计算机新闻学TextAnalysis.ppt

Partitioning clustering Keep adjusting clusters until convergence E.g., k-means Shandong University Text Analysis For News： Shandong University Text Analysis READING:A full-text visualization of the Iraq war logs Shandong University Text Analysis This is a picture of the 11,616 SIGACT (“significant action”) reports from December 2006,the bloodiest month of the war. 1.Interpreting the Iraq War, December 2006 2.How to do it, and what we can and can’t learn from this picture 3.Much more is possible /blog/2010/12/a-full-text-visualization-of-the-iraq-war-logs/ READING:A full-text visualization of the Iraq war logs Interpreting the Iraq War, December 2006 Shandong University Text Analysis 1.Each report is a dot. 2.Each dot is labelled by the three most “characteristic” words in that report. 3.Documents that are “similar” have edges drawn between them. 4.Dots with edges between them are pulled closer together. 5.Clusters are labelled by the words that are most “characteristic” of the reports in that cluster. 6.Color each report/dot by the “incident type”. Shandong University Text Analysis Interpreting the Iraq War, December 2006 The central cluster is blue, the color for the “criminal event” type, and the documents within it all include the word “corpse.” There are a heartbreaking number of them, because this was the height of the Iraqi civil war. Sub-clusters include various modifiers such as “shot.” READING:A full-text visualization of the Iraq war logs READING:A full-text visualization of the Iraq war logs Shandong University Text Analysis Interpreting the Iraq War, December 2006 The blue murders merge into the green “enemy action” reports. At the interface we have “civ, killed, shot,” which are apparently reports of civilians wounded in battle. Enemy actions also have their own clusters labelled with “mortar,” “female,” “officer,” and “injured.” READING:A full-text visualization of the Iraq war logs Shandong University Text Analysis Interpreting the Iraq Wa