule based classification. A feature used to discriminate between music and environmental sound is noise frame ratio and is found that it provides better results. We further extend our work so as to discriminate music signal into a particular genre i.e. music genre classification, this classification algorithm also makes use of rule based classification scheme. In case of music genre classification we make use of a new feature called beat histogram, this feature is used to classify the signal labelled as music into different taals of tabla, this can be considered as third section. The performance of this method is tested on an audio database "Swarsanchay", which consists of 500 sound files belonging to four classes of audio data types which are speech, music, silence and environmental sound. (4 pages)

Enhancing accessibility of web-based SVG buttons: An optimization method and best practices

Visually summarizing web pages is an attractive approach that provides users an effective and friendly interface to identify desired contents at a first glance for search and re-finding tasks. Using dominant images in web pages is generally reliable for this purpose. However, dominant images are often unavailable in many web pages. To solve this problem, we first propose a new approach to summarize those web pages without any dominant images by retrieving relevant external images from the Internet. However, relevant external images are sometimes unreliable. To take the advantages of these two kinds of images, we further propose a clustering based algorithm to select the best summarization among all of internal and external images. This algorithm leverages relevance and dominance of images as the prior information. Experimental results show that our approach achieves 0.098 and 0.082 NDCG1 gain on a human labeled data set, compared with relevant external image and dominant image, respectively. Our user study also indicates that the images selected by our algorithm are useful as the summarization of web pages.

Outcomes are not different for patient-matched versus nonmatched treatment in subjects with chronic recurrent low back pain: a randomized clinical trial

The current web technologies make intensive the use of colors in web pages. Nowadays, colors are essential in the design of interfaces and play a central role in the distinction and comprehension of information. However, this affects colorblind users, i.e., those who have difficulties in recognizing or distinguishing colors. This paper presents a user study involving colorblind people to empirically investigate several aspects related to the recoloring of web interfaces. We aim to detect limitations, barriers, and needs about these users’ interaction with web pages. Our employed evaluation investigates indicators of satisfaction (contentment) and pleasantness (enjoyable) for several scenarios of interface recoloring adaptation. We found a ranking of application for interface adaptation techniques with the use of recoloring algorithms. The obtained results reveal the advantages of considering the colorblind individual’s needs and preferences for the development of adaptive systems. Our contribution can enhance web interface accessibility based on user interface adaptation techniques.

Abstract. The shapes of ice crystals play an important role in global precipitation formation and the radiation budget. Classifying ice crystal shapes can improve our understanding of in-cloud conditions and these processes. Existing classification methods rely on features such as the aspect ratio of ice crystals, environmental temperature, and so on, which bring high instability to the classification performance, or employ supervised machine learning algorithms that heavily rely on human labeling. This poses significant challenges, including human subjectivity in classification and a substantial labor cost in manual labeling. In addition, previous deep learning algorithms for ice crystal classification are often trained and evaluated on datasets with varying sizes and classification schemes, each with distinct criteria and a different number of categories, making it difficult to make a fair comparison of algorithm performance. To overcome these limitations, a contrastive semi-supervised learning (CSSL) algorithm for the classification of ice crystals is proposed. The algorithm consists of an upstream unsupervised learning network tasked with extracting meaningful representations from a large number of unlabeled ice crystal images, and a downstream supervised network is fine-tuned with a small subset of labeled images of the entire dataset to perform the classification task. To determine the minimum number of ice crystal images that require human labeling while maintaining the algorithm performance, the algorithm is trained and evaluated on datasets with varying sizes and numbers of categories. The ice crystal data used in this study were collected during the NASCENT campaign at Ny-Ålesund and CLOUDLAB project on the Swiss Plateau using a holographic imager mounted on a tethered balloon system. In general, the CSSL algorithm outperforms a purely supervised algorithm in classifying 19 categories. Approximately 154 h of manual labeling can be avoided using just 11 % (2048 images) of the training set for fine tuning, sacrificing only 3.8 % in overall precision compared to a fully supervised model trained on the entire dataset. In the four-category classification task, the CSSL algorithm also outperforms the purely supervised algorithm. When fine-tuned on just 2048 images (25 % of the dataset), it achieves an overall accuracy of 89.6 %, nearly matching the 91.0 % accuracy of the supervised algorithm trained on 8192 images. When tested on the unseen CLOUDLAB dataset, CSSL shows superior generalization, improving accuracy by an average of 2.19 %. Our analysis also reveals that both CSSL and purely supervised algorithms exhibit inherent instability when trained on small dataset sizes, and the performance difference between them converges as the training set size exceeds 2048 samples. These results highlight the strength and practical effectiveness of CSSL in comparison to purely supervised methods and the potential of the CSSL algorithm to perform well on datasets collected under different conditions.

Research on how the Internet is diffusing across the population has broadened from questions about who uses the medium to what people do during their time online. With this change in focus comes a need for more detailed data on people's online actions. The author provides a method for coding and classifying users' online information-seeking behavior. The author presents an exhaustive list of ways in which a user may arrive at a web page. The proposed methodology includes enough nuanced information to distinguish among different search actions and links. In its entirety, the coding scheme makes it possible to understand many details about the users' sequence of actions simply by looking at the spreadsheet containing the information proposed in this article. The author also demonstrates the utility of this scheme with findings from a study on the information-seeking behavior of 100 randomly selected Internet users to exemplify the utility of this coding and classification scheme.

In this paper, we deal with the problem of analyzing and classifying web documents in a given domain by information filtering agents. We present the ontology-based web content mining methodology that contains such main stages as creation of ontology for the specified domain, collecting a training set of labeled documents, building a classification model in this domain using the constructed ontology and a classification algorithm, and classification of new documents by information agents via the induced model. We evaluated the proposed methodology in two specific domains: the chemical domain (web pages containing information about production of certain chemicals), and Yahoo! collection of web news documents divided into several categories. Our system receives as input the domain-specific ontology, and a set of categorized web documents, and then perfroms concept generalization on these documents. We use a key-phrase extractor with integrated ontology parser for creating a database from input documents and use it as a training set for the classification algorithm. The system classification accuracy is estimated using various levels of ontology.

Defining decision region borders properly is a major task of classification algorithms. In this paper, the border feature detection and adaptation (BFDA) algorithm is introduced for this purpose. The BFDA is a novel classification scheme, especially useful for the classification of remote sensing images. The method exploits the powerful discrimination capability of the 1-nearest neighborhood (1-NN) method with the border feature vectors. The first part of the algorithm consists of generating border feature vectors using class centers and misclassified training vectors. With this approach, a manageable number of border feature vectors are obtained. The second part of the algorithm involves the adaptation of the border feature vectors with a technique similar to the learning vector quantization (LVQ) algorithm. The performance of the BFDA was compared with other classification algorithms including support vector machines (SVMs) and several statistical classification techniques.

This communication presents two new classification algorithms based on time varying higher order statistics (TVHOS). These two algorithms benefit from the advantageous localization properties provided by TVHOS4 for instantaneous frequency laws disrupted by multiplicative noise. The classification scheme used is a bank of several normalized TVHOS4 correlators. Simulations illustrate successful performance of the classification algorithms for different situations and especially if the SNR is higher than -7 dB.