HERMAN
Improved Food Nutrition Search Algorithm
Herman Franklin LLC is developing a food information search algorithm for quick and reliable food nutrition insights. Instead of Googling individual food items for specific nutrition information, like "protein content of a Ballpark hotdog", you can get a sorted list of the options that best meet your criteria. The technology behind this innovation includes natural language processing (NLP) methods for knowledge extraction. Future work will include development of a more robust suite of knowledge extraction AI.
Current Status
This project is available for development. Although this entire project is subject to the protection of intellectual property laws, we are more than happy to arrange a meeting to showcase some of our proud accomplishments to interested parties. Please contact us.
Milestones
Business Milestones
-
Submitted a project pitch in 2020 to the NSF Seed Fund for an SBIR/STTR grant, which was accepted and turned into an invitation for a full proposal submission. For investment opportunities please contact us.
-
Hired a video production team to produce a product mockup
-
Hired a grant writer for assistance with the NSF SBIR/STTR grant
-
Set a domain placeholder for the product: foodpooch.com
Technical Milestones
-
Scraped nutrition information text for over 81 brands spanning 41 websites resulting in 2519 observations in either text or image format.
-
Homogenized the data set by converting 396 image format nutrition information observations into text.
-
Parsed about over 1900 observations using regular expressions to create a training data set. This dataset will be protected by intellectual property laws.
-
In progress: We are developing an evaluation metric for training and testing our model.