A Big Linked Data Toolkit for Social Media Analysis and Visualization Based on W3C Web Components

摘要

Social media generates a massive amount of data at a very fast pace. Objective information such as news, and subjective content such as opinions and emotions are intertwined and readily available. This data is very appealing from both a research and a commercial point of view, for applications such as public polling or marketing purposes. A complete understanding requires a combined view of information from different sources which are usually enriched (e.g. sentiment analysis) and visualized in a dashboard. In this work, we present a toolkit that tackles these issues on different levels: (1) to extract heterogeneous information, it provides independent data extractors and web scrapers; (2) data processing is done with independent semantic analysis services that are easily deployed; (3) a configurable Big Data orchestrator controls the execution of extraction and processing tasks; (4) the end result is presented in a sensible and interactive format with a modular visualization framework based on Web Components that connects to different sources such as SPARQL and ElasticSearch endpoints. Data workflows can be defined by connecting different extractors and analysis services. The different elements of this toolkit interoperate through a linked data principled approach and a set of common ontologies. To illustrate the usefulness of this toolkit, this work describes several use cases in which the toolkit has been successfully applied.