1995); (3) analyze news wires and transcripts of radio and television broadcasts to find and summarize descriptions of terrorist activities (MUC-4 1992; MUC-3 1991); (4) monitor technical articles describing microelectronic chip fabrication to capture information on chip sales, manufacturing advances, and the development or use of chip-processing technologies (MUC-5 1994); (5) analyze newspaper articles with the goal of finding and summarizing business joint ventures (MUC-5 1994); and (6) support the automatic classification of legal documents (Holowczak and Adam 1997).
This result is especially impressive because the handcrafted patterns had placed the University of Massachusetts system among the top performers in the MUC-3 performance evaluation (Lehnert et al.
Evaluating Message-Understanding Systems: An Analysis of the Third Message-Understanding Conference (MUC-3).
Hughes Trainable Text Skimmer: Description of the TTS System as Used for MUC-3. In Proceedings of the Third Message-Understanding Conference (MUC-3), 155-162.
University of Massachusetts: Description of the CIRCUS System as Used in MUC-3. In Proceedings of the Third Message-Understanding Conference (MUC-3), 223-233.
(The K in MUCK was dropped from the names of more recent conferences.) A change of topic and text type was made for the third and fourth conferences, called MUC-3 and MUC-4 [7, 8].
MUC-3 and MUC-4 involved a largely stable group of about 17 research sites.(2) These groups met three times to discuss their progress, problems, and proposed plans in preparing for future system evaluations.
DARPA's interest in IE, inspired by the progress shown by MUC-3, prompted the funding of several large-scale IE projects using various approaches, although all worked on the same tasks.
Several measures were developed for MUC-3 and MUC-4 .
* "Prior to MUC-3, no one would have considered that parsing fragments would work, now everyone does it."
Shallow knowledge is a term coined by MUC-3 participants to describe their approaches to the knowledge engineering bottleneck.
(A discussion by several MUC-3 participants appears in the MUC-3 proceedings .) Of the three problems, co-reference resolution is by far the most challenging.