Sentiment Analysis 2018, Guide, Progetti e Ricerche di Lingua Inglese

Scarica Sentiment Analysis 2018 e più Guide, Progetti e Ricerche in PDF di Lingua Inglese solo su Docsity!

C O N T U R S I A L E S S I A C O P P O L A A L E S S A N D R A D A M A S C O A R M A N D O S C A P O L A T I E L L O V I N C E N Z O S E V E R I N O L U C I A

• (^) Volume : refers to

amount of data

generated every

second

• (^) Variety : refers to

the different types

of data that are

generated,

collected and used

• (^) Velocity : refers to

the speed with

which data is

SENTIMENT ANALYSIS

Analysis of textual information, with the aim of revealing emotions,

sentiments and opinions of a specified entity (product, person,

argument, etc ).

The idea is that investigating a small text, the system is able to define the general sentiment expressed by itself; so this “sentiment” let us to analyze the text and to “extract” from it its negative, positive or neutral meaning.

• (^) Versatility : can result helpful in many fields as brand reputation, politics and social science.
• (^) Authenticity of the data : it is able to listen to emotions which are given spontaneously and, therefore, they reflect the expectations and the feelings of the users.
• (^) Very cheap in terms of time and costs. ADVANTEGE S LIMITS
• (^) Inability to recognize complex texts : the automated system not able to catch emotional complex concepts like irony or humour which could be classified like positive feelings.
• (^) Problems linked at grammatical errors, the punctuation, the abbreviations and the usage of metaphor.

CATEGORIZATION OF TEXTS the assignment of each document (in our case each tweet) to a specific category, based on the sentiment expressed in them. Ontological Dictionary Machine-learning

DIFFERENCE BETWEEN SENTIMENT

ANALYSIS AND WORDCLOUDS

We are not interested in single words, but in feelings or emotions with individual tweets

BAR CHARTS

• (^) After the date of the ship’s docking ”positive" tweets have increased
• (^) Tweets cataloged as “trust" increased and became more frequent than those cataloged as “fear”
• (^) Tweets cataloged as “joy" increased and became more frequent than those cataloged as “disgust”

CONCLUSION We have noticed that the prevailed tweets contain a "negative" feeling, and that as can be seen from the Worldclouds, certainly not few words express a