Extractive multi-document summarization using multilayer networks

Abstract

Huge volumes of textual information has been produced every single day. In order to organize and understand such large datasets, in recent years, summarization techniques have become popular. These techniques aims at finding relevant, concise and non-redundant content from such a big data. While network methods have been adopted to model texts in some scenarios, a systematic evaluation of multilayer network models in the multi-document summarization task has been limited to a few studies. Here, we evaluate the performance of a multilayer-based method to select the most relevant sentences in the context of an extractive multi document summarization (MDS) task. In the adopted model, nodes represent sentences and edges are created based on the number of shared words between sentences. Differently from previous studies in multidocument summarization, we make a distinction between edges linking sentences from different documents (inter-layer) and those connecting sentences from the same document (intra-layer). As a proof of principle, our results reveal that such a discrimination between intra-and inter-layer in a multilayered representation is able to improve the quality of the generated summaries. This piece of information could be used to improve current statistical methods and related textual models.it involves paraphrasing sections of the source document and, for this reason, it requires natural language generation tools. In addition, abstractive methods may reuse clauses or phrases from original documents [4].In this work, we target our analysis on extractive summarization applied to a set of documents (MDS).The most traditional employed methods to select relevant sentences for extractive summarization are divided into the following major classes: methods based on word frequency, sentence clustering and machine learning. In recent years, a new class of methods based on network theory have been proposed to analyze texts. Applications of network models in text analysis include the study of scientific documents [5,6,7,8], stylometry [9,10,11,12], sense discrimination [13,14] and several other applications [15,16]. The problem of creating single-document extractive summaries has benefited from these previous network models of texts [17,18,19]. It has been claimed that network features overcome other traditional statistical methods when they are used to identify the most central sentences [17,18]. While most of the studies applying network concepts in summarization have been limited to the single-document counterpart, here we evaluate the usefulness of networks in the multi-document scenario. In particular, the main objective of this paper is to probe whether a discrimination between intra-and inter-layer edges is able to improve the characterization of documents modeled as multilayer complex networks. This is an important feature to be considered in the models because a sentence connected to many other sentences from other documents may indicate a high relevance of the approached topics.In the adopted method, nodes ...

A Review on Text Summarization Techniques

“…It is further suggested that the syntactic parsing can enhance the performance of summarizers. Tohalino and Amancio (2018) discuss a multilayer approach based extractive summarizer where several measurements such as degree, strength, page rank, accessibility, symmetry, shortest path, absorption time, etc., are used to weight the edges in the network of documents. They find that the distinction between intra-and inter-layer edges can play a major role in improving the results of a summarizer.…”

Culture under Complex Perspective: A Classification for Traditional Chinese Cultural Elements Based on NLP and Complex Networks

“…Sentences in the same document are connected at the same layer, while sentences in different documents are connected at different layers. is method effectively improves the quality of summary generation [21]. Marinho et al analyzed the author's attribution of documents based on the motif structure in complex networks.…”

Word synonym relationships for text analysis: A graph-based approach

“…Text graph representation has been used for keyword extraction [3,[11][12][13][14][15][16][17][18][19]], text summarization [32], and language classification [33]. Modeling text as graphs has also been used for PLOS ONE text semantic analysis including information retrieval [34,35] and authorship attribution analysis [36,37], and word sense disambiguation [38,39].…”