BOOTStrep Annotation Scheme: Encoding Information for Text Mining
Lists
Only you see this
Abstract
Annotation of information in corpora is an important aspect of text mining. It bridges between the information hidden in natural language texts and the semantic search queries for the information desired by users. Due to the complex nature of the information needed for text mining, it is essential to design comprehensive annotation schemes to encode and organise the unstructured information contained in text into certain structured forms that can be processed by computers. In the EU BOOTStrep Project 1, which aims to pull together existing biological databases and various terminological repositories and implement a text analysis system to populate a Bio-Lexicon and a Bio-Ontology to support text mining, an annotation scheme has been under construction for encoding multilayer information pertaining to text mining applications. In particular, it is aimed at supporting the interoperability among a set of text annotator tools. In practice, this annotation scheme is built into a type system of a software architecture, named the Unstructured Information Management System (UIMA), which provides a software architecture for the integration and interoperation of NLP annotation tools. Although this scheme is targeted at the biological domain and hence contains some biology-specific semantic elements, the major part of the annotation scheme covers generic language annotation, such as part-of-speech, syntactic parse, co-reference resolution, etc. Therefore, our scheme can easily be ported to other domains or generic language analysis tasks. Coupled with UIMA, this scheme provides a practical means of capturing information from texts for various purposes, such as text mining, NLP, corpus linguistics, etc. 1.
BibTeX
@inproceedings{piao2007bootstrep,
author = {Scott Piao and Ekaterina Buyko and Yoshimasa Tsuruoka and Katrin Tomanek and Jin-Dong Kim and John McNaught and Udo Hahn and Jian Su and Sophia Ananiadou},
title = {{BOOTStrep Annotation Scheme: Encoding Information for Text Mining}},
booktitle = {Research},
year = {2007},
}Loading related content...