We are piloting a classification use case with a customer that may be of interest you.
- Do you have internal labeled data?
- This can be patent or non patent literature that you manually labeled as relevant/irrelevant, or with other proprietary technical categories.
- Do you reuse this data often, or perform labeling often?
- Use cases include - labeling weekly patent alerts, analyzing internal or external portfolios, etc
- Do you wish to save time?
- If this is a recurring task - do you wish you could speed it up somehow?
- Do you wish to reuse this data for new projects and other use cases?
- This is really your secret sauce - these labels capture your technical expertise.
- What do you do after one project - delete or file all this work away?
We provide a cloud or desktop solution that chews up your data to train a smart AI CLASSIFIER.
You can feed it a new document to classify using your proprietary labels.
Our CLASSIFIER spits out the correct labels with ~95% accuracy for binary classes, and ~98% accuracy for the top 2 non-binary classes.
We also provide an 'estimate confidence' for each label, to help you curate the weaker estimates faster.
Put your knowledge to work!