Sentiment Analysis on Social Network
Transcript
Sentiment Analysis on Social Network
Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/293654834 SentimentAnalysisonSocialNetwork ConferencePaper·February2016 READS 91 3authors,including: VitoSantarcangelo GiuseppeOddo CentroStudi,Buccino(SA),Italy CentroStudi,Buccino(SA),Italy 40PUBLICATIONS15CITATIONS 12PUBLICATIONS1CITATION SEEPROFILE SEEPROFILE Availablefrom:VitoSantarcangelo Retrievedon:26July2016 Sentiment Analysis on Social Network a cura di Ing. Vito Santarcangelo, Dott. Antonio Ruoto e Ing. Giuseppe Oddo Matera, 06/02/2016 Convegno “Ingegneria del Sentimento” OSINT United States Department of Defense defines the Open Source Intelligence (OSINT) as “The intelligence discipline that pertains to intelligence produced from publicly available information that is collected, exploited, and disseminated in a timely manner to an appropriate audience for the purpose of addressing a specific intelligence and information requirement”. BIG DATA Big data means large and heterogeneous datasets obtained from web and local systems. INFORMATION AND OPINIONS Blogs, social networks and social communities represent the mainly used tools for information and opinions sharing. OPINION MINING OSINT is also strictly related to Opinion Mining (also known as Sentiment Analysis), a discipline aiming at retrieving the opinion of a subject from web contents, reaching also the reputation analysis scope. OPINION MINING SYSTEM STEPS: Target Definition (1) OSINT Extraction (2) Sentiment Analysis (3) Score Return (4) After choosing a keyword/phrase (1), a crawler extracts contents related to the user input from OSINT data (2). Sentiment Analysis (3) examines(2)the polarity of the filtered data extracted thanks to the use of a Sentimental Thesaurus, associating a polarity (negative, neutral, positive) for each term of the extracted excerpts, determining the Score (4). KEYWORD (1) OSINT EXTRACTION (2) EXCERPTS SENTIMENT ANALYSIS (3) (3) SCORE (4) DB FiCloud 2015 AIN APPROACH ADJECTIVES (A), INTENSIFIERS (I) and NEGATIONS (N) AIN APPROACH – FILTERING PHASE If a tweet does not contain one or more adjectives/adverbs of our thesaurus, it is dropped from the tweet array. AIN - ITALIAN THESAURUS -1 -2 -1 1 AIN THESAURUS AIN LOGIC AIN THESAURUS EXCERPTS NON MOLTO DECADENTE -1 -1 -2 0,5 -1 AIN + SEMANTIC - ITALIAN THESAURUS F,E FP, A FP, L F AIN SEMANTIC THESAURUS AIN THESAURUS (BOOK) In press by RCE MULTIMEDIA Graphic developed by Antonio Ruoto www.ainthesaurus.it HOW THE SYSTEM WORKS USER INPUT WEB CRAWLER AIN THESAURUS + SEMANTIC NETWORK SENTIMENT ANALYSIS SYSTEM REPUTATION OUTPUT PARSER TEXT ANALYZER DATA EXTRACTION COMPARISON WITH SENTISTRENGHT COMPARISON WITH SENTISTRENGHT AIN FEATURES: NON MOLTO DECADENTE -1 -1 -USE OF NEGATION COMBINED WITH INTENSIFIER -USE OF SEMANTIC APPROACH -2 0,5 -USE OF THREE METRICS (DOMAIN EXPERT, COMMON USER, AVERAGE) -1 INSTAGRAM SOCIAL NETWORK USAGE HASHTAG QUERY HASHTAG QUERY USAGE OF INSTAGRAM AIN ON INSTAGRAM 64,283,404 posts about adjectives (30/01/2016) Usage of positive adjectives : 46% of posts Thanks to Michele Di Lecce for the support in the extraction Sentiment Hashtag Distribution Instagram as Positive Egocentric Social Network based on photos Semantic Hashtag Distribution Instagram as Positive Egocentric Social Network based on photos Figurative People Art Most used Hashtags Bello 2,020,475 VS Brutto 82,391 Triste 1,028,314 VS Felice 599,446 Nero 935,425 VS Bianco 675,072 Instagram as Positive Egocentric Social Network based on photos SOME POSSIBLE APPLICATIONS: REPUTATION ANALYSIS, CUSTOMER SATISFACTION SYSTEM Sentiment Analysis on social networks could improve the service quality level of public administrations, as it referred to people opinions of the web. It can be also useful to obtain the reputation score about a person, an office, a project, subject, theme of Public Administration realizing an indirect customer satisfaction system. The integration with public opinions of websites and blogs gives relevancy, accuracy and quality to this kind of applications. EXAMPLE OF APPLICATION The user feedbacks can be integrated with the reputation analysis of OSINT DATA (as the App “Erice Amico Comune” of “Comune di Erice”). CONCLUSION & FUTURE WORKS An improvement in ambiguity management (e.g. “caro” that means dear and expensive) could be obtained through the examination of the context using a semantic network. The font style (bold, italic) and the colour (red, black) of the text could also be an input to evaluate. The analysis of visual objects could indicate the feelings of the users (e.g. using the keyword Favignana, the recognition of the sea in the photo communicates good feelings, instead of cockroaches near the hotel that communicate bad feelings). Vs REFERENCES For more information and dataset visit http://www.researchgate.net/profile/Vito_Santarcangelo Thanks for the attention!