Semantic Analysis of Tag Similarity Measures in Collaborative Tagging Systems - Computer Science > Digital LibrariesReport as inadecuate




Semantic Analysis of Tag Similarity Measures in Collaborative Tagging Systems - Computer Science > Digital Libraries - Download this document for free, or read online. Document in PDF available to download.

Abstract: Social bookmarking systems allow users to organise collections of resourceson the Web in a collaborative fashion. The increasing popularity of thesesystems as well as first insights into their emergent semantics have made themrelevant to disciplines like knowledge extraction and ontology learning. Theproblem of devising methods to measure the semantic relatedness between tagsand characterizing it semantically is still largely open. Here we analyze threemeasures of tag relatedness: tag co-occurrence, cosine similarity ofco-occurrence distributions, and FolkRank, an adaptation of the PageRankalgorithm to folksonomies. Each measure is computed on tags from a large-scaledataset crawled from the social bookmarking system del.icio.us. To provide asemantic grounding of our findings, a connection to WordNet a semantic lexiconfor the English language is established by mapping tags into synonym sets ofWordNet, and applying there well-known metrics of semantic similarity. Ourresults clearly expose different characteristics of the selected measures ofrelatedness, making them applicable to different subtasks of knowledgeextraction such as synonym detection or discovery of concept hierarchies.



Author: Ciro Cattuto, Dominik Benz, Andreas Hotho, Gerd Stumme

Source: https://arxiv.org/







Related documents