C2S Analytics team helped an IT giant Microsoft instrumentally to address auto clustering the customer support cases by leveraging the Machine Learning and NLP techniques, initially we loaded significant amount of training data to train the machine learning modules to get the accuracy to 95%.
Microsoft receives customer support related enquires from varies channels in the form of calls, blogs, auto generated text files, or online. Microsoft support team manually assess each case and moves these cases to respective taxonomy bucket for a resolution. There is a huge delay to resolve these tickets since manual intervention. Hence Microsoft decided to auto cluster these cases to speed up to improve customer support.
C2S Technologies together worked with Microsoft to solve client problem in three approaches.
During phase I, we treated this project as POC to gain the customer confidence to make sure whether we have right tools, right team, and right knowledge to solve customer problems using one taxonomy data. This phase we exhibited our Machine Learning techniques and NLP techniques to map a case to new taxonomy or existing taxonomy.
We increased accuracy from 70% to 85% by leveraging more training data, this proved us that our algorithms were working right.
We have built a front end tools with rich user experience by integrating Machine Learning algorithms.
HDInsights & Machine Learning, NLP + Text Mining. Supervised training for classification.
95% of Customer Support cases are auto clustered and created with right taxonomies without human intervention. Cost savings and support efficiency increased.
• Eliminated 95% of human interventions.
• Taxonomy auto creation.
• Saved 2 business days to for each customer support ticket life.
• Increased customer support efficiency and saved $$ per each ticket.