Let’s take a look at some ML results for document classification and extracting unstructured data from documents:
- Dropbox: 87% accuracy after training on more than 300,000 invoices
- Cloudscan: 84% accuracy after training on more than 326,000 invoices
- Jatana.ai, #2: 75% accuracy after training on 153 documents using one ML method
- Jatana.ai, #3: 85% accuracy after training on more than 14,000 documents using another ML method
- GLYNT: 98% accuracy after training on 2 – 7 invoices
Other companies, such as butter.ai, acquired by Box, have used similar methods to accomplish the these tasks.
For example, healthcare records contains both precise numeric data, such as lab results, and wordy paragraphs of text, the clinical notes made by the doctor. Some ML techniques, often used by NLP specialists will be used to create summaries and extract context from the text. GLYNT can be used in the other parts of the record, delivering the precise numerical data with rigor. The mix is the winner.
So, when we’re asked the question “Is GLYNT NLP?” it won’t surprise you to hear that our answer is “Yes. And tell me about the problem you want to solve.”
