(1)What is Sentence Centrality and Centroid-based Summarization ?

  Extractive summarization works by choosing a subset of the sentences in the original documents. This process can be viewed as identifying the most central sentences in a (multi-document) cluster that give the necessary and sufficient amount of information related to the main theme of the cluster.

  The centroid of a cluster is a pseudo-document which consists of words that have tf×idf scores above a predefined threshold, where tf is the frequency of a word in the cluster, and idf values are typically computed over a much larger and similar genre data set.

  In centroid-based summarization (Radev, Jing, & Budzikowska, 2000), the sentences that contain more words from the centroid of the cluster are considered as central. This is a measure of how close the sentence is to the centroid of the cluster.

(2)Centrality-based Sentence Salience:

  All of our approaches are based on the concept of prestige in social networks. A social network is a mapping of relationships between interacting entities (e.g. people, organizations, computers). Social networks are represented as graphs, where the nodes represent the entities and the links represent the relations between the nodes.

  A cluster of documents can be viewed as a network of sentences that are related to each other. We hypothesize that the sentences that are similar to many of the other sentences in a cluster are more central (or salient) to the topic.

  There are two points to clarify in this definition of centrality:

  1.How to define similarity between two sentences.

  2.How to compute the overall centrality of a sentence given its similarity to other sentences.

  To define similarity, we use the bag-of-words model to represent each sentence as an N-dimensional vector, where N is the number of all possible words in the target language. For each word that occurs in a sentence, the value of the corresponding dimension in the vector representation of the sentence is the number of occurrences of the word in the sentence times the idf of the word. The similarity between two sentences is then defined by the cosine between two corresponding vectors:

  A cluster of documents may be represented by a cosine similarity matrix where each entry in the matrix is the similarity between the corresponding sentence pair.

  Figure 1 shows a subset of a cluster used in DUC 2004, and the corresponding cosine similarity matrix. Sentence ID dXsY indicates the Y th sentence in the Xth document.

                                             Figure 1: Intra-sentence cosine similarities in a subset of cluster d1003t from DUC 2004.

  This matrix can also be represented as a weighted graph where each edge shows the cosine similarity between a pair of sentence (Figure 2).

Figure 2: Weighted cosine similarity graph for the cluster in Figure 1.

(3)Degree Centrality:

  Since we are interested in significant similarities, we can eliminate some low values in this matrix by defining a threshold so that the cluster can be viewed as an (undirected) graph.

  Figure 3 shows the graphs that correspond to the adjacency matrices derived by assuming the pair of sentences that have a similarity above 0.1, 0.2, and 0.3, respectively, in Figure 1 are similar to each other. Note that there should also be self links for all of the nodes in the graphs since every sentence is trivially similar to itself. Although we omit the self links for readability, the arguments in the following sections assume that they exist.

                                                             -----------------------------------------------------------------------

                                                               -----------------------------------------------------------------------

Figure 3: Similarity graphs that correspond to thresholds 0.1, 0.2, and 0.3, respectively, for the cluster in Figure 1.

  A simple way of assessing sentence centrality by looking at the graphs in Figure 3 is to count the number of similar sentences for each sentence. We define degree centrality of a sentence as the degree of the corresponding node in the similarity graph. As seen in Table 1, the choice of cosine threshold dramatically influences the interpretation of centrality. Too low thresholds may mistakenly take weak similarities into consideration while too high thresholds may lose many of the similarity relations in a cluster.

Table 1: Degree centrality scores for the graphs in Figure 3. Sentence d4s1 is the most central sentence for thresholds 0.1 and 0.2.

JRSmith©2014 - Feedback

Learning LexRank——Graph-based Centrality as Salience in Text Summarization(一)的更多相关文章

  1. Deep Learning of Graph Matching 阅读笔记

    Deep Learning of Graph Matching 阅读笔记 CVPR2018的一篇文章,主要提出了一种利用深度神经网络实现端到端图匹配(Graph Matching)的方法. 该篇文章理 ...

