Sentiment Analysis

Sentiment analysis is a REST API for opinion mining (also known as sentiment analysis) of spanish comments. It analizes a comment, tweet, post or sentence and determines sentiment polarity such as negative, positive or neutral according to the vocabulary used. With this API you can:

• Analize a single comment and obtain sentiment polarity (positive, negative, neutral)
• Analize multiple comments and obtain sentiment polarity of each of these comments
• For a set of comments retrieve overall-sentiment polarity
• For a set of comments retrieve statistics about sentiment orientation
• For a set of comments retrieve a bag of words (folksonomy) related to sentiment orientation

Version

0.1.0

Download

download source file from bit bucket [spribo_content/sentiment-analysis]

sentiment analysis in github

sentiment analysis in bitbucket

###Dependencies:

python (2.7)
pattern (2.6)
django (1.6.3)
djangorestframework (2.3.13)
django-rest-swagger (0.1.14)

Deployment

###Unix-based: It is highly recommended to use virtualenv:

sudo pip install virtualenv

Create the directory for deployment and activate it:

sudo virtualenv /your_path/deployment_venv
cd /your_path/deployment_venv
source bin/activate

Copy sentiment directory into deployment_env directory, project should look like:

deployment_env
|-- sentiment
|    |-- manage.py
| |-- tmp.db
| |-- sentiment
| |-- comments
| | `-- classifier
| |-- static-files
| |-- rest-framework-swagger
|-- bin
|-- lib
|-- include

Install dependencies:

pip install pattern
pip install django
pip install djangorestframework
pip install django-rest-swagger

####Using GUNICORN for dynamic content:

pip install gunicorn
cd sentiment
gunicorn sentiment.wsgi:application

And now our service is running in http://localhost:8000 or http://127.0.0.1:8000

To chance port use:

gunicorn sentiment.wsgi:application --bind=127.0.0.1:8001

####Using NGINX for static content (pretty browsable interface)

For Linux

sudo apt-get install nginx

For Mac OS, firt install git then homebrew:

sudo mkdir /usr/local
sudo chown -R `_username_` /usr/local
curl -L http://github.com/mxcl/homebrew/tarball/master | tar xz --strip 1 -C /usr/local
brew install git
cd /usr/local
git init
git remote add origin git://github.com/mxcl/homebrew.git
git pull origin master

Then install NGINX:

brew install nginx

Copy the STATIC_ROOT of settings.py

STATIC_ROOT = os.path.join(BASE_DIR, 'static-files')

Here BASE_DIR is the direction of deployment directory.(This addres is printed in the bash as /context_address/deployment_env/static-files) Then, modify config nginx file (make backup of config file, just in case)

cd /usr/local/etc/nginx/
vim nginx.config

Modify followig parameters:

server{
  listen 8000;
  server_name 'your-fully-qualified-domain-name.com';
    
}

location /{
  proxy_pass http://127.0.0.1:8001;
}

location /static/{
  autoindex on;
  alias 'here you put static_root directory'
}

To see more options of NGINX configuration see nginx conf

Then link config file

mkdir sites-enabled
sudo ln nginx.config sites-enabled/any_project_name

Start servers, first nginx:

sudo start nginx

or:

launchctl load ~/Library/LaunchAgents/homebrew.mxcl.nginx.plist

And run gunicorn as daemon:

gunicorn sentiment.wsgi:application --bind=127.0.0.1:8001 --daemon

###Deploying in Windows with Apache and mod_wsgi: Deploying with Apache

---

Usage

Actually this API support two methods: GET and POST.

###GET: URL:

http://yourdomain.com/comments

QUERY_PARAMS

comment: commet string  (mandatory)
classifier_type: SVM or MNB (optional)
no_classes: 2 or 5 (optional)

EXAMPLE_URL :

http://localhost:8000/comments?comment=%27este+es+un+comentario+muy+malo%27&classifier_type=MNB&no_classes=2

You can also use from bash:

curl http://localhost:8000/comments?comment=%27este+es+un+buen+comentario%27

###POST:

Either in the browser or in bash:

curl -X POST http://localhost:8000/comments -d 'JSON_FORMAT' -H "Content-Type: application/json"

JSON FORMAT:

{
     "control": {
          "classifier": "automatic", 
          "no_classes": "default", 
          "response": {"folksonomies":true, "comments":"full"}
     }, 
     "data": [
          {
               "comment": "primer comentario", 
               "id": 1
          }, 
          {
               "comment": "segundo comentario", 
               "id": 2
          }, 
          {
               "comment": "n-esimo comentario", 
               "id": 99
          }
     ]
}

---

We define 2 parts: control and data

Control:

• classifier "automatic" for automatic selection of classifiers "SVM" classifier 88% precision less than 50 words "MNB" classifier 77% precision any amount of words

• no_classes "default" means 5 classes (muy_negativo, negativo, neutro, positivo, muy_positivo) "quick" means 2 classes (negativo, positivo)

• response "folksonomies" if true returns folksonomies (remember json true/false are lowercase) "comments" if "ids_only" returns ids_only polarity, if "full" returns comment by comment polarity if "nothing" returns nothing

####Data:

• All comments with their respectives id's

Request example:

curl -X POST http://127.0.0.1:8000/comments -d '{"control": {"classifier": "automatic", "no_classes": "default", "response": {"folksonomies":true, "comments":"full"}}, "data": [{"comment": "primer comentario", "id": 1}, {"comment": "segundo comentario", "id": 2}, {"comment": "n-esimo comentario", "id": 99}]}' -H "Content-Type: application/json"

Response example:

{
    "Overall Sentiment": 2.0, 
    "Statistics": {
        "positivos": 0, 
        "neutros": 0, 
        "muy_positivos": 3, 
        "muy_negativos": 0, 
        "negativos": 0
    }, 
    "Comments": [
        {
            "comment": "primer comentario", 
            "polarity": "muy_positivo", 
            "id": 1, 
            "ranking": 5
        }, 
        {
            "comment": "segundo comentario", 
            "polarity": "muy_positivo", 
            "id": 2, 
            "ranking": 5
        }, 
        {
            "comment": "n-esimo comentario", 
            "polarity": "muy_positivo", 
            "id": 99, 
            "ranking": 5
        }
    ], 
    "folksonomies": [
        {
            "muy_negativos": {}
        }, 
        {
            "negativos": {}
        }, 
        {
            "neutros": {}
        }, 
        {
            "positivos": {}
        }, 
        {
            "muy_positivos": {
                "n-esimo comentario": 1, 
                "comentario": 3, 
                "n-esimo": 1
            }
        }
    ]
}

License

BSD