DOCSLIB.ORG

Assisted Annotation Inception Platform Motivation Project

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Assisted Annotation Inception Platform Motivation Project INCEPTION CORPUS-BASED DATA SCIENCE FROM SCRATCH Richard Eckart de Castilho, Jan-Christoph Klie, Naveen Kumar, Beto Boullosa, Iryna Gurevych Ubiquitous Knowledge Processing Lab (UKP-TUDA) Dept. of Computer Science Technische Universität Darmstadt Motivation Project Workflow Assisted Annotation • Corpus-based data science is seeing $ % & rapid adoption in science and industry ### Expert Trained “Crowd” ### • Domain-specific annotated corpora are Sources Subcorporation Export / Freeze required in many domains Identify relevant annotation units Project Setup Curation • Current corpus processes for manual Annotation Define data sources Annotate extracted units Manual comparison Define annotation scheme With categories from KB Automatic aggregation corpus annotation do not scale Define knowledge scheme Linking to instances in KB Assign permissions $ KB Population Human in the loop Define schema (classes/properties) Recommenders Active Learning INCEpTION is an infrastructure-ready Edit instances Add qualifiers to statements ● Continually learn from the users actions ● Aims at reducing the time to learn by human-in-the-loop annotation platform ● Asynchronous training does not slow asking specific feedback from the user down user interface ● Using uncertainty-sampling strategy Monitoring ● Automatic evaluation to avoid inaccurate ● Compatible with any recommender that combining machine learning and human ! Assign workload Monitor progress " predictions (configurable quality threshold) provides a confidence score Manager Evaluate quality Curator ● Built-in recommenders: Dictionary-based, ● User can freely switch between active expertise for rapid domain adaptation OpenNLP-based sequence classifier (part- learning and normal annotation of-speech, named entities, …) INCEpTION Platform Knowledge Bases EXtensibility INCEpTION aims to support three functionalities which are INCEpTION is a modular architecture providing many commonly required for text annotation projects but typically extension points where new functionality can be added and not available in a single tool: corpus creation; text existing functionality can be changed - a selection of these annotation; knowledge management. is shown below: The platform additionally provides assistive features such as Annotation Annotation Project machine learning recommenders to help users working on Editors Layers Importers/Exporters these tasks more efficiently. Annotation Editor Annotation Data Formats Sidebars Feature types Integrating these functionalities into a single comprehensive Data Model Entity and Fact Linking platform permits addressing tasks typically not found in Annotation Editor ● RDF-based data model: ● Linking of annotations to KB classes and Recommenders Event Logging generic annotation platforms, such as entity linking, classes, properties, instances, literals instances via an auto-complete field Extensions knowledge base population, cross-document coreference ● Support for class hierarchies ● Optional contextual ranking of candidates ● Classes can have instances ● Linked annotations shown on the KB page annotation, etc. ● Support for typed properties including when a class/instance is selected The architecture modular architecture is realized using the domain and range restrictions ● Special support for linking factual ● Editors for different value types (string, statements in text documents to Spring framework. Dependency injection and events are numeric, boolean, KB resources) subject/predicate/object triples in the KB used to achieve a loose coupling between the modules. Diversity Relevance Feedback Corpus Creation Search Semantic Search Interoperability and Integration Active Learning Statistics Distant Supervision INCEpTION Within the NLP and text mining landscape, Integration goes beyond interoperability. E.g. when an external text mining an annotation platform like INCEpTION only platform wants to delegate annotation to the INCEpTION platform, it needs to be Knowledge Cross-Document covers a part of the overall text analysis able to automatically set up annotation projects, import data, monitor the ongoing Annotation Relations Management needs. Therefore, it is important that the annotations and finally retrieve the annotated data for further use. Mentions Concepts platform is open and interoperable with Relations Entity Linking Entities Automatic Event Linking Completion external services and resources. Suggestions Syntax Events Deduplication Infrastructures & Platforms Legacy WebAnno TSV CLARIN TCF CoNLL-U CoNLL ● Supports the OpenMinTeD AERO protocol formats ● Remote configuration of annotation projects ● Import of documents for annotation Knowledge Resources ● Export of annotated documents Data Formats UIMA CAS Wikipedia Binary+XMI ● RDF-based knowledge resources are ● Notifications on key events (e.g. document or ● Supports a range of different formats for supported project finished) via web hooks importing and exporting annotated text corpora ● Knowledge bases are accessed via SPARQL ● Support of the TCF format enables Open Development DBPedia ● Configurable schema mapping allows interoperability with the CLARIN-D WebLicht Text supporting many different knowledge annotation services resources (RDFS, OWL, SKOS, …) ● Full compatibility with CLARIN’s WebAnno including the import of entire annotation projects ● Support for entity linking if knowledge base While most annotation tools are built in annotation projects, ● UIMA CAS data format enables interoperability Custom KB has full-text search capabilities ... INCEpTION is an infrastructure software project and is not with NLP pipelines like DKPro Core associated with any single annotation project. Acquiring INCEpTION early adopters and aligning with their use-cases is a key part Annotation Services Annotation Schemata DKPro TC of our mission. ● Connect external NLP services and machine ● Flexible type system configuration. This motivates our open development philosophy: learning tools to support the user during the ● Use UIMA type system descriptor to auto- Python- annotation process Authentication based configure an annotation project. ● Programming language independent HTTP- machine ● Support for external authentication ● DKPro Core types pre-configured (token, learning based protocol sentence, POS, lemma, morphological ● All code is open and publicly available on GitHub under mechanisms via HTTP headers ● Protocol supports re-training the external tools features, dependencies, coreference ● UIMA CAS XMI data format; supported in Python ● Allows delegating authentication to a reverse the liberal Apache License 2.0 chains…) ... via DKPro PyCAS; in Java via Apache UIMA proxy (e.g. Apache HTTPD) and using any of ● All development-related tasks and issues are publicly its authentication mechanisms, e.g. LDAP, managed and discussed via GitHub SAML2/Shibboleth ● Enables single-sign-on scenarios ● Internal and private communication is kept at a minimum System-level integration Data-level interoperability This work has received funding from the German Research Foundation under the grants No. EC 503/1-1 and GU 798/21-1 Download INCEpTION at (INCEpTION) and as part of the RTG Adaptive Preparation of Information from Heterogeneous Sources (AIPHES) under grant https://inception-project.github.io No. GRK 1994/1. We thank Wei Ding, Peter Jiang, Marcel de Boer and Michael Bugert for their valuable contributions..
Load more

Recommended publications
  • 9.1.1 Lesson 5
    NYS Common Core ELA & Literacy Curriculum Grade 9 • Module 1 • Unit 1 • Lesson 5 D R A F T 9.1.1 Lesson 5 Introduction In this lesson, students continue to practice reading closely through annotating text. Students will begin learning how to annotate text using four annotation codes and marking their thinking on the text. Students will reread the Stage 1 epigraph and narrative of "St. Lucy’s Home for Girls Raised by Wolves," to “Neither did they” (pp. 225–227). The text annotation will serve as a scaffold for teacher- led text-dependent questions. Students will participate in a discussion about the reasons for text annotation and its connection to close reading. After an introduction and modeling of annotation, students will practice annotating the text with a partner. As student partners practice annotation, they will discuss the codes and notes they write on the text. The annotation codes introduced in this lesson will be used throughout Unit 1. For the lesson assessment, students will use their annotation as a tool to find evidence to independently answer, in writing, one text-dependent question. For homework, students will continue to read their Accountable Independent Reading (AIR) texts, this time using a focus standard to guide their reading. Standards Assessed Standard(s) RL.9-10.1 Cite strong and thorough textual evidence to support analysis of what the text says explicitly as well as inferences drawn from the text. Addressed Standard(s) RL.9-10.3 Analyze how complex characters (e.g., those with multiple or conflicting motivations) develop over the course of a text, interact with other characters, advance the plot or develop the theme.
