NLP Project for Summarizing Comments

I am planning to do my latest natural language processing project (using NLTK), and my area of ​​interest is Commenting on summation from social media websites like Facebook. For example, I'm trying to do something like this:

Random Facebook comments in the picture:

  • Wow! Beautiful.
  • The look is really beautiful.
  • Very pretty, Nice.

Now all these comments will be matched (using a template based comment summarization technique) something like this:

3 people consider this photo "beautiful".

The output will consist of the word "beautiful" as it is more commonly used in comments than the word "pretty" (and the fact that Beautiful and pretty are synonyms). To accomplish this task, I'm going to use approaches such as tracking Keyword Frequency and Keyword Score (In this scenario, “Beautiful” and “Pretty” “have a very close score.” Is this the best way to do this?

So far with my research, I could come up with the following docs, but none of these articles deal with such a generalization of comments:

What are other papers in this area that address a similar issue?

Also, I also want my compiler to improve with every summation task. How did I apply machine learning in this regard?

+3


source to share


1 answer


Model model clustering is what you are looking for.

Searching Google Scholars for "topic model clustering" will give you many links to topic model clustering.



To understand them, you should be familiar with approaches to the following tasks, besides the basics of machine learning in general.

  • Clustering: clustering cosine distances, clustering k-values
  • Ranking: PageRank, TF-IDF, mutual information ratio, maximum margin relevance
+8


source







All Articles