Acronym:

CATACX: Cog-Affective social media Text Analytics for Customer eXperience analysis  

From: 2016-06-01 To: 2017-05-31 (Completed)

BICRO 

PoC 

- 

doc. dr. sc. Jan Šnajder

- 

The analysis of customer opinions towards products and services is crucial for any business. Much of communication that a company has with its customers, traditionally done through dedicated customer support channels, is now taking place on social networks. Besides to communicate with the company, customers use social networks to communicate among themselves and, often critically, discuss company’s products and services. Furthermore, social networks are characterized by high speed, high volume, interactivity, availability, and informality, and directness of communication. As such, they are an invaluable source of information for the analysis of customer opinions and customer experience (CX). The goal of the project is to develop and validate an innovative solution for textual analysis of user comments from social networks, aiming to support a detailed and comprehensive analysis of user experience. We will base our solution on state-of-the-art natural language processing and machine learning techniques. A distinguishing feature of our solution is the multidimensional analysis of text subjectivity, which includes both cognitive (attitudes, reasons, sentiment) and affective (emotions) aspects of online communication. Such an analysis, carried out over a large number user comments, will provide insights about the experiences that the users have with certain products and services, what they expect, what thrills or disappoints them, and what they perceive as a problem. These are valuable insights for improving the customer experience and improving business results. Within this project, we will develop the prototype at the level of laboratory-tested technology. The prototype will be specialized for analyzing comments of telecom customers written in Croatian language. The development will include the collection and processing of data from social networks, the development of fine-grained opinion mining models, and the validation of models by CX experts. Acknowledging the inherent complexity of natural language processing, the proposed solution poses a scientific and technological challenge, and requires innovation that goes beyond straightforward extensions of existing state of the art. Although the risk is mitigated by the expertise of the R&D team and their previous results, a proof of concept is necessary as there is no warrant that the solution is technically feasible.