||A US based consumer products company
|Problem & Context
||The client engaged SPA to conduct regular alerts in an area where the volume of patent data was growing rapidly. Manual screening and selection of appropriate patent records was very time consuming. In an effort to improve the productivity, SPA team engaged the AI/ML Data analytics team to develop a tool which would expedite the screening without compromising the quality.
|SPA Approach & Deliverables
- The technical team collaborated with the AI/ML Data analytics team to develop an Auto-screen tool to help save time on the weekly alerts.
- The tool was designed in such a way that the end user would have complete control over filtering criteria. The user could add/modify existing criteria.
- The challenge posed was solved by using hybrid-AI and keyword based approach. Setting up criteria was made easy by using machine assisted keywords (not only synonyms but also equivalent keywords in respect of context). This was executed with the help of trained domain specific ML models. This approach provided quick/accurate results.
- A facility was provided for the user to review the results and adjust the criteria before committing the changes to the target dataset.
- Auto-tagging features using AI were also added to the application. This enabled the end user to switch between semi or auto mode at any given point of time.
- This tool helped in reducing the turn around time of alerts by ~40%.
|Resources / Cost
- One Chemical Engineer and AI/ML Data analytics team