Ask HN: Summarize Content Machine Learning

1 points by cwt 10y ago ↗ HN

Are there any projects out there that attempt to use Machine Learning (other technologies welcome too) to create a summary of given content?

Specifically, I am thinking in terms of reading comprehension. A list of prompts given with content and the expectation is the computer answers the prompts based on the content provided. But a project that attempts to summarize/parse content would be awesome.

3 comments

[ 5.2 ms ] story [ 15.6 ms ] thread

jeffmould 10y ago ↗

I was doing some research on natural language algorithms a while back. Not sure if these links are of any help for what you are looking for:

https://github.com/26medias/node-nlpsum

https://www.quora.com/How-do-I-extract-key-phrases-from-onli...

codyguy 10y ago ↗

ThatNeedle is working on it and showing promising results. For now there is only an NLP API for retail.

signaler 10y ago ↗

A very simple example is that of Google which gives you a Wikipedia summary of a topic, but that's too simple.

In your case you are looking for some way to heatmap certain keywords, or even the tone of a piece of content. You could go further and see the context of the content (which is presumably some form of document which can be parsed).

There are innumerable things online to do this. My first port of call is to trawl Github and find a repo on there which does this.

Just be careful of online services which are hoovering up your query and making off with your data...