Hypergraph Modelization of a Syntactically Annotated English Wikipedia Dump

Hypergraph Modelization of a Syntactically Annotated English Wikipedia Dump Edmundo-Pavel Soriano-Morales, Julien Ah-Pine, Sabine Loudcher Universit´e de Lyon, Laboratoire ERIC, France {edmundo.soriano-morales, julien.ah-pine, sabine.loudcher}@univ-lyon2.fr Abstract Wikipedia, the well known internet encyclopedia, is nowadays a widely used source of information. To leverage its rich information, already parsed versions of Wikipedia have been proposed. We present an annotated dump of the English Wikipedia. This dump draws upon previously released Wikipedia parsed dumps. Still, we head in a different direction. In this parse we focus more into the syntactical characteristics of words: aside from the classical Part-of-Speech (PoS) tags and dependency parsing relations, we provide the full constituent parse branch for each word in a succinct way. Additionally, we propose a hypergraph network representation of the extracted linguistic information. The proposed modelization aims to take advantage of the information stocked within our parsed Wikipedia dump. We hope that by releasing these resources, researchers from the concerned communities will have a ready-to-experiment Wikipedia corpus to compare and distribute their work. We render public our parsed Wikipedia dump as well as the tool (and its source code) used to perform the parse. The hypergraph network and its related metadata is also distributed. Keywords: corpus, wikipedia, syntactic language network 1. Introduction and Related Work Today, the broad reach of Wikipedia in Text Mining (TM) and Natural Language Processing (NLP) research is indis- putable. Several recent approaches and tools have been conducted based on the explicit and implicit knowledge contained in it. Certainly, Wikipedia provides a common ground for researchers and developers to test and compare their results. Wikipedia has been used as a source of valuable data as well as a common background corpus to perform exper- iments and compare results for diverse NLP/TM related tasks. For example, concerning the first case, in the area of Information Extraction, Wikipedia’s infoboxes structured information is used in (Wu and Weld, 2010) as a valuable resource to complement and improve their open IE system. Along the same line, (Charton and Torres-Moreno, 2010) extracted metadata from Wikipedia while leveraging its in- ternal structure in order to produce a semantically anno- tated corpus. Moving on to the Information Retrieval field, features extracted from Wikipedia can also help to better predict the performance of a query (Katz et al., 2014) in a given corpus. In the second case, as a background col- lection for experiments, a document-aligned version of En- glish and Italian Wikipedia has been used to determine the quality between word’s translations (Vuli´c et al., 2011). Wikipedia, being such a popular resource already has var- ious off-the-shelf parsed snapshots (or dumps). These parsed dumps allow researchers to focus more into their approaches than into the extraction and transformation of Wikipedia’s data. We briefly describe certain relevant parses found in the literature. A relevant Wikipedia parsed dump example comes from (Jordi Atserias and Attardi, 2008). Their work provides a balanced amount of syntactic and semantic information. In short, the dump includes each word’s Part-of-Speech (PoS) tag, their dependency relations as well …

Content truncated to 500 words