G

github_contest

Settings | Report Duplicate

0

I Use This!

Inactive

Commits : Listings

Analyzed about 5 hours ago. based on code collected about 5 hours ago.

Commit Message	Contributor	Files Modified	Lines Added	Lines Removed	Code Location	Date
Jan 13, 2025 — Jan 13, 2026 Showing page 1 of 7 Search / Filter on:
Merge branch 'graph' of [email protected]:jeremybarnes/github_contest	Jeremy Barnes	More...				over 16 years ago
getting towards basic indexing	Jeremy Barnes	More...				over 16 years ago
reorganized some of the interfaces and improved the query code; still not being used	Jeremy Barnes	More...				over 16 years ago
moving towards more advanced functionality	Jeremy Barnes	More...				over 16 years ago
improved syntax for accessing nodes with a primary key	Jeremy Barnes	More...				over 16 years ago
modified example to load the rest of the data	Jeremy Barnes	More...				over 16 years ago
github import test now working properly (if slowly)	Jeremy Barnes	More...				over 16 years ago
fixed reference counting problem due to default assign on AttributeRef not being implemented; further improved test cases	Jeremy Barnes	More...				over 16 years ago
all attribute tests are now passing	Jeremy Barnes	More...				over 16 years ago
second test (reference counted attributes) is passing basic functionality	Jeremy Barnes	More...				over 16 years ago
gradual progress on attributes; first test is running	Jeremy Barnes	More...				over 16 years ago
gradually fleshing out graph data structures	Jeremy Barnes	More...				over 16 years ago
use random forests (from the latest jml update)	Jeremy Barnes	More...				over 16 years ago
merged in changes from other machine	Jeremy Barnes	More...				over 16 years ago
started work on a generic graph data structure to allow more efficient mining	Jeremy Barnes	More...				over 16 years ago
added missing data file	Jeremy Barnes	More...				over 16 years ago
added follower graph to the information; minor improvements	Jeremy Barnes	More...				over 16 years ago
added collaborator information; feature seems informative but no major effect on the results	Jeremy Barnes	More...				over 16 years ago
no real changes	Jeremy Barnes	More...				over 16 years ago
don't keep copying prob-results.txt	Jeremy Barnes	More...				over 16 years ago
added number of missing watchers and forks feature based upon data from hintly's submission	Jeremy Barnes	More...				over 16 years ago
removed user and author matches from coocs and made them only suggest repos that are watched at least twice. Not much difference; probably slightly worse	Jeremy Barnes	More...				over 16 years ago
don't allow same author or same name results in in_cluster to improve diversity	Jeremy Barnes	More...				over 16 years ago
use both keywords and watcher SVD for repo clustering	Jeremy Barnes	More...				over 16 years ago
using keyword vector for clustering repos; looks worse (we lose a lot of repos suggested by repo clustering)	Jeremy Barnes	More...				over 16 years ago
use binary features for keyword svd; big improvement	Jeremy Barnes	More...				over 16 years ago
used tf-idf not tf for svd; works less well	Jeremy Barnes	More...				over 16 years ago
more keyword features; still not making much of a difference	Jeremy Barnes	More...				over 16 years ago
finished SVD for keywords and various improvements to the preprocessing; added very basic features; still not much good	Jeremy Barnes	More...				over 16 years ago
added a probability propagation candidate generator (similar to the cooc and cooc2 generators). Slow and not much improvement. Thanks to Mark Lodato for the idea.	Jeremy Barnes	More...				over 16 years ago

←
1
2
3
4
5
6
7
→