This paper introduces automatic text summarization on Internet.Firstly,some main approaches to automatic text summarization are presented.Second,it discusses the new demands of automatic summarization for text on Internet and some related information.It also describes the process of text summarization on Internet and evaluation of it.Lastly,it draws a conclusion and prospect on the research of auto text summarization on Internet.

Automatic Text Summarization for Web Pages on Internet

Classification algorithm is one of the key techniques to affect text automatic classification system's performance, play an important role in automatic classification research area. This paper comparatively analyzed k-NN. VSM and hybrid classification algorithm presented by our research group. Some 2000 pieces of Internet news provided by ChinaInfoBank are used in the experiment. The result shows that the hybrid algorithm's performance presented by the groups is superior to the other two algorithms.

An Experimental Comparative Study on Three Classification Algorithms

Data mining concepts and techniques when applied to world-wide web with its existing technologies are termed as web mining. With flooding of information on World-wide web it has become necessary to apply some strategy so that valuable knowledge can be extracted and consequently returned to the user. This paper proposes a multi- agent based Web mining model which is designed to improve the search efficiency of the web pages in compare to the ranked list of keywords based search engine. The proposed model generates valuable information from the web logs of the server by dividing the mining task into several parallel agents which coordinately work together and classify the web documents by which the mining efficiency is improved greatly.

Optimizing Web Mining using Multi-agent System

Text Classification has been addressed by purely statistical approaches that utilize the frequency of occurrence of significant terms as well as by tapping a range of semantic features conveyed by the text. Both approaches have proved their strengths, yet each has its own limitations when applied to corpuses with different sizes and expressive styles. This raises two interesting problemsgiven a corpus, how to automate the process of (i) finding an optimum blend of statistical and contextual contributions for the most appropriate classification, and (ii) determining the relative importance of different kinds of contextual features that are employed? In this paper, we address these issues by developing a Collaborative Document Classification (CDC) system that adapts according to a given corpus, the weighted contributions of statistical features, an array of lexical-semantic features derived from the WordNet ontology and categorical-semantic features obtained from the hierarchical organization of Wikipedia category pages.

A Framework for Collaborative Document Classification with GA-SVM

This paper introduced how to configure a cluster environment for Cognos BI in Connection Metrics and how to troubleshoot the issues. Cognos Clustering greatly enhance the load capacity of the report server, improve the performance, effectiveness and capacity, make the server more stable, ensure the user quantity concurrency

Cognos Clustering in IBM Connections Metrics

Web document summarization (WDS) is becoming one of the hot subjects in the text summarization field due to the rapidly increasing number of documents on Web. WDS is different from traditional text summarization because it must process hyperlinked texts. This paper first analyses the features of Web documents, then gives a definition for WDS, and finally presents an algorithm for WDS based on sentences extraction. Each sentence's weight is a weighted sum of words' weight and its sentence-structure's weight. The former weight is adjusted by document class graph and latter weight considers both the Web formats and hyperlink attributes. The weight proportion of words and structures is learned by machine learning approach. Experiments on 2,000 Web documents show that our algorithm is feasible.

Research on Web Document Summarization

For resolving the low classification efficiency to domain-specific documents for traditional text categorization algorithms (like KNN, SVM and etc), this paper presents a new text classifier with high performance oriented domain-specific documents. The algorithm is mainly depended on the weight and weight factor of hierarchy feature words in documents. Different hierarchy feature words which have different weights are collected by processing to corpus and classification tree. The weight factor is gained by machine learning method to corpus and knowledge base. Classification experiment to metrological documents shows that new classifier outperforms the KNN.

Research on Domain-Specific Text Classifier

For dissolving the imperfection that traditional text summarization method is poor to express domain-specific document meanings, techniques called post-processing to summaries are studied. The post-processing includes: elimination of redundancy; adjustment to coarse summary by clustering document paragraphs; generalization of summary sentences; filling domain-specific knowledge by means of constructing knowledge base. Finally, experiments on clothing field documents are done and software named CTS is developed based on the above theories. Compared with MS Word 2003, the experimental results show that our approaches are effective and efficient. The fruit can be introduced easily into mobile devices to explorer Web pages according its domain.

Post-Processing of Automatic Text Summarization for Domain-Specific Documents

A multi-agent based Web mining model is designed for the improvement of the efficiency of keywords based search engine. The model divides mining task into several parallel agents which coordinately work together, and the mining efficiency is improved greatly. Evolving from HITS, algorithm named Grabber in the model removes Link Farm pages in the expansion of root set, makes anchor text similarity calculation when crawling link page, and chooses pages by a brief conceptual analysis of page content. With the overcoming of the shortcomings of only text analysis or link analysis, Grabber enhances the search engine in understanding the user interest and crawling more Web pages to meet the needs of the users.

A Multi-Agent Based Web Mining Model

For the improvement of recall and precision of search engine based on key words, this thesis makes a study of web pages acquirement technology based on conception topic. Initially it analyses the shortcoming of HITS algorithm and develop a new algorithm based on conception topic. The new algorithm reforms HITS by getting rid of link farms when extending the root set, it overcomes the weakness of which is based on text analysis or link analysis, it promotes the search engine's intellectual judgment of the users' interest and makes the crawling web pages more in need of the users' requirement. Experiments in fashion field show the new algorithm is better than Google.

Xuefei Li

Papers

Research on Web Document Summarization

Research on Domain-Specific Text Classifier

Post-Processing of Automatic Text Summarization for Domain-Specific Documents

A Multi-Agent Based Web Mining Model

Study on Web Content Acquirement Technology Based on Conception Topic