Page 1 of 7 SENTIMENT ANALYZER Sede opertiva: Piazza Vermicelli 87036 Rende (CS), Italy
Page 2 of 7 TABLE OF CONTENTS 1 APP documentation... 3 1.1 HOW IT WORKS... 3 1.2 Input data... 4 1.3 Output data... 4 1.4 Basic workflow example... 5 2 API documentation... 7
Page 3 of 7 1 APP DOCUMENTATION 1.1 HOW IT WORKS Sentiment Analyzer APP is a normalization task that enables to recognize sentiments in sentences from selected text fields.
Page 4 of 7 1.2 INPUT DATA This APP allows you: to select fields from which take text to analyze; specify the language of the text selected. Input data shows in Figure 1 return sentiments from the txt field given in output from the previous app in the workflow, by processing the English natural language contained in this field. Figure 1 input data. 1.3 OUTPUT DATA This APP gives back, for each single sentence found in the source text, a set of five elements: a) global_polarity: this field shows the sentiment polarity by considering a whole source text, non a single sentence; b) sentiment: this field shows sentiment words (like happy, happiness, good, nice, ); c) polarity: this field shows the sentiment polarity of a sentence in a scale of numeric values (generally negative integer for bad sentiments/opinions and positive integer for good sentiments/opinions); d) sentence: the text from which algorithms extracts sentiment and entities (it could be a tweet from twitter, or a facebook post, or any other text source); e) entities: a comma separated list of entities found in the source text. This set of elements is given back in an excel spreadsheet or a data cube whose metadata depend on the data extracted and presented in the output. You can see an example of output data in the following figure:
Page 5 of 7 Figure 2 - example of output data This example is a sample of output where you can see a tweet created on Friday the 30th of March and containing the hashtag #Obama in which polarity and global polarity have a value of 2 and the sentiment value is like. You can also find as result some entities that the NLP algorithms found in a tweet sentence. 1.4 BASIC WORKFLOW EXAMPLE Sentiment - Analyzer needs some extraction app before in the workflow that can get some text to process, so you can create a simple flow as shown in the following Figure 3. You can set input parameter and watch the results by the watcher button as shown in the following Figure 4. You can also obtain the results by the to excel app as shown in the following Figure 5.: Figure 3 - simple sentiment-analizer workflow example Figure 4 - The watcher panel.
Page 6 of 7 Figure 5 - the excel format of results.
Page 7 of 7 2 API DOCUMENTATION For information about how to use Sentiment Analyzer API in your application, send us a message to info@altiliagroup.com.