ASPECT-BASED SENTIMENT ANALYSIS

Sentiment analysis is part of Natural Language Processing (NLP) area and it is the process of identifying and extracting subjective information from text. Sentiment analysis is widely applied in web-related applications, for the monitoring of social media and, in general, for the monitoring of all contents where people/consumers express and opinion. Sentiment analysis aims to determine the attitude of a writer in a document by means of Machine Learning approaches.

Aspect Based Sentiment Analysis (ABSA) is a new research area of NLP that aims to detect the polarity of a written text in regards to a particular aspect/feature. This topic is of particular interest for user of monitoring tool for detecting on complex/high-dimensional dataset the topics where the writer's polarity is more different from a particular clustering to another. While some approaches have been presented to deal with this task, many issues are still unanswered: for example, the portability of a methodology from one language to another and the need of tagged text to perform the algorithm training.

 


HIT Research Centre will be involved in:
- Defining appropriate market domains for the research investigations
- Identifying target entities
- Defining datasets format
- Developing algorithms and procedures for aspect term/category extraction
- Developing algorithms and procedures for aspect term/category polarity
- Testing the algorithms performance
- Validating the procedure in a Support Decision System framework"



Project Start Date: 1st May 2016
Project Duration: 28th Feb 2017
Project Coordinator: Alessandro Beghi
Total Project Funding: 35.200,00 €
Partners: Statwolf LTD
Official website: hit.psy.unipd.it/aspect-based-sentiment-analysis