opinion mining Flashcards
What is the reason of opinion mining and sentiment analys
Opinion mining is extracting opinions from evaluating text produced by users, in order to understand what people think about products, how they make decision etc.
How can one use opinion mining results for their benefit
Search engines: search, present and aggregate opinion, improve results
business intelligence: customer opinion, news monitoring
Individuals: purphasing products/services
Marketing: place ads
What is named entity recognition
Information extraction technique where one tries to extract info about person name, organisation, location, time and date. In order to identify opinion holders.
What are the three main types of opinion mining tasks
Sentiment classification - “thought provking, interesting and genereally good” pos/neg on doc or sentence level
Explicit feature-based opinion mining - “the graphic card on the computer is poor”, specific opinions about a feature
Comparative sentence and relation mining - “ the battery life on this computer is better than that computer”
In what chategories can comparitive opinions be classified
Non-equal gradable comparison (X är bättre än Y, unequal)
Equative comparison (X är lika bra som Y, Equally)
Superlative comparison (X är bäst, no vs.)
Non-gradable (X är annorlunda Y, uncomparible)
Opinion mining task on document level
Assumes that one document is about one subject with only one opinion
opinion mining task on sentence level
Assumes that each sentence contain only 1 opinion written by one person
in sentiment analysis, which class of words suggests subjectivity and opinion
adjective and adverbs