Our client is a leading Indian telecommunication service provider and a mobile network operator. The company offers products such as sim cards, mobile devices, and internet dongles. It also provides services that include online bill payments, plans, voice, internet data, applications, and subscription services in prepaid and post-paid format. Their current active mobile user count sums up to 400 million subscribers.
The client had a knowledge management system (KMS) in place to complement the customer support agents while troubleshooting any service-related issues for subscribers. Client had a search feature in the KMS that did not give meaningful results, as well as error logs that made it difficult to troubleshoot or conduct analytics. Publishing a new article in the KMS was quite the process, almost taking an entire day, and rebuilding the index was a tough task. The client was looking to transform their search feature in the KMS to get the right results for faster dispute handling.
PureSoftware, with the expertise, saw an opportunity to leverage the world’s best open-source engine, Elasticsearch. So, an ideal knowledge management system should have a Google-like search mechanism where end users can instantly find content suggestions as soon as they start typing in the search bar.
As a solution, the team of experts implemented restful APIs over the Elasticsearch APIs to provide google search like capabilities. Modifications were also carried out in the Restful web APIs on top of Elasticsearch to customize their solution. Elasticsearch helped to keep the log information of APIs which enabled the level-4 team to search exceptional scenarios like google test search. The data from the legacy system was also migrated to the new solution by creating various SSIS packages.
Publishing a new article in the KMS reduced form 1 day to 5-7 milliseconds.
New published content in the KMS is immediately available on search and accessible to end-user.
End-user can see recommendations and top search suggestions with most frequent data, just like Google.
Recommendations such as “Did you mean” and suggestions such as spelling correction in the search feature greatly simplified the end-user experience.