Multi-document summarization via submodularity

Abstract

Multi-document summarization is becoming an important issue in the Information Retrieval community. It aims to distill the most important information from a set of documents to generate a compressed summary. Given a set of documents as input, most of existing multi-document summarization approaches utilize different sentence selection techniques to extract a set of sentences from the document set as the summary. The submodularity hidden in the term coverage and the textual-unit similarity motivates us to incorporate this property into our solution to multidocument summarization tasks. In this paper, we propose a new principled and versatile framework for different multidocument summarization tasks using submodular functions (Nemhauser et al. in Math. Prog. 14(1): 1978) based on the term coverage and the textual-unit similarity which can be efficiently optimized through the improved greedy algorithm. We show that four known summarization tasks, including generic, query-focused, update, and comparative summarization, can be modeled as different variations derived from the proposed framework. Experiments on benchmark summarization data sets (e.g., DUC04-06, TAC08, TDT2 corpora) are conducted to demonstrate the efficacy and effectiveness of our proposed framework for the general multi-document summarization tasks.

Generating event storylines from microblogs

“…One type of summarization systems select representative sentences, e.g. with significant frequency [40], or structural centroid in sentence graph [18,13]. Another type is based on matrix decomposition [17,35,31].…”

“…Multi-Document Summarization using Submodularity (MSSF) [18]: a multi-document summarization framework based on Submodularity;…”

META: Multi-resolution Framework for Event Summarization

“…Some research efforts have been working on providing various summarization methods [8,13,4,9,17]. Each of them defines its own way of summarizing events/documents.…”

Personal Social Screen--A Dynamic Privacy Assignment System for Social Sharing in Complex Social Object Networks

“…Intuitively, by adding one element to a larger set T , the value increment of f can never be larger than that by adding one element to a smaller set S. This intuitive diminishing property exists in different areas, e.g., in social network, adding one new friend cannot increase more social influence for a more social group than for a less social group. Submodularity modeling has been employed into multiple research areas, e.g., document summarization [14] and news recommendation [15]. The budgeted maximum coverage problem is then described as: given a set of elements E where each element is associated with an influence and a cost defined over a domain of these elements and a budget B, the goal is to find out a subset of E which has the largest possible influence while the total cost does not exceed B.…”