Sentiment Analysis is the process of computationally identifying and categorizing opinions expressed in a piece of text, especially in order to determine whether the writer's attitude towards a particular topic, product, etc., is positive, negative, or neutral.
This package is an implementation of the Naive Bayes Algorithm To Determine the sentiment of a particular statement, a book review, chat, speech and so on. It marks a sentence as positive, negative or neutral depending on the kind of words that are used, this can help in automatically selecting a review, comment or chat that has the best intentions in any situation.
We instantiate the SentimentAnalyzerTest class below by passing in the SentimentAnalyzer object (class) found in the file: 'SentimentAnalyzer.class.php'.
This class must be injected as a dependency into the constructor as shown below
$sat = new SentimentAnalyzerTest(new SentimentAnalyzer());
Training The Sentiment Analysis Algorithm with words found in the trainingSet directory
The File 'data.neg' contains a list of sentences that's been marked 'Negative'. We use the words in this file to train the algorithm on how a negative sentence/sentiment might be structured.
Likewise, the file 'data.pos' contains a list of 'Positive' sentences and the words are also used to train the algorithm on how to score a sentence or document as 'Positive'.
The trainAnalyzer method below accepts three parameters: + param 1: The Location of the file where the training data are located + param 2: Used to describe the 'type' of file [param 1] is; used to indicate whether the supplied file contians positive words or not + param 3: Enter a less than or equal to 0 here if you want all lines in the file to be used as a training set. Enter any other number if you want to use exactly those number of lines to train the algorithm
$sat->trainAnalyzer('../trainingSet/data.neg', 'negative', 5000); //training with negative data
$sat->trainAnalyzer('../trainingSet/data.pos', 'positive', 5000); //trainign with positive data
The analyzeSentence method accepts a sentence as parameter and score it as a positive, negative or neutral sentiment. it returns an array that looks like this:
array
(
'sentiment' => '[the sentiment value returned]',
'accuracy' => array
(
'positivity'=> 'A floating point number showing us the probability of the sentence being positive',
'negativity' => 'A floating point number showing us the probability of the sentence being negative',
),
)
An example is shown below:
$sentence1 = 'while the performances are often engaging , this loose collection of largely improvised numbers would probably have worked better as a one-hour tv documentary . ';
$sentence2 = 'edited and shot with a syncopated style mimicking the work of his subjects , pray turns the idea of the documentary on its head , making it rousing , invigorating fun lacking any mtv puffery . ';
$sentimentAnalysisOfSentence1 = $sat->analyzeSentence($sentence1);
$resultofAnalyzingSentence1 = $sentimentAnalysisOfSentence1['sentiment'];
$probabilityofSentence1BeingPositive = $sentimentAnalysisOfSentence1['accuracy']['positivity'];
$probabilityofSentence1BeingNegative = $sentimentAnalysisOfSentence1['accuracy']['negativity'];
$sentimentAnalysisOfSentence2 = $sat->analyzeSentence($sentence2);
$resultofAnalyzingSentence2 = $sentimentAnalysisOfSentence2['sentiment'];
$probabilityofSentence2BeingPositive = $sentimentAnalysisOfSentence1['accuracy']['positivity'];
$probabilityofSentence2BeingNegative = $sentimentAnalysisOfSentence1['accuracy']['negativity'];`
The AnalyzeDocument method accepts the path to a text file as parameter. It analyzes the file and scores it as either a positive or a negative sentiment. It also returns an array with the same keys as the analyzeSentence method.
An example is demonstrated below
$documentLocation = '../trainingSet/review.txt';
$sentimentAnalysisOfDocument = $sat->analyzeDocument($documentLocation);
$resultofAnalyzingDocument = $sentimentAnalysisOfDocument['sentiment'];
$probabilityofDocumentBeingPositive = $sentimentAnalysisOfDocument['accuracy']['positivity'];
$probabilityofDocumentBeingNegative = $sentimentAnalysisOfDocument['accuracy']['negativity'];