Generating descriptive multi‐document summaries of geo‐located entities using entity type models

Abstract

In this article, we investigate the application of entity type models in extractive multi-document summarization using automatic caption generation for images of geo-located entities (e.g., Westminster Abbey) as an application scenario. Entity type models contain sets of patterns aiming to capture the ways geo-located entities are described in natural language. They are automatically derived from texts about geo-located entities of the same type (e.g., churches, lakes). We integrate entity type models into a multi-document summarizer and use them to address the 2 major tasks in extractive multidocument summarization: sentence scoring and summary composition. We experiment with 3 different representation methods for entity type models: signature words, n-gram language models, and dependency patterns. We evaluate the summarizer with integrated entity type models relative to (a) a summarizer using standard text-related features commonly used in text summarization and (b) the Wikipedia location descriptions. Our results show that entity type models significantly improve the quality of output summaries over that of summaries generated using standard summarization features and Wikipedia summaries. The representation of entity type models using dependency patterns is superior to the representations using signature words and n-gram language models.

Generating natural language tags for video information management

“…This work contrasts to previous approaches in several aspects: first, sentences are generated from scratch, instead of retrieving (11,27) or summarising existing text fragments associated with an image (3,7). Second, textual descriptions are generated for specific and real contents of videos, whereas related (but subtly different) work in automatic caption generation created news text (12) or encyclopedic text (2) that was contextually relevant but not closely pertinent to the specific content of images.…”

From puzzles to portraits: Enhancing situation awareness during natural disasters using a design science approach

“…In recent years, many techniques have been developed to conduct information retrieval from social media data (Chua et al., 2019; Hasan et al., 2019), which include topic modeling using sentiment analysis (Deng et al., 2017; Wu & Cui, 2018), social network analysis using semantics and emotion mining (Kryvasheyeu et al., 2015), message tracking through a viral diffusion model (Goel et al., 2016), query‐based text extraction using location‐based entity models (Aker & Gaizauskas, 2015; Pournarakis et al., 2017), and generating spatial temporal graphs of semantic relations between concepts (Xu et al., 2017). Although these approaches provide a unique perspective and a set of tools to derive useful information from a given type of social media data, most of these approaches rely on ad hoc modeling techniques that require strong modeling assumptions.…”

Long story short: finding health advice with informative summaries on health social media

“…Text summarization, which is a data abstraction process, enables users to procure a sense of the content of a document without reading every single sentence in the full text (Mani and Maybury, 1999; Gambhir and Gupta, 2017). Over the past decade, several advancements in text summarization types have been reported, including single-document (Zhongyuwei and Gao, 2014; Cheng and Lapata, 2016), multi-document (Piwowarski et al , 2012; Aker and Gaizauskas, 2015), feature-based opinion (Huang and Cheng, 2015) and biomedical (Plaza et al , 2011; Mishra et al , 2014) text summarization. Only a few studies have considered extracting summaries from health social media.…”