Chris Mellish's software and documentation archive.
Includes several bits of code and tutorial notes.
A Java class library which performs basic NLG lexicalisation and realisation.
The interactive software for building punning riddles, created on the STANDUP project, is available for downloading. Some of its component parts are also available as Java APIs.
The SumTime-Meteo Corpus
Now available on the SIGGEN resources wiki, this is one of the few available corpora that provide a raw data model and textual reports on the data, making it ideal for NLG research.
A corpus of identifying references to objects in visual domains.
A large lexical database, created on the STANDUP project, is available for downloading. It is derived from WordNet, but also contains other data (e.g. phonetic representations). There is also software for modifying the database.
The TUNA Annotated Bibliography on the Generation of Referring Expressions and related problems
Annotated online bibliography of papers dealing with automatic generation of eferring expressions and psycholinguistic research into the problem. Compiled as part of the TUNA Project.
SumTime system evaluation: Online summary
A summary of a recent evaluation experiment on the SumTime weather report generation system, showing that the system's performance compares extremely favourably to that of human authors.