    [Show full text]
  • A Super-Lightweight Annotation Tool for Experts
    SLATE: A Super-Lightweight Annotation Tool for Experts Jonathan K. Kummerfeld Computer Science & Engineering University of Michigan Ann Arbor, MI, USA [email protected] Abstract we minimise the time cost of installation by imple- menting the entire system in Python using built-in Many annotation tools have been developed, covering a wide variety of tasks and pro- libraries. Third, we follow the Unix Tools Philos- viding features like user management, pre- ophy (Raymond, 2003) to write programs that do processing, and automatic labeling. However, one thing well, with flat text formats. In our case, all of these tools use Graphical User Inter- (1) the tool only does annotation, not tokenisation, faces, and often require substantial effort to automatic labeling, file management, etc, which install and configure. This paper presents a are covered by other tools, and (2) data is stored in new annotation tool that is designed to fill the a format that both people and command line tools niche of a lightweight interface for users with like grep can easily read. a terminal-based workflow. SLATE supports annotation at different scales (spans of char- SLATE supports annotation of items that are acters, tokens, and lines, or a document) and continuous spans of either characters, tokens, of different types (free text, labels, and links), lines, or documents. For all item types there are with easily customisable keybindings, and uni- three forms of annotation: labeling items with cat- code support. In a user study comparing with egories, writing free-text labels, or linking pairs other tools it was consistently the easiest to in- of items.
    [Show full text]
  • Active Learning with a Human in the Loop
    M T R 1 2 0 6 0 3 MITRETECHNICALREPORT Active Learning with a Human In The Loop Seamus Clancy Sam Bayer Robyn Kozierok November 2012 The views, opinions, and/or findings contained in this report are those of The MITRE Corpora- tion and should not be construed as an official Government position, policy, or decision, unless designated by other documentation. This technical data was produced for the U.S. Government under Contract No. W15P7T-12-C- F600, and is subject to the Rights in Technical Data—Noncommercial Items clause (DFARS) 252.227-7013 (NOV 1995) c 2013 The MITRE Corporation. All Rights Reserved. M T R 1 2 0 6 0 3 MITRETECHNICALREPORT Active Learning with a Human In The Loop Sponsor: MITRE Innovation Program Dept. No.: G063 Contract No.: W15P7T-12-C-F600 Project No.: 0712M750-AA Seamus Clancy The views, opinions, and/or findings con- Sam Bayer tained in this report are those of The MITRE Corporation and should not be construed as an Robyn Kozierok official Government position, policy, or deci- sion, unless designated by other documenta- tion. Approved for Public Release. November 2012 c 2013 The MITRE Corporation. All Rights Reserved. Abstract Text annotation is an expensive pre-requisite for applying data-driven natural language processing techniques to new datasets. Tools that can reliably reduce the time and money required to construct an annotated corpus would be of immediate value to MITRE’s sponsors. To this end, we have explored the possibility of using active learning strategies to aid human annotators in performing a basic named entity annotation task.
    [Show full text]
  • The Influence of Text Annotation Tools on Print and Digital Reading Comprehension
    28 The influence of text annotation tools on print and digital reading comprehension The Influence of Text Annotation Tools on Print and Digital Reading Comprehension Gal Ben-Yehudah Yoram Eshet-Alkalai The Open University of Israel The Open University of Israel [email protected] [email protected] Abstract Recent studies that compared reading comprehension between print and digital displays found a disadvantage for digital reading. A separate line of research has shown that effective use of learning strategies and active learning tools (e.g. annotation tools: highlighting, underlining and note- taking) can improve reading comprehension. Here, the influence of annotation tools on comprehension and their utility in diminishing the gap between print and digital reading was investigated. In a 2 X 2 experimental design, ninety three undergraduates (mean age 29.6 years, 72 women) read an academic essay in print or in digital format, with or without the use of annotation tools. In general, the results reinforce previous research reports on the inferiority of digital compared to print reading comprehension. Our findings indicate that for factual-level questions using annotation tools had no effect on comprehension in both formats. For the inference-level questions, however, use of annotation tools improved comprehension of printed text but not of digital text. In other words, text-annotation in the digital format had no effect on reading comprehension. In the digital era, digital reading is the norm rather than the exception; thus, it is essential to improve our understanding of the factors that influence the comprehension of digital text. Keywords: digital reading, print reading, annotation tools, learning strategies, self regulated learning, monitoring.
