关键词: 主观题, 自然语言理解, 语义脉络, 单向贴近度, 动态规划

Abstract:

The present research status both at home and abroad are introduced and analyzed. By using the string matching method of the single similar degree which is based on the closeness theory of fuzzy mathematics, and combining with dynamic programming algorithms, a subjective automatic assessment algorithm based on a semantic skeleton is designed and implemented. Taking sentence as the basic semantic unit, this approach separates standard answers into some keywords representing the points for score, adds a synonym chain for the keywords to match the students’ answers which can make expression much more complete and accurate, and then calculates the score according to the matching degree. The string matching is conducted in accordance with the order of words by using dynamic programming algorithms to avoid the mistakes caused by mechanical matching, which is carried out only according to words. Then the main idea and the flow of the algorithm are given. Finally, the feasibility of this algorithm is proved through the analysis of an example.

Key words: subjective;natural language understanding;semantic skeleton;single similar degree;dynamic programming