From a classed large-scale corpus, key words are extracted from web pages indexed by specialists, and based on the domainal iuhomogeneity and critical region of key words, subject degree is brought up and calculated by statistical model to cue text content' subject concept. Based on subject degree, a key words auto-indexing system is constructed, whose average score is 8.08. Aimed at the difficulty in testing key words indexing, a meticulous, integrated and manipulable test method, multi-feature synthetical test, is put forward.
展开▼