    [Show full text]
  • A Multi-Layer Markup Language for Geospatial Semantic Annotations Ludovic Moncla, Mauro Gaio
    A multi-layer markup language for geospatial semantic annotations Ludovic Moncla, Mauro Gaio To cite this version: Ludovic Moncla, Mauro Gaio. A multi-layer markup language for geospatial semantic annotations. 9th Workshop on Geographic Information Retrieval, Nov 2015, Paris, France. 10.1145/2837689.2837700. hal-01256370 HAL Id: hal-01256370 https://hal.archives-ouvertes.fr/hal-01256370 Submitted on 1 Jan 2018 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. A Multi-Layer Markup Language for Geospatial Semantic Annotations ∗ † Ludovic Moncla Mauro Gaio Universidad de Zaragoza LIUPPA - EA 3000 C/ María de Luna, 1 - Zaragoza, Spain Av. de l’Université - Pau, France [email protected] [email protected] ABSTRACT 1 Motivation and Background In this paper we describe a markup language for semantically Two categories of markup languages can be considered in annotating raw texts. We define a formal representation of geospatial tasks, those dedicated to the encoding of spatial text documents written in natural language that can be ap- data (e.g., gml, kml) and those dedicated to the annota- plied for the task of Named Entities Recognition and Spatial tion of spatial or spatio-temporal information in texts, such Role Labeling.
    [Show full text]
  • BOEMIE Ontology-Based Text Annotation Tool
    BOEMIE ontology-based text annotation tool Pavlina Fragkou, Georgios Petasis, Aris Theodorakos, Vangelis Karkaletsis, Constantine D. Spyropoulos Software and Knowledge Engineering Laboratory Institute of Informatics and Telecommunications National Center for Scientific Research (N.C.S.R.) “Demokritos” P. Grigoriou and Neapoleos Str., 15310 Aghia Paraskevi Attikis, Greece E-mail: {fragou, petasis, artheo, vangelis, costass}@ iit.demokritos.gr Abstract The huge amount of the available information in the Web creates the need of effective information extraction systems that are able to produce metadata that satisfy user’s information needs. The development of such systems, in the majority of cases, depends on the availability of an appropriately annotated corpus in order to learn extraction models. The production of such corpora can be significantly facilitated by annotation tools that are able to annotate, according to a defined ontology, not only named entities but most importantly relations between them. This paper describes the BOEMIE ontology-based annotation tool which is able to locate blocks of text that correspond to specific types of named entities, fill tables corresponding to ontology concepts with those named entities and link the filled tables based on relations defined in the domain ontology. Additionally, it can perform annotation of blocks of text that refer to the same topic. The tool has a user-friendly interface, supports automatic pre-annotation, annotation comparison as well as customization to other annotation schemata. The annotation tool has been used in a large scale annotation task involving 3000 web pages regarding athletics. It has also been used in another annotation task involving 503 web pages with medical information, in different languages.
    [Show full text]
  • PDF Annotation with Labels and Structure
    PAWLS: PDF Annotation With Labels and Structure Mark Neumann Zejiang Shen Sam Skjonsberg Allen Institute for Artificial Intelligence {markn,shannons,sams}@allenai.org Abstract the extracted text stream lacks any organization or markup, such as paragraph boundaries, figure Adobe’s Portable Document Format (PDF) is placement and page headers/footers. a popular way of distributing view-only docu- Existing popular annotation tools such as BRAT ments with a rich visual markup. This presents a challenge to NLP practitioners who wish to (Stenetorp et al., 2012) focus on annotation of user use the information contained within PDF doc- provided plain text in a web browser specifically uments for training models or data analysis, designed for annotation only. For many labeling because annotating these documents is diffi- tasks, this format is exactly what is required. How- cult. In this paper, we present PDF Annota- ever, as the scope and ability of natural language tion with Labels and Structure (PAWLS), a new processing technology goes beyond purely textual annotation tool designed specifically for the processing due in part to recent advances in large PDF document format. PAWLS is particularly suited for mixed-mode annotation and scenar- language models (Peters et al., 2018; Devlin et al., ios in which annotators require extended con- 2019, inter alia), the context and media in which text to annotate accurately. PAWLS supports datasets are created must evolve as well. span-based textual annotation, N-ary relations In addition, the quality of both data collection and freeform, non-textual bounding boxes, all and evaluation methodology is highly dependent of which can be exported in convenient for- on the particular annotation/evaluation context in mats for training multi-modal machine learn- which the data being annotated is viewed (Joseph ing models.
