Opinion mining and sen-ment analysis W3C SEMINAR, JUNE 19 2012 TH David Osimo, Tech4i2.com www.crossover-‐project.eu #pmod
THE CHALLENGES Making sense of a thousand voices across diﬀerent pla:orms Iden-fying “good ideas” Reduce uncertainty on possible impact of policies by ge@ng early and real-‐-me feedback
POSSIBLE SOLUTION Argument mapping soCware helps organising in a logical way these policy statements, by explicita-ng the logical links between them.. Vo-ng Advise Applica-ons help voters understanding which poli-cal party (or other voters) have closer posi-ons to theirs. Automated content analysis helps processing large amount of qualita-ve data.
FUTURE CHALLENGES Reduc-on of human eﬀort The detec-on of spam and fake reviews, The limits of collabora-ve ﬁltering, which tends to iden-fy most popular concepts and to overlook most innova-ve / out of the box thinking The risk of a ﬁlter bubble (pariser 2011) The asymmetry in availability of opinion mining soCware, which can currently be aﬀorded only by organisa-ons and government, but not by ci-zens. In other words, government have the means today to monitor public opinion in ways that are not available to the average ci-zens. While content produc-on and publica-on has democra-zed, content analysis has not. The integra-on of opinion with behaviour and implicit data, in order to validate and provide further analysis into the data beyond opinion expressed The con-nuous need for beTer usability and user-‐friendliness of the tools, which are currently usable mainly by data analysts
Current free Top Current research Short term future Long termtools market research future tools researchfiltering opinion Machine · Statistical + Semantic · Visual representation· Multilingualbased on rating; learning + analysis through · Audiovisual opinion audiovisualassessing human lexicon/corpus of words mining opinion miningsentiments based analysis with known sentiment · Real-time opinion · Usable, peer-on keywords; mining to-peer opinionvisual word for sentiment · Machine learning mining tools forcounting classification algorithms citizensArgument · Identification of policy · Natural language · Non-bipolarmapping and VAA opinionated material to interfaces assessment of be analysed · SNA applied to opinion · Computer-generated opinion and expertise · Automatic reference corpuses in · Bipolar assessment of irony detection political/governance opinions field · Multilingual reference · Visual mapping of corpora bipolar opinion · Recommendation · Identification of highly algorythms rated experts
COMMENT THE ROADMAP hTp://www.crossover-‐project.eu/ResearchRoadmap.aspx