Optmizing Solr boost factors with genetic algorithms
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
app
public
routes
views
.gitignore
LICENSE
README.md
app.js
index.js
package.json
train.js

README.md

Boost Genetics

This implements a proof of concept genetic optimization for Apache Solr boost factor optimization problems. A rough outline for that can be found the related blog post.

Running the application

Field usage

  • Host: The host where Solr can be found (e.g. localhost)
  • Port: The port where Solr can be found (e.g. 8080)
  • Path: The base path to the select where handler, the query parameter has to be the last one (e.g. /solr/rfcs/select?defType=dismax&rows=50&fl=id&q=)
  • Fields: The query fields which should be optimized, actual factors replaced by placeholders (e.g. &qf=title^0%20keywords^1%20abstract^2%20sec1^3%20text^4%20titles^5%20references^6%20author^7)
  • Iterations: The amount of generations to run the optimization through - 50 seems to be a good start
  • Domain: The allowed value range for the factors - must match to the amount of placeholders in your fields string (e.g. [[1,100],[1,100],[1,100],[1,100],[1,100],[1,100],[1,100],[1,100]])

Licence:

Apache License, Version 2.0 (see LICENSE)