    [Show full text]
  • YEDDA: a Lightweight Collaborative Text Span Annotation Tool
    YEDDA: A Lightweight Collaborative Text Span Annotation Tool Jie Yang, Yue Zhang, Linwei Li, Xingxuan Li Singapore University of Technology and Design fjie yang, linwei li, xingxuan [email protected] yue [email protected] Abstract Raw Text Annotated Text In this paper, we introduce YEDDA, a Multi-Annotation lightweight but efficient and comprehen- Analysis Results sive open-source tool for text span an- Annotator 1 notation. YEDDA provides a systematic Annotator 2 Annotator Admin solution for text span annotation, rang- Interface Toolkits Administrator ing from collaborative user annotation to administrator evaluation and analysis. It Detailed Pairwise Annotator n Annotation Report overcomes the low efficiency of traditional Feedback text annotation tools by annotating entities through both command line and shortcut Figure 1: Framework of YEDDA. keys, which are configurable with custom labels. YEDDA also gives intelligent rec- ommendations by learning the up-to-date process but rarely consider the post-annotation annotated text. An administrator client quality analysis, which is necessary due to the is developed to evaluate annotation qual- inter-annotator disagreement. In addition to the ity of multiple annotators and generate de- annotation quality, efficiency is also critical in tailed comparison report for each annota- large-scale annotation task, while being relatively tor pair. Experiments show that the pro- less addressed in existing annotation tools (Ogren, posed system can reduce the annotation 2006; Stenetorp et al., 2012). Besides, many tools time by half compared with existing anno- (Ogren, 2006; Chen and Styler, 2013) require a tation tools. And the annotation time can complex system configuration on either local de- be further compressed by 16.47% through vice or server, which is not friendly to new users.
    [Show full text]
  • Close Reading Strategies for Difficult Text
    St. Catherine University SOPHIA Masters of Arts in Education Action Research Papers Education 5-2016 Close Reading Strategies for Difficultext: T The Effects on Comprehension and Analysis at the Secondary Level Kimberly Goblirsch St. Catherine University, [email protected] Follow this and additional works at: https://sophia.stkate.edu/maed Part of the Education Commons Recommended Citation Goblirsch, Kimberly. (2016). Close Reading Strategies for Difficultext: T The Effects on Comprehension and Analysis at the Secondary Level. Retrieved from Sophia, the St. Catherine University repository website: https://sophia.stkate.edu/maed/166 This Action Research Project is brought to you for free and open access by the Education at SOPHIA. It has been accepted for inclusion in Masters of Arts in Education Action Research Papers by an authorized administrator of SOPHIA. For more information, please contact [email protected]. Running head: CLOSE READING STRATEGIES OF DIFFICULT TEXT 1 Close Reading Strategies for Difficult Text: The Effects on Comprehension and Analysis at the Secondary Level An Action Research Report by Kimberly Goblirsch CLOSE READING STRATEGIES OF DIFFICULT TEXT 2 Close Reading Strategies for Difficult Text: The Effects On Comprehension and Analysis at the Secondary Level Submitted on May 6, 2016 in fulfillment of final requirements for the MAED degree Saint Catherine University St. Paul, Minnesota Advisor____________________________ Date_______________________ CLOSE READING STRATEGIES OF DIFFICULT TEXT 3 Abstract Literacy skills are an essential component in transitioning from learning to read to reading to learn. But reading to learn has several levels of difficulty. This research focuses on exploring in what ways close reading strategies of difficult text can impact comprehension and analysis.
    [Show full text]
  • Collaborative Annotation for Reliable Natural Language Processing
    Collaborative Annotation for Reliable Natural Language Processing Kar¨enFort February 12, 2016 14:47 2 Contents 1 Introduction 9 1.1 Natural Language Processing and Manual Annotation: Dr Jekyll and Mr Hyjide? . .9 1.1.1 Where Linguistics Hides . .9 1.1.2 What is Annotation? . 10 1.1.3 New Forms, Old Issues . 11 1.2 Rediscovering Annotation . 13 1.2.1 A Rise in Diversity and Complexity . 13 1.2.2 Redefining Manual Annotation Costs . 15 2 Annotating Collaboratively 17 2.1 The Annotation Process (Re)visited . 17 2.1.1 Building Consensus . 17 2.1.2 Existing Methodologies . 18 2.1.3 Preparatory Work . 20 2.1.4 Pre-campaign . 24 2.1.5 Annotation . 27 2.1.6 Finalization . 29 2.2 Annotation Complexity . 31 2.2.1 Examples Overview . 31 2.2.2 What to Annotate? . 33 2.2.3 How to Annotate? . 35 2.2.4 The Weight of the Context . 38 2.2.5 Visualization . 39 2.2.6 Elementary Annotation Tasks . 41 2.3 Annotation Tools . 42 2.3.1 To be or not to be an Annotation Tool . 43 2.3.2 Much more than Prototypes . 44 2.3.3 Addressing the new Annotation Challenges . 46 2.3.4 The Impossible Dream Tool . 49 2.4 Evaluating the Annotation Quality . 50 2.4.1 What is Annotation Quality? . 50 2.4.2 Understanding the Basics . 50 2.4.3 Beyond Kappas . 55 2.4.4 Giving Meaning to the Metrics . 57 3 4 CONTENTS 3 Crowdsourcing Annotation 63 3.1 What is Crowdsourcing and Why Should we be Interested in it? 63 3.1.1 A Moving Target .
