How to embed a speech part marker (POS)

Question

I am looking for the best PHP way to scan a large number of text entries (ads) and pull out keywords - does anyone know about Part-of-Speech? Is there any way PHP-ish does this?

I look at a lot of online ads, but none of them have categories! To speed up the categorization process, I want to install the Part of of Speech tagger (http://en.wikipedia.org/wiki/Part-of-speech_tagging). In principle, these are cool algorithmic text analysis packages that can tell me which words are nouns (for example, “Apartment”, “Car”, “Dog”, etc.) and which words are undesirable, for example, if, and, etc., BUT ...

There are online tagged services - one from Yahoo, which seems to be getting less love these days, and the other is XEROX. However, I am really interested in installing my own library / software and connecting it to my web application.

Does anyone know a good way to set POS tags that work with a PHP web application? I am dying to understand this, so any information, advice or other wisdom you have is really appreciated!

Here is a list of LOTs of different PICs: http://www-nlp.stanford.edu/links/statnlp.html#Taggers (Look under "POS Taggers")

Thanks for reading this!

Answer 1

Ian Barber implemented Brill Tagger in PHP, which he presents in his PHP / ir , where he describes its use for tweet analysis.

Answer 2

I am currently using the Brill tag. This works to some extent, although I would like to figure out how to contribute to his set of rules. It makes a lot of mistakes, but it gives about 85% of the exact data. My only problem is that she is SLOW!

It is correct when he takes into account words with a double meaning, however, there are many unaccounted conventions, such as contrasting sentences of unions, for example, I can say something negative about someone, but after the decimal point, say something that flips the polarity to positive or not. The computer cannot see idioms.