Introduction

The explosion of social media has created unprecedented opportunities for citizens to publicly voice their opinions. People regularly share their ideas, opinions and personal statements via social media channels. Analyzing those content created by people plays a significant role for market researchers who take decisions on brands and marketing campaigns and consumers who are always seeking for high quality products and content.

Since all the above described content is very large in scale and heterogeneous, it is hard to evaluate them manually to understand perceptions. Thus researchers have identified the approaches to implement the autonomous systems for perception analysis and opinion mining which are mainly built on top of natural language processing methods which is called sentiment analysis. Maintaining a high accuracy level while minimizing computational power is a challenge in current mechanisms used for sentiment analysis.

Introducing an alternate mechanism to capture and analyze the users’ perception effectively and efficiently will be focused in this project. Since the users can convey their perception with various methods, the data capturing mechanisms can be diversified in many areas like  questionnaire with likert items, star rating, voice clips and video clip analysis, emotion capturing and etc. Since all above mentioned mechanisms have their own drawbacks and strengths the human psychological aspects and efficient survey techniques will be used in choosing data capturing mechanisms and designing GUIs for end user applications.

Due to the high demand for accurate and efficient perception analysis, we will introduce an accurate Perception Analysis Engine with multiple end user applications to collect data from people. Basically in this project we will also focus on innovative end user applications which can be used to provide instant feedback or opinion.  

These end user applications will generate enormous amount of data. The perception analysis engine will be architected to process these data either real time or as batch processes in an optimal and scalable way.Also we will design a data representation layer which can represent interconnections and relationships between stored data. The data storage will be designed such that it supports optimal data mining and reliability.

Currently there exists some systems which can be used to achieve domain specific advantages in perception analysis. We will work to integrate them into this new perception analysis engine to give a complete aggregated perception analysis platform.