In this task, an approach for single document summaries based on local topic identification and word frequency is proposed. In recent years, there has been increased interest in automatic summarization. The physical features are often used and have been successfully applied to this field; it also has some disadvantages of non-redundancy, structure and coherence. Therefore, we introduced logical structure feature which has been successfully applied in multi-document summarization (MDS), and we designed a system to accomplish this task. Documents can be clustered into local topic after sentences similarity is calculated, which can be sorted by the scoring. Then sentences from all local topics are selected by computing the word frequency. Using this proposed method, the information redundancy of each local topic and among local topic is reduced. The information coverage ratio and structure of the summarization is improved.

Single Document Summarization Based on Local Topic Identification and Word Frequency

In the last years different methodologies have been proposed to systematize the Web development processes. Some of them are supported by CASE tools that help the user in all or in part of the phases of the application development. A number of these tools are used with didactic aims, for teaching the development process of Web applications. Nevertheless, to our knowledge, none of the currently available CASE tools neither supports the online cooperative design, nor allows the modelling of a process with the supervision of a teacher. Our technological solution is based on Rich Internet Applications (RIAs), i.e., Web applications with multimedia, high levels of interactivity, and collaborative work support. This document presents a Rich Internet application for the automatic code generation of Web applications that allows different modellers to design Web applications in a cooperative and supervised way, as well as to automatically generate its code. This work is inspired by the Web modelling language WebML.

An e-Learning collaborative RIA environment for Web Application Modeling and Code Generation

Sub-topic clustering is a crucial step in multi-document summarization. The traditional k-means clustering method is not effective for topic clustering because the number of clusters k must be given in advance. This paper describes a new method for sub-topic clustering based on semi-supervised learning: the method firstly partition the set of sentences into disjoint subsets, each of which contained sentences covering exactly one topic, and labels the sentences which have high scores in the topic, then use the method of constrained-k-means to decide the number of topics, and finally get the sub-topic sets by k-Means clustering. This algorithm can dynamically generate the number of k-means clustering, and the experiment result indicates that the accuracy of clustering is improved

A New Sub-topics Clustering Method Based on Semi-supervised Learing

Query-directed multi-document summarization aims to provide a more effective characterization of a document set accounting to the user's information need when generating a summary. In this paper, we propose a practical approach for this task by identifying the sentences with high query-relevant and high information density. This is implemented by mining two kinds of features for each sentence: the power of correlation with the query and the power of global connectivity. While the first is executed by computing semantic similarity between the sentence and the query, and the other is executed by using semantic graph. Then these two kinds of features are blessed to score each sentence. At last with the help of MMR for reducing redundancy, we get the summary. Experimental results indicate that this method is encouraging for both those retrieved documents that correspondingly concentrating to one subject and retrieved documents who have many sub-topics and comparatively being related to the query.

The Implementation of a Query-Directed Multi-document Summarization System

In this paper, we present a novel technique to create a special domain news collection system from really simple syndication (RSS) news sites through summarization and classification. The main aim of this research is to build a self-sufficient news collection system in disaster domain. In this news collection system, we used new strategies and algorithms to mine news from RSS sites, recognized and collected disaster news using automatic summarization and classification. The most striking dissimilarity between our study and previous work is that we use a novel summary approach to improve the classification performance. This paper discusses the effect of summarization and classification model on system performance. Results show that our method yields a better performance in this field. Copyright © 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Create Special Domain News Collections through Summarization and Classification

Sentences Optimum Selection for Multi-Document Summarization

With the rapid development of the Internet, multi documents summarization is becoming a very hot research topic. In order to generate a summarization that can effectively characterize the original information from documents, this paper proposes a multi documents summarization approach based on the physical features and logical structure of the document set. This method firstly clusters similar sentences into several Logical Topics (LTs), and then orders these topics according to their physical features of multi documents. After that, sentences used for the summarization are extracted from these LTs, and finally the summarization is generated via certain sorting algorithms. Our experiments show that the information coverage rate of our method is 8. 83 % higher than those methods based solely on logical structures, and 14.31% higher than Top-N method.

Summarization based on physical features and logical structure of multi documents

Search Result Clustering Method Based on SOM and LSI

The update summary as defined for the DUC2007 new task aims to capture evolving information of a single topic over time. It delivers focused information to a user who has already read a set of older documents covering the same topic. This paper presents a novel manifold-ranking frame based on iterative feedback mechanism to this summary task. The topic set is extended by using the summarization of previous timeslices and the first sentences of documents in current timeslice. Iterative feedback mechanism is applied to model the dynamically evolving characteristic and represent the relay propagation of information in temporally evolving data. Modified manifold-ranking process also can naturally make use of both the relationships among all the sentences in the documents and relationships between the topic and the sentences. The ranking score for each sentence obtained in the manifold-ranking process denotes the importance of sentence biased towards topic, and then the greedy algorithm is employed to rerank the sentences for removing the redundant information. The summary is produced by choosing the sentences with high ranking score. Experiments on dataset of DUC2007 update task demonstrate the encouraging performance of the proposed approach.

/pdf/iterative-feedback-based-manifold-ranking-for-update-summary-4t55o465fj.pdf

Qin Bing

Papers

Sentences Optimum Selection for Multi-Document Summarization

Summarization based on physical features and logical structure of multi documents

Search Result Clustering Method Based on SOM and LSI

Iterative Feedback Based Manifold-Ranking for Update Summary