Comments for Pavel Surmenok

Comment on Modeling of customer, order and address entities by Javier H. Mora

Javier H. Mora — Sat, 13 Apr 2019 17:14:00 +0000

“The IsChangeable attribute defines whether the address can be changed by the customer later. I consider that an address should be unchangeable if it is related to an order.”

Thanks, I read all the article just for that sentence haha. Great article btw!

Comment on Improving DeepSpell Code by Alex Skasgard

Alex Skasgard — Tue, 05 Feb 2019 16:03:00 +0000

Can you share new approach? @yaakov_belch:disqus

Comment on Chatbot Architecture by surmenok

surmenok — Fri, 21 Sep 2018 05:07:00 +0000

There was research showing that using word+POS tag works better than just a word for text classification tasks.

Comment on Chatbot Architecture by Roma Jain

Roma Jain — Wed, 19 Sep 2018 02:58:00 +0000

Awesome article!
Should the words tagged as entities be used instead of words for intent classification?

Comment on How to Run Text Summarization with TensorFlow by David Hansen

David Hansen — Wed, 30 May 2018 20:12:00 +0000

Hello, I’m joining the discussion a year (maybe two) late.

First things first–many thanks to Pavel for the blog–the clearest explanation I have read.

My question:
Summarizing the first two sentences of an article is a good step and like others in this post stream I wish to apply textsum beyond sentences to full article. Wondering if anyone has had a modicum of success in summarizing beyond two sentences. Also wondering if it feasible to treat each paragraph (first two sentences of each) in a larger article as separate “articles” (round robin fashion) to generate summary. If this is possible it might open way to sum across many full length articles.

Full disclosure: I am not well versed in writing or running code but manage to stumble through, so I may be blindly dreaming of solution that is not possible.

Comment on How to Run Text Summarization with TensorFlow by TD

TD — Tue, 30 Jan 2018 05:14:00 +0000

Excellent, Thanks Pavel. I will look for this link.

Comment on How to Run Text Summarization with TensorFlow by surmenok

surmenok — Mon, 29 Jan 2018 22:52:00 +0000

Hi Tridib,
I haven’t played with textsum for a while.
I think there could have been some improvement in text summarization algorithms since my post, and it makes sense to look for better models. There were links to Salesforce and IBM research in comments to this article.

Comment on How to Run Text Summarization with TensorFlow by Tridib dutta

Tridib dutta — Mon, 29 Jan 2018 22:45:00 +0000

Hi Pavel,
Nice post. I am also trying to use textsum in my project. But so far facing lot of issues. I am just wondering whether you have further played around with it. Eventually I would like to use it once I get over the issues I am facing. Thanks

Comment on Natural Language Pipeline for Chatbots by surmenok

surmenok — Tue, 16 Jan 2018 03:03:00 +0000

Another idea. Entity recognition can be done by assigning a tag to each word (e.g. https://en.wikipedia.org/wiki/Inside%E2%80%93outside%E2%80%93beginning_(tagging) )
If you perform this kind of entity tagging first, then you can just a sequence of pairs (word, tag) into classification model instead of a sequence of words.
This way you don’t increase complexity of the model too much.

Comment on Natural Language Pipeline for Chatbots by Kaushik Govindarajan

Kaushik Govindarajan — Tue, 16 Jan 2018 02:52:00 +0000

That sounds like good approach. So while converting my words to vectors I need to add additional features for the entities. While predicting, the NER has to be done first and based on its output I can predict the intent also. But the only thing that concerns me is that the features will keep increasing as my entities increase and hence the classification algorithm may not perform well.
Thanks