Show HN: TextBlob, Natural language processing made simple in Python

Yay #1: a nice wrapper around NLTK. NLTK is great but its API is not very Pythonic or comfortable. Pleasant facades over it are a great help for Python NLP.<p>Yay #2: an actually interesting programming-related article on HN. These get rarer every day, losing their place to gossips about what Snowden remarked following some or another NSA official&#x27;s remarks about Snowden&#x27;s even earlier remarks.

Just a quick word on Pattern[1].<p>TextBlob is probably just using the en module, I would suggest everyone take a look at the other modules in particular the web module should you be doing any light data scraping. It has nice wrappers around BeautifulSoup and Scrapy among others, jumping into BeautifulSoup and Scrapy can be daunting for beginners.<p>[1] <a href="http://www.clips.ua.ac.be/pages/pattern" rel="nofollow">http:&#x2F;&#x2F;www.clips.ua.ac.be&#x2F;pages&#x2F;pattern</a>

I&#x27;ve had good fun playing around with this, it&#x27;s certainly made NLP more approachable.<p>One issue though is that it seems to choke with certain characters.<p>For instance the character £ it seems to complain with this error message:<p>&gt;&gt;&gt; TextBlob(&quot;£&quot;) Traceback (most recent call last): File &quot;&lt;stdin&gt;&quot;, line 1, in &lt;module&gt; File &quot;&#x2F;home&#x2F;eterm&#x2F;nlp&#x2F;local&#x2F;lib&#x2F;python2.7&#x2F;site-packages&#x2F;text&#x2F;blob.py&quot;, line 340, in __repr__ return unicode(&quot;{cls}(&#x27;{text}&#x27;)&quot;.format(cls=class_name, text=self.raw)) UnicodeDecodeError: &#x27;ascii&#x27; codec can&#x27;t decode byte 0xc2 in position 10: ordinal not in range(128)

The NodeBox linguistics module is another nice wrapper around NLTK (and other natural language processing libraries). I used it for extracting actions and details from sentences, but it&#x27;s also great for spelling correction, pluralization, part-of-speech tagging and other common NLP tasks.<p><a href="http://nodebox.net/code/index.php/Linguistics" rel="nofollow">http:&#x2F;&#x2F;nodebox.net&#x2F;code&#x2F;index.php&#x2F;Linguistics</a>

Both for my study and side job I work on NLP with python.<p>Sorry, but I think this thing is very much overrated by the HN crowd. There are many such libraries and this one adds exactly nothing. I also don&#x27;t see how this is easier to use than, lets say, Pattern.<p>Try and add new functionality. One new functionality could be to use an ontology to calculate the distance between two words. Then you can do other cool things with that and place it in your module.

This looks great! NLTK is incredible but definitely can be a bit intimidating. Very cool to have a wrapper around it.<p>I&#x27;m curious to see exactly how it works and so I&#x27;ll certainly check out the source when I have a bit more time. Thanks for posting this.

If you could add a blob.target and a default vectorizer, you could use scikits learn to offer some nice classification and regression. It&#x27;s pretty easy to do that with what you have now, but some of those concepts are a little foreign if you haven&#x27;t done text classification before, like me before yesterday. Particularly the part of speech tagging- using those as features could be powerful alongside n-grams.

For the Google Translate functionality, does this pass the request through an intermediary service or direct to the API?

So after poking around with this for a bit, I will say that it DEFINITELY is vulnerable to Python2&#x27;s string handling warts. Constructing a `TextBlob` out of a string with non-ASCII characters doesn&#x27;t seem to work. I created another virtualenv with Python 3 and it works quite well.

I played with this a few days ago. It is a nice wrapper for NLTK. You probably want to, at some point, read the free NLTK book online.<p>Edit: and it also uses pattern.

Can someone explain what this does in layman&#x27;s terms? I&#x27;m a biz guy, not a coder, but I&#x27;m interested in the use cases. thanks

This is awesome. I looked, but couldn&#x27;t find out: is there a word sense disambiguation layer somewhere hidden in here?

Curious as to what training algorithms you used for Sentiment Analysis? Also can I add my domain specific training set?

This looks great, thanks for sharing.<p>Any thoughts or relevant benchmarks you would like to share about its speed?

Awesome! Thanks for posting. Are you the hacker that put it together?

Seems to have an incredibly easy interface. Will test it. Well done!

Thanks, I plan on using this.

Awesome. I could use this.

This looks great