    [Show full text]
  • Observations on Annotations
    Georg Rehm Observations on Annotations Abstract: The annotation of textual information is a fundamental activity in Lin- guistics and Computational Linguistics. This article presents various observations on annotations. It approaches the topic from several angles, including Hyper- text, Computational Linguistics and Language Technology, Artificial Intelligence and Open Science. Annotations can be examined along different dimensions. In terms of complexity, they can range from trivial to highly sophisticated, in terms of maturity from experimental to standardised. Annotations can be annotated themselves using more abstract annotations. Primary research data such as, e.g., text documents can be annotated on different layers concurrently, which are inde- pendent but can be exploited using multi-layer querying. Standards guarantee the interoperability and reusability of data sets. The chapter concludes with four final observations, formulated as research questions or rather provocative remarks on the current state of annotation research. Keywords: Evaluation, Levels of Annotation, Markup, Semantic Web, Artificial Intelligence, Computational Linguistics, Digital Humanities, Digital Publishing 1 Introduction The annotation of textual information is one of the most fundamental activities in Linguistics and Computational Linguistics including neighbouring fields such as, among others, Literary Studies, Library Science and Digital Humanities (Ide and Pustejovsky 2017; Bludau et al. 2020). Horizontally, data annotation plays an increasingly important role in Open Science, in the development of NLP/NLU prototypes (Natural Language Processing/Understanding), more application- and solution-oriented Language Technologies (LT) and systems based on neural tech- nologies in the area of Artificial Intelligence (AI). This article reflects on more than two decades of research in the wider areaof annotation including multi-layer annotations (Witt et al.
    [Show full text]
  • Simplified Dependency Annotations with GFL-Web
    Simplified Dependency Annotations with GFL-Web Michael T. Mordowanec Nathan Schneider Chris Dyer Noah A. Smith School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213, USA [email protected], {nschneid,cdyer,nasmith}@cs.cmu.edu Abstract The Fragmentary Unlabeled Dependency Grammar (FUDG) formalism (Schneider et al., We present GFL-Web, a web-based in- 2013) was proposed as a simplified framework for terface for syntactic dependency annota- annotating dependency syntax. Annotation effort tion with the lightweight FUDG/GFL for- is reduced by relaxing a number of constraints malism. Syntactic attachments are spec- placed on traditional annotators: partial fragments ified in GFL notation and visualized as can be specified where the annotator is uncertain a graph. A one-day pilot of this work- of part of the structure or wishes to focus only flow with 26 annotators established that on certain phenomena (such as verbal argument even novices were, with a bit of training, structure). FUDG also offers mechanisms for able to rapidly annotate the syntax of En- excluding extraneous tokens from the annotation, glish Twitter messages. The open-source for marking multiword units, and for describing tool is easily installed and configured; it coordinate structures. FUDG is written in an is available at: https://github.com/ ASCII-based notation for annotations called Mordeaux/gfl_web Graph Fragment Language (GFL), and text-based tools for verifying, converting, and rendering GFL 1 Introduction annotations are provided. High-quality syntactic annotation of natural lan- Although GFL offers a number of conveniences guage is expensive to produce. Well-known large- to annotators, the text-based UI is limiting: the scale syntactic annotation projects, such as the existing tools require constant switching between Penn Treebank (Marcus et al., 1993), the En- a text editor and executing commands, and there glish Web Treebank (Bies et al., 2012), the Penn are no tools for managing a large-scale annotation Arabic Treebank (Maamouri et al., 2004), and effort.
    [Show full text]