How Bosch Power Tools leverages User Data and AI...
summary
Bosch Power Tools crawls reviews from 150 e‑shops, aggregates ~160k labeled docs, translates everything to English with Azure cognitive services, and runs a production ML pipeline on Azure/Databricks. Supervised aspect‑based sentiment models (PyTorch + Hugging Face, fine‑tuned on Bosch data, ~80% accuracy) tag 161 predefined aspects; unsupervised BERTopic clusters reveal new issues (pilot found warranty/serial‑number registration failures). Outputs: anonymization, 150‑word cluster summaries, dashboards and automated emails to stakeholders.
AI-written summary · about
chapters (12)
- 0:00Bosch Power Tools project intro
- 2:00Project overview & team
- 3:09Aspect-based sentiment analysis
- 5:55Data preprocessing & model tuning
- 8:02ML pipeline on Azure
- 9:05In-house AI benefits & dashboard
- 11:33Supervised AI limitations & NLP shift
- 12:58Unsupervised NLP pipeline
- 15:48Topic insights & business actions
- 16:11Infrastructure & production
- 18:19Technologies & tools used
- 19:47Q&A
related talks
51mTaking Models to the Next Level with Azure Machine Learning
29mBosch DataOps: NLP Development on Kubernetes
21mFast and Efficent Text Analysis using Azure's Language Cognitive Service
15mAI Gardening: Harvesting and Pruning the AI Tree
50mAzure AI, Power new possibilities for every organization
57mTaking Models to the Next Level with Azure Machine Learning
16mWatts Up? Leveraging AI & Data for Energy Management
30mAI in Power BI