Systems and methods for summarizing online articles for consumption on a user device are disclosed herein. The system extracts the main body of an article's text from the HTML code of an online article. The system may then classify the extracted article into one of several different categories and removes duplicate articles. The system breaks down the article into its component sentences, and each sentence is classified into one of three categories: (1) potential candidate sentences that may be included in the generated summary; (2) weakly rejected sentences that will not be included in the summary but may be used to generate the summary; and (3) strongly rejected sentences that are not included in the summary. Finally, the system applies a document summarizer to generate quickly readable article summaries, for viewing on the user device, using relevant sentences from the article while maintaining the coherence of the article.

Systems and methods for generating summaries of documents

Music regional classification, which is an important branch of music automatic classification, aims at classifying folk songs according to different regional style. Chinese folk songs have developed various regional musical styles in the process of its evolution. Regional classification of Chinese folk songs can promote the development of music recommendation systems which recommending proper style of music to users and improve the efficiency of the music retrieval system. However, the accuracy of existing music regional classification systems is not high enough, because most methods do not consider temporal characteristics of music for both features extraction and classification. In this paper, we proposed an approach based on conditional random field (CRF) which can fully take advantage of the temporal characteristics of musical audio features for music regional classification. Considering the continuity, high dimensionality and large size of the audio feature data, we employed two ways to calculate the label sequence of musical audio features in CRF, which are Gaussian Mixture Model (GMM) and Restricted Boltzmann Machine (RBM). The experimental results demonstrated that the proposed method based on CRF-RBM outperforms other existing music regional classifiers with the best accuracy of 84.71% on Chinese folk songs datasets. Besides, when the proposed methods were applied to the Greek folk songs dataset, the CRF-RBM model also performs the best.

Regional classification of Chinese folk songs based on CRF model

Music regional style classification aims to divide the traditional folk songs according to different geographies Research on the regional style classification of music contributes to digging into the creation rules of traditional folk songs and is of great significance to geographical type retrieval This paper combines the temporal model to encode the musical structures based on the analysis of the characteristics of Chinese traditional folk songs Conditional Random Field (CRF) is utilized to establish the model of folk songs for the first time The state function and the transfer function of the CRF are calculated by the classical clustering algorithm Gaussian Mixture Model (GMM) to estimate the label sequence In contrast to previous music classification methods (such as SVM, KNN, etc), the accuracy of the regional style classification of folk songs increases in a range of 46% ~ 1813%, and the best performance is achieved by the GMM-CRF model in our experiment

The Regional Style Classification of Chinese Folk Songs Based on GMM-CRF Model

Group incremental adaptive clustering based on neural network and rough set theory for crime report categorization

Novelty detection in news events has long been a difficult problem. A number of models performed well on specific data streams but certain issues are far from being solved, particularly in large data streams from the WWW where unpredictability of new terms requires adaptation in the vector space model. We present a novel event detection system based on the Incremental Term Frequency-Inverse Document Frequency (TF-IDF) weighting incorporated with Locality Sensitive Hashing (LSH). Our system could efficiently and effectively adapt to the changes within the data streams of any new terms with continual updates to the vector space model. Regarding miss probability, our proposed novelty detection framework outperforms a recognised baseline system by approximately 16% when evaluating a benchmark dataset from Google News.

/pdf/an-improved-system-for-sentence-level-novelty-detection-in-143cg83gt1.pdf

An Improved System For Sentence-Level Novelty Detection In Textual Streams

Traditional Clustering is a powerful technique for revealing the hot topics among Web information. However, it failed to discover the trending events coming out gradually. In this paper, we propose a novel method to address this problem which is modeled as detecting the new cluster from time-streaming documents. Our approach concludes three parts: the cluster definition based on Multi- Representation Index Tree (MI-Tree), the new cluster detecting process and the metrics for measuring a new cluster. Compared with the traditional method, we process the newly coming data first and merge the old clustering tree into the new one. Our algorithm can avoid that the documents owning high similarity were assigned to different clusters. We designed and implemented a system for practical application, the experimental results on a variety of domains demonstrate that our algorithm can recognize new valuable cluster during the iteration process, and produce quality clusters.

/pdf/new-trending-events-detection-based-on-the-multi-15g5yh2e8y.pdf

New Trending Events Detection based on the Multi-Representation Index Tree Clustering

In this paper, we perform an exploring on music characteristics of Chinese folk music, in order to do targeted music restoration and perform digital reproduction on music segments Starting from the point of music classification and music feature selection, we firstly choose SVM as the classifier according to experiment results of different classifiers Then we introduce two common filter-filter methods: ReliefF-PCA and ReliefF-CA, and put forward to two heuristic filter-wrapper methods: HBS and HFS At last, we apply these feature selection methods on our music data set and test their performance through experiments The results show that HBS and HFS algorithms can effectively reduce the dimension of features vector and improve the classification accuracy, while other common feature selection methods perform poorer than them

HBS and HFS feature selection methods for Chinese folk music classification

Incremental Document Clustering is a powerful technique for large-scale topic discovery from incremental documentation set. Indexing tree algorithm is advanced in efficiency. However, it tended to process spherical data. To address this problem, we present a novel Multi-Representation Indexing Tree (MRIT) algorithm for constructing a hierarchy that satisfies arbitrary shape clusters with a good performance. Compared with the Indexing tree algorithm, a cluster is decomposed into several sub clusters and is represented as a union of the sub clusters rather than the center of the cluster. Similarity of a document to one cluster is the distance to the nearest neighbor among the cluster's representative points. The experimental results on a variety of domains demonstrate that our algorithm can produce a quality cluster. It's insensitive to document input order, and efficient in terms of computational time.

Incremental document clustering using Multi-representation Indexing Tree

Traditional Clustering is a powerful technique for revealing the "hot" topics among documents However, it's hard to discover the new type events coming out gradually In this paper, we propose a novel model for detecting new clusters from time-streaming documents It consists of three parts: the cluster definition based on Multi-Representation Index Tree (MI-Tree), the new cluster detecting process and the metrics for measuring a new cluster Compared with the traditional method, we process the newly coming data first and merge the old clustering tree into the new one This algorithm can avoid this effect: the documents enjoying high similarity were assigned to different clusters We designed and implemented a system for practical application, the experimental results on a variety of domains demonstrate that our algorithm can recognize new valuable clusters during the iteration process, and produce quality clusters

Hui Song

Papers

New Trending Events Detection based on the Multi-Representation Index Tree Clustering

HBS and HFS feature selection methods for Chinese folk music classification

Incremental document clustering using Multi-representation Indexing Tree

New Cluster Detection Based on Multi-Representation Index Tree Text Clustering