gensim lda模型中的主题权重
我们在Gensim软件包中是否有任何功能来计算主题的权重? 我需要这个重量将它们与0.05(p_value)进行比较,以了解哪个是均值的。
do we have any function in gensim package to calculate the weights of topics ?
I need this weight to compare them with 0.05 (p_value) to understand which one is mean full.
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
目前尚不清楚“主题的权重”是什么意思。您可以在评论中澄清或编辑您的问题吗?
核心 Gensim LDA 方法,例如
get_document_topics()
(docs)已经以(主题,概率)
对列表的形式返回结果,并允许您提供可选的minimum_probability
不返回任何概率较低的主题。这可能正是您想要的。但请记住,尽管许多研究领域的惯例是“零假设产生这些结果的机会小于 0.05”,但这个阈值并没有什么神奇或真实的地方——只是方便和方便。习惯。 LDA 报告的概率较低的主题仍然对最终用户有用,或者对下游分析(如分类器)有帮助。
It's not clear what you mean by "weights of topics". Can you clarify, in a comment, or edit to your question?
Core Gensim LDA methods like
get_document_topics()
(docs) already return results as a list of(topic, probability)
pairs, and allow you to supply an optionalminimum_probability
to not return any topics with a lower probability. That may be roughly what you want.But keep in mind, despite the convention in many fields of research that "the chance of the null hypothesis generating these results is less than 0.05", there's nothing magical or true about that threshold – just convenient & customary. Topics that LDA reports with lower-probability could still be useful to end-users, or helpful contributors to downstream analyses (like classifiers).