Related content / leveraging the full-text-index for some 'AI'


#1

hey,

love Dynalist versatility, simplicity and speed! for me it holds the promise of a better version of note + information management.
as Dynalist is server-based, I think it would be awesome to use that as leverage to do some ‘AI’-stuff on the backside, and implement it in the very decent, clutterless UI-form of Dynalist.
– as Dynalist already relies centrally on having a full index (and great search-capabilities), it doesn´t seem too far fetched to use that to implement a function like ‘related content’.

… this would immediately propel Dynalist on par with high-quality apps in the field of info-management like devonThink (or to some lesser degree Evernote pro, which also provides a ‘related content’ function).

thx for considering! oliver


#2

Hi Oliver,

Thanks for the suggestion!

As we have no experience with devonThink or Evernote Pro, could you briefly explain how the “related content” function works?


#3

hey Erica, sure!

it´s rather easy – and one doesn´t neccessarily need the reference to those apps.
(actually one could also think of blogs, online publications etc. and their ‘related content’ function).

but to the point: this function would mean that based on a ‘simple’ (this is the AI part) comparison of text-elements/lexems, Dynalist would – on a click – show things like:

  • these notes/items are related
  • these documents are relevant (or: these documents could be the right place to file this item; <- this was another feature request in here ;-]

Dynalist already has this awe-some way to quickly find ‘related content’ by typing in a link and then essentially giving instant search results for any text typed; in a way ‘related content’ could be seen as nothing else than a – more intelligent - shortcut or a more sophisticated version for this. as it would mean there is a way to quickly access any related notes as based on an (intelligent) text-comparison of the whole note with other notes/documents. the other part of helpful ‘magic’ would be a ranked/weighted shortlist of notes that match the pattern of the existing note. – and there it is!

I guess the hard part is that you need things like linguistic analysis for this, and thesauri and the like. at the same time – and although I am not a coder – I am quite confident there are existing frameworks out there that can be implemented ‘rather easily’.

hope this is clear enough? otherwise happy to elaborate!

I´d see giving myself over totally to Dynalist in terms of research (and other structured) notes if it would allow this kind of interaction (also seeing that an extensive use of Dynalist will produce huge text-corpus and by implication redundancy and confusion; … just like any other bulk-information container :wink: )

best! oliver