CAJ | 학술논문

句子打分算法作为抽取式自动摘要方法中最为关键的组成部分，成为近年来自动摘要领域的热点问题。本文提出了一种基于电阻网络图的摘要句子打分方法，该方法将句子作为节点，句子之间的相似度作为边，通过基尔霍夫定律在文本相似度图上计算两节点间的电阻距离，并在此基础上提出一种基于电能消耗的节点重要度评估方法来抽取文章句子形成摘要。本文还给出了一个基于该模型的高效算法，最后通过实验分析验证了通过该模型和算法来生成文本摘要的高效性和有效性。
구자타분산법작위추취식자동적요방법중최위관건적조성부분，성위근년래자동적요영역적열점문제。본문제출료일충기우전조망락도적적요구자타분방법，해방법장구자작위절점，구자지간적상사도작위변，통과기이곽부정률재문본상사도도상계산량절점간적전조거리，병재차기출상제출일충기우전능소모적절점중요도평고방법래추취문장구자형성적요。본문환급출료일개기우해모형적고효산법，최후통과실험분석험증료통과해모형화산법래생성문본적요적고효성화유효성。
the sentence scoring algorltt~m as the most crucml part m automatic extraction summarization methods, has been the hotspot in the field of automatic Summarization in recent years. This paper puts forward a sentences scoring methods based on the resistance of the network diagram, this method regard a sentence as a node, the similarity between sentences as edge, then calculating resistance distance of two nodes in text similarity graph through the kirchhoffs law, and based on this, puts forward a kind of based on electricity consumption of node importance evaluation method to extract the sentence to form summarization. This paper also gives a efficient algorithm based on this model, the experimental analysis verified the efficiency and effectiveness of the model and algorithm.