Dear community,
I'm trying to analyze a document based on multiple terms. I have 80 terms and 137 documents. I would like to reduce the number of terms and try to cluster different words into words that reflect the same concept. For example, the terms reduce cost, reduce expense, cut cost, technology, etc can be grouped under one concept which is cost minimization.
I have computed the U,d,V terms of the singular value decompositio (SVD) of the term-document matrix, and I have chosen to reduce the terms into 5 instead of 80. So I have the following dimension of these matrices U (80*5) d(5*5) and V^(t)(5*137)
I would like to know what is the next step ?
Many thanks for your great help
You can search for how to use svd in topic modeling
Hope something like the following link will help you
https://hackernoon.com/advanced-topic-modeling-tutorial-how-to-use-svd-and-nmf-in-python-to-find-topics-in-text
Samah Jradi Singular value decomposition (SVD) can be used to minimize the dimensions of the term-document frequency matrix to increase performance. SVD reduces the dimension of the matrix, making it more compact and informative. A large number of SVD dimensions generally results in a better summary of the data.
Hi, the next step is usually to plot the new space and interpret the proximities.
You can also built a similarity matrix terms/documents to find topics
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix