General Architecture for Text Engineering
| Company type | Open source software project |
|---|---|
| Industry | Natural language processing, Software development |
| Founded | 1995 |
| Headquarters | Sheffield , United Kingdom |
Area served | Worldwide |
| Products | Natural language processing tools, ANNIE information extraction system, GATE Developer, GATE Mímir |
| Website | gate |
General Architecture for Text Engineering (GATE) is a comprehensive Java-based collection of natural language processing (NLP) tools tailored for applications such as multilingual information extraction. Originating in 1995 at the University of Sheffield, it now supports a worldwide community comprising researchers, enterprises, educators, and learners.
As of 2011, the SourceForge mailing list for GATE had surpassed 880 members, with total downloads exceeding 111,000 since 2005. The seminal publication titled "GATE: A framework and graphical development environment for robust NLP tools and applications" has garnered upwards of 2,000 citations, underscoring its significance within NLP research. Key references include Manu Konchady’s "Building Search Applications: Lucene, LingPipe, and Gate" and Graham Wilcock’s "Introduction to Linguistic Annotation and Text Analytics."
The GATE team and its user community have contributed to numerous European research initiatives such as Transitioning Applications to Ontologies, SEKT, NeOn, Media-Campaign, Musing, Service-Finder, LIRICS, and the KnowledgeWeb, which have collectively expanded GATE’s functionalities and industrial applicability.
Features
Among GATE’s components is ANNIE (A Nearly-New Information Extraction System), an information extraction framework composed of modular elements including a tokenizer, gazetteer, sentence splitter, part of speech tagger, a transducer for named entities, and a coreference resolution module. ANNIE either delivers fundamental extraction capabilities or acts as a base for more targeted NLP tasks.
The platform supports multiple languages such as English, Chinese, Arabic, Bulgarian, French, German, Hindi, Italian, Cebuano, Romanian, Russian, and Danish.
Functionality is extended through plugins that integrate with machine learning frameworks like Weka, RASP, MAXENT, SVM Light, LIBSVM, along with a proprietary perceptron algorithm. Additional extensions facilitate ontology management with resources such as WordNet, enable search engine queries including Google and Yahoo, offer alternative part of speech taggers like Brill and TreeTagger, and support processing of social media content such as tweets.
GATE accommodates various input types ranging from plain text (TXT), HTML, and XML to Microsoft Word (Doc) and PDF. Data storage options include Java Serial format and databases such as PostgreSQL, Lucene, and Oracle accessed through JDBC connections.
Text annotation manipulation within GATE is handled by JAPE transducers, empowering users to create customized text processing pipelines as detailed in the GATE User Guide.
GATE Developer
The user interface presents documents alongside their corresponding annotations. Hyperlinks embedded in HTML documents are highlighted in pink. Annotation sets are displayed in the right-hand panel while the bottom panel lists individual annotations. The central area contains the annotation editor for detailed modification.
GATE Mímir
GATE Mímir serves as a powerful tool for indexing and retrieving extensive linguistic, semantic, and ontological datasets. While the raw data can be valuable on its own, its true potential is unlocked through effective search capabilities. This system supports complex queries that integrate textual content, structural information, and SPARQL, facilitating sophisticated semantic and linguistic data exploration.
See also
- Unstructured Information Management Architecture (UIMA)
- OpenNLP
- Pheme, an EU project spearheaded by the GATE team aimed at early identification of misinformation on social media