DSP Implements AI-powered Sentiment Analysis to improve customer satisfaction

DSP offers Professional and Managed Services, and providing high-quality customer service is key to our work. We needed a detailed analysis of activities within tickets, based on emails sent by our customers, to monitor their experience. This required a solution to provide a high-level analysis of email activities and the ability to identify potential issues within ticket resolution at a more detailed level. Due to our growing customer base and the high volume of tickets raised by our customers, we knew the solution needed advanced automation with minimal human intervention.

Our approach involved the following: 

• Data extraction from source application: The data used in the sentiment analysis are emails exchanged between DSP and our customers during ticket resolution. The emails need to be extracted from the source application: our ticketing application. After extraction, the data needs to be ingested into a database storage or a local storage location.
  
  
• Data cleaning: After extracting and ingesting the data, the data is cleaned to remove noise from the dataset. Data cleaning could involve removing duplicate records, text cleaning, information extraction, etc. We leveraged Oracle Generative AI models for the text cleaning process, which helped improve the accuracy of our solution.
  
  
• Sentiment analysis: After the data cleaning step, the data is ready to be processed on a sentiment analysis model. Sentiment analysis is the process of using AI to classify texts based on the emotions detected in the text. A text can be classified as positive, negative or neutral when run through a sentiment analysis model. In our case, the emails from our ticketing application are analysed and classified using a sentiment analysis model.
  
  
• Data visualisation: The results from the sentiment analysis are visualised using Microsoft Power BI. These visuals give a high-level analysis of ticket activities and give the option to drill into the data. Having a visual representation of the solution helps to present the result to technical and non-technical decision-makers. 

This automated and scalable solution has streamlined what was once a labour-intensive, manual review process. By integrating AI-driven sentiment analysis into our support services, DSP has further strengthened its ability to proactively respond to customer needs.