FinnForm - The Finnish language machine learning platform

Project Skills

  • coding
  • engineering

Project Blog

The brief

Large amounts of data are generated during different stages of the customer service process, an asset on which Provad wishes to capitalize. Provad is developing a technology platform for the #1 customer service experience in the world. One of the main objectives of this platform is to capture all relevant information from any structured or unstructured data interactions, as well as to make the most of this continuously accumulating data.

The background

Provad is a rapidly growing company in the customer service industry, with customers such as … Provad supplies an integrated multi-channel (RPA-) platform, which improves the speed and quality of Provad’s customers’ customer service. Provad’s long term goal is to create a new business paradigm of connectedness and the integrating technology for the #1 customer service experience. This technology stack will include AI/machine learning.

The problem

It is a challenge to tap, analyze and use the huge amounts of digital data originating from customer service interactions. Equally challenging is to further automate the input of information in order to help customer service workers do a better job. Similarly, the analysis of big data also helps forecast the future and find relevant trends and patterns.
The primary technological means are:

  • Robotic process automation, for service speed, quality and automation of data input.
  • Artificial intelligence and machine learning (AI, e.g. neural networks) for data input analysis, continuous learning and inference. AI also improves decision making and quality customer care by providing the right information.

The specific target for the project is to automatically understand Finnish written language at the customer service interface (e.g. email, chat). A scan on available solutions, commercial and open source, is first needed, followed by experimenting with a selected technology in a narrow application area. If successful, a set of further experiments will be prepared to validate the concept and to spot main challenges. The experiments can be done using authentic data from Provad’s customers.