The India Voter Registration Roll project was centered around the extraction and transformation of data gathered from different types of files. Intsurfing was requested to bring all hard-to-process data about Indian voters into a single file.

Within the scope of this project, our team had to deal with:

• Non-standardized data - information was presented in different file types - PDF, images, photos of hand-written samples - all files having a different structure.
• Data volume - our team had to process 1 billion records and transform them into a readable format.
• Different languages - the processing engine had to be able to recognize all 22 separate official Indian languages.

By the end of the project, the Client received a file with a poll of Indian voters. By leveraging cutting-edge technology and our deep expertise in big data, we extracted and transformed information about the Indian population harvested from open resources and wrapped it in a readable format. Thus:

• The adaptive engine we developed scanned various files (PDF files, screenshots, images with hand-written notes, etc.) and converted the captured information into an easy-to-process format.
• The engine was also capable of recognizing data in different Indian languages.
• Intsurfing brought all the pieces of data together to create a single database of voters with 1 billion records.
• Our team provided ongoing support and updated the file once a year to delete irrelevant information and add new data.

• .NET Core 2.0
• Amazon AWS
• Tesseract OCR
• iText PDF