Question

Briefly discuss polarity classification methodologies for opinion mining based on the document level, sentence level and aspect level.

Answer 1

Opinion mining also known as sentiment analysis refers to the area of research that attempts to make an automatic system to determine human opinion from the text written in natural language.

Opinion mining or sentiment analysis is a type of natural language processing for following or tracking the feelings, attitudes or appraisal of the people about a particular idea, product or service. The basic components of the opinion are:

Opinion holder: is a person that provides a particular or specific opinion on an object.

Object: it is an entity on which opinion is expressed by a person.

Opinion: it is a view, idea, appraisal or sentiment of an object done by a person.

Opinion mining is a growing and promising research field, especially with the use of social media, connected with the scrutiny or scanning of the content or opinion to decide if a person is of positive or negative sentiments. Opinionated texts from blogs, discussion forums, reviews and social networking sites such as twitter and facebook, etc. are regarded as a source for identifying acknowledgment or dissatisfaction of people towards certain products, services or topics under discussion.

POLARITY CLASSIFICATION FOR OPINION MINING

ON DOCUMENT LEVEL-  Document-level opinion mining is about classifying the general opinion presented by a person as positive, negative or neutral about a certain object. The assumption derived at the document level is that each document focuses on a single object and holds opinions from a single opinion holder. It involves three steps: in the first step adjectives are taken out along with a word that provides appropriate information. In the second step, semantic orientation is captured from words of known polarity. In the third step, classifies a review as recommended or not. The aim is to test whether selected groups can produce a good result when opinion mining is perceived as document level, associated with two topics: positive or negative.

SENTENCE LEVEL-  this level of opinion mining is associated with two tasks. The first one is to identify whether the given sentence is opinionated or objective. The second one is finding the opinion of the subjective sentence as positive, negative, or neutral. The assumption taken at the sentence level is that a sentence contains only one opinion for e.g, the picture quality of this phone is good. However, the assumption is not true in many cases.

ASPECT LEVEL-  The first task of opinion mining at this level is to identify object features that have been commented on by an opinion holder. Determine whether the opinions on the feature are positive, negative, or neutral. Group feature synonyms and produce a feature-based opinion summary of multiple reviews.