Interoperability of Annotation Schemes: Using the Pepper Framework to Display AWA Documents in the ANNIS Interface

Interoperability of Annotation Schemes: Using the Pepper framework to display AWA documents in the ANNIS interface Talvany Carlotto∗, Zuhaitz Beloki†, Xabier Artola†, Aitor Soroa† ∗Center for Language and Cognition Groningen, University of Groningen, Groningen, Netherlands, †IXA NLP Group, University of the Basque Country, Donostia, Basque Country ∗talvanynet@gmail.com, †xabier.artola, zuhaitz.beloki, a.soroa@ehu.es Abstract Natural language processing applications are frequently integrated to solve complex linguistic problems, but the lack of interoperability between these tools tends to be one of the main issues found in that process. That is often caused by the different linguistic formats used across the applications, which leads to attempts to both establish standard formats to represent linguistic information and to create conversion tools to facilitate this integration. Pepper is an example of the latter, as a framework that helps the conversion between different linguistic annotation formats. In this paper, we describe the use of Pepper to convert a corpus linguistically annotated by the annotation scheme AWA into the relANNIS format, with the ultimate goal of interacting with AWA documents through the ANNIS interface. The experiment converted 40 megabytes of AWA documents, allowed their use on the ANNIS interface, and involved making architectural decisions during the mapping from AWA into relANNIS using Pepper. The main issues faced during this process were due to technical issues mainly caused by the integration of the different systems and projects, namely AWA, Pepper and ANNIS. Keywords: Interoperability, AWA, Pepper, relANNIS, ANNIS 1. Introduction Natural language processing (NLP) applications demand the integration of many basic linguistic processing mod- ules (Tokenization, POS tagging, Named Entity Recogni- tion and Classification, Syntactic Parsing, Coreference Res- olution, etc.) that perform linguistic analysis at several levels. Sophisticated applications such as event extraction systems, for example, are built upon those basic modules, which are typically combined to build complex process- ing workflows. However, application developers face many challenges when combining NLP modules, which are often tailored to the particular needs of the task at hand and, thus, difficult to integrate. Arguably, one of the main problems encountered when combining NLP modules is the lack of interoperability among them, as each module uses different mechanisms and formats to represent their output, i.e., their annotations. Nowadays, there exist many linguistic annotation schemes that define logical levels of annotation independent of the annotation’s physical format, and are thus able to represent linguistic annotations on many levels. For instance, GATE (Cunningham et al., 1996), AWA (Artola et al., 2009), PAULA (Dipper and G¨otze, 2005) or NAF (Fokkens et al., 2014) are some of the most relevant projects where stand- off annotation is used to deal with the combination of mul- tiple overlapping hierarchies characteristic of the multidi- mensional nature of linguistic information. There have been also attempts to establish standards for representing linguistic information with the aim of facilitat- ing effective resource management and integration of NLP tools (Ide and Romary, 2004). However, as desirable as it might be, expecting such a unique representation standard to be adopted by the NLP community is unrealistic. The …

Content truncated to 500 words