Automatic Computing Semantic Relatedness Using Absolute Depth and Link Content on WikiRelationNet

• Main Authors: You-wei Chen, 陳佑瑋
• Other Authors: Shih-Hung Wu
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/85757964031977392897

碩士 === 朝陽科技大學 === 資訊工程系碩士班 === 99 === In this paper, we develop a new way to calculate the semantic realatedness between two terms. We propose a new semantic network, WikiRelationNet, and a new way to calculate semantic relatedness on it. The WikiRelationNet is built based on the entities and links between entities in Wikipedia. An absolute depth concept is also proposed to covert a network graph into a tree-like graph. By measuring the distance and link content of the nodes on WikiRelationNet, we can get the semantic relatedness between two nodes. Our system is evaluated on a manually collected test set, WordSimilarity-353 collection, and the Spearman correlation coefficient between the results of our system and the manual annotated data is reported. We conduct several experiments and show how to calculate the semantic relatedness on WikiRelationNet.