Show HN: TF-IDF based text analyzer

bookshrink.com

14 points·peter_l_downs··5 comments

It's a bit of a paradox that the head of the page reads "BookShrink - summary made simple" when, further down, the user is told that "this will not generate a summary of what you type in. [...] Summary is something different." I find it confusing.

Another thing is that the results area should be used for the results, not error messages. For instance, i entered a text with less than 10 sentences (which is illegal) and couldn't tell what was wrong. Perhaps highlight the error (ie use colour) and show it above the form.

This is rather interesting, it could be marketed in the autoblog arena with ease!

Inconsistencies have been fixed! Thanks for pointing it out - you were 100% correct.

I'm still working on the error handling - you're right, there should be a better way to do it than what I have now.

So, for the idf component, what is the corpus of other documents you're comparing the input text against?

Doesn't work with Cyrillic text. I got only dots as summary.

This should work now. I can't verify how accurate it will be, as I don't read or speak a language with a Cyrillic alphabet, and there were quite a few modifications to the algorithm that had to do with English text only, but you'll at least get a result back at this point. Let me know if it's at all accurate!

Reply on news.ycombinator.com