Commit e36fc4dd authored by J. Fernando Sánchez's avatar J. Fernando Sánchez
Browse files

Add new taiger plugin

parent 45dde044
# Senpy Plugin Taiger
Service that analyzes sentiments from social posts written in Spanish or English.
Proxy for two of Taiger's sentiment analysis services for social media posts:
* taiger-plugin: proxy for a service that normalizes the text, and gives both a polarity and a polarity value for the text. It works for Spanish and English text.
* taiger3c-plugin: it uses a simpler service that only returns a polarity (positive, negative or none). It only works for Spanish.
## Usage
......@@ -8,18 +11,19 @@ Service that analyzes sentiments from social posts written in Spanish or English
To use this plugin, you should use a GET Requests with the following possible params:
Params:
- Input: text to analyse.(required)
- Endpoint: Enpoint to the Taiger service.
## Example of Usage
Example request:
```
http://senpy.cluster.gsi.dit.upm.es/api/?algo=sentiment-taiger&inputText=This%20is%20amazing
```
curl http://senpy.cluster.gsi.dit.upm.es/api/?algo=sentiment-taiger&inputText=This%20is%20amazing
Example respond: This plugin follows the standard for the senpy plugin response. For more information, please visit [senpy documentation](http://senpy.readthedocs.io). Specifically, NIF API section.
#Or, for the taiger3c plugin:
curl http://senpy.cluster.gsi.dit.upm.es/api/?algo=sentiment-taiger3c&inputText=Me%20encanta
```
For example, this would be the example respond for the request done.
This plugin follows the senpy schema and vocabularies, please visit [senpy documentation](http://senpy.readthedocs.io). Specifically, the NIF API section.
It should look like this:
```
{
......@@ -53,7 +57,8 @@ For example, this would be the example respond for the request done.
}
```
As it can be seen, this plugin analyzes sentiment givin three categories or tags: `marl:Positive`, `marl:Neutral` or `marl:Negative`, that will be held in the `marl:hasPolarity` field. Moreover, the plugin retrieves a `marl:polarityValue`.
As can be seen, this plugin analyzes sentiment giving three categories or tags: `marl:Positive`, `marl:Neutral` or `marl:Negative`, that will be held in the `marl:hasPolarity` field.
Moreover, the plugin retrieves a `marl:polarityValue` (a value between -1 and 1).
This plugin supports **python2.7** and **python3**.
![alt GSI Logo][logoGSI]
......
......@@ -8,4 +8,4 @@ services:
volumes:
- ".:/senpy-plugins"
environment:
TAIGER_ENDPOINT: 'http://34.244.91.7:8080/sentiment/classifyPositivity'
TAIGER_ENDPOINT: "${TAIGER_ENDPOINT:-http://somedi-taiger.hopto.org:5406/es_sentiment_analyzer_3classes}"
# -*- coding: utf-8 -*-
import time
import requests
import json
import string
import os
from os import path
import time
from senpy.plugins import SentimentPlugin
from senpy.models import Results, Entry, Entity, Topic, Sentiment, Error
TAIGER_ENDPOINT = os.environ.get("TAIGER3C_ENDPOINT", 'http://somedi-taiger.hopto.org:5406/es_sentiment_analyzer_3classes')
class TaigerPlugin3cats(SentimentPlugin):
'''
Service that analyzes sentiments from social posts written in Spanish or English.
Example request:
http://senpy.cluster.gsi.dit.upm.es/api/?algo=sentiment-taiger3c&inputText=This%20is%20amazing
'''
name = 'sentiment-taiger3c'
author = 'GSI UPM'
version = "0.1"
maxPolarityValue = -1
minPolarityValue = 1
def _polarity(self, value):
if 'NONE' == value:
polarity = 'marl:Neutral'
value = 0
elif 'N' == value:
polarity = 'marl:Negative'
value = -1
elif 'P' == value:
polarity = 'marl:Positive'
value = 1
else:
raise ValueError('unknown polarity: {}'.format(value))
print(value, 'whatsup')
return polarity, value
def analyse_entry(self, entry, params):
txt = entry['nif:isString']
api = TAIGER_ENDPOINT
parameters = {
'text': txt
}
try:
r = requests.get(
api, params=parameters, timeout=3)
agg_polarity, value = self._polarity(r.text.strip())
except requests.exceptions.Timeout:
raise Error("No response from the API")
except Exception as ex:
raise Error("There was a problem with the endpoint: {}".format(ex))
if not agg_polarity:
raise Error('No category in response: {}'.format(ar.text))
self.log.debug(agg_polarity)
agg_opinion = Sentiment(
id="Opinion0",
marl__hasPolarity=agg_polarity,
marl__polarityValue=value,
)
agg_opinion.prov(self)
entry.sentiments.append(agg_opinion)
yield entry
test_cases = [
{
'params': {
'algo': 'sentiment-taiger',
'intype': 'direct',
'expanded-jsonld': 0,
'informat': 'text',
'prefix': '',
'plugin_type': 'analysisPlugin',
'urischeme': 'RFC5147String',
'outformat': 'json-ld',
'conversion': 'full',
'language': 'en',
'apikey': '00000',
'algorithm': 'sentiment-taiger'
},
'input': 'I hate to say this',
'expected': {
'sentiments': [
{'marl:hasPolarity': 'marl:Negative'}],
},
'responses': [
{
'url': TAIGER_ENDPOINT,
'body': 'N',
}
]
},
{
'params': {
'algo': 'sentiment-taiger',
'intype': 'direct',
'expanded-jsonld': 0,
'informat': 'text',
'prefix': '',
'plugin_type': 'analysisPlugin',
'urischeme': 'RFC5147String',
'outformat': 'json-ld',
'conversion': 'full',
'language': 'en',
'apikey': '00000',
'algorithm': 'sentiment-taiger'
},
'input': 'This is amazing',
'expected': {
'sentiments': [
{'marl:hasPolarity': 'marl:Positive'}],
},
'responses': [
{
'url': TAIGER_ENDPOINT,
'body': 'P',
}
]
},
{
'params': {
'algo': 'sentiment-taiger',
'intype': 'direct',
'expanded-jsonld': 0,
'informat': 'text',
'prefix': '',
'plugin_type': 'analysisPlugin',
'urischeme': 'RFC5147String',
'outformat': 'json-ld',
'conversion': 'full',
'language': 'en',
'apikey': '00000',
'algorithm': 'sentiment-taiger'
},
'input': 'The pillow is in the wardrobe',
'expected': {
'sentiments': [
{'marl:hasPolarity': 'marl:Neutral'}],
},
'responses': [
{
'url': TAIGER_ENDPOINT,
'body': 'NONE',
}
]
}
]
if __name__ == '__main__':
from senpy import easy_test
easy_test(debug=False)
......@@ -25,8 +25,8 @@ class TaigerPlugin(SentimentPlugin):
name = 'sentiment-taiger'
author = 'GSI UPM'
version = "0.1"
maxPolarityValue = 0
minPolarityValue = -10
maxPolarityValue = -1
minPolarityValue = 1
def _polarity(self, value):
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment