If enabled Program-Y calculates the positivity an the subjectivity of each sentence you ask it, and also keepsa running score of the overall conversation. The more you converse with it in a specific tone, the more it will startto respond to that tone.

The sentiment analyser keeps 2 scores

• Positivity
• Subjectivity

Polirity, as the name suggests refers to how positive the question and/or the running conversation is.The more positive the questions, the higher the score, the more negative the lower the score.

The positivity score is a floating point numerical value between -1.0 and 1.0. The bot returns a textual valueto support grammar processing with the following values

• < -0.9 - EXTREMELY NEGATIVE
• > -0.9 and score <= -0.7 - VERY NEGATIVE
• > -0.7 and score <= -0.5 - QUITE NEGATIVE
• > -0.5 and score <= -0.3 - NEGATIVE
• > -0.3 and score <= -0.1 - SOMEWHAT NEGATIVE
• > -0.1 and score <= 0.1 - NEUTRAL
• > 0.1 and score <= 0.3 - SOMEWHAT POSITIVE
• > 0.3 and score <= 0.5 - POSITIVE
• > 0.5 and score <= 0.7 - QUITE POSITIVE
• > 0.7 and score <= 0.9 - VERY POSITIVE
• > 0.9 - EXTREMELY POSITIVE

#SubjectivityA subjectivity score refers to how subjectivity or objective is the question and/or conversation. Objective refers to the factthe questions reference actual people, places objects, where as a more subjective conversations are more based on feeling and mood.

Subjectivity is a numerical floating point value between 0.0 and 1.0. The bot returns a textual valueto support grammar processing with the following values

• score == 0.0 - COMPLETELY OBJECTIVE
• score > 0.0 and score <= 0.2 - MOSTLY OBJECTIVE
• score > 0.2 and score <= 0.4 - SOMEWHAT OBJECTIVE
• score > 0.4 and score <= 0.6 - NEUTRAL
• score > 0.6 and score <= 0.8 - SOMEWHAT SUBJECTIVE
• score > 0.8 and score < 1.0 - MOSTLY SUBJECTIVE
• score == 1.0 - COMPLETELY SUBJECTIVE

At anytime during the conversation you can get the current positivity and subjectivity scores.These are stored in 2 variables which can be accessed via the 'get' template tag. As an example in AIML

<category><pattern>CONVERSATION SENTIMENT</pattern><template>Current positivity is <getname="positivity" />, and subjectivity is<getname="subjectivity" />.</template></category>

You can check if sentiment is enabled using the following 'SENTIMENT ENABLED' grammar.This will return either 'SENTIMENT ENABLED' or 'SENTIMENT DIABLED'. For example

<category><pattern>IS SENTIMENT ENABLED</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT ENABLED</extension></template></category>

To get the textual version of sentiment scores, use the extention as follows

<category><pattern>CURRENT SENTIMENT TEXT</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT CURRENT TEXT</extension></template></category>

This will return the following text

SENTIMENT SCORES POSITIVITY SUBJECTIVITY .

E.g.

SENTIMENT SCORES POSITIVITY SOMEWHAT POSITIVE SUBJECTIVITY NEUTRAL.

To get the numeric version of sentiment scores, use the extention as follows

<category><pattern>CURRENT SENTIMENT NUMERIC</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT CURRENT NUMERIC</extension></template></category>

SENTIMENT SCORES POSITIVITY SUBJECTIVITY .

E.g.

SENTIMENT SCORES POSITIVITY 0.5 SUBJECTIVITY 0.6.

###PreviousWe can extract the sentiment of previous parts of the conversation. The following call to the extensiongives us the sentiment for the current conversation

<category><pattern>CALCULATE SENTIMENT FEELING OVERALL</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT FEELING OVERALL</extension></template></category>

Where as the following call to the extension will extract the sentiment score for the last nth sentence

<category><pattern>CALCULATE SENTIMENT FEELING LAST *</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT FEELING LAST <star /></extension></template></category>

##Adhoc Sentiment ScoreYou can get the sentiment for a specific sentence using the following call to the extension

<category><pattern>CALCULATE SENTIMENT SCORE *</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT SCORE <star /></extension></template></category>

This will return the following text string

SENTIMENT SCORES POSITIVITY SUBJECTIVITY .

E.g.

SENTIMENT SCORES POSITIVITY VERY NEGATIVE SUBJECTIVITY MOSTLY SUBJECTIVE.

###Numeric to Text ValuesIf you have the raw numeric values for positivity and/or subjectivity you can pull the text valuethrough a call to the extension with the following grammars. The numeric value should bedenormalised first so 0.1 should 0 dot 1.

<category><pattern>GET SENTIMENT POSITIVITY *</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT POSITIVITY <star /></extension></template></category><category><pattern>GET SENTIMENT SUBJECTIVITY *</pattern><template><extensionpath="programy.sentiment.extension.SentimentExtension" >SENTIMENT POSITIVITY <star /></extension></template></category>