Home首页 · Paths路径 · Counts, clusters, coordinates计数、分组、坐标

Counts, clusters, coordinates计数、分组、坐标

Three ways of reading the same short headlines about a training program. Each link opens a lab at a suggested setting.用三种方式读同一批关于培训项目的短标题。每个链接按一个建议设定打开实验。

  1. Shared strings

    Opposite claims, little words such as the and for kept. The agreement score stays high because the headlines share most words. Word order, and the verb that flips the claim, is a small part of the count.相反的主张,the、for 这类小词留下。一致分数仍然高,因为标题共享大部分词。词序,和那个把主张翻过来的动词,只占计数的一小部分。

    Open the bag-of-words lab打开词袋实验

  2. How many groups

    One folder puts every clipping in the same pile. Move the slider to 3 on that page and see the lists separate. The number of folders was not written in the clippings.一个文件夹把每一行剪报放进同一堆。在那一页把滑块调到 3,看清单分开。文件夹的个数没有写在剪报里。

    Open the topic-count lab at K = 1打开 K = 1 的主题个数实验

  3. A map can disagree with the counts地图可以和计数不一致

    Different words, the same kind of claim. The count score is zero. The places chosen by hand put the synonyms near each other, so the map score is not zero.用词不同,主张同类。计数分数为零。亲手选定的位置把同义词放在附近,所以地图分数不是零。

    Open the word map

All paths全部路径