  2. Learning Context Graph for Person Search

    Learning Context Graph for Person Search 2019-06-24 09:14:03 Paper:http://openaccess.thecvf.com/cont ...

  3. Learning Conditioned Graph Structures for Interpretable Visual Question Answering

    Learning Conditioned Graph Structures for Interpretable Visual Question Answering 2019-05-29 00:29:4 ...

  4. 《Deep Learning of Graph Matching》论文阅读

    1. 论文概述 论文首次将深度学习同图匹配(Graph matching)结合,设计了end-to-end网络去学习图匹配过程. 1.1 网络学习的目标(输出) 是两个图(Graph)之间的相似度矩阵 ...

  5. DAG-GNN: DAG Structure Learning with Graph Neural Networks

    目录 概 主要内容 代码 Yu Y., Chen J., Gao T. and Yu M. DAG-GNN: DAG structure learning with graph neural netw ...

  6. Graph Based SLAM 基本原理

    作者 | Alex 01 引言 SLAM 基本框架大致分为两大类:基于概率的方法如 EKF, UKF, particle filters 和基于图的方法 .基于图的方法本质上是种优化方法,一个以最小化 ...

  7. 论文解读( N2N)《Node Representation Learning in Graph via Node-to-Neighbourhood Mutual Information Maximization》

    论文信息 论文标题:Node Representation Learning in Graph via Node-to-Neighbourhood Mutual Information Maximiz ...

  8. 论文解读(GMT)《Accurate Learning of Graph Representations with Graph Multiset Pooling》

    论文信息 论文标题:Accurate Learning of Graph Representations with Graph Multiset Pooling论文作者:Jinheon Baek, M ...

  9. Learning Latent Graph Representations for Relational VQA

    The key mechanism of transformer-based models is cross-attentions, which implicitly form graphs over ...

随机推荐

  1. Java三大主流框架概述(转载)

    转自:http://www.douban.com/note/320140839/ Struts.Hibernate和Spring是我们Java开发中的常用关键,他们分别针对不同的应用场景给出最合适的解 ...

  2. 12种超酷HTML5 SVG和CSS3浮动标签效果

    这是一组效果很炫酷的SVG和CSS3表单浮动标签特效.这组浮动标签特效共12种效果,这些浮动标签效果部分在元素的伪元素上使用CSS transitions和CSS animations完毕,一部分则使 ...

  3. Uncaught TypeError: Cannot read property 'post' of undefined

  4. Java泛型方法定义及泛型类型推断

    泛型的推断 @Test public void test3(){ //类型推断时使用两个类型的最小公倍数 int x1 = add(3,4); Number x2 = add(3.5,4); Obje ...

  5. 设计模式之Facade模式

    Facade(外观)模式为子系统中的各类(或结构与方法)提供一个简明一致的界面,隐藏子系统的复杂性,使子系统更加容易使用.他是为子系统中的一组接口所提供的一个一致的界面. 在遇到以下情况使用Facad ...

  6. 程序员带你学习安卓开发,十天快速入-对比C#学习java语法

    关注今日头条-做全栈攻城狮,学代码也要读书,爱全栈,更爱生活.提供程序员技术及生活指导干货. 如果你真想学习,请评论学过的每篇文章,记录学习的痕迹. 请把所有教程文章中所提及的代码,最少敲写三遍,达到 ...

  7. JS手动创建标签

    代码: <html> <head> <title>js标签属性的添加</title> <script > function setxxx() ...

  8. 雷鸟(Thunderbird)收取Gmail出错,收到警告邮件

    如题,每次打开thunderbird都会收到以下提醒邮件,后来发现,只要将thunderbird中和google服务有关的插件卸载掉就可以了 We prevented the sign-in atte ...

  9. rest例子

    http://www.xdemo.org/spring-restful/(可用) http://www.open-open.com/lib/view/open1389075258125.html(有例 ...

  10. web开发基础(同步更新中)

    1/Get与Post的区别 GET是我们都熟悉的.它用于请求网页文本.当你在浏览器输入harvard.edu,它会直接访问Harvard的web服务器,去GET /. 第二个最有名的是POST,它经常 ...