Welcome to the MMAX Annotation Tool Homepage.
This web site (http://mmax.eml-research.de) covers primarily the new MMAX2 annotation tool. Information regarding its predecessor MMAX (including a list of projects using MMAX) can be found here.
Please direct all questions, comments and suggestions
regarding this web site and the featured annotation tool(s) to [email protected].
Download MMAX2: Click here to download an evaluation version of the latest version of MMAX2 (1.0 beta 4 b). (Released February 9, 2005, updated February 17, 2005).
Note: If you downloaded MMAX2 1.0 beta 4 already, click here
to update to version 1.0 beta 4 b. Follow the instructions in the file
README.txt. This version fixes some display problems with discontinuous
markables and odd but rare attribute window behaviour.
This version contains an anonymous evaluation key valid until end of March. Note: If you download MMAX2 after end of March, just send a mail to the above address, and we will mail you a new key.
Different forms of licenses (for academic teaching, academic research and commerical use) are available. Contact us at [email protected] for details.
Do you want to know why MMAX2 is not free? Click here!
Some of MMAX2's features:
MMAX2 features separation of base data and annotations (i.e. markables) on the file level by means of stand-off annotation.
In contrast to its predecessor, MMAX2 now supports arbitrarily many levels of annotation, each of which resides in a separate file.
Arbitrarily many relations between markables (both intra- and inter-level) can now be defined, annotated and browsed.
The following screen shot shows how a set relation can be used to model and visualize coreference.
Similarly, a pointer relation can be used to link a bridging expression to its bridging antecedent.
MMAX2 now also supports attribute-dependent rendering of markables.
In the above screen shots, e.g., coref-level markables that are
actually part of a coreference set are rendered in italics, while
singleton coref-level markables are rendered in plain text.
The screen shot below (taken from the SPAAC project) demonstrates a different type of visualization:
Each utterance is 'tagged' in the display with its associated speech
act tag (left) and its syntactic type (right). In addition, utterances
with a negative polarity appear in red, those with a positive polarity
in green.
MMAX2 comes with a Markable Browser
add-on which allows for easy browsing through markable levels in either
alphabetic or document order. The markable browser also has a KWIC-Index view for easily inspecting markables in their context.
MMAX2
now contains functionality to edit a project's base data, i.e. words.
Supported editing commands include modification, deletion and insertion
of elements. This functionality is necessary for correcting transcription and tokenization errors, but it also allows insertion of zero elements and splitting off of clitics.
The
MMAX2 Project Wizard allows you to create a MMAX2 project from raw
text. Apart from tokenization and XML-conversion, the wizard can also
create markables based on structural information in the input file. See
the updated MMAX2 Quick Start Guide (also included in the tool
download) for details.
MMAX: Multi-Modal Annotation in XML
Download MMAX1
Note: This section is about an old version
of MMAX. This version is stable and will remain to be available for
free download. However, the tool will not be developed any further. The only version that is being actively
developed is now MMAX2.
Click here to download the complete MMAX package (latest version 0.94 build 77) (tar-gzipped) (incl. Apache Xalan and Xerces).
This is required for the first installation of MMAX1.
Click here to download the latest version (0.94 build 77) only.
Note: From version 0.865 on, MMAX will contain beta-versions of the MMAX Discourse API.
Last Update: April 14th, 2003 (0.94 build 77) Note: Please reference this web site (http://mmax.eml-research.de) if you use the MMAX tool in your projects.
You might also want to include a reference to a MMAX-related paper from our publications section.
The following is a (still incomplete) list of non-EML projects using or evaluating MMAX (last modification: August 13, 2004)
Research Projects
NEW MULI: Multilinguale Informationsstruktur
Baumann, Stephan et al. (2004): Multi-dimensional annotation of linguistic corpora for investigating information structure.
Kruijff-Korbayová, Ivana and Kruijff, Geert-Jan (2004): Discourse-Level Annotation for Investigating Information Structure.
VENEX (no web site yet)
Poesio, Massimo et al. (2004): The VENEX corpus of anaphora and deixis in spoken and written Italian
DIALOG: Tutorial Dialogue with a Mathematics Assistance System
Wolska, Magdalena et al. (2004): An annotated corpus of tutorial dialogs on mathematical theorem proving
GNOME
Poesio, Massimo (2004): The MATE/GNOME Scheme for Anaphoric Annotation, Revisited
The Potsdam Commentary Corpus
NEW Stede, Manfred (2004): The Potsdam Commentary Corpus
Dipper,
Stefanie and Götze, Michael and Stede, Manfred (2004): Simple
Annotation Tools for Complex Annotation Tasks: an Evaluation
ICT 'Mission Rehearsal Excercise Project'
Garg, Saurabh et al. (2004): Evaluation of Transcription and Annotation tools for a Multi-modal, Multi-party dialogue corpus
ANANAS: Annotation Anaphorique pour l' Analyse Sémantique de Corpus
Gardent, Claire and Manuélian, Hélène and Kow, Eric (2003): Which bridges for bridging definite descriptions?
COMMOn-REFs (no current web site)
Vieira, Renata et al. (2003): From concrete to virtual annotation mark-up language: The case of COMMON-REFs
IM2.MDM
Popescu-Belis, Andrei et al. (2004): Building and Using a Corpus of Shallow Dialog Annotated Meetings
University courses related to MMAX
Ivana Kruijff-Korbayova, Uni Saarbrücken: Resolution of anaphoric reference
<< back
|
