Supply Chain bot for Retailers





Retailers always need to have an efficient supply chain to serve the customers with the right product at the right price and at the right time. Managing the supply chain thus becomes imperative in this hyper competitive market. A typical Supply Chain IT ecosystem is made up of a plethora of applications and reporting systems. These systems manage the business processes around vendors, distribution centers, inventory, physical items, replenishment etc. Apart from this we do have a lot of dashboards, decision support systems to assist decision making as well. With Machine Learning becoming pervasive, retailers are heading in the direction of infusing algorithmic decision making in their business processes. With all these systems, one had to be agile and sift through systems to get a pointed answer for his or her hypothesis.

This is where an intelligent assistant or a cognitive agent or a bot comes into picture. The bot can assist a business user to fetch a pointed answer for a pointed question on a business domain. It can accept a query in plain natural language like English and thus remove the technicalities from a business user in fetching the desired information from IT systems. Sounds good so far!!. At the same time can the bot understand and respond to business questions containing domain-specific jargon words which are not present in the English vocabulary? Can it be multilingual ? Can it exhibit context awareness? Can it understand and mimic a human assistant ?

This talk focuses on building a bot solution for the above scenario. It will articulate on the approach used, the open source technologies leveraged and the challenges faced while building the intelligent supply chain bot.

Outline of the talk:

  • Introduction to bots and the supply chain use-case [2 mins]
  • Tech stack used [2 mins]
  • ML pipeline
    • Training data [1 min]
    • Data transformation and cleansing [2 mins]
    • Automated annotation pipeline [2 mins]
    • SpaCy models for Named Entity Recognition [3 mins]
    • Intent classifiers [2 mins]
    • Context awareness [3 mins]
    • Bot as a web service [3 mins]
    • User feedback capture [1 mins]
    • Model retraining [2 mins]
  • Conclusion and final thoughts [2 mins]

Key takeaways:

This talk will be helpful to those who are looking to build an intelligent chatbot and want to look beyond commercial software or frameworks provided by cloud providers.


Basic knowledge of Python and Machine Learning

Video URL:

Speaker Info:

Niharika currently works on building Natural Language Processing based solutions for the retail industry and believes in leveraging AI to help people lead better lives!

Sridhar is a ML enthusiast and has worked on ML and DL based solutions for the last 5 years and has more than 15 years of retail industry experience.

Speaker Links:

Niharika - LinkedIn

Sridhar - LinkedIn

Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Intermediate
Last Updated: