The Munchies Index™ by Stonewright.AI

Proposals Round 1
#1

Name of Project: The Munchies Index™ by Stonewright.AI

Proposal in a Sentence: Gathering social content to build a next iteration of the Munchies Index™.

Description of Product and Market:

The Munchies Index™ is a collection of natural language processing scripts engineered to learn from the fast growing cannabis community. The scripts are built around a database of 2.90M social media posts, pulled multiple times per month every month since May 2019, and currently totaling over 82.57M words. The current scripts include:

  • Data gathering from a social platform API
  • Comparative data gathering from Google Trends
  • A system of 17 tuned filters to suppress noise topics to reveal signal patterns.
  • A custom defined lemmatization document of 40,000 words.
  • Scripted analysis including hierarchical clustering, n-grams, frequency rankings, feature & comparison clouds, and more.
  • Featured ranks of top snack foods, top fruit flavors, and the capability to easily mine for others.

The legal cannabis industry is experiencing rapid growth worldwide. According to Grandview Research, the industry’s projected compound annual growth rate (CAGR) for 2021-2028 is 26.7%. Stonewright.AI has spent several years building brand visibility in the cannabis community while gathering data and testing various tools for product market fit.

A promising next generation for the technology will have three main features:

  • Significantly expanding the database of historic and on-going cannabis community social data. I have already targeted several hundred million additional words of user generated content. Algovera’s grant will fund a major new data gathering effort.
  • A formalized Munchies Index™ report to continue building on awareness within the food & drink sector.
  • The alpha release of a rebuilt Stonewright.AI to feature some limited degree of data interactivity.

Grant Deliverables:

Collaborate with a contractor to supplement the Munchies Index™ set with a large volume of relevant social data. The new data gathered will be from a timeframe of May 2019 to present.

Develop preliminary designs for a rebuilt Stonewright.AI including some form of data access for users.

Funding Requested: $1,000

3 Likes
Teaching Pre-Trained Language Models to Master Idiomatic Expressions (aka Munchies Index)
#2

Awesome work! Thanks for presenting your project yesterday during the Townhall.

1 Like
#3

Thanks a lot, Hitesh. The townhall was a fun conversation. Many projects to be excited about.

#4

I’ve updated some of the code this week in order to share a text mining image. The four terms featured here were chosen to offer some perspective on language usage within the cannabis community.

We could perform further comparative analysis using Google Trends data as a proxy for mainstream term usage. We could also isolate interesting peaks and summarize the adjacent language patterns. For example, if we wanted to better understand the usage “learning” in mid-2019, we would filter on the target date range and produce a dendrogram to show the topics most commonly associated with “learning”.

Thanks again for your interest in the Munchies Index™!

term_mining_1-22-22
term_mining_1-22-221895×685 75.6 KB

2 Likes