Create a text document on any topic(eg. my school days)
.
Answers
Get started
Open in app
Responses (13)
To respond to this story,
get the free Medium app.
Open in app
Alejandro Cadavid Romero
Alejandro Cadavid Romero
12 months ago
Nice article…one advice, in order to make the notebook or the experiment reproducible, please add the helper functions that you used.
10
Dimitra Papagianni
Dimitra Papagianni
about 1 year ago
This is an incomplete project. Such a waste of time
8
Mikołaj Nurek
Mikołaj Nurek
about 1 year ago
Very cool article!
I’m trying to replicate it on my own data and it’s going great, but I have one problem: how did you create the topic term and document topic matrix needed for pyLDAvis to visualize the clusters? I’m confused because there is no such data directly in GSDMM class on github
4
1
Simon meunier
Simon meunier
over 1 year ago
Really cool article, thanks man!!
3
Omar Souaidi
Omar Souaidi
12 months ago
Good article, very useful!
1
Syaamantak Das
Syaamantak Das
over 1 year ago
Hi, Thanks for such a nice article. I have three specific question — (i) What length of words can be considered as short text ? (ii) Can we use any form of supervised learning on LDA for STTM ? Let’s say I have short texts with average length of 100…...
Read More
1
1
Rishi
Rishi
over 1 year ago
Hi Matyas,
This was a great read of today. Thank you!
Do you think we have any unsupervised approaches for STTM?
1
1
bryce peake
bryce peake
about 1 year ago
rom topic_allocation i
Photo by Hello I’m Nik on Unsplash
Topic Modeling aims to find the topics (or clusters) inside a corpus of texts (like mails or news articles), without knowing those topics at first. Here lies the real power of Topic Modeling, you don’t need any labeled or annotated data, only raw texts, and from this chaos Topic Modeling algorithms will find the topics your texts are about!
In this post we will describe the intuition and logic behind the most popular approach for Topic Modeling, the LDA, and see its limitation on short texts. Given this post is about Short Text Topic Modeling (STTM) we will not dive into the details of LDA. The reader willing to deepen his knowledge of LDA can find great articles and useful resources about LDA here and here.
Then, in a second part, we will present a new approach for STTM and finally see in a third part how to easily apply it (fit/predict ) on a toy dataset and evaluate its performance.
The reader already familiar with LDA and Topic Modeling may want to skip the first part and directly go to the second and third ones which present a new approach for Short Text Topic Modeling and its Python coding .
hope you liked my answer mark it brainliest and give it a thanks pls pls pls.