-
@ mleku
2025-05-19 06:09:50
#realy #devstr #progressreport
i have completed the revision of the filter and sort algorithms for the proximity/sequence search strategy for the full text search function
it now analyses the groups of indexes per event to find the number and length of sequential results, as well as the distance of these sequences, and then sorts them by ascending order of distance (shortest distance first) and descending order of items in sequence
this should work the best within the requirements i have set, which is to make a language agnostic full text search that adequately generates a relevance score based on simple word matches
now, back to the fiat mine for a work session, got a few things to sort out, one of them being a retarded API of some anime list website that doesn't have a valid openapi 2 schema and requires me to write a manual http request
after hearing about the giant bill on heroku i'm going to also shift the steam playtime data query to using the steam API directly, for the same reasons
and lastly, to implement the query to my colleague's full text analysis server to add its scores to the scores generated by the mostly set intersection based comparisons