spark word2vec窗口大小

时间:2015-08-26 16:30:06

标签: machine-learning apache-spark word2vec

是否有更改 //Filter on known fields var keyboard = Query<Product>.Create(p=>p.Category=="Keyboard"); var keyboards = repository.Get(keyboard); //Or filter on dynamic fields var filter = Query<Product>.Create("Rating", OperationType.GreaterThan, 4) var filteredKeyboards = repository.Get(filter); //You can also combine two queries togather var filterdKeyboards2 = repository.Get(keyboard.And(filter)) //Order it on known fields var orderedKeyboard = keyboard.OrderBy(o=>o.Asc(p=>p.Name)); var orderedKeyboards = repository.Get(orderedKeyboard); //Or order by on dynamic fields var userOrdering = keyboard.OrderBy(o=>o.Asc("Name")); var orderedKeyboards2 = repository.Get(userOrdering); 中使用的窗口大小或是否永久固定为5?这似乎是一个相当重要的特征。

我在这里看不到这个选项: https://spark.apache.org/docs/1.4.1/api/scala/index.html#org.apache.spark.mllib.feature.Word2Vec

1 个答案:

答案 0 :(得分:5)

多数民众赞成正确,看看Word2Vec代码,我们可以看到它是一个私有的val。如果你想,你可以覆盖Word2Vec来改变窗口大小。我已经创建了一个JIRA(https://issues.apache.org/jira/browse/SPARK-10299)以允许设置窗口大小,因为1.5已经处于RC阶段,这可能直到1.6时才会出现。