Distinguishing Fact From Opinion And Events From Meta-Events

A major problem In automatical ly anaiy~dng the t e x t o f m~lltary messaq;les In m~lar to synthes ize data b ~ l e e le ments is separating f ac t from opinion, I.e., I d e n t ~ n 9 factual s ta tements versus eva luat ive commentary such 4,; degree Of bel ie f or conf idence In the parameters o f s part icular event , comparat ive evaiual~ons o f ent i t les and events, predict ions o f fu ture events , etc. This paper desmlbes a m~iel which accmmta fo r the range o f f a c tual to eva luat ive Information In the message t raf f ic , and dlocusses a means for repreaenUng such Ioformation In the con tex t o f an oxperlmentsJ system for automated data b a l e geners t lo~