swotai
commented
3 years ago If we try to clarify what we are aiming to do (Anju's term: what's the "ask"):
Given a PDF file of memorandum/addendum/decision, We want to summarize into the following (think of this as the additional data columns that we can add to Legistar table for each agenda item #)
- Filename (given, no need to extract from PDF)
- Keywords (comma separated)
- Other items:
- e.g. referred address, related organization/government departments, others?
Currently doing tf-idf.
I have previous code in the Text-analysis folder on Github as well as some code based on Anju's colab code that does some text summarization and text modeling, but needs refinement. A dependency is how we read/cleanup the initial text from the PDF.