5 ms·

We’re building a graph of this same underlying data at https://alma.app https://alma.app (with a lot of enrichments) to help people discover and donate to nonpr

We’re building a graph of this same underlying data at https://alma.app https://alma.app (with a lot of enrichments) to help people discover and donate to nonprofits. E.g. here’s Stanford: https://alma.app/charities/941156365-the-board-of-trustees-of-the-leland-stanford-junior-university https://alma.app/charities/941156365-the-board-of-trustees-o... Would love to know what types of analysis people would like to see about organizations and their relationships?

johnymontana 7y ago
Could you clarify what you mean by “building a graph”? Are you combining the 990 data plus other enrichments in a graph database?
- danhilltech 7y ago
  Yep basically - looking at the network of grants (who funds who), people (who works where over their careers), social media and news (who’s referenced or cited alongside whom) and so on. You start to see interesting clusters of nonprofits that are funded by the same groups and work together extensively.
  - johnymontana 7y ago
    Neat. What technologies are you using to store and analyze the data? Are you using any graph algorithms, like PageRank, community detection, etc?
    - danhilltech 7y ago
      DGraph and elasticsearch with some redis caching. Yea, a version of pagerank is coming soon! Love any other ideas...
walterbell 7y ago
> Would love to know what types of analysis people would like to see about organizations and their relationships? Related public records, e.g. court filings, municipal or other co-investments alongside the nonprofits, adjacent (time or geo) legislation/policy changes, rotating doors of nonprofit, gov, commercial.
- danhilltech 7y ago
  Interesting thanks!
schwanksta 7y ago
This is neat! I thought about extracting the grants (still might), but full-text seemed like good bang for the buck. Your tools sound like they might be very useful for reporters. Have you given any thought to that? We love mapping these sorts of connections.
- danhilltech 7y ago
  Hey, first up really amazing work you’re doing, hugely inspiring for us! Thank you. Whilst our focus has been delivering a consumer layer on top of all this data, yea, very open to exposing our underlying graph to others. Want to drop me a note at dan at alma.app? As you mention elsewhere, half the battle is cleaning the data and getting quality.

johnymontana 7y ago

Could you clarify what you mean by “building a graph”? Are you combining the 990 data plus other enrichments in a graph database?

danhilltech 7y ago
Yep basically - looking at the network of grants (who funds who), people (who works where over their careers), social media and news (who’s referenced or cited alongside whom) and so on. You start to see interesting clusters of nonprofits that are funded by the same groups and work together extensively.
- johnymontana 7y ago
  Neat. What technologies are you using to store and analyze the data? Are you using any graph algorithms, like PageRank, community detection, etc?
  - danhilltech 7y ago
    DGraph and elasticsearch with some redis caching. Yea, a version of pagerank is coming soon! Love any other ideas...

walterbell 7y ago

> Would love to know what types of analysis people would like to see about organizations and their relationships? Related public records, e.g. court filings, municipal or other co-investments alongside the nonprofits, adjacent (time or geo) legislation/policy changes, rotating doors of nonprofit, gov, commercial.

danhilltech 7y ago
Interesting thanks!

schwanksta 7y ago

This is neat! I thought about extracting the grants (still might), but full-text seemed like good bang for the buck. Your tools sound like they might be very useful for reporters. Have you given any thought to that? We love mapping these sorts of connections.

danhilltech 7y ago
Hey, first up really amazing work you’re doing, hugely inspiring for us! Thank you. Whilst our focus has been delivering a consumer layer on top of all this data, yea, very open to exposing our underlying graph to others. Want to drop me a note at dan at alma.app? As you mention elsewhere, half the battle is cleaning the data and getting quality.