计算机工程
計算機工程
계산궤공정
COMPUTER ENGINEERING
2015年
2期
151-156
,共6页
词语相似度%词语语义%义原深度%概念
詞語相似度%詞語語義%義原深度%概唸
사어상사도%사어어의%의원심도%개념
word similarity%word semantic%depth of sememe%concept
现有词语相似度计算方法未深入考虑义原之间的距离与义原深度的主次关系，或直接指定含具体词概念的相似度，导致计算结果不够精确。针对该问题，通过义原之间的距离限制义原深度对义原相似度的影响，分析统计《知网》中概念的义项表达式，使用第一基本义原(能反映具体词本质)替换概念义项表达式中出现的具体词，从而提出一种改进的词语语义相似度计算算法。实验结果表明，该算法能有效提高词汇相似度计算的精确度。
現有詞語相似度計算方法未深入攷慮義原之間的距離與義原深度的主次關繫，或直接指定含具體詞概唸的相似度，導緻計算結果不夠精確。針對該問題，通過義原之間的距離限製義原深度對義原相似度的影響，分析統計《知網》中概唸的義項錶達式，使用第一基本義原(能反映具體詞本質)替換概唸義項錶達式中齣現的具體詞，從而提齣一種改進的詞語語義相似度計算算法。實驗結果錶明，該算法能有效提高詞彙相似度計算的精確度。
현유사어상사도계산방법미심입고필의원지간적거리여의원심도적주차관계，혹직접지정함구체사개념적상사도，도치계산결과불구정학。침대해문제，통과의원지간적거리한제의원심도대의원상사도적영향，분석통계《지망》중개념적의항표체식，사용제일기본의원(능반영구체사본질)체환개념의항표체식중출현적구체사，종이제출일충개진적사어어의상사도계산산법。실험결과표명，해산법능유효제고사회상사도계산적정학도。
The current word similarity calculation does not consider in depth with the primary and secondary relationship between the distance and the depth of sememes. In addition,concept similarity is specified directly when the conceptual description expression contains specific words,which leads to unreasonable. The depth of sememes impacts on the word similarity is limited by the distance of sememes. It analyzes the statistical meanings of the concept expression in“HowNet”. Besides,word similarity calculation uses the first basic sememe that can reflect the essence of the word to replace the specific words that appear in the conceptual description expression. Based on the above,an improved algorithm of word semantic similarity is proposed in this paper. Experimental results show that the improved algorithm effectively improves the precision of word similarity calculations.
