Train ML Model for Agenda Meeting Meta Data

[TeddyCr]

Context

We are working on building a ML pipeline to power new functionality on engage.town:

• Automatic Topic Categorization
  This issue relates to the development of ML algorithm

File

TBD

Tasks

• Read over contents and see notebook at: https://github.com/ylou/CouncilML/tree/master/doc_embedder
• Contact @eselkin for reference on ML models and research already completed
• Download JSONs with labeled data for years of agenda items here.
• If you need a refresher on embeddings, see this google post

[eselkin]

Search should not be contained in ML and should be a simple Elasticsearch on the contents and tags.
Also, it's not a model, since we are doing NLP using fasttext. It should be a pipeline.

[TeddyCr]

Thanks specifying Eli. :)

[owenjoseph]

@TeddyCr @eselkin should this be closed?

[owenjoseph]

Moving to "future development